TVS/Zener Theory and Design Considerations Handbook

TVS/Zener Theory and Design Considerations
Handbook
HBD854/D
Rev. 0, Jun−2005
© SCILLC, 2005
Previous Edition © 2001 as Excerpted from DL150/D
“All Rights Reserved’’
http://onsemi.com
1
Technical Information, Application Notes and Articles
Zener Diode Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Zener Diode Fabrication Techniques . . . . . . . . . . . . . . . 8
Reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Zener Diode Characteristics . . . . . . . . . . . . . . . . . . . . . 18
Temperature Compensated Zeners . . . . . . . . . . . . . . . 30
Basic Voltage Regulation Using Zener Diodes . . . . . 34
Zener Protective Circuits and Techniques:
Basic Design Considerations . . . . . . . . . . . . . . . . . 44
Zener Voltage Sensing Circuits and Applications . . . 54
Miscellaneous Applications of
Zener Type Devices . . . . . . . . . . . . . . . . . . . . . . . . . 61
Transient Voltage Suppression . . . . . . . . . . . . . . . . . . 63
AN784 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
AN843 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Design Considerations and Performance of
Temperature Compensated Zener Diodes . . . . . . 97
MOSORBs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
AR450 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Measurement of Zener Voltage to Thermal
Equilibrium with Pulsed Test Current . . . . . . . . . . 119
http://onsemi.com
2
ZENER DIODE THEORY
INTRODUCTION
now see. Silicon is a tetravalent element, one with four
valence electrons in the outer shell; all are virtually locked
into place by the covalent bonds of the crystal lattice
structure, as shown schematically in Figure 1a. When
controlled amounts of donor impurities (Group V elements)
such as phosphorus are added, the pentavalent phosphorus
atoms entering the lattice structure provide extra electrons
not required by the covalent bonds. These impurities are
called donor impurities since they “donate” a free electron
to the lattice. These donated electrons are free to drift from
negative to positive across the crystal when a field is applied,
as shown in Figure 1b. The “N” nomenclature for this kind
of conductivity implies “negative” charge carriers.
In P-type conductivity, the charges that carry electric
current across the crystal act as if they were positive charges.
We know that electricity is always carried by drifting
electrons in any material, and that there are no mobile
positively charged carriers in a solid. Positive charge
carriers can exist in gases and liquids in the form of positive
ions but not in solids. The positive character of the current
flow in the semiconductor crystal may be thought of as the
movement of vacancies (called holes) in the covalent lattice.
These holes drift from positive toward negative in an electric
field, behaving as if they were positive carriers.
P-type conductivity in semiconductors result from adding
acceptor impurities (Group III elements) such as boron to
silicon to the semiconductor crystal. In this case, boron
atoms, with three valence electrons, enter the tetravalent
silicon lattice. Since the covalent bonds cannot be satisfied
by only three electrons, each acceptor atom leaves a hole in
the lattice which is deficient by one electron. These holes
readily accept electrons introduced by external sources or
created by radiation or heat, as shown in Figure 1c. Hence
the name acceptor ion or acceptor impurity. When an
external circuit is connected, electrons from the current
source “fill up” these holes from the negative end and jump
from hole to hole across the crystal or one may think of this
process in a slightly different but equivalent way, that is as
the displacement of positive holes toward the negative
terminal. It is this drift of the positively charged holes which
accounts for the term P-type conductivity.
When semiconductor regions of N- and P-type
conductivities are formed in a semiconductor crystal
adjacent to each other, this structure is called a PN junction.
Such a junction is responsible for the action of both zener
diodes and rectifier devices, and will be discussed in the next
section.
The zener diode is a semiconductor device unique in its
mode of operation and completely unreplaceable by any
other electronic device. Because of its unusual properties it
fills a long-standing need in electronic circuitry. It provides,
among other useful functions, a constant voltage reference
or voltage control element available over a wide spectrum of
voltage and power levels.
The zener diode is unique among the semiconductor
family of devices because its electrical properties are
derived from a rectifying junction which operates in the
reverse breakdown region. In the sections that follow, the
reverse biased rectifying junction, some of the terms
associated with it, and properties derived from it will be
discussed fully.
The zener diode is fabricated from the element silicon.
Special techniques are applied in the fabrication of zener
diodes to create the required properties.
This manual was prepared to acquaint the engineer, the
equipment designer and manufacturer, and the experimenter
with the fundamental principles, design characteristics,
applications and advantages of this important
semiconductor device.
SEMICONDUCTOR THEORY
The active portion of a zener diode is a semiconductor PN
junction. PN junctions are formed in various kinds of
semiconductor devices by several techniques. Among these
are the widely used techniques known as alloying and
diffusion which are utilized in fabricating zener PN
junctions to provide excellent control over zener breakdown
voltage.
At the present time, zener diodes use silicon as the basic
material in the formation of their PN junction. Silicon is in
Group IV of the periodic table (tetravalent) and is classed as
a “semiconductor” due to the fact that it is a poor conductor
in a pure state. When controlled amounts of certain
“impurities” are added to a semiconductor it becomes a
better conductor of electricity. Depending on the type of
impurity added to the basic semiconductor, its conductivity
may take two different forms, called P- and N-type
respectively.
N-type conductivity in a semiconductor is much like the
conductivity due to the drift of free electrons in a metal. In
pure silicon at room temperature there are too few free
electrons to conduct current. However, there are ways of
introducing free electrons into the crystal lattice as we shall
http://onsemi.com
3
junction interface toward the negative N-type side.
Conversely, electrons in the N-type are readily attracted by
the positive polarity in the other direction.
When a PN junction is reverse biased, the P-type side is
made more negative than the N-type side. (See Figure 2b.)
At voltages below the breakdown of the junction, there is
very little current flow across the junction interface. At first
thought one would expect no reverse current under reverse
bias conditions, but several effects are responsible for this
small current.
Under this condition the positive holes in the P-type
semiconductor are repelled from the junction interface by
the positive polarity applied to the N side, and conversely,
the electrons in the N material are repelled from the interface
by the negative polarity of the P side. This creates a region
extending from the junction interface into both P- and
N-type materials which is completely free of charge carriers,
that is, the region is depleted of its charge carriers. Hence,
this region is usually called the depletion region.
Although the region is free of charge carriers, the P-side
of the depletion region will have an excess negative charge
due to the presence of acceptor ions which are, of course,
fixed in the lattice; while the N-side of the depletion region
has an excess positive charge due to the presence of donor
ions. These opposing regions of charged ions create a strong
electric field across the PN junction responsible for the
creation of reverse current.
The semiconductor regions are never perfect; there are
always a few free electrons in P material and few holes in N
material. A more significant factor, however, is the fact that
great magnitudes of electron-hole pairs may be thermally
generated at room temperatures in the semiconductor. When
these electron-hole pairs are created within the depletion
region, then the intense electric field mentioned in the above
paragraph will cause a small current to flow. This small
current is called the reverse saturation current, and tends to
maintain a relatively constant value for a fixed temperature
at all voltages. The reverse saturation current is usually
negligible compared with the current flow when the junction
is forward biased. Hence, we see that the PN junction, when
not reverse biased beyond breakdown voltage, will conduct
heavily in only one direction. When this property is utilized
in a circuit we are employing the PN junction as a rectifier.
Let us see how we can employ its reverse breakdown
characteristics to an advantage.
As the reverse voltage is increased to a point called the
voltage breakdown point and beyond, current conduction
across the junction interface increases rapidly. The break
from a low value of the reverse saturation current to heavy
conductance is very sharp and well defined in most PN
junctions. It is called the zener knee. When reverse voltages
greater than the voltage breakdown point are applied to the
PN junction, the voltage drop across the PN junction
remains essentially constant at the value of the breakdown
voltage for a relatively wide range of currents. This region
ELECTRONS ARE LOCKED
IN COVALENT BONDS
Si
Si
Si
Si
Si
Si
Si
(a) Lattice Structures of
Pure Silicon
Si
Si
LOCKED COVALENT
BOND ELECTRONS
Si
P
Si
−
Si
Si
APPLIED FIELD
FREE ELECTRON
FROM PHOSPHOROUS
ATOM DRIFTS TOWARD
APPLIED POSITIVE POLE.
(b) N-Type Silicon
Si
INCOMPLETED
COVALENT BOND
Si
+
B
Si
Si
−
Si
Si
APPLIED FIELD
THIS ELECTRON JUMPS INTO
HOLE LEFT BY BORON ATOM.
HOLE POSITION IS DISPLACED
TO RIGHT. THIS RESULTS IN
A DRIFT OF HOLES TOWARD
THE NEGATIVE POLE, GIVING
THEM THE CHARACTER OF
MOBILE POSITIVE CHARGES.
(c) P-Type Silicon
Figure 1. Semiconductor Structure
THE SEMICONDUCTOR DIODE
In the forward-biased PN junction, Figure 2a, the P region
is made more positive than the N region by an external
circuit. Under these conditions there is a very low resistance
to current flow in the circuit. This is because the holes in the
positive P-type material are very readily attracted across the
http://onsemi.com
4
P
CHARGES FROM BOTH P AND N REGIONS
DRIFT ACROSS JUNCTION AT VERY LOW
APPLIED VOLTAGES.
N
(a) Forward-Based PN
Junction
LARGE
CURRENT
APPLIED FIELD
N
P
VERY
SMALL
CURRENT
AT APPLIED VOLTAGES BELOW THE CRITICAL
BREAKDOWN LEVEL ONLY A FEW CHARGES
DRIFT ACROSS THE INTERFACE.
(b) Reverse-Biased PN
Junction
SEVERAL VOLTS
APPLIED FIELD
Figure 2. Effects of Junction Bias
beyond the voltage breakdown point is called the zener
control region.
VREV
VBREAKDOWN
ZENER CONTROL REGION: VOLTAGE
BREAKDOWN MECHANISMS
Figure 3 depicts the extension of reverse biasing to the
point where voltage breakdown occurs. Although all PN
junctions exhibit a voltage breakdown, it is important to
know that there are two distinct voltage breakdown
mechanisms. One is called zener breakdown and the other is
called avalanche breakdown. In zener breakdown the value
of breakdown voltage decreases as the PN junction
temperature increases; while in avalanche breakdown the
value of the breakdown voltage increases as the PN junction
temperature increases. Typical diode breakdown
characteristics of each category are shown in Figure 4. The
factor determining which of the two breakdown
mechanisms occurs is the relative concentrations of the
impurities in the materials which comprise the junction. If
two different resistivity P-type materials are placed against
two separate but equally doped low-resistivity pieces of
N-type materials, the depletion region spread in the low
resistivity P-type material will be smaller than the depletion
region spread in the high resistivity P-type material.
Moreover, in both situations little of the resultant depletion
width lies in the N material if its resistivity is low compared
to the P-type material. In other words, the depletion region
always spreads principally into the material having the
highest resistivity. Also, the electric field (voltage per unit
IREV
SLOPE
Figure 3. Reverse Characteristic Extended
to Show Breakdown Effect
length) in the less resistive material is greater than the
electric field in the material of greater resistivity due to the
presence of more ions/unit volume in the less resistive
material. A junction that results in a narrow depletion region
will therefore develop a high field intensity and breakdown
by the zener mechanism. A junction that results in a wider
depletion region and, thus, a lower field intensity will break
down by the avalanche mechanism before a zener
breakdown condition can be reached.
The zener mechanism can be described qualitatively as
follows: because the depletion width is very small, the
application of low reverse bias (5 volts or less) will cause a
field across the depletion region on the order of 3 x 105V/cm.
A field of such high magnitude exerts a large force on the
http://onsemi.com
5
4
VREV (VOLTS)
3
2
1
30
25
VREV (VOLTS)
20
15
10
5
IREV
IREV
25°C
65°C
65°C
25°C
(A)
(B)
ZENER BREAKDOWN
OF A PN FUNCTION
AVALANCHE BREAKDOWN OF A PN FUNCTION
Figure 4. Typical Breakdown Diode Characteristics. Note Effects of Temperature for Each Mechanism
before combining with a carrier of opposite conductivity).
Therefore, when temperature is increased, the increased
lattice vibration shortens the distance a carrier travels before
colliding and thus requires a higher voltage to get it across
the depletion region.
As established earlier, the applied reverse bias causes a
small movement of intrinsic electrons from the P material to
the potentially positive N material and intrinsic holes from
the N material to the potentially negative P material (leakage
current). As the applied voltage becomes larger, these
electrons and holes increasingly accelerate. There are also
collisions between these intrinsic particles and bound
electrons as the intrinsic particles move through the
depletion region. If the applied voltage is such that the
intrinsic electrons do not have high velocity, then the
collisions take some energy from the intrinsic particles,
altering their velocity. If the applied voltage is increased,
collision with a valence electron will give considerable
energy to the electron and it will break free of its covalent
bond. Thus, one electron by collision, has created an
electron-hole pair. These secondary particles will also be
accelerated and participate in collisions which generate new
electron-hole pairs. This phenomenon is called carrier
multiplication. Electron-hole pairs are generated so quickly
and in such large numbers that there is an apparent avalanche
or self-sustained multiplication process (depicted
graphically in Figure 5). The junction is said to be in
breakdown and the current is limited only by resistance
external to the junction. Zener diodes above 7 to 8 volts
exhibit avalanche breakdown.
As junction temperature increases, the voltage breakdown
point for the avalanche mechanism increases. This effect can
be explained by considering the vibration displacement of
atoms in their lattice increases, and this increased
displacement corresponds to an increase in the probability
that intrinsic particles in the depletion region will collide
with the lattice atoms. If the probability of an intrinsic
particle-atom collision increases, then the probability that a
given intrinsic particle will obtain high momentum
valence electrons of a silicon atom, tending to separate them
from their respective nuclei. Actual rupture of the covalent
bonds occurs when the field approaches 3 x 105V/cm. Thus,
electron-hole pairs are generated in large numbers and a
sudden increase of current is observed. Although we speak
of a rupture of the atomic structure, it should be understood
that this generation of electron-hole pairs may be carried on
continuously as long as an external source supplies
additional electrons. If a limiting resistance in the circuit
external to the diode junction does not prevent the current
from increasing to high values, the device may be destroyed
due to overheating. The actual critical value of field causing
zener breakdown is believed to be approximately
3 x 105V/cm. On most commercially available silicon
diodes, the maximum value of voltage breakdown by the
zener mechanism is 8 volts. In order to fabricate devices
with higher voltage breakdown characteristics, materials
with higher resistivity, and consequently, wider depletion
regions are required. These wide depletion regions hold the
field strength down below the zener breakdown value
(3 x 105V/cm). Consequently, for devices with breakdown
voltage lower than 5 volts the zener mechanism
predominates, between 5 and 8 volts both zener and an
avalanche mechanism are involved, while above 8 volts the
avalanche mechanism alone takes over.
The decrease of zener breakdown voltage as junction
temperature increases can be explained in terms of the
energies of the valence electrons. An increase of temperature
increases the energies of the valence electrons. This weakens
the bonds holding the electrons and consequently, less applied
voltage is necessary to pull the valence electrons from their
position around the nuclei. Thus, the breakdown voltage
decreases as the temperature increases.
The dependence on temperature of the avalanche
breakdown mechanism is quite different. Here the depletion
region is of sufficient width that the carriers (electrons or
holes) can suffer collisions before traveling the region
completely i.e., the depletion region is wider than one
mean-free path (the average distance a carrier can travel
http://onsemi.com
6
REVERSE-BIASED
PN JUNCTION IN AVALANCHE
LARGE CURRENT
N
P
WHEN THE APPLIED VOLTAGE IS
ABOVE THE BREAKDOWN POINT, A
FEW INJECTED ELECTRONS RECEIVE
ENOUGH ACCELERATION FROM THE
FIELD TO GENERATE NEW ELECTRONS
BY COLLISION. DURING THIS PROCESS
THE VOLTAGE DROP ACROSS THE
JUNCTION REMAINS CONSTANT.
R
S
RS ABSORBS EXCESS VOLTAGE.
CONSTANT VOLTAGE DROP
Figure 5. PN Junction in Avalanche Breakdown
a different symbol. For the leakage current region, i.e.
non-conducting region, between 0 volts and VZ, the reverse
current is denoted by the symbol IR; but for the zener control
region, VR ≥ VZ, the reverse current is denoted by the
symbol IZ. IR is usually specified at a reverse voltage
VR ≈ 0.8 VZ.
The PN junction breakdown voltage, VZ, is usually called
the zener voltage, regardless whether the diode is of the
zener or avalanche breakdown type. Commercial zener
diodes are available with zener voltages from about
1.8 V − 400 V. For most applications the zener diode is
operated well into the breakdown region (IZT to IZM). Most
manufacturers give an additional specification of IZK
(= 5 mA in Figure 6) to indicate a minimum operating
current to assure reasonable regulation.
This minimum current IZK varies in the various types of
zener diodes and, consequently, is given on the data sheets.
The maximum zener current IZM should be considered the
maximum reverse current recommended by the
manufacturer. Values of IZM are usually given in the data
sheets.
Between the limits of IZK and IZM, which are 5 mA and
1400 mA (1.4 Amps) in the example of Figure 6, the voltage
across the diode is essentially constant, and ≈ VZ. This
plateau region has, however, a large positive slope such that
the precise value of reverse voltage will change slightly as
a function of IZ. For any point on this plateau region one may
calculate an impedance using the incremental magnitudes of
the voltage and current. This impedance is usually called the
zener impedance ZZ, and is specified for most zener diodes.
Most manufacturers measure the maximum zener
impedance at two test points on the plateau region. The first
is usually near the knee of the zener plateau, ZZK, and the
latter point near the midrange of the usable zener current
excursion. Two such points are illustrated in Figure 6.
This section was intended to introduce the reader to a few
of the major terms used with zener diodes. A complete
description of these terms may be found in chapter four. In
chapter four a full discussion of zener leakage, DC
breakdown, zener impedance, temperature coefficients and
many other topics may be found.
decreases, and it follows that the low momentum intrinsic
particles are less likely to ionize the lattice atoms. Naturally,
increased voltage increases the acceleration of the intrinsic
particles, providing higher mean momentum and more
electron-hole pairs production. If the voltage is raised
sufficiently, the mean momentum becomes great enough to
create electron-hole pairs and carrier multiplication results.
Hence, for increasing temperature, the value of the
avalanche breakdown voltage increases.
VOLT-AMPERE CHARACTERISTICS
The zener volt-ampere characteristics for a typical 30 volt
zener diode is illustrated in Figure 6. It shows that the zener
diode conducts current in both directions; the forward
current IF being a function of forward voltage VF. Note that
IF is small until VF ≈ 0.65 V; then IF increases very rapidly.
For VF > 0.65 V IF is limited primarily by the circuit
resistance external to the diode.
IZK = 5 mA
VZ
5
IR
ZZ K
IZT
0
420 mA
0.5
ZZT
IZM
1.40 A
30
20
10
VR
(VOLTS)
0
REVERSE
CHARACTERISTIC
0.5
1
VF
(VOLTS)
1
1.5
1.5
REVERSE CURRENT (AMPS)
10
I F (AMPS)
15
FORWARD
CHARACTERISTIC TYPICAL
Figure 6. Zener Diode Characteristics
The reverse current is a function of the reverse voltage VR
but for most prNO TAGactical purposes is zero until the
reverse voltage approaches VZ, the PN junction breakdown
voltage, at which time the reverse current increases very
rapidly. Since the reverse current is small for VR < VZ, but
great for VR > VZ each of the current regions is specified by
http://onsemi.com
7
ZENER DIODE FABRICATION TECHNIQUES
INTRODUCTION
have the same surge carrying capability as diffused diodes.
The manufacturing process begins with the growing of high
quality silicon crystals.
Crystals for ON Semiconductor zener diodes are grown
using the Czochralski technique, a widely used process
which begins with ultra-pure polycrystalline silicon. The
polycrystalline silicon is first melted in a nonreactive
crucible held at a temperature just above the melting point.
A carefully controlled quantity of the desired dopant
impurity, such as phosphorus or boron is added. A high
quality seed crystal of the desired crystalline orientation is
then lowered into the melt while rotating. A portion of this
seed crystal is allowed to melt into the molten silicon. The
seed is then slowly pulled and continues to rotate as it is
raised from the melt. As the seed is raised, cooling takes
place and material from the melt adheres to it, thus forming
a single crystal ingot. With this technique, ingots with
diameters of several inches can be fabricated.
A brief exposure to the techniques used in the fabrication
of zener diodes can provide the engineer with additional
insight using zeners in their applications. That is, an
understanding of zener fabrication makes the capabilities
and limitations of the zener diode more meaningful. This
chapter discusses the basic steps in the fabrication of the
zener from crystal growing through final testing.
ZENER DIODE WAFER FABRICATION
The major steps in the manufacture of zeners are provided
in the process flow in Figure 1. It is important to point out
that the manufacturing steps vary somewhat from
manufacturer to manufacturer, and also vary with the type of
zener diode produced. This is driven by the type of package
required as well as the electrical characteristics desired. For
example, alloy diffused devices provide excellent low
voltage reference with low leakage characteristics but do not
SILICON CRYSTAL
GROWING
WAFER
PREPARATION
OXIDE
PASSIVATION
WAFER THINNING
ANODE
METALLIZATION
JUNCTION
FORMATION
CATHODE
METALLIZATION
WAFER
TESTING
WAFER DICING
TEST
LEAD
FINISH
ASSEMBLY
MARK
TEST
PACKAGE
SHIP
Figure 1. General Flow of the Zener Diode Process
http://onsemi.com
8
SILICON DIOXIDE SELECTIVELY REMOVED
SILICON DIOXIDE GROWTH
SiO2
Si
(a)
(b)
DOPANT ATOMS DEPOSITED ONTO
THE EXPOSED SILICON
DOPANT ATOMS DIFFUSE INTO SILICON
BUT NOT APPRECIABLY INTO THE SILICON DIOXIDE
(c)
(d)
Figure 2. Basic Fabrication Steps in the Silicon Planar Process: a) oxide formation, b) selective oxide removal,
c) deposition of dopant atoms, d) junction formation by diffusion of dopant atoms.
Dopant is then introduced onto the wafer surface using
various techniques such as aluminum alloy for low voltage
devices, ion-implantation, spin-on dopants, or chemical
vapor deposition. Once the dopant is deposited, the
junctions are formed in a subsequent high temperature (1100
to 1250 degrees celcius are typical) drive-in. The resultant
junction profile is determined by the background
concentration of the starting substrate, the amount of dopant
placed at the surface, and amount of time and temperature
used during the dopant drive-in. This junction profile
determines the electrical characteristics of the device.
During the drive-in cycle, additional passivation oxide is
grown providing additional protection for the devices.
After junction formation, the wafers are then processed
through what is called a getter process. The getter step
utilizes high temperature and slight stress provided by a
highly doped phosphosilicate glass layer introduced into the
backside of the wafers. This causes any contaminants in the
area of the junction to diffuse away from the region. This
serves to improve the reverse leakage characteristic and the
stability of the device. Following the getter process, a second
photo resist step opens the contact area in which the anode
metallization is deposited.
Metal systems for ON Semiconductor’s zener diodes are
determined by the requirements of the package. The metal
systems are deposited in ultra-clean vacuum chambers
utilizing electron-beam evaporation techniques. Once the
metal is deposited, photo resist processing is utilized to form
the desired patterns. The wafers are then lapped to their final
thickness and the cathode metallization deposited using the
same e-beam process.
Once the single-crystal silicon ingot is grown, it is tested
for doping concentration (resistivity), undesired impurity
levels, and minority carrier lifetime. The ingot is then sliced
into thin, circular wafers. The wafers are then chemically
etched to remove saw damage and polished in a sequence of
successively finer polishing grits until a mirror-like defect
free surface is obtained. The wafers are then cleaned and
placed in vacuum sealed wafer carriers to prevent any
contamination from getting on them. At this point, the
wafers are ready to begin device fabrication.
Zener diodes can be manufactured using different
processing techniques such as planar processing or mesa
etched processing. The majority of ON Semiconductor
zener diodes are manufactured using the planar technique as
shown in Figure 2.
The planar process begins by growing an ultra-clean
protective silicon dioxide passivation layer. The oxide is
typically grown in the temperature range of 900 to 1200
degrees celcius. Once the protective layer of silicon dioxide
has been formed, it must be selectively removed from those
areas into which dopant atoms will be introduced. This is
done using photolithographic techniques.
First a light sensitive solution called photo resist is spun
onto the wafer. The resist is then dried and a photographic
negative or mask is placed over the wafer. The resist is then
exposed to ultraviolet light causing the molecules in it to
cross link or polymerize becoming very rigid. Those areas
of the wafer that are protected by opaque portions of the
mask are not exposed and are developed away. The oxide is
then etched forming the exposed regions in which the dopant
will be introduced. The remaining resist is then removed and
the wafers carefully cleaned for the doping steps.
http://onsemi.com
9
a solder disc is placed into each cavity and then a die is put
in on top. A solder disc is put in on top of the die. Another
assembly boat containing only leads is mated to the boat
containing the leads, die, and two solder discs. The boats are
passed through the assembly furnace; this operation requires
only one pass through the furnace.
After assembly, the leads on the Surmetic 30s, 40s and
MOSORBs are plated with a tin-lead alloy making them
readily solderable and corrosion resistant.
The quality of the wafers is closely monitored throughout
the process by using statistical process control techniques
and careful microscopic inspections at critical steps. Special
wafer handling equipment is used throughout the
manufacturing process to minimize contamination and to
avoid damaging the wafers in any way. This further
enhances the quality and stability of the devices.
Upon completion of the fabrication steps, the wafers are
electrically probed, inspected, and packaged for shipment to
the assembly operations. All ON Semiconductor zener
diode product is sawn using 100% saw-through techniques
stringently developed to provide high quality silicon die.
Double Slug (DO-35 and DO-41)
Double slugs receive their name from the dumet slugs, one
attached to one end of each lead. These slugs sandwich the
pre-tinned die between them and are hermetically sealed to
the glass envelope or body during assembly. Figure 4 shows
typical assembly.
The assembly begins with the copper clad steel leads
being loaded into assembly “boats.” Every other boat load
of leads has a glass body set over the slug. A pre-tinned die
is placed into each glass body and the other boat load of leads
is mated to the boat holding the leads, body and die. These
mated boats are then placed into the assembly furnace where
the total mass is heated. Each glass body melts; and as the
boat proceeds through the cooling portion of the furnace
chamber, the tin which has wetted to each slug solidifies
forming a bond between the die and both slugs. The glass
hardens, attaching itself to the sides of the two slugs forming
the hermetic seal. The above illustrates how the diodes are
completely assembled using a single furnace pass
minimizing assembly problems.
The encapsulated devices are then processed through lead
finish. This consists of dipping the leads in molten tin/lead
solder alloy. The solder dipped leads produce an external
finish which is tarnish-resistant and very solderable.
ZENER DIODE ASSEMBLY
Surmetic 30, 40 and MOSORB
The plastic packages (Surmetic 30, 40 and MOSORBs)
are assembled using oxygen free high conductivity copper
leads for efficient heat transfer from the die and allowing
maximum power dissipation with a minimum of external
heatsinking. Figure 3 shows typical assembly. The leads are
of nail head construction, soldered directly to the die, which
further enhances the heat dissipating capabilities of the
package.
The Surmetic 30s, 40s and MOSORBs are basically
assembled in the same manner; the only difference being the
MOSORBs are soldered together using a solder disc
between the lead and die whereas the Surmetic 30s and
Surmetic 40s utilize pre-soldered leads.
Assembly is started on the Surmetic 30 and 40 by loading
the leads into assembly boats and pre-soldering the nail
heads. After pre-soldering, one die is then placed into each
cavity of one assembly boat and another assembly boat is
then mated to it. Since the MOSORBs do not use
pre-soldered leads, the leads are put into the assembly boat,
OFHC COPPER LEAD,
SOLDER PLATED
PLASTIC
(THERMO SET)
ENCAPSULATED
LEAD, STEEL, CU CLAD
SOLDER DIPPED
NAILHEAD LEAD
SLUG DUMET
GLASS SLEEVE
Sn Pb
ZENER DIE
PASSIVATED
ZENER DIE
NAILHEAD LEAD
OFHC COPPER LEAD,
SOLDER PLATED
Figure 3. Double-Slug Plastic
Zener Construction
Figure 4. Double Slug Glass
Zener Construction
http://onsemi.com
10
ZENER DIODE TEST, MARK AND PACKAGING
parameters as the units are being tested to ensure that the lot
is testing well to the process average and compared against
other lots of the same voltage.
After testing, the units are marked as required by the
specification. The markers are equipped to polarity orient
the devices as well as perform 100% redundant test prior to
packaging.
After marking, the units are packaged either in “bulk”
form or taped and reeled or taped and ammo packed to
accommodate automatic insertion.
Double Slug, Surmetic 30, 40 and MOSORB
After lead finish, all products are final tested, whether
they are double slug or of Surmetic construction, all are 100
percent final tested for zener voltage, leakage current,
impedance and forward voltage drop.
Process average testing is used which is based upon the
averages of the previous lots for a given voltage line and
package type. Histograms are generated for the various
http://onsemi.com
11
RELIABILITY
INTRODUCTION
distributions will be centered within the specification limits
with a product distribution of plus or minus Six Sigma about
mean. Six Sigma capability, shown graphically in Figure 1,
details the benefit in terms of yield and outgoing quality
levels. This compares a centered distribution versus a 1.5
sigma worst case distribution shift.
New product development at ON Semiconductor requires
more robust design features that make them less sensitive to
minor variations in processing. These features make the
implementation of SPC much easier.
A complete commitment to SPC is present throughout
ON Semiconductor. All managers, engineers, production
operators, supervisors and maintenance personnel have
received multiple training courses on SPC techniques.
Manufacturing has identified 22 wafer processing and 8
assembly steps considered critical to the processing of zener
products. Processes, controlled by SPC methods, that have
shown significant improvement are in the diffusion,
photolithography and metallization areas.
To better understand SPC principles, brief explanations
have been provided. These cover process capability,
implementation and use.
ON Semiconductor’s Quality System maintains
“continuous product improvement” goals in all phases of the
operation. Statistical process control (SPC), quality control
sampling, reliability audits and accelerated stress testing
techniques monitor the quality and reliability of its products.
Management and engineering skills are continuously
upgraded through training programs. This maintains a
unified focus on Six Sigma quality and reliability from the
inception of the product to final customer use.
STATISTICAL PROCESS CONTROL
ON Semiconductor’s Discrete Group is continually
pursuing new ways to improve product quality. Initial design
improvement is one method that can be used to produce a
superior product. Equally important to outgoing product
quality is the ability to produce product that consistently
conforms to specification. Process variability is the basic
enemy of semiconductor manufacturing since it leads to
product variability. Used in all phases of
ON Semiconductor’s
product
manufacturing,
STATISTICAL PROCESS CONTROL (SPC) replaces
variability with predictability. The traditional philosophy in
the semiconductor industry has been adherence to the data
sheet specification. Using SPC methods assures the product
will meet specific process requirements throughout the
manufacturing cycle. The emphasis is on defect prevention,
not detection. Predictability through SPC methods requires
the manufacturing culture to focus on constant and
permanent improvements. Usually these improvements
cannot be bought with state-of-the-art equipment or
automated factories. With quality in design, process and
material selection, coupled with manufacturing
predictability, ON Semiconductor can produce world class
products.
The immediate effect of SPC manufacturing is
predictability through process controls. Product centered and
distributed well within the product specification benefits ON
Semiconductor with fewer rejects, improved yields and lower
cost. The direct benefit to ON Semiconductor’s customers
includes better incoming quality levels, less inspection time
and ship-to-stock capability. Circuit performance is often
dependent on the cumulative effect of component variability.
Tightly controlled component distributions give the customer
greater circuit predictability. Many customers are also
converting to just-in-time (JIT) delivery programs. These
programs require improvements in cycle time and yield
predictability achievable only through SPC techniques. The
benefit derived from SPC helps the manufacturer meet the
customer’s expectations of higher quality and lower cost
product.
Ultimately, ON Semiconductor will have Six Sigma
capability on all products. This means parametric
-6σ -5s -4σ -3σ -2σ -1σ 0
1σ 2σ 3σ 4σ
5σ 6σ
Standard Deviations From Mean
Distribution Centered
At ± 3 σ 2700 ppm defective
99.73% yield
Distribution Shifted ± 1.5
66810 ppm defective
93.32% yield
At ± 4 σ 63 ppm defective
99.9937% yield
6210 ppm defective
99.379% yield
At ± 5 σ 0.57 ppm defective
99.999943% yield
233 ppm defective
99.9767% yield
At ± 6 σ 0.002 ppm defective
99.9999998% yield
3.4 ppm defective
99.99966% yield
Figure 1. AOQL and Yield from a Normal Distribution of
Product With 6σ Capability
PROCESS CAPABILITY
One goal of SPC is to ensure a process is CAPABLE.
Process capability is the measurement of a process to
produce
products
consistently
to
specification
requirements. The purpose of a process capability study is
to separate the inherent RANDOM VARIABILITY from
ASSIGNABLE CAUSES. Once completed, steps are taken
to identify and eliminate the most significant assignable
causes. Random variability is generally present in the
system and does not fluctuate. Sometimes, these are
considered basic limitations associated with the machinery,
materials, personnel skills or manufacturing methods.
Assignable cause inconsistencies relate to time variations in
yield, performance or reliability.
http://onsemi.com
12
? ?
?
?
?
? ?
? ?
?
?
PREDICTION
PREDICTION
TIME
TIME
SIZE
SIZE
Process “under control” − all assignable causes are
removed and future distribution is predictable.
Figure 2. Impact of Assignable Causes on Process Predictable
In control assignable
causes eliminated
Lower
Specification Limit
TIME
Upper
Specification Limit
TIME
Out of control
(assignable causes present)
SIZE
In control and capable
(variation from random
variability reduced)
In control but not capable
(variation from random variability
excessive)
SIZE
Figure 3. Difference Between Process Control and Process Capability
SPC IMPLEMENTATION AND USE
Traditionally, assignable causes appear to be random due
to the lack of close examination or analysis. Figure 2 shows
the impact on predictability that assignable cause can have.
Figure 3 shows the difference between process control and
process capability.
A process capability study involves taking periodic
samples from the process under controlled conditions. The
performance characteristics of these samples are charted
against time. In time, assignable causes can be identified and
engineered out. Careful documentation of the process is key
to accurate diagnosis and successful removal of the
assignable causes. Sometimes, the assignable causes will
remain unclear requiring prolonged experimentation.
Elements which measure process variation control and
capability are Cp and Cpk respectively. Cp is the
specification width divided by the process width or Cp =
(specification width) / 6σ. Cpk is the absolute value of the
closest specification value to the mean, minus the mean,
divided by half the process width or Cpk = | closest
specification — X / 3σ.
At ON Semiconductor, for critical parameters, the process
capability is acceptable with a Cpk = 1.33. The desired
process capability is a Cpk = 2 and the ideal is a Cpk = 5.
Cpk, by definition, shows where the current production
process fits with relationship to the specification limits. Off
center distributions or excessive process variability will
result in less than optimum conditions.
The Discrete Group uses many parameters that show
conformance to specification. Some parameters are
sensitive to process variations while others remain constant
for a given product line. Often, specific parameters are
influenced when changes to other parameters occur. It is
both impractical and unnecessary to monitor all parameters
using SPC methods. Only critical parameters that are
sensitive to process variability are chosen for SPC
monitoring. The process steps affecting these critical
parameters must be identified also. It is equally important to
find a measurement in these process steps that correlates
with product performance. This is called a critical process
parameter.
Once the critical process parameters are selected, a sample
plan must be determined. The samples used for
measurement
are
organized
into
RATIONAL
SUBGROUPS of approximately 2 to 5 pieces. The
subgroup size should be such that variation among the
samples within the subgroup remain small. All samples must
come from the same source e.g., the same mold press
operator, etc.. Subgroup data should be collected at
appropriate time intervals to detect variations in the process.
As the process begins to show improved stability, the
interval may be increased. The data collected must be
carefully documented and maintained for later correlation.
Examples of common documentation entries would include
operator, machine, time, settings, product type, etc..
http://onsemi.com
13
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
1
154
153
UCL = 152.8
152
151
X = 150.4
150
149
LCL = 148.0
148
147
UCL = 7.3
7
6
5
4
R = 3.2
3
2
1
LCL = 0
0
Figure 4. Example of Process Control Chart Showing Oven Temperature Data
problems with piece to piece variability related to the
process. The X chart can often identify changes in people,
machines, methods, etc. The source of the variability can be
difficult to find and may require experimental design
techniques to identify assignable causes.
Some general rules have been established to help
determine when a process is OUT-OF-CONTROL. Figure
5a shows a control chart subdivided into zones A, B, and C
corresponding to 3 sigma, 2 sigma, and 1 sigma limits
respectively. In Figure 5b through Figure 5e four of the tests
that can be used to identify excessive variability and the
presence of assignable causes are shown. As familiarity with
a given process increases, more subtle tests may be
employed successfully.
Once the variability is identified, the cause of the
variability must be determined. Normally, only a few factors
have a significant impact on the total variability of the
process. The importance of correctly identifying these
factors is stressed in the following example. Suppose a
process variability depends on the variance of five factors A,
B, C, D and E. Each has a variance of 5, 3, 2, 1 and 0.4
respectively.
Once the plan is established, data collection may begin.
The data collected will generate X and R values that are
plotted with respect to time. X refers to the mean of the
values within a given subgroup, while R is the range or
greatest value minus least value. When approximately 20 or
more X and R values have been generated, the average of
these values is computed as follows:
X = (X + X2 + X3 + ...)/K
R = (R1 + R2 + R3 + ...)/K
where K = the number of subgroups measured.
The values of X and R are used to create the process
control chart. Control charts are the primary SPC tool used
to signal a problem. Shown in Figure 4, process control
charts show X and R values with respect to time and
concerning reference to upper and lower control limit
values. Control limits are computed as follows:
R
R
X
X
upper
lower
upper
lower
control
control
control
control
limit
limit
limit
limit
= UCLR = D4 R
LCLR = D3 R
= UCLX = X + A2 R
= LCLX = X − A
Since:
Where D4, D3 and A2 are constants varying by sample
size, with values for sample sizes from 2 to 10 shown in
the following partial table:
Control charts are used to monitor the variability of
critical process parameters. The R chart shows basic
σ tot =
σA2 + σB2 + σC2 + σD2 + σE2
52 + 32 + 22 + 12 + (0.4)2 = 6.3
σ tot =
n
2
3
4
5
6
7
8
9
10
D4
3.27
2.57
2.28
2.11
2.00
1.92
1.86
1.82
1.78
D3
*
*
*
*
*
0.08
0.14
0.18
0.22
A2
1.88
1.02
0.73
0.58
0.48
0.42
0.37
0.34
0.31
* For sample sizes below 7, the LCLR would technically be a negative number; in those cases there is no lower control limit;
this means that for a subgroup size 6, six “identical” measurements would not be unreasonable.
http://onsemi.com
14
be identified and eliminated in the most expeditious manner
possible.
After assignable causes have been eliminated, new
control limits are calculated to provide a more challenging
variability criteria for the process. As yields and variability
improve, it may become more difficult to detect
improvements because they become much smaller. When all
assignable causes have been eliminated and the points
remain within control limits for 25 groups, the process is
said to be in a state of control.
Now if only D is identified and eliminated then;
s tot =
52
+
32
+
22
+
(0.4)2
= 6.2
This results in less than 2% total variability improvement.
If B, C and D were eliminated, then;
σ tot =
52 + (0.4)2 = 5.02
This gives a considerably better improvement of 23%. If
only A is identified and reduced from 5 to 2, then;
σ tot =
22 + 32 + 22 + 12 + (0.4)2 = 4.3
Identifying and improving the variability from 5 to 2 gives
us a total variability improvement of nearly 40%.
Most techniques may be employed to identify the primary
assignable cause(s). Out-of-control conditions may be
correlated to documented process changes. The product may
be analyzed in detail using best versus worst part
comparisons or Product Analysis Lab equipment.
Multi-variance analysis can be used to determine the family
of variation (positional, critical or temporal). Lastly,
experiments may be run to test theoretical or factorial
analysis. Whatever method is used, assignable causes must
SUMMARY
ON Semiconductor is committed to the use of
STATISTICAL PROCESS CONTROLS. These principles,
used throughout manufacturing, have already resulted in
many significant improvements to the processes. Continued
dedication to the SPC culture will allow ON Semiconductor
to reach the Six Sigma and zero defect capability goals. SPC
will further enhance the commitment to TOTAL
CUSTOMER SATISFACTION.
UCL
A
ZONE A (+ 3 SIGMA)
UCL
B
ZONE B (+ 2 SIGMA)
ZONE C (+ 1 SIGMA)
C
CENTERLINE
ZONE C (− 1 SIGMA)
C
ZONE B (− 2 SIGMA)
B
ZONE A (− 3 SIGMA)
A
LCL
Figure 5a. Control Chart Zones
A
Figure 5b. One Point Outside Control Limit
Indicating Excessive Variability
UCL
A
B
B
C
C
C
C
B
B
A
LCL
A
LCL
Figure 5c. Two Out of Three Points in Zone A or
Beyond Indicating Excessive Variability
UCL
LCL
Figure 5d. Four Out of Five Points in Zone B or
Beyond Indicating Excessive Variability
UCL
A
B
C
C
B
A
LCL
Figure 5e. Seven Out of Eight Points in Zone C or
Beyond Indicating Excessive Variability
http://onsemi.com
15
RELIABILITY STRESS TESTS
HIGH TEMPERATURE STORAGE LIFE (HTSL)
High temperature storage life testing is performed to
accelerate failure mechanisms which are thermally
activated through the application of extreme temperatures.
Typical Test Conditions: TA = 70°C to 200°C, no bias,
t = 24 to 2500 hours
Common Failure Modes: Parametric shifts in leakage
Common Failure Mechanisms: Bulk die and diffusion
defects
Military Reference: MIL-STD-750, Method 1032
The following gives brief descriptions of the reliability
tests commonly used in the reliability monitoring program.
Not all of the tests listed are performed on each product.
Other tests may be performed when appropriate. In addition
some form of preconditioning may be used in conjunction
with the following tests.
AUTOCLAVE (aka, PRESSURE COOKER)
Autoclave is an environmental test which measures
device resistance to moisture penetration and the resultant
effects of galvanic corrosion. Autoclave is a highly
accelerated and destructive test.
Typical Test Conditions: TA = 121°C, rh = 100%, p =
1 atmosphere (15 psig), t = 24 to 96 hours
Common Failure Modes: Parametric shifts, high
leakage and/or catastrophic
Common Failure Mechanisms: Die corrosion or
contaminants such as foreign material on or within the
package materials. Poor package sealing
INTERMITTENT OPERATING LIFE (IOL)
The purpose of this test is the same as SSOL in addition
to checking the integrity of both wire and die bonds by
means of thermal stressing.
Typical Test Conditions: TA = 25°C, Pd = Data Book
maximum rating, Ton = Toff = Δ of 50°C to 100°C, t =
42 to 30000 cycles
Common Failure Modes: Parametric shifts and
catastrophic
Common Failure Mechanisms: Foreign material, crack
and bulk die defects, metallization, wire and die bond
defects
Military Reference: MIL-STD-750, Method 1037
HIGH HUMIDITY HIGH TEMPERATURE BIAS
(H3TB or H3TRB)
This is an environmental test designed to measure the
moisture resistance of plastic encapsulated devices. A bias
is applied to create an electrolytic cell necessary to
accelerate corrosion of the die metallization. With time, this
is a catastrophically destructive test.
Typical Test Conditions: TA = 85°C to 95°C, rh = 85%
to 95%, Bias = 80% to 100% of Data Book max. rating,
t = 96 to 1750 hours
Common Failure Modes: Parametric shifts, high
leakage and/or catastrophic
Common Failure Mechanisms: Die corrosion or
contaminants such as foreign material on or within the
package materials. Poor package sealing
Military Reference: MIL-STD-750, Method 1042
MECHANICAL SHOCK
This test is used to determine the ability of the device to
withstand a sudden change in mechanical stress due to
abrupt changes in motion as seen in handling, transportation,
or actual use.
Typical Test Conditions: Acceleration = 1500 g’s,
Orientation = X1, Y1, Y2 plane, t = 0.5 msec, Blows = 5
Common Failure Modes: Open, short, excessive
leakage, mechanical failure
Common Failure Mechanisms: Die and wire bonds,
cracked die, package defects
Military Reference: MIL-STD-750, Method 2015
HIGH TEMPERATURE REVERSE BIAS (HTRB)
MOISTURE RESISTANCE
The purpose of this test is to align mobile ions by means
of temperature and voltage stress to form a high-current
leakage path between two or more junctions.
Typical Test Conditions: TA = 85°C to 150°C, Bias =
80% to 100% of Data Book max. rating, t = 120 to 1000
hours
Common Failure Modes: Parametric shifts in leakage
Common Failure Mechanisms: Ionic contamination on
the surface or under the metallization of the die
Military Reference: MIL-STD-750, Method 1039
The purpose of this test is to evaluate the moisture
resistance of components under temperature/humidity
conditions typical of tropical environments.
Typical Test Conditions: TA = −10°C to 65°C, rh = 80%
to 98%, t = 24 hours/cycle, cycle = 10
Common Failure Modes: Parametric shifts in leakage
and mechanical failure
Common Failure Mechanisms: Corrosion or
contaminants on or within the package materials. Poor
package sealing
Military Reference: MIL-STD-750, Method 1021
http://onsemi.com
16
SOLDERABILITY
and transitions between temperature extremes. This testing
will also expose excessive thermal mismatch between
materials.
Typical Test Conditions: TA = −65°C to 200°C, cycle
= 10 to 1000
Common Failure Modes: Parametric shifts and
catastrophic
Common Failure Mechanisms: Wire bond, cracked or
lifted die and package failure
Military Reference: MIL-STD-750, Method 1051
The purpose of this test is to measure the ability of device
leads/terminals to be soldered after an extended period of
storage (shelf life).
Typical Test Conditions: Steam aging = 8 hours, Flux
= R, Solder = Sn60, Sn63
Common Failure Modes: Pin holes, dewetting,
non-wetting
Common Failure Mechanisms: Poor plating,
contaminated leads
Military Reference: MIL-STD-750, Method 2026
THERMAL SHOCK (LIQUID TO LIQUID)
SOLDER HEAT
The purpose of this test is to evaluate the ability of the
device to withstand both exposure to extreme temperatures
and sudden transitions between temperature extremes. This
testing will also expose excessive thermal mismatch
between materials.
Typical Test Conditions: TA = 0°C to 100°C, cycles
= 10 to 1000
Common Failure Modes: Parametric shifts and
catastrophic
Common Failure Mechanisms: Wire bond, cracked or
lifted die and package failure
Military Reference: MIL-STD-750, Method 1056
This test is used to measure the ability of a device to
withstand the temperatures as may be seen in wave soldering
operations. Electrical testing is the endpoint criterion for this
stress.
Typical Test Conditions: Solder Temperature = 260°C,
t = 10 seconds
Common Failure Modes: Parameter shifts, mechanical
failure
Common Failure Mechanisms: Poor package design
Military Reference: MIL-STD-750, Method 2031
STEADY STATE OPERATING LIFE (SSOL)
VARIABLE FREQUENCY VIBRATION
The purpose of this test is to evaluate the bulk stability of
the die and to generate defects resulting from manufacturing
aberrations that are manifested as time and stress-dependent
failures.
Typical Test Conditions: TA = 25°C, PD = Data Book
maximum rating, t = 16 to 1000 hours
Common Failure Modes: Parametric shifts and
catastrophic
Common Failure Mechanisms: Foreign material, crack
die, bulk die, metallization, wire and die bond defects
Military Reference: MIL-STD-750, Method 1026
This test is used to examine the ability of the device to
withstand deterioration due to mechanical resonance.
Typical Test Conditions: Peak acceleration = 20 g’s,
Frequency range = 20 Hz to 20 kHz, t = 48 minutes.
Common Failure Modes: Open, short, excessive
leakage, mechanical failure
Common Failure Mechanisms: Die and wire bonds,
cracked die, package defects
Military Reference: MIL-STD-750, Method 2056
TEMPERATURE CYCLING (AIR TO AIR)
The purpose of this test is to evaluate the ability of the
device to withstand both exposure to extreme temperatures
http://onsemi.com
17
ZENER DIODE CHARACTERISTICS
At first glance the zener diode is a simple device
consisting of one P-N junction with controlled breakdown
voltage properties. However, when considerations are given
to the variations of temperature coefficient, zener
impedance, thermal time response, and capacitance, all of
which are a function of the breakdown voltage (from 1.8 to
400 V), a much more complicated picture arises. In addition
to the voltage spectrum, a variety of power packages are on
the market with a variation of dice area inside the
encapsulation.
This chapter is devoted to sorting out the important
considerations in a “typical” fashion. For exact details, the
data sheets must be consulted. However, much of the
information contained herein is supplemental to the data
sheet curves and will broaden your understanding of zener
diode behavior.
Specifically, the following main subjects will be detailed:
Basic DC Volt-Ampere Characteristics
Impedance versus Voltage and Current
Temperature Coefficient versus Voltage and Current
Power Derating
Mounting
Thermal Time Response − Effective Thermal Impedance
Surge Capabilities
Frequency Response − Capacitance and Switching
Effects
FORWARD CURRENT
INTRODUCTION
REVERSE VOLTAGE
FORWARD CHARACTERISTIC
REVERSE CURRENT
LEAKAGE REGION
BREAKDOWN
REGION
FORWARD VOLTAGE
Figure 1. Typical Zener Diode DC V-I Characteristics
(Not to Scale)
While the common form of the diode equation suggests
that IR is constant, in fact IR is itself strongly temperature
dependent. The rapid increase in IR with increasing
temperature dominates the decrease contributed by the
exponential term in the diode equation. As a result, the
forward current increases with increasing temperature.
Figure 2 shows a forward characteristic temperature
dependence for a typical zener. These curves indicate that
for a constant current, an increase in temperature causes a
decrease in forward voltage. The voltage temperature
coefficient values are typically in the range of −1.4 to
−2 mV/°C.
BASIC ZENER DIODE DC VOLT-AMPERE
CHARACTERISTICS
1000
500
I F , FORWARD CURRENT (mA)
Reverse and forward volt-ampere curves are represented
in Figure 1 for a typical zener diode. The three areas −
forward, leakage, and breakdown − will each be examined.
FORWARD DC CHARACTERISTICS
The forward characteristics of a zener diode are
essentially identical with an “ordinary” rectifier and is
shown in Figure 2. The volt-ampere curve follows the basic
diode equation of IF = IReqV/KT where KT/q equals about
0.026 volts at room temperature and IR (reverse leakage
current) is dependent upon the doping levels of the P-N
junction as well as the area. The actual plot of VF versus IF
deviates from the theoretical due to slightly “fixed” series
resistance of the lead wire, bonding contacts and some bulk
effects in the silicon.
200
100
50
20
10
TJ = 150°C
100°C
25°C
−55°C
5
2
1
0.3
0.4
0.5
0.6
0.7
0.8
0.9
VF, FORWARD VOLTAGE (VOLTS)
Figure 2. Typical Forward Characteristics of
Zener Diodes
http://onsemi.com
18
1
LEAKAGE DC CHARACTERISTICS
1000
500
I Z , ZENER CURRENT (mA)
When reverse voltage less than the breakdown is applied
to a zener diode, the behavior of current is similar to any
back-biased silicon P-N junction. Ideally, the reverse current
would reach a level at about one volt reverse voltage and
remain constant until breakdown is reached. There are both
theoretical and practical reasons why the typical V-I curve
will have a definite slope to it as seen in Figure 3.
Multiplication effects and charge generation sites are
present in a zener diode which dictate that reverse current
(even at low voltages) will increase with voltage. In
addition, surface charges are ever present across P-N
junctions which appear to be resistive in nature.
The leakage currents are generally less than one
microampere at 150°C except with some large area devices.
Quite often a leakage specification at 80% or so of
breakdown voltage is used to assure low reverse currents.
20
10
5
2
1
0.5
0.2
0.1
25
T = −55°C
26
25°C
27
100°C
28
29
150°C
30
31
32
VZ, ZENER VOLTAGE (VOLTS)
Figure 4. Typical Zener Characteristic
Variation with Temperature
An exaggerated view of the knee region is shown in
Figure 5. As can be seen, the breakdown or avalanche
current does not increase suddenly, but consists of a series
of smoothly rising current versus voltage increments each
with a sudden break point.
10000
1000
TJ = 150°C
CURRENT
100
10
ZENER CURRENT
IR , REVERSE LEAKAGE CURRENT (nA)
T = TJ
T = TA
200
100
50
25°C
1
−55°C
0.1
0
4
6
12
16
20
VOLTAGE
EXAGGERATED V-I
OF KNEE REGION
VR, REVERSE VOLTAGE (VOLTS)
Figure 3. Typical Leakage Current
versus Voltage
ZENER VOLTAGE
VOLTAGE BREAKDOWN
Figure 5. Exaggerated V-I Characteristics
of the Knee Region
At some definite reverse voltage, depending on the doping
levels (resistivity) of the P-N junction, the current will begin
to avalanche. This is the so-called “zener” or “breakdown”
area and is where the device is usually biased during use. A
typical family of breakdown curves showing the effect of
temperature is illustrated in Figure 4.
Between the minimum currents shown in Figure 4 and the
leakage currents, there is the “knee” region. The avalanche
mechanism may not occur simultaneously across the entire
area of the P-N junction, but first at one microscopic site,
then at an increasing number of sites as further voltage is
applied. This action can be accounted for by the
“microplasma discharge” theory and correlates with several
breakdown characteristics.
At the lowest point, the zener resistance (slope of the
curve) would test high, but as current continues to climb, the
resistance decreases. It is as though each discharge site has
high resistance with each succeeding site being in parallel
until the total resistance is very small.
In addition to the resistive effects, the micro plasmas may
act as noise generators. The exact process of manufacturing
affects how high the noise will be, but in any event there will
be some noise at the knee, and it will diminish considerably
as current is allowed to increase.
Since the zener impedance and the temperature
coefficient are of prime importance when using the zener
diode as a reference device, the next two sections will
expand on these points.
http://onsemi.com
19
ZENER IMPEDANCE
100
ZENER CURRENT (mA)
The slope of the VZ − IZ curve (in breakdown) is defined
as zener impedance or resistance. The measurement is
generally done with a 60 Hz (on modern, computerized
equipment this test is being done at 1 kHz) current variation
whose value is 10% in rms of the dc value of the current.
(That is, ΔIZ peak to peak = 0.282 IZ.) This is really not a
small signal measurement but is convenient to use and gives
repeatable results.
The zener impedance always decreases as current
increases, although at very high currents (usually beyond IZ
max) the impedance will approach a constant. In contrast,
the zener impedance decreases very rapidly with increasing
current in the knee region. On Semiconductor specifies most
zener diode impedances at two points: IZT and IZK. The term
IZT usually is at the quarter power point, and IZK is an
arbitrary low value in the knee region. Between these two
points a plot of impedance versus current on a log-log scale
is close to a straight line. Figure 6 shows a typical plot of ZZ
versus IZ for a 20 volt−500 mW zener. The worst case
impedance between IZT and IZK could be approximated by
assuming a straight line function on a log-log plot; however,
at currents above IZT or below IZK a projection of this line
may give erroneous values.
0.1
2
1
3
4
5
6
7
8
ZENER VOLTAGE (VOLTS)
Figure 7. Zener Current versus Zener Voltage
(Low Voltage Region)
Possibly the plots shown in Figure 8 of zener impedance
versus voltage at several constant IZ’s more clearly points
out this effect. It is obvious that zener diodes whose
breakdowns are about 7 volts will have remarkably low
impedance.
200
Z Z , DYNAMIC IMPEDANCE (OHMS)
ZENER IMPEDANCE (OHMS)
1
0.01
1000
ZZK(MAX)
10
APPROXIMATE MAXIMUM LINE
100
ZZT(MAX)
10
100
70
50
1 mA
TA = 25°C
IZ(ac) = 0.1 IZ(dc)
30
20
10 mA
10
7
5
20 mA
3
2
2
3
5
7
10
20
30
50 70 100
200
VZ, ZENER VOLTAGE (VOLTS)
1
0.1
1
10
Figure 8. Dynamic Zener Impedance (Typical)
versus Zener Voltage
100
ZENER CURRENT (mA)
However, this is not the whole picture. A zener diode
figure of merit as a regulator could be ZZ/VZ. This would
give some idea of what percentage change of voltage could
be expected for some given change in current. Of course, a
low ZZ/VZ is desirable. Generally zener current must be
decreased as voltage is increased to prevent excessive power
dissipation; hence zener impedance will rise even higher and
the “figure of merit” will become higher as voltage
increases. This is the case with IZT taken as the test point.
However, if IZK is used as a comparison level in those
devices which keep a constant IZK over a large range of
voltage, the “figure of merit” will exhibit a bowl-shaped
curve − first decreasing and then increasing as voltage is
increased. Typical plots are shown in Figure 9. The
conclusion can be reached that for uses where wide swings
of current may occur and the quiescent bias current must be
high, the lower voltage zener will provide best regulation,
Figure 6. Zener Impedance versus
Zener Current
The impedance variation with voltage is much more
complex. First of all, zeners below 6 volts or so exhibit “field
emission” breakdown converting to “avalanche” at higher
currents. The two breakdowns behave somewhat differently
with “field emission” associated with high impedance and
negative temperature coefficients and “avalanche” with
lower impedance and positive temperature coefficients.
A V-I plot of several low voltage 500 mW zener diodes is
shown in Figure 7. It is seen that at some given current
(higher for the lower voltage types) there is a fairly sudden
decrease in the slope of ΔV/ΔI. Apparently, this current is the
transition from one type of breakdown to the other. Above
6 volts the curves would show a gradual decrease of ΔV/ΔI
rather than an abrupt change, as current is increased.
http://onsemi.com
20
9.3 volts at 75°C for all useful current levels (disregarding
impedance effects).
As was mentioned, the situation is further complicated by
the normal deviation of TC at a given current. For example,
for 7.5 mA the normal spread of TC (expressed in %/°C) is
shown in Figure 11. This is based on limited samples and in
no manner implies that all On Semiconductor zeners
between 2 and 12 volts will exhibit this behavior. At other
current levels similar deviations would occur.
100
400 mW, ZZK(MAX) AT 0.25 mA
0.85 mA
1.3 mA
10 W, ZZK(MAX) AT 1 mA
10
12.5 mA
3.8 mA
SEE NOTE BELOW
1.7 mA
1
+0.08
TEMPERATURE COEFFICIENT (%/C)
°
ZENER IMPEDANCE (MAX)/ZENER VOLTAGE
but for low power applications, the best performance could
be obtained between 50 and 100 volts.
400 mW, ZZT(MAX)
250 mA
0.1
10
75 mA
17 mA
28 mA
10 W, ZZT(MAX)
30
50
70
90
110
130
150
ZENER VOLTAGE (VOLTS)
(NOTE: CURVE IS APPROXIMATE, ACTUAL
ZZ(MAX) IS ROUNDED OFF TO NEAREST
WHOLE NUMBER ON A DATA SHEET)
Figure 9. Figure of Merit: ZZ(Max)/VZ versus VZ
(400 mW & 10 W Zeners)
+0.04
+0.02
−0.02
−0.04
−0.06
MAX
2
4
6
8
10
12
14
Figure 11. Temperature Coefficient Spread
versus Zener Voltage
Obviously, all of these factors make it very difficult to
attempt any calculation of precise voltage shift due to
temperature. Except in devices with specified maximum
T.C., no “worse case” design is possible. Information
concerning the On Semiconductor temperature
compensated or reference diodes is given in Chapter 4.
Typical temperature characteristics for a broad range of
voltages is illustrated in Figure 12. This graphically shows
the significant change in voltage for high voltage devices
(about a 20 volt increase for a 100°C increase on a 200 volt
device).
6
VZ REFERENCE AT IZ = IZT
& TA = 25°C
3
2
1
100
0.01 mA
10 mA
0
30 mA
−1
0.1 mA
Δ V Z (+25 °C TO +125 °C)
TC , TEMPERATURE COEFFICIENTS (mV/C)
°
0
ZENER VOLTAGE (VOLTS)
7
1 mA
−2
−3
MIN
TYPICAL
−0.08
−0.10
Below three volts and above eight volts the zener voltage
change due to temperature is nearly a straight line function
and is almost independent of current (disregarding
self-heating effects). However, between three and eight
volts the temperature coefficients are not a simple affair. A
typical plot of TC versus VZ is shown in Figure 10.
4
IZT = 7.5 mA
0
TEMPERATURE COEFFICIENT
5
MAX
TYPICAL
MIN
+0.06
2
3
4
5
6
7
8
9
10
11
12
VZ, ZENER VOLTAGE (VOLTS)
Figure 10. Temperature Coefficient versus
Zener Voltage at 25°C Conditions Typical
Any attempt to predict voltage changes as temperature
changes would be very difficult on a “typical” basis. (This,
of course, is true to a lesser degree below three volts and
above eight volts since the curve shown is a typical one and
slight deviations will exist with a particular zener diode.) For
example, a zener which is 5 volts at 25°C could be from 4.9
to 5.05 volts at 75°C depending on the current level.
Whereas, a zener which is 9 volts at 25°C would be close to
10
1
NOTE: DV IS + ABOVE 5 VOLTS
− BELOW 4.3 VOLTS
BETWEEN 4.3 & 5 VOLTS
VARIES ABOUT + 0.08 VOLTS
0.1
0.01
1
2 3
5
10
50
100
200
ZENER VOLTAGE (VOLTS)
Figure 12. Typical Temperature
Characteristics
http://onsemi.com
21
1,000
PD , MAXIMUM POWER DISSIPATION (WATTS)
POWER DERATING AND MOUNTING
The zener diode like any other semiconductor has a
maximum junction temperature. This limit is somewhat
arbitrary and is set from a reliability viewpoint. Most
semiconductors exhibit an increasing failure rate as
temperature increases. At some temperature, the solder will
melt or soften and the failure rate soars. The 175°C to 200°C
junction temperature rating is quite safe from solder failures
and still has a very low failure rate.
In order that power dissipated in the device will never cause
the junction to rise beyond 175°C or 200°C (depending on the
device), the relation between temperature rise and power
must be known. Of course, the thermal resistance (RθJA or
RθJL) is the factor which relates power and temperature in the
well known “Thermal Ohm’s Law’’ relation:
and
where
TJ
TA
TL
RθJA
RθJL
PZ
ΔT = TJ − TA = RθJAPZ
(1)
ΔT = TJ − TL = RθJLPZ
(2)
L = LEAD LENGTH
TO HEAT SINK
L = 1″
1
L = 1/8″
L = 3/8″
0.75
0.50
0.25
0
20
40
60
80
100
120
140
160
200
Figure 14. Power Temperature
Derating Curve
A lead mounted device can have its power rating
increased by shortening the lead length and “heatsinking”
the ends of the leads. This effect is shown in Figure 15, for
the 1N4728, 1 watt zener diode.
Each zener has a derating curve on its data sheet and
steady state power can be set properly. However,
temperature increases due to pulse use are not so easily
calculated. The use of “Transient Thermal Resistance”
would be required. The next section expounds upon
transient thermal behavior as a function of time and surge
power.
= Junction temperature
= Ambient temperature
= Lead temperature
= Thermal resistance junction to ambient
= Thermal resistance junction to lead
= Zener power dissipation
RqJL, JUNCTION−TO−LEAD
THERMAL RESISTANCE (°C/W)
175
150
125
100
75
50
25
0
500
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
L, LEAD LENGTH TO HEAT SINK (INCH)
400
Figure 15. Typical 1N4728 Thermal
Resistance versus Lead Length
300
THERMAL TIME RESPONSE
Early studies of zener diodes indicated that a “thermal
time constant” existed which allowed calculation of
temperature rise as a function of power pulse height, width,
and duty cycle. More precise measurements have shown that
temperature response as a function of time cannot be
represented as a simple time constant. Although as shown in
the preceding section, the steady state conditions are
analogous in every way to an electrical resistance; a simple
“thermal capacitance” placed across the resistor is not the
true equivalent circuit. Probably a series of parallel R-C
200
100
0
25
180
TL, LEAD TEMPERATURE (°C)
Obviously, if ambient or lead temperature is known and
the appropriate thermal resistance for a given device is
known, the junction temperature could be precisely
calculated by simply measuring the zener dc current and
voltage (PZ = IZVZ). This would be helpful to calculate
voltage change versus temperature. However, only
maximum and typical values of thermal resistance are given
for a family of zener diodes. So only “worst case” or typical
information could be obtained as to voltage changes.
The relations of equations 1 and 2 are usually expressed
as a graphical derating of power versus the appropriate
temperature. Maximum thermal resistance is used to
generate the slope of the curve. An example of a 400
milliwatt device derated to the ambient temperature and a 1
watt device derated to the lead temperature are shown in
Figures 13 and 14.
PD , POWER DISSIPATION (MILLIWATTS)
1.25
50
75
100
125
150
TA, AMBIENT TEMPERATURE (°C)
175
200
Figure 13. 400 mW Power Temperature
Derating Curve
http://onsemi.com
22
networks or lumped constants representing a thermal
transmission line would be more accurate.
Fortunately a concept has developed in the industry
wherein the exact thermal equivalent circuit need not be
found. If one simply accepts the concept of a thermal
resistance which varies with time in a predictable manner,
the situation becomes very practical. For each family of
zener diodes, a “worst case” transient thermal resistance
curve may be generated.
The main use of this transient RθJL curve is when the zener
is used as a clipper or a protective device. First of all, the
power wave shape must be constructed. (Note, even though
the power-transient thermal resistance indicates reasonable
junction temperatures, the device still may fail if the peak
current exceeds certain values. Apparently a current
crowding effect occurs which causes the zener to short. This
is discussed further in this section.)
PEAK
TEMPERATURE RISE
AVERAGE
TEMPERATURE RISE
AMBIENT
TEMPERATURE
PEAK POWER (PP)
T
T1
Figure 17. Relation of Junction Temperature to
Power Pulses
This method will predict the temperature rise at the end of
the power pulse after the chain of pulses has reached
equilibrium. In other words, the average power will have
caused an average temperature rise which has stabilized, but
a temperature “ripple” is present.
TRANSIENT POWER-TEMPERATURE EFFECTS
A typical transient thermal resistance curve is shown in
Figure 16. This is for a lead mounted device and shows the
effect of lead length to an essentially infinite heatsink.
To calculate the temperature rise, the power surge wave
shape must be approximated by its rectangular equivalent as
shown in Figure 17. In case of an essentially non-recurrent
pulse, there would be just one pulse, and ΔT = RθT1 Pp. In
the general case, it can be shown that
Example: (Use curve in Figure 16)
Then
ΔT = [DRθJA (ss) + (1 − D) RθT1 + T + RθT1 − RθT] PP
where
D
RθT1
= Duty cycle in percent
= Transient thermal resistance at the time
equal to the pulse width
= Transient thermal resistance at the time
RθT
equal to pulse interval
RθT1 + T = Transient thermal resistance at the time
equal to the pulse interval
= plus one more pulse width.
RθJA(ss) or RθJL(ss) = Steady state value of thermal
RθJA(ss) or RθJL(ss) = resistance
RθJL(t), JUNCTION-TO-LEAD TRANSIENT
THERMAL RESISTANCE (°C/W)
100
70
50
30
20
ÉÉ
ÉÉ
L
L
HEAT SINK
ÉÉ
ÉÉ
And
1
3 5
10
20
50 100 200
If no other considerations were present, it would be a
simple matter to specify a maximum junction temperature
no matter what pulses are present. However, as has been
noted, apparently other fault conditions prevail. The same
group of devices for which the transient thermal curves were
generated were tested by subjecting them to single shot
power pulses. A failure was defined as a significant shift of
leakage or zener voltage, or of course opens or shorts. Each
device was measured before and after the applied pulse.
Most failures were shifts in zener voltage. The results are
shown in Figure 18.
Attempts to correlate this to the transient thermal
resistance work quite well on a typical basis. For example,
assuming a value for 1 ms of 90 watts and 35 watts at 10 ms,
the predicted temperature rise would be 180°C and 190°C.
But on a worst case basis, the temperature rises would be
about one half these values or junction temperatures, on the
order of 85°C to 105°C, which are obviously low.
Apparently at very high power levels a current restriction
occurs causing hot spots. There was no apparent correlation
L = 1″
L = 1/32″
500 1000 2000 5000 10k
RθT1 = 1.8°C/W
RθT = 5.8°C/W
RθT1 + T = 6°C/W
SURGE FAILURES
FOR θJL(t) VALUES AT PULSE WIDTHS
LESS THAN 3.0 ms, THE ABOVE
CURVE CAN BE EXTRAPOLATED
DOWN TO 10 μs AT A CONTINUING
SLOPE OF 1/2
3
2
PP = 5 watt (Lead length 1/32″)
D = 0.1
T1 = 10 ms
T = 100 ms
RθJA(ss) = 12°C/W (for 1/32″ lead length)
ΔT = [0.1 x 12 + (1 − 0.1) 6 + 1.8 − 5.8] 5
ΔT = 13°C
Or at TA = 25°, TJ = 38°C peak
10
7
5
T1
AVERAGE POWER = T PP
30k
PW, PULSE WIDTH (ms)
Figure 16. Typical Transient Thermal
Resistance (For Axial Lead Zener)
http://onsemi.com
23
FREQUENCY AND PULSE CHARACTERISTICS
of zener voltage or current on the failure points since each
group of failures contained a mixture of voltages.
The zener diode may be used in applications which
require a knowledge of the frequency response of the device.
Of main concern are the zener resistance (usually specified
as “impedance”) and the junction capacitance. The
capacitance curves shown in this section are typical.
POWER (WATTS)
1000
100
TYPICAL
ZENER CAPACITANCE
Since zener diodes are basically PN junctions operated in
the reverse direction, they display a capacitance that
decreases with increasing reverse voltage. This is so because
the effective width of the PN junction is increased by the
removal of charges (holes and electrons) as reverse voltage
is increased. This decrease in capacitance continues until the
zener breakdown region is entered; very little further
capacitance change takes place, owing to the now fixed
voltage across the junction. The value of this capacitance is
a function of the material resistivity, ρ, (amount of doping −
which determines VZ nominal), the diameter, D, of junction
or dice size (determines amount of power dissipation), the
voltage across the junction VC, and some constant, K. This
relationship can be expressed as:
WORSE CASE
10
1
0.00001
0.0001
0.001
0.01
0.1
1
TIME OF PULSE (SECONDS)
Figure 18. One Shot Power Failure Axial
Lead Zener Diode
VOLTAGE VERSUS TIME
Quite often the junction temperature is only of academic
interest, and the designer is more concerned with the voltage
behavior versus time. By using the transient thermal
resistance, the power, and the temperature coefficient, the
designer could generate VZ versus time curves. The
On Semiconductor zener diode test group has observed
device voltages versus time until the thermal equilibrium
was reached. A typical curve is shown for a lead mounted
low wattage device in Figure 19 where the ambient
temperature was maintained constant. It is seen that voltage
stabilizes in about 100 seconds for 1 inch leads.
Since information contained in this section may not be
found on data sheets it is necessary for the designer to
contact the factory when using a zener diode as a surge
suppressor. Additional information on transient suppression
application is presented elsewhere in this book.
CC =
ZENER VOLTAGE (VOLTS)
165
164
163
162
161
0.1
1
10
KD4
n KD4
pV
ρ VCC
After the junction enters the zener region, capacitance
remains relatively fixed and the AC resistance then
decreases with increasing zener current.
TEST CIRCUIT CONSIDERATIONS: A capacitive
bridge was used to measure junction capacitance. In this
method the zener is used as one leg of a bridge that is
balanced for both DC at a given reverse voltage and for AC
(the test frequency 1 MHz). After balancing, the variable
capacitor used for balancing is removed and its value
measured on a test instrument. The value thus indicated is
the zener capacitance at reverse voltage for which bridge
balance was obtained. Figure 20 shows capacitance test
circuit.
Figure 21 is a plot of junction capacitance for diffused
zener diode units versus their nominal operating voltage.
Capacitance is the value obtained with reverse bias set at
one-half the nominal VZ. The plot of the voltage range from
6.8 V to 200 V, for three dice sizes, covers most
On Semiconductor diffused-junction zeners. Consult
specific data sheets for capacitance values.
Figures 22, 23, and 24 show plots of capacitance versus
reverse voltage for units of various voltage ratings in each
of the three dice sizes. Junction capacitance decreases as
reverse voltage increases to the zener region. This change in
capacitance can be expressed as a ratio which follows a
one-third law, and C1/C2 = (V2/V1)1/3. This law holds only
from the zener voltage down to about 1 volt, where the curve
begins to flatten out. Figure 25 shows this for a group of low
wattage units.
166
160
0.01
Ǹ
100
TIME (SECONDS)
Figure 19. Zener Voltage (Typical) versus
Time for Step Power Pulses
(500 mW Lead Mounted Devices)
http://onsemi.com
24
0.1 μF
DC
POWER
SUPPLY
HP
NO. 712A
1k
1%
VDC
1k
1%
1k
1%
1k
1%
10/50
pF
ZENER
X UNDER
TEST
SIGNAL
GEN
TEK
NO. 190A
VAC
100 W
HI-GAIN
DIFF SCOPE
NULL IND
TEK
TYPE D
C1
1 MHz
IV
BAL READ
S1
CAP
DECADE
0−.09 mF
100 pF
STEPS
ΔC 1%
R = ZR R
L/C
METER
TEK
140
C2
10/150
pF
Figure 20. Capacitance Test Circuit
1,000
LOW WATTAGE UNITS
AVG. FOR 10 UNITS EACH
C Z , CAPACITANCE (PICOFARADS)
C Z , CAPACITANCE (PICOFARADS) @ VZ/ 2
10,000
HIGH WATTAGE
MEDIUM WATTAGE
1,000
100
LOW WATTAGE
10
1
10
100
100
50 VOLTS
100 VOLTS
1
10
100
VZ, NOMINAL UNIT VOLTAGE (VOLTS)
VR, REVERSE VOLTAGE (VOLTS)
Figure 21. Capacitance versus Voltage
Figure 22. Capacitance versus Reverse Voltage
10,000
MEDIUM WATTAGE UNITS
AVG. FOR 10 UNITS EACH
C Z , CAPACITANCE (PICOFARADS)
C Z , CAPACITANCE (PICOFARADS)
20 VOLTS
10
1,000
10,000
10 VOLTS
1,000
20 VOLTS
50 VOLTS
100 VOLTS
100
10
10 VOLTS
1
10
100
20 VOLTS
1,000
50 VOLTS
100 VOLTS
100
1,000
HIGH WATTAGE UNITS
AVG. FOR 10 UNITS EACH
10 VOLTS
1
10
VR, REVERSE VOLTAGE (VOLTS)
VR, REVERSE VOLTAGE (VOLTS)
Figure 23. Capacitance versus
Reverse Voltage
Figure 24. Capacitance versus
Reverse Voltage
http://onsemi.com
25
100
C Z, CAPACITANCE (PICOFARADS)
1,000
LOW WATTAGE
UNITS
10 VOLTS
100
50 VOLTS
100 VOLTS
10
0.1
1
100
10
VR, REVERSE VOLTAGE (VOLTS)
Figure 25. Flattening of Capacitance Curve at
Low Voltages
mA
DC
DC
SUPPLY
HP
712A
1k
READ
Rx =
SIGNAL
GEN
HP
650A
0.1 μF
1k
600
S1 A
10M
E1 − E2
E2
SET
E1
READ
100 pF
RX = ZZ
E2
SET
S1 B
AC
VTVM
HP
400H
R2
1Ω
DC
VTVM
HP
412A
Figure 26. Impedance Test Circuit
ZENER IMPEDANCE
(3) Measure the voltage across the entire network by
switching S1. The ratio of these two AC voltages is
then a measure of the impedance ratio. This can be
expressed simply as RX = [(E1 − E2)/E2] R2.
Zener impedance appears primarily as composed of a
current-dependent
resistance
shunted
by
a
voltage-dependent capacitor. Figure 26 shows the test
circuit used to gather impedance data. This is a
voltage-impedance ratio method of determining the
unknown zener impedance. The operation is as follows:
Section A of S1 provides a dummy load consisting of a
10-M resistor and a 100 pF capacitor. This network is
required to simulate the input impedance of the AC VTVM
while it is being used to measure the AC IR drop across R2.
This method has been found accurate up to about three
megahertz; above this frequency, lead inductances and strap
capacitance become the dominant factors.
(1) Adjust for desired zener IZDC by observing IR drop
across the 1-ohm current-viewing resistor R2.
(2) Adjust IZAC to 100 μA by observing AC IR drop across
R2.
http://onsemi.com
26
HIGH FREQUENCY AND SWITCHING
CONSIDERATIONS
Figure 27 shows typical impedance versus frequency
relationships of 6.8 volt 500 mW zener diodes at various DC
zener currents. Before the zener breakdown region is
entered, the impedance is almost all reactive, being provided
by a voltage-dependent capacitor shunted by a very high
resistance. When the zener breakdown region is entered, the
capacitance is fixed and now is shunted by
current-dependent resistance. For comparison, Figure 27
also shows the plot for a 680 pF capacitor XC, a 1K 1%
nonreactive resistor, R, and the parallel combination of these
two passive elements, ZT.
At frequencies about 100 kHz or so and switching speeds
above 10 microseconds, shunt capacitance of zener diodes
begins to seriously effect their usefulness. The upper photo
of Figure 28 shows the output waveform of a symmetrical
peak limiter using two zener diodes back-to-back. The
capacitive effects are obvious here. In any application where
the signal is recurrent, the shunt capacitance limitations can
be overcome, as lower photo of Figure 28 shows. This is
done by operating fast diodes in series with the zener. Upon
application of a signal, the fast diode conducts in the forward
direction charging the shunt zener capacitance to the level
where the zener conducts and limits the peak. When the
signal swings the opposite direction, the fast diode becomes
back-biased and prevents fast discharge of shunt
capacitance. The fast diode remains back-biased when the
signal reverses again to the forward direction and remains
off until the input signal rises and exceeds the charge level
of the capacitor. When the signal exceeds this level, the fast
diode conducts as does the zener. Thus, between successive
cycles or pulses the charge in the shunt capacitor holds off
the fast diode, preventing capacitive loading of the signal
until zener breakdown is reached. Figures 29 and 30 show
this method applied to fast-pulse peak limiting.
Z Z , ZENER IMPEDANCE (OHMS)
10,000
1,000
IZMA
0.25
0.50
XC , 680 pF
ZT
R, 1K 1% DC
1K & 680 pF
1.00
100
2
10
1
10
2.50
5
10
20
100
1 kHz
10 kHz
100 kHz
1 MHz
10 MHz
FREQUENCY (Hz)
Figure 27. Zener Impedance
versus Frequency
RS
ei
eo
5 V/cm
0.5 μs/cm
RS
ei
5 V/cm
0.5 μs/cm
Figure 28. Symmetrical Peak Limiter
http://onsemi.com
27
eo
2 V/cm
200 Ω
eo
ei
10 V
Z
50 Ω
eo
ei
20 ns/cm
Figure 29. Shunt Clipper
2 V/cm
200 Ω
ei
eo
ei
eo
50 Ω
0.001
20 ns/cm
Figure 30. Shunt Clipper with Clamping Network
http://onsemi.com
28
Figure 31 is a photo of input-output pulse waveforms
using a zener alone as a series peak clipper. The smaller
output waveform shows the capacitive spike on the leading
edge. Figure 32 clearly points out the advantage of the
clamping network.
0
2 V/cm
eo
10 VZ
ei
50 Ω
200 Ω
eo
ei
20 ns/cm
Figure 31. Basic Series Clipper
.001
2 V/cm
eo
10 VZ
ei
ei
50 Ω
20 ns/cm
Figure 32. Series Clipper with Clamping Network
http://onsemi.com
29
200 Ω
eo
TEMPERATURE COMPENSATED ZENERS
TEMPERATURE CHARACTERISTICS OF THE P-N
JUNCTION AND COMPENSATION
INTRODUCTION
A device which provides reference voltages in a special
manner is a reference diode.
As was discussed in the preceding chapters, the zener
diode has the unique characteristic of exhibiting either a
positive or a negative temperature coefficient, or both. By
properly employing this phenomenon in conjunction with
other semiconductor devices, it is possible to manufacture a
zener reference element exhibiting a very low temperature
coefficient when properly used. This type of low
temperature coefficient device is referred to as a reference
diode.
The voltage of a forward biased P-N junction, at a specific
current, will decrease with increasing temperature. Thus, a
device so biased displays a negative temperature coefficient
(Figure 1). A P-N junction in avalanche (above 5 volts
breakdown) will display a positive temperature coefficient;
that is, voltage will increase as temperature increases. Due
to energy levels of a junction which breaks down below
5 volts, the temperature coefficient is negative.
It follows that various combinations of forward biased
junctions and reverse biased junctions may be arranged to
achieve temperature compensation. From Figure 2 it can be
seen that if the absolute value of voltage change (ΔV) is the
same for both the forward biased diode and the zener diode
where the temperature has gone from 25°C to 100°C, then
the total voltage across the combination will be the same at
both temperatures since one ΔV is negative and the other
positive. Furthermore, if the rate of increase (or decrease) is
the same throughout the temperature change, voltage will
remain constant. The non-linearity associated with the
voltage temperature characteristics is a result of this rate of
change not being a perfect match.
INTRODUCTION TO REFERENCE DIODES
The temperature characteristics of the zener diode are
discussed in a previous chapter, where it was shown that
change in zener voltage with temperature can be significant
under severe ambient temperature changes (for example, a
100 V zener can change 12.5 volts from 0 to 125°C). The
reference diode (often called the temperature compensated
zener or the TC zener) is specially designed to minimize
these specific temperature effects.
Design of temperature compensated zeners make possible
devices with voltage changes as low as 5 mV from −55 to
+100°C, consequently, the advantages of the temperature
compensated zener are obvious. In critical applications, as
a voltage reference in precision dc power supplies, in high
stability oscillators, in digital voltmeters, in frequency
meters, in analog-to-digital converters, or in other precision
equipment, the temperature compensated zener is a
necessity.
Conceivably temperature compensated devices can be
designed for any voltage but present devices with optimum
voltage temperature characteristics are limited to specific
voltages. Each family of temperature compensated zeners is
designed by careful selection of its integral parts with special
attention to the use conditions (temperature range and
current). A distinct operating current is associated with each
device. Consequently, changes from the specified operating
current can only degrade the voltage-temperature
relationships. This will be discussed in more detail later.
The device “drift” or voltage-time stability is critical in
some reference applications. Typically zeners and TC
zeners offer stability of better than 500 parts per million per
1000 hours.
VREF = VZ + ΔVZ + VD − ΔVD
THE METHODS OF TEMPERATURE
COMPENSATION
The effect of temperature is shown in Figure 1. The
forward characteristic does not vary significantly with
reverse voltage breakdown (zener voltage) rating. A change
in ambient temperature from 25° to 100°C produces a shift
in the forward curve in the direction of lower voltage (a
negative temperature coefficient — in this case about
150 mV change), while the same temperature change
produces approximately 1.9 V increase in the zener voltage
(a positive coefficient). By combining one or more silicon
diodes biased in the forward direction with the P-N biased
zener diode as shown in Figure 3, it is possible to
compensate almost completely for the zener temperature
coefficient. Obviously, with the example shown, 13
junctions would be needed. Usually reference diodes are low
voltage devices, using zeners with 6 to 8 volts breakdown
and one or two forward diodes.
http://onsemi.com
30
450
IF (mA)
150 mV
FORWARD
CHARACTERISTIC
TYPICAL
(ALL TYPES)
100°C
25°C
150
VZ (VOLTS)
20
30
300
10
0.5
1
1.5
VF (VOLTS)
15
IZ (mA)
1.9 V
100°C
30
25°C
45
Figure 1. Effects of Temperature on Zener Diode Characteristics
DIRECTION OF CURRENT FLOW
FORWARD-BIASED
PN JUNCTION
+
REVERSE-BIASED
ZENER JUNCTION
PACKAGE
OUTLINE
25°C
100°C 25°C
−
−ΔV
−V
100°C
7.5 mA
25°C
−
7.5 mA
+V
+ΔV
100°C
+ΔV
−
100°C 25°C
Figure 2. Principle of Temperature Compensation
+
SILICON JUNCTION
DIODES
ZENER DIODES
−
Figure 3. Zener Temperature Compensation with Silicon Forward Junctions
http://onsemi.com
31
TEMPERATURE COEFFICIENT STABILITY
In ac regulator and clipper circuits where zener diodes are
normally connected cathode to cathode, the forward biased
diode during each half cycle can be chosen with the correct
forward temperature coefficient (by stacking, etc.) to
correctly compensate for the temperature coefficient of the
reverse-biased zener diode. It is possible to compensate for
voltage drift with temperature using this method to the
extent that zener voltage stabilities on the order of
0.001%/°C are quite feasible.
This technique is sometimes employed where higher
wattage devices are required or where the zener is
compensated by the emitter base junction of a transistor
stage. Consider the example of using discrete components,
1N4001 rectifier and ON Semiconductor 5 Watt zener, to
obtain compensated voltage-temperature characteristics.
Examination of the curve in Figure 4 indicates that a 10 volt
zener diode exhibits a temperature coefficient of
approximately +5.5 mV/°C. At a current level of 100 mA a
temperature coefficient of approximately −2.0 mV/°C is
characteristic of the 1N4001 rectifiers. A series connection
of three silicon 1N4001 rectifiers produces a total
temperature coefficient of approximately −6 mV/°C and a
total forward drop of approximately 2.17 volts at 25°C. The
combination of three silicon rectifiers and the 10 volt zener
diode produces a device with a coefficient of approximately
−0.5 mV/°C and a total breakdown voltage at 100 mA of
approximately 12.2 volts. Calculation shows this to be a
temperature stability of −0.004%/°C.
ǒ* 0.512.2mVńV 5CǓ
Figure 5 shows the voltage-temperature characteristics of
the TC diode. It can be seen that the voltage drops slightly
with increasing temperature.
6.326
7
mV/ C
4
°
−1
−2
0
1
2
3
4
5
6
7
8
9
10
25
62
100
−6
This non-linearity of the voltage temperature
characteristic leads to a definition of a representative design
parameter ΔVZ. For each device type there is a specified
maximum change allowable. The voltage temperature
stability measurement consists of voltage measurement at
specified temperatures (for the 1N821 Series the
temperatures are −55, 0, +25, +75, and +100°C). The voltage
readings at each of the temperatures is compared with
readings at the other temperatures and the largest voltage
change between any of the specified temperatures
determines the exact device type. For devices registered
prior to complete definition of the voltage temperature
stability measurement, the allowable maximum voltage
change over the temperature range is derived from the
calculation converting %/°C to mV over the temperature
range. Under this standard definition, %/°C is merely a
nomenclature and the meaningful allowable voltage
deviation to be expected becomes the designed parameter.
0
−3
−10
Figure 5. Voltage versus Temperature,
Typical for ON Semiconductor 1N827
Temperature Compensated Zener Diode
ONE
FORWARD
1
−55
TEMPERATURE (°C)
TWO
FORWARDS
2
6.318
−5
THREE
FORWARDS
ALLOY-DIFFUSED
JUNCTION
3
6.320
6.314
100
DIFFUSED
JUNCTION
6.322
6.316
ZENER VOLTAGE (10 mA AT 25°C)
5
mV CHANGE FROM 25°C VOLTAGE
VOLTAGE (VOLTS)
6.324
The temperature compensated zener employs the
technique of specially selected dice. This provides optimum
voltage temperature characteristics by close control of dice
resistivities.
6
6
5
4
3
2
1
0
−1
−2
−3
−4
11 12 13
VOLTS
Figure 4.
http://onsemi.com
32
CURRENT
−6.6
Thus far, temperature compensated zeners have been
discussed mainly with regard to temperature and voltage.
However, the underlying assumption throughout the
previous discussion was that current remained constant.
There is a significant change in the temperature
coefficient of a unit depending on how much above or below
the test current the device is operated.
A particular unit with a 0.01%/°C temperature coefficient
at 7.5 mA over a temperature range of −55°C to +100°C
could possibly have a 0.0005%/°C temperature coefficient
at 11 mA. In fact, there is a particular current which can be
determined for each individual unit that will give the lowest
TC.
Manufacturing processes are designed so that the yields of
low TC units are high at the test specification for current. A
unit with a high TC at the test current can have a low TC at
some other current. A look at the volt-ampere curves at
different temperatures illustrates this point clearly (see
Figure 6).
−6.6
−6.5
−6.4
VOLTAGE (VOLTS)
−6.3
−6.2
−6.1
−6
CURRENT
−5.9
+100
ΔIMAX
Figure 7. Effects of Poorly Regulated Current
The three volt-ampere curves do not usually cross over at
exactly the same point. However, this does not take away
from the argument that current regulation is probably the
most
critical
consideration
when
using
temperature-compensated units.
ZENER IMPEDANCE AND CURRENT
REGULATION
Zener impedance is defined as the slope of the V-I curve
at the test point corresponding to the test current. It is
measured by superimposing a small ac current on the dc test
current and then measuring the resulting ac voltage. This
procedure is identical with that used for regular zeners.
Impedance changes with temperature, but the variation is
usually small and it can be assumed that the amount of
current regulation needed at +25°C will be the same for other
temperatures.
As an example, one might want to determine the amount
of current regulation necessary for the device described
below when the maximum deviation in voltage due to
current variation is ±5 millivolts.
IB
IA
C
ΔVC
−6
CURRENT
ΔV
MAX
+100
A
VOLTAGE (VOLTS)
−6.3
−6.2
−6.1
−55
−5.9
B
ΔVB
−6.4
25°C
25°C
−55
−6.5
IC
Figure 6. Voltage-Ampere Curves Showing
Crossover at A
VZT = 6.32 V
IZT = 7.5 mA
ZZT = 15 Ω @ +25°C
ΔV = ΔI⋅VZZT
0.005 = ΔI⋅V15
0.005
= 0.33 mA
ΔI =
15
If the three curves intersect at A, then operation at IA
results in the least amount of voltage deviation due to
temperature from the +25°C voltage. At IB and IC there are
greater excursions (ΔVB and ΔVC) from the +25°C voltage
as temperature increases or decreases.
THE EFFECTS OF POOR CURRENT
REGULATION
Therefore, the current cannot vary more than 0.33 mA.
The amount of current regulation necessary is:
If current shifts (randomly or as a function of
temperature), then an area of operation can be defined for the
temperature compensated zener.
Once again the curves are drawn, this time a shaded area
is shown on the graph. The upper and lower extremities
denote the maximum current values generated by the current
supply while the voltage extremes at each current are shown
by the left and right sides of the area, shown in Figure 7.
0.33
x 100% = 4.5% regulation.
7.5
If the device of Figure 5 is considered to be the device used
in the preceding discussion, it becomes apparent that on the
average more voltage variation is due to current fluctuation
than is due to temperature variation. Therefore, to obtain a
truly stable reference source, the device must be driven from
a constant current source.
http://onsemi.com
33
BASIC VOLTAGE REGULATION USING ZENER DIODES
BASIC CONCEPTS OF REGULATION
going to have to be for the same VO, and not until the ratio
of RS to RR reaches infinity will the output be held entirely
constant for variation in VI. This, of course, is an
impossibility, but it does stress the fact that the regulation
improves as the output impedance becomes lower and lower.
Where the output impedance of Figure 1 is given by
The purpose of any regulator circuit is to minimize output
variations with respect to variations in input, temperature,
and load requirements. The most obvious use of a regulator
is in the design of a power supply, but any circuit that
incorporates regulatory technique to give a controlled
output or function can be considered as a regulator. In
general, to provide a regulated output voltage, electronic
circuitry will be used to pass an output voltage that is
significantly lower than the input voltage and block all
voltage in excess of the desired output. Allocations should
also be made in the regulation circuitry to maintain this
output voltage for variation in load current demand.
There are some basic rules of thumb for the electrical
requirements of the electronic circuitry in order for it to
provide regulation. Number one, the output impedance
should be kept as low as possible. Number two, a controlling
reference needs to be established that is relatively
insensitive to the prevailing variables. In order to illustrate
the importance of these rules, an analysis of some simple
regulator circuits will point out the validity of the
statements. The circuit of Figure 1 can be considered a
regulator. This circuit will serve to illustrate the importance
of a low output impedance.
The resistors RS and RR can be considered as the source
and regulator impedances, respectively.
The output of the circuit is:
ǒ
RRRL
VO = VI x
RR + RL
RRRRLL
RSS +
R
RRR)
+R
RLL
RO =
+
RR
−
The resistance RR represents the internal impedance of the
battery. For this circuit, the output is
VO = VR + VI
Figure 1. Shunt Resistance Regulator
For a given incremental change in VI, the changes in VO
will be:
ǒR
RS
+
RL
1
R
RR L
RS
+1
RRRR )
Ǔ
V
RS
RS
+
+1
RL
RR
(4)
Then for incremental changes in the input VI, the changes
in VO will be dependent on the second term of equation
(4), which again makes the regulation dependent on the
ratio of RS to RR. Where changes in the output voltage
or the regulation of the circuit in Figure 1 were directly
and solely dependent upon the input voltage and output
impedance, the regulation of circuit 2 will have an output
that varies about the reference source VR in accordance
with the magnitude of battery resistance RR and its
fluctuations for changes in VI. Theoretically, if a perfect
battery were used, that is, VR is constant and RR is zero,
the circuit would be a perfect regulator. In other words,
in line with the basic rules of thumb the circuit exhibits
optimum regulation with an output impedance of zero, and
a constant reference source.
For regulator application, a zener diode can be used
instead of a battery with a number of advantages. A battery’s
resistance and nominal voltage will change with age and
load demand; the ON Semiconductor zener diode
characteristics remain unchanged when operating within its
−
ΔVO = ΔVI
−
Figure 2. Regulator with Battery Reference Source
VO
−
VO
VR
Ǔ
RL
RL
VI
RS
RR
+
RS
+
VI
(3)
It is apparent from this relation that as regulation is
improving with RS increasing and RR decreasing the output
impedance RO is decreasing, and is approximately equal to
RR as the ratio is 10 times or greater. The regulation of this
circuit can be greatly improved by inserting a reference
source of voltage in series with RR such as Figure 2.
(1)
VI
=
RS
RS
+
+1
RL
RR
+
RSR R
RS + RR
(2)
Assuming RL fixed at some constant value, it is obvious
from equation (2) that in order to minimize changes in VO
for variations in VI, the shunt resistor RR should be made as
small as possible with respect to the source resistor RS.
Obviously, the better this relation becomes, the larger VI is
http://onsemi.com
34
specified limits. Any voltage value from a couple of volts to
hundreds of volts is available with zener diodes, where
conventional batteries are limited in the nominal values
available. Also, the zener presents a definite size advantage,
and is less expensive than a battery because it is permanent
and need not be regularly replaced. The basic zener diode
shunt regulator circuit is shown in Figure 3.
device in the breakover region in order to maintain the
zener voltage reference. The minimum current can be either
chosen at some point beyond the knee or found on the
manufacturer’s data sheet (IZK). The basic voltage loop
equation for this circuit is:
+
The minimum zener current will occur when VI is
minimum, VZ is maximum, and IL is maximum, then
solving for RS, we have:
+
RS
RZ
VI
ZENER
DIODE
VO
RL
RS =
VZ
−
(5)
VI = (IZ + IL)RS + VZ
VI(min) − VZ(max)
IZ(min) + IL(max)
(6)
Having found RS, we can determine the maximum power
dissipation PZ for the zener diode.
−
Figure 3. Basic Zener Diode Shunt Regulator
PZ(max) = IZ(max) VZ(max)
(7)
VI(max) − VZ(min)
RS
(8)
Where:
Depending upon the operating conditions of the device, a
zener diode will exhibit some relatively low zener
impedance RZ and have a specified breakover voltage of VZ
that is essentially constant. These inherent characteristics
make the zener diode suited for voltage regulator
applications.
IZ(max) =
− IL(min)
Therefore:
PZ(max) =
DESIGNING THE ZENER SHUNT REGULATOR
ƪ
ƫ
VI(max)*
− VZ(min)
VZ(min)
VI(max)
VZ(max)
*− IIL(min)
L(min)
RS
RS
(9)
Once the basic regulator components values have been
determined, adequate considerations will have to be given to
the variation in VO. The changes in VO are a function of four
different factors; namely, changes in VI, IL, temperature, and
the value of zener impedance, RZ. These changes in VO can
be expressed as:
For any given application of a zener diode shunt regulator,
it will be required to know the input voltage variations and
output load requirements. The calculation of component
values will be directly dependent upon the circuit
requirements. The input may be constant or have maximum
and minimum values depending upon the natural regulation
or waveform of the supply source. The output voltage will
be determined by the designer’s choice of VZ and the circuit
requirements. The actual value of VZ will be dependent
upon the manufacturer’s tolerance and some small variation
for different zener currents and operating temperatures.
For all practical purposes, the value of VZ as specified on
the manufacturer’s data sheet can be used to approximate
VO in computing component values. The requirement for
load current will be known and will vary within some given
range of IL(min) to IL(max).
The design objective of Figure 3 is to determine the proper
values of the series resistance, RS, and zener power
dissipation, PZ. A general solution for these values can be
developed as follows, when the following conditions are
known:
VI (input voltage) from VI(min) to VI(max)
VO (output voltage) from VZ(min) to VZ(max)
IL (load current) from IL(min) to IL(max)
ΔVO =
ΔVI
R SR Z
−
ΔIL + TCΔTVZ
RS + RZ
RS
1 + RS
+
RZ
RL
(10)
The value of ΔVO as calculated with equation (10) will
quite probably be slightly different from the actual value
when measured empirically. For all practical purposes
though, this difference will be insignificant for regulator
designs utilizing the conventional commercial line of zener
diodes.
Obviously to precisely predict ΔVO with a given zener
diode, exact information would be needed about the zener
impedance and temperature coefficient throughout the
variation of zener current. The “worst case” change can only
be approximated by using maximum zener impedance and
with typical temperature coefficient.
The basic zener shunt regulator can be modified to
minimize the effects of each term in the regulation equation
(10). Taking one term at a time, it is apparent that the
regulation or changes in output ΔVO will be improved if the
magnitude of ΔVI is reduced. A practical and widely used
technique to reduce input variation is to cascade zener shunt
regulators such as shown in Figure 4.
The value of RS must be of such a value so that the zener
current will not drop below a minimum value of IZ(min).
This minimum zener current is mandatory to keep the
http://onsemi.com
35
+
+
RS1
VI
RS2
Z1
RL′
VO′ = VI′ = VZ1
Z2
RL
VO
−
−
Figure 4. Cascaded Zener Shunt Regulators Reduce
ΔVO by Reducing ΔVI to the Succeeding Stages
high voltage device. So to speak, this technique will kill two
birds with one stone, as it can also be used to minimize
temperature induced variations of the regulator.
In most regulator applications, the single most detrimental
factor affecting regulation is that of variation in junction
temperature. The junction temperature is a function of both
the ambient temperature and that of self heating. In order to
illustrate how the overall temperature coefficient is
improved with series lower voltage zener, a mathematical
relationship can be developed. Consider the diagram of
Figure 5.
This, in essence, is a regulator driven with a pre-regulator
so that the over all regulation is the product of both. The
regulation or changes in output voltage is determined by:
ΔVO =
ΔVZI
RS2RZ2
−
RS2 + RZ2
RS2
RS2
1+
+
RL
RZ2
(11)
ΔIL + TC2 ΔTVZ2
Where:
ΔVZ1 = ΔVO′ =
RS1RZ1
ΔVI
−
RS1 + RZ1
RS1
RS1
1+
+
RL′
RZ1
ΔIL′ + TC1 ΔTVZ1
RL′ = RS2 +
+
(12)
Z1
RLRZ2
and IL′ = IL + IZ2
RL + RZ2
Z2
VI
The changes in output with respect to changes in input for
both stages assuming the temperature and load are constant
is
ΔVO
ΔVO
=
= Regulation of second stage
ΔVZ1
ΔVO′
+
RS
(14)
ΔVO′
ΔVO
ΔVO
x
= Combined regulation
=
ΔVI
ΔVI
ΔVO′
(15)
VO
Zn
−
−
Figure 5. Series Zener Improve Dynamic Impedance
and Temperature Coefficient
(13)
ΔVO′
= Regulation of first stage
ΔVI
RL
With the temperature coefficient TC defined as the % change
per °C, the change in output for a given temperature range
will equal some overall TC x ΔT x Total VZ. Such as
ΔVO(ΔT) = TC ΔT (VZ1 + VZ2 + . . . + VZN)
Obviously, this technique will vastly improve overall
regulation where the input fluctuates over a relatively wide
range. As an example, let’s say the input varies by ±20% and
the regulation of each individual stage reduces the variation
by a factor of 1/20. This then gives an overall output
variation of ±20% × (1/20)2 or ±0.05%.
The next two factors in equation (10) affecting regulation
are changes in load current and temperature excursions. In
order to minimize changes for load current variation, the
output impedance RZRS/(RZ + RS) will have to be reduced.
This can only be done by having a lower zener impedance
because the value of RS is fixed by circuit requirements.
There are basically two ways that a lower zener impedance
can be achieved. One, a higher wattage device can be used
which allows for an increase in zener current of which will
reduce the impedance. The other technique is to series lower
voltage devices to obtain the desired equivalent voltage, so
that the sum of the impedance is less than that for a single
(16)
Obviously, the change in output will also be equal to the
sum of the changes as attributed from each zener.
ΔVO(ΔT) = ΔT(TC1VZ1 + TC2VZ2 + . . . + TCNVZN)
(17)
Setting the two equations equal to each other and solving
for the overall TC, we get
TCΔT(VZ1 + VZ2 + . . . + VZN) = ΔT(TC1VZ1
+ TC2VZ2 + . . . + TCNVZN)
(18)
TC1 VZ1 + TC2VZ2 + . . . + TCNVZN
VZ1 + VZ2 + . . . + VZN
(19)
TC =
For equation (19) the overall temperature coefficient for
any combination of series zeners can be calculated. Say for
instance several identical zeners in series replace a single
higher voltage zener. The new overall temperature
coefficient will now be that of one of the low voltage
devices. This allows the designer to go to the manufacturer’s
http://onsemi.com
36
problems because individual voltages are additive and the
devices all carry the same current and the extent that this
technique can be used is only restricted by the feasibility of
circuit parameters and cost. On the other hand, caution must
be taken when attempting to parallel zener diodes. If the
devices are not closely matched so that they all break over
at the same voltage, the low voltage device will go into
conduction first and ultimately carry all the current. In order
to avoid this situation, the diodes should be matched for
equal current sharing.
data sheet and select a combination of low TC zener diodes
in place of the single higher TC devices. Generally speaking,
the technique of using multiple devices will also yield a
lower dynamic impedance. Advantages of this technique are
best demonstrated by example. Consider a 5 watt diode with
a nominal zener voltage of 10 volts exhibits approximately
0.055% change in voltage per degree centrigrade, a 20 volt
unit approximately 0.075%/°C, and a 100 volt unit
approximately 0.1%/°C. In the case of the 100 volt diode,
five 20 volt diodes could be connected together to provide
the correct voltage reference, but the overall temperature
coefficient would remain that of the low voltage units, i.e.
0.075%/°C. It should also be noted that the same series
combination improves the overall zener impedance in
addition to the temperature coefficient. A 20 volt, 5 watt
ON Semiconductor zener diode has a maximum zener
impedance of 3 ohms, compared to the 90 ohms impedance
which is maximum for a 100 volt unit. Although these
impedances are measured at different current levels, the
series impedance of five 20 volt zener diodes is still much
lower than that of a single 100 volt zener diode at the test
current specified on the data sheet.
For the ultimate in zener shunt regulator performance, the
aforementioned techniques can be combined with the proper
selection of devices to yield an overall improvement in
regulation. For instance, a multiple string of low voltage
zener diodes can be used as a preregulator, with a series
combination of zero TC reference diodes in the final stage
such as Figure 6.
The first stage will reduce the large variation in VI to some
relatively low level, i.e. ΔVZ. This ΔVZ is optimized by
utilizing a series combination of zeners to reduce the overall
TC and ΔVZ. Because of this small fluctuation of input to the
second stage, and if RL is constant, the biasing current of the
TC units can be maintained at their specified level. This will
give an output that is very precise and not significantly
affected by changes in input voltage or junction temperature.
+
EXTENDING POWER AND CURRENT RANGE
The most common practice for extending the power
handling capabilities of a regulator is to incorporate
transistors in the design. This technique is discussed in detail
in the following sections of this chapter. The second
disadvantage to the basic zener shunt regulator is that
because the device does not have a gain function, a feedback
system is not possible with just the zener resistor
combination. For very precise regulators, the design will
normally be an electronic circuit consisting of transistor
devices for control, probably a closed loop feedback system
with a zener device as the basic referencing element.
The concept of regulation can be further extended and
improved with the addition of transistors as the power
absorbing elements to the zener diodes establishing a
reference. There are three basic techniques used that
combine zener diodes and transistors for voltage regulation.
The shunt transistor type shown in Figure 7 will extend the
power handling capabilities of the basic shunt regulator, and
exhibit marked improvement in regulation.
+
+
RS
IZ
IB
VI
Q1
+
RB
RS2
RS1
IC
Z
IL
RL
VO
VBE
Z1
TC1
−
Z2
VI
Z3
Figure 7. Basic Transistor Shunt Regulator
VO
RL
In this configuration the source resistance must be large
enough to absorb the overvoltage in the same manner as in
the conventional zener shunt regulator. Most of the shunt
regulating current in this circuit will pass through the
transistor reducing the current requirements of the zener
diode by essentially the dc current gain of the transistor hFE.
Where the total regulating shunt current is:
TC2
Z4
−
−
−
Figure 6. Series Zeners Cascaded With Series
Reference Diodes for Improved Zener
Shunt Regulation
IS = IZ + IC = IZ + IB hFE
The basic zener shunt regulator exhibits some inherent
limitations to the designer. First of all, the zener is limited to
its particular power dissipating rating which may be less
than the required amount for a particular situation. The total
magnitude of dissipation can be increased to some degree by
utilizing series or parallel units. Zeners in series present few
where
IZ = IB + IRB and IB >> IRB
therefore
http://onsemi.com
37
IS ≈ IZ + IZ hFE = IZ (1 + hFE)
(20)
EMITTER FOLLOWER REGULATOR
The output voltage is the reference voltage VZ plus the
forward junction drop from base to emitter VBE of the
transistor.
VO = VZ + VBE
Another basic technique of transistor-zener regulation is
that of the emitter follower type shown in Figure 8.
(21)
The values of components and their operating condition
is dictated by the specific input and output requirements and
the characteristics of the designer’s chosen devices, as
shown in the following relations:
RS =
ƪ
IZ(min) [1+hFE(min)] + IL(max)
ƪ
PDZ = IZ(max) VZ(max)
(24)
VV
* − VO(min)
V O(min)
I(max)
I(max)
R
RSS
PDQ =
ƪ
VO
+
+
+
ǒƫ
This circuit has the desirable feature of using a series
transistor to absorb overvoltages instead of a large fixed
resistor, thereby giving a significant improvement in
efficiency over the shunt type regulator. The transistor must
be capable of carrying the entire load current and
withstanding voltages equal to the input voltage minus the
load voltage. This, of course, imposes a much more stringent
power handling requirement upon the transistor than was
required in the shunt regulator. The output voltage is a
function of the zener reference voltage and the base to
emitter drop of Q1 as expressed by the equation (28).
(25)
Ǔ
11
− IL(min)
* I L(min) 1 + )
h h
FE(min)
FE(min)
(26)
Ǔ
ǒƫ
VZ(max)
1
− IL(min)
* I L(min) 1 + )
hFE(min)
h FE(min)
VV
* − VVO(min)
I(max)
I(max)
O(min)
R
RSS
−
RL
Figure 8. Emitter Follower Regulator
hence
PDZ =
IL
Z1
(23)
R
RSS
VBE
−
IZ
VI(min) − VZ(max)
IZ(min)
VV
*−VVO(min)
I(max)
I(max)
O(min)
RB
IRB
(22)
when
IZ(max) =
−
+
RS
VI
VI(min) − VO(max)
RB =
Q1
IC
−
ƫ
VO = VZ − VBE
(28)
The load current is approximately equal to the transistor
collector current, such as shown in equation (29).
IL(max) ≈ IC(max)
(27)
Ǔ
− IL(min)
* I L(min)ǒ1V888888
O(max)
(29)
The designer must select a transistor that will meet the
following basic requirements:
Regulation with this circuit is derived in essentially the
same manner as in the shunt zener circuit, where the output
impedance is low and the output voltage is a function of the
reference voltage. The regulation is improved with this
configuration because the small signal output impedance is
reduced by the gain of Q1 by 1/hFE.
One other highly desirable feature of this type of regulator
is that the output is somewhat self compensating for
temperature changes by the opposing changes in VZ and
VBE for VZ ≈ 10 volts. With the zener having a positive
2 mV/°C TC and the transistor base to emitter being a
negative 2 mV/°C TC, therefore, a change in one is cancelled
by the change in the other. Even though this circuit is a very
effective regulator it is somewhat undesirable from an
efficiency standpoint. Because the magnitude of RS is
required to be large, and it must carry the entire input
current, a large percentage of power is lost from input to
output.
PD ≅ (VI(max) − VO)IL(max)
IC(max) ≈ IL(max)
BVCES ≥ (VI(max) − VO)
(30)
Depending upon the designer’s choice of a transistor and
the imposed circuit requirements, the operation conditions
of the circuit are expressed by the following equations:
VZ = VO + VBE
= VO + IL(max)/gFE(min) @ IL(max)
RS =
VI(min) − VZ − VCE(min) @ IL(max)
IL(max)
(31)
Where VCE(min) is an arbitrary value of minimum
collector to emitter voltage and gFE is the transconductance.
This is sufficient to keep the transistor out of saturation,
which is usually about 2 volts.
http://onsemi.com
38
VCE(min) @ IL(max)
IL(max)/hFE(min) @ IL(max) + IZ(min)
RB =
IZ(max) =
transistor with the zener for the shunt control element as
illustrated in Figure 10.
(32)
RS
VI(max) −VZ
RB + RZ
(33)
CONSTANT
CURRENT
SOURCE
(34)
PDZ = IZ(max)VZ
Z1
VI
There are two primary factors that effect the regulation
most in a circuit of this type. First of all, the zener current
may vary over a considerable range as the input changes
from minimum to maximum and this, of course, may have
a significant effect on the value of VZ and therefore VO.
Secondly, VZ and VBE will both be effected by temperature
changes which are additive on their effect of output voltage.
This can be seen by altering equation (28) to show changes
in VO as dependent on temperature, see equation (36).
VO(ΔT) = ΔT[(+TC) VZ − (−TC) VBE]
Q2
RL
+
Figure 10. Series Pass Regulator
This is the third basic technique used for transistor-zener
regulators. This technique or at least a variation of it, finds
the widest use in practical applications. In this circuit the
transistor Q1 is still the series control device operating as an
emitter follower. The output voltage is now established by
the transistor Q2 base to emitter voltage and the zener
voltage. Because the zener is only supplying base drive to
Q2, and it derives its bias from the output, the zener current
remains essential constant, which minimizes changes in VZ
due to IZ excursions. Also, it may be possible (VZ ≈ 10 V)
to match the zener to the base-emitter junction of Q2 for an
output that is insensitive to temperature changes. The
constant current source looks like a very high load
impedance to the collector of Q2 thus assuming a very high
voltage gain. There are three primary advantages gained
with this configuration over the basic emitter follower:
(36)
RS
−
−
Q1
IB = K
VO
RL
FORWARD
BIAS DIODE
1. The increased voltage gain of the circuit with the
addition of Q2 will improve regulation for changes in
both load and input.
2. The zener current excursions are reduced, thereby
improving regulation.
3. For certain voltages the configuration allows good
temperature compensation by matching the
temperature characteristics of the zener to the
base-emitter junction of Q2.
TC ZENER
+
VO
RB
The effects of these detrimental factors can be minimized
by replacing the bleeder resistor RB with a constant current
source and the zener with a reference diode in series with a
forward biased diode (see Figure 9).
VI
IC2
(35)
Actual PDQ = (VI(max) − VO) IL(max)
CONSTANT
CURRENT
SOURCE
−
Q1
+
Figure 9. Improved Emitter Follower Regulator
The constant current source can be either a current limiter
diode or a transistor source. The current limiter diode is
ideally suited for applications of this type, because it will
supply the same biasing current irregardless of collector to
base voltage swing as long as it is within the voltage limits
of the device. This technique will overcome changes in VZ
for changes in IZ and temperature, but changes in VBE due
to load current changes are still directly reflected upon the
output. This can be reduced somewhat by combining a
The series pass regulator is superior to the other transistor
regulators thus far discussed. It has good efficiency, better
stability and regulation, and is simple enough to be
economically practical for a large percentage of
applications.
http://onsemi.com
39
REGULATING
POWER
ELEMENT
REFERENCE
CONTROL
UNIT
INPUT
AND
AMPLIFIER
LOAD
OUTPUT
ERROR
DETECTION
Figure 11. Block Diagram of Regulator with Feedback
EMPLOYING FEEDBACK FOR
OPTIMUM REGULATION
In this circuit, the zener establishes a reference level for
the differential amplifier composed of Q4 and Q5 which will
set the base drive for the control transistor Q3 to regulate the
series high gain transistor combination of Q1 and Q2. The
differential amplifier samples the output at the voltage
dividing network of R8, R9, and R10. This is compared to the
reference voltage provided by the zener Z1. The difference,
if any, is amplified and fed back to the control elements. By
adjusting the potentiometer, R9, the output level can be set
to any desired value within the range of the supply. (The
output voltage is set by the relation VO = VZ[(RX +
RY)/RX].) By matching the transistor Q4 and Q5 for
variations in VBE and gain with temperature changes and
incorporating a temperature compensated diode as the
reference, the circuit will be ultra stable to temperature
effects. The regulation and stability of this circuit is very
good, and for this reason is used in a large percentage of
commercial power supplies.
The regulators discussed thus far do not employ any
feedback techniques for precise control and compensation
and, therefore, find limited use where an ultra precise
regulator is required. In the more sophisticated regulators
some form of error detection is incorporated and amplified
through a feedback network to closely control the power
elements as illustrated in the block diagram of Figure 11.
Regulating circuits of this type will vary in complexity
and configuration from application to application. This
technique can best be illustrated with a couple of actual
circuits of this type. The feedback regulators will generally
be some form of series pass regulator, for optimum
performance and efficiency. A practical circuit of this type
that is extensively utilized is shown in Figure 12.
+
+
Q1
R2
R7
RX
Z1
Q2
Q4
R1
R8
Q5
R9
VI
RL
VO
R3
RY
Q3
C1
R4
R5
R10
R6
−
−
Figure 12. Series Pass Regulator with Error Detection and Feedback
Amplification Derived from a Differential Amplifier
http://onsemi.com
40
+
+
Q1
Q2
R1
R6
REFERENCE
AMPLIFIER
VI
RL
R7
R2
VO
R4
Q3
R8
R3
R5
D1
−
−
Figure 13. Series Pass Regulator with Temperature Compensated Reference Amplifier
CONSTANT CURRENT SOURCES FOR
REGULATOR APPLICATIONS
Another variation of the feedback series pass regulator is
shown in Figure 13. This circuit incorporates a stable
temperature compensated reference amplifier as the primary
control element.
This circuit also employs error detection and amplified
feedback compensation. It is an improved version over the
basic series pass regulator shown in Figure 10. The series
element is composed of a Darlington high gain
configuration formed by Q1 and Q2 for an improved
regulation factor. The combined gain of the reference
amplifier and Q3 is incorporated to control the series unit.
This reduced the required collector current change of the
reference amplifier to control the regulator so that the bias
current remains close to the specified current for low
temperature coefficient. Also the germanium diode D1 will
compensate for the base to emitter change in Q3 and keep the
reference amplifier collector biasing current fairly constant
with temperature changes. Proper biasing of the zener and
transistor in the reference amplifier must be adhered to if the
output voltage changes are to be minimized.
Several places throughout this chapter emphasize the
need for maintaining a constant current level in the various
biasing circuits for optimum regulation. As was mentioned
previously in the discussion on the basic series pass
regulator, the current limiter diode can be effectively used
for the purpose.
Aside from the current limiter diode a transistorized
source can be used. A widely used technique is shown
incorporated in a basic series pass regulator in Figure 14.
The circuit is used as a preregulated current source to
supply the biasing current to the transistor Q2. The constant
current circuit is seldom used alone, but does find wide use
in conjunction with voltage regulators to supply biasing
current to transistors or reference diodes for stable
operation. The Zener Z2 establishes a fixed voltage across
RE and the base to emitter of Q3. This gives an emitter
current of IE = (VZ − VBE)/RE which will vary only slightly
for changes in input voltage and temperature.
−
−
Q1
RE
Z2
Z1
VI
RL
RB2
VO
Q3
CONSTANT
CURRENT
SOURCE
Q2
RB1
+
+
Figure 14. Constant Current Source Incorporated in a Basic Regulator Circuit
http://onsemi.com
41
IMPEDANCE CANCELLATION
Equation 40 merely states that the change in reference
voltage with input tends to zero when the zener impedance
tends also to zero, as expected.
The figure of merit equation can be applied to the circuits
of Figure 16 and 17 to explain impedance cancellation. The
Change Factor equations for each leg and the reference
voltage VR are:
One of the most common applications of zener diodes is
in the general category of reference voltage supplies. The
function of the zener diode in such applications is to provide
a stable reference voltage during input voltage variations.
This function is complicated by the zener diode impedance,
which effectively causes an incremental change in zener
breakdown voltage with changing zener current.
RZ
ΔVZ
=
= RA
ΔVI
R1 + RZ
(41)
CFV2 =
ΔV2
R3
=
= RB
ΔVI
R2 + R3
(42)
ΔVR
RZ
R3
=
=
= RA − RB
ΔVI
R1 + RZ
R2 + R3
(43)
R2
R1
+
CFVZ =
VI
Z1
TC1
−
CFVR =
Figure 15. Impedance Cancellation with An
Uncompensated Zener
It is possible, however, by employing a bridge type circuit
which includes the zener diode and current regulating
resistance in its branch legs, to effectively cancel the effect
of the zener impedance. Consider the circuit of Figure 15 as
an example. This is the common configuration for a zener
diode voltage regulating system. The zener impedance at 20
mA of a 1N4740 diode is typically 2 ohms. If the supply
voltage now changes from 30 V to 40 V, the diode current
determined by R1 changes from 20 to 30 mA; the average
zener impedance becomes 1.9 ohms; and the reference
voltage shifts by 19 mV. This represents a reference change
of .19%, an amount far too large for an input change of 30%
in most reference supplies.
The effect of zener impedance change with current is
relatively small for most input changes and will be neglected
for this analysis. Assuming constant zener impedance, the
zener voltage is approximated by
V′Z = VZ + Z(I′Z − IZ)
where
Then
I
R1
VI′ ΔVI
VZ
Figure 16. Standard Voltage
Regulation Circuit
R2
R1
(37)
VR
VI′ ΔVI
V′Z is the new zener voltage
VZ is the former zener voltage
I′Z is the new zener current
IZ is the new zener current flowing at VZ
RZ is the zener impedance
V2
VZ
R3
ΔVZ = ZΔIZ
Let the input voltage VI in Figure 15 increase by an
amount ΔVI
Then ΔI =
Also ΔI =
ΔVI − ΔVZ
R1
ΔVZ
RZ
Figure 17. Impedance Cancellation Bridge
(38)
(39)
Solving ΔVIRZ − ΔVZRZ − ΔVZR1 = 0
Or
ΔVZ
RZ
=
ΔVI
R1 + RZ
(40)
http://onsemi.com
42
It is seen that γVR can be minimized by setting RB = RA.
Note that it is not necessary to match R3 to RZ and R2 to
R1. Thus R3 and R2 can be large and hence dissipate low
power. This discussion is assuming very light load currents.
Since the design is to minimize CFVR, RB can be set equal
to RA. The Input Regulation Factors are:
γVZ =
ΔVZ
ΔVI
γV2 =
γVR =
ΔVR
ΔVI
ǒ Ǔ
VII
Vi
VRZ
ǒ Ǔ
VII
Vi
VZZ
ΔV2
ΔVI
=
1
VZ
1+
VI
ǒ Ǔ
VII
Vi
V2Z
ǒ Ǔ
R
V1I
Vi
RZZ
(45)
=1
1
=
1+
ǒ Ǔǒ Ǔ ǒ
VVZI
Vi
VIZ
RV1I
Vi Z
RZ
(44)
1V I
Vi RBZ
1−
RA
(46)
Ǔ
http://onsemi.com
43
ZENER PROTECTIVE CIRCUITS AND TECHNIQUES:
BASIC DESIGN CONSIDERATIONS
INTRODUCTION
of the input voltage transient and the total circuit impedance
minus the load current. The worst case occurs when load
current is zero and may be expressed as follows:
The reliability of any system is a function of the ability of
the equipment to operate satisfactorily during moderate
changes of environment, and to protect itself during
otherwise damaging catastrophic changes. The silicon zener
diode offers a convenient, simple but effective means of
achieving this result. Its precise voltage sensitive
breakdown characteristic provides an accurate limiting
element in the protective circuit. The extremely high
switching speed possible with the zener phenomenon allows
the circuit to react faster by orders of magnitude that
comparable mechanical and magnetic systems.
By shunting a component, circuit, or system with a zener
diode, the applied voltage cannot exceed that of the
particular device’s breakdown voltage. (See Figure 1.)
A device should be chosen so that its zener voltage is
somewhat higher than the nominal operating voltage but
lower than the value of voltage that would be damaging if
allowed to pass. In order to adequately incorporate the zener
diode for circuit protection, the designer must consider
several factors in addition to the required zener voltage. The
first thing the designer should know is just what transient
characteristics can be anticipated, such as magnitude,
duration, and the rate of reoccurrence. For short duration
transients, it is usually possible to suppress the voltage spike
and allow the zener to shunt the transient current away from
the load without a circuit shutdown. On the other hand, if the
over-voltage condition is for a long duration, the protective
circuit may need to be complimented with a disconnect
element to protect the zener from damage created by
excessive heating. In all cases, the end circuit will have to be
designed around the junction temperature limits of the
device.
The following sections illustrate the most common zener
protective circuits, and will demonstrate the criteria to be
followed for an adequate design.
IZ(max) =
VI(max) − VZ
RS
(1)
RS
+
POWER
SUPPLY
Z
LOAD
−
Figure 1. Basic Shunt Zener Transient
Protection Circuit
The maximum power dissipated by the zener is
PZ(max) = IZ(max) VZ(max)=
VI(max) − VZ
VZ(max)
RS
(2)
Also, more than one device can be used, i.e., a series
string, which will reduce the percentage of total power to be
dissipated per device by a factor equal to the number of
devices in series. The number of diodes required can be
found from the following expression:
Number =
PZ(max)
PZ (allowable per device)
(3)
Any fraction of a zener must be taken as the next highest
whole number. This design discussion has been based upon
the assumption that the transient is of a single shot,
non-recurrent type. For all practical purposes it can be
considered non-recurrent if the “off period” between
transients is at least four times the thermal time constant of
the device. If the “off period” is shorter than this, then the
design calculations must include a factor for the duty cycle.
This is discussed in detail in Chapter 4. In Chapter 4 there are
also some typical curves relating peak power, pulse duration
and duty cycle that may be appropriate for some designs.
Obviously, the factor that limits the feasibility of the basic
zener shunt protective circuit is the pulse durations “t”. As
the duration increases, the allowable peak power for a given
configuration decreases and will approach a steady state
condition.
BASIC PROTECTIVE CIRCUITS
FOR SUPPLY TRANSIENTS
The simple zener shunt protection circuit shown in
Figure 1 is widely used for supply voltage transient
protection where the duration is relatively short. The circuit
applies whether the load is an individual component or a
complete circuit requiring protection. Whenever the input
exceeds the zener voltage, the device avalanches into
conduction clamping the load voltage to VZ. The total
current the diode must carry is determined by the magnitude
http://onsemi.com
44
LOAD
RS
FUSE
+
POWER
SUPPLY
VS
R
C
−
Figure 2. Overvoltage Protection with Zener Diodes and Fuses
would be reached before the fuse would begin to protect the
load.
When the anticipated transients expected to prevail for a
specific situation are of long duration, a basic zener shunt
becomes impractical, in such a case the circuit can be
improved by using a complementary disconnect element.
The most common overload protective element is without a
doubt the standard fuse. The common fuse adequately
protects circuit components from over-voltage surges, but at
the same time must be chosen to eliminate “nuisance fusing”
which results when the maximum current rating of the fuse
is too close to the normal operational current of the circuit.
220
B"
FUSE CURRENT (mA)
200
AN EXAMPLE PROBLEM: SELECTING A
FUSE-ZENER COMBINATION
180
160
ZENER DIODE WITH,
RESISTIVE LOAD
140
RESISTIVE
LOAD ONLY
120
A"
100
ZENER BREAKDOWN,
VOLTAGE
80
Consider the case illustrated in Figure 2. Here the load
components are represented by a parallel combination of R
and C, equivalent to many loads found in practice. The
maximum capacitor voltage rating is usually the
circuit-voltage limiting factor due to the cost of high voltage
capacitors. Consequently, a protective circuit must be
designed to prevent voltage surges greater than 1.5 times
normal working voltage of the capacitor. It is common,
however, for the supply voltage to increase to 135% normal
for long periods. Examination of fuse manufacturers’
melting time-current curves shows the difficulty of trying to
select a fuse which will melt rapidly at overload (within one
or two cycles of the supply frequency to prevent capacitor
damage), and will not melt when subjected to voltages close
to overload for prolonged periods.
By connecting a zener diode of correct voltage ratings
across the load as shown, a fuse large enough to withstand
normal current increases for long periods may be chosen.
The sudden current increase when zener breakdown occurs
melts the fuse rapidly and protects the load from large
surges. In Figure 3, fuse current was plotted against supply
voltage to illustrate the improvement in load protection
obtained with zener-fuse combinations. Fuse current “A”
would be selected to limit current resulting from voltage
surges above 112 V to 90 mA, which would melt the fuse in
100 ms. It is a simple matter, however, to select a fuse which
melts in 30 ms at 200 mA but is unaffected by 100 mA
currents. The zener connection allows fuse current “B” to be
selected, eliminating this design problem and providing a
faster, more reliable protective circuit. If the same fuse was
used without the zener diode, a supply voltage of 210 volts
60
60
70
NORMAL LOAD VOLTAGE
80
90
100
110
120
130
140
VS, SUPPLY VOLTAGE (VOLTS)
Figure 3. Fuse Current versus
Supply Voltage
Selection of the correct power rating of zener diodes to be
used for surge protection depends upon the magnitude and
duration of anticipated surges. Often in circuits employing
both fuses and zener diodes, the limiting surge duration will
be the melting time of the fuse. This, in turn, depends on the
nature of the load protected and the length of time it will
tolerate an overload.
As a first solution to the example problem, consider a
zener diode with a nominal breakdown voltage of 110 volts
measured at a test current (IZT) of 110 mA. Since the fuse
requires about 200 mA to melt and 100 mA are drawn
through the load at this voltage, the load voltage will never
exceed the zener breakdown voltage on slowly rising inputs.
Transients producing currents of approximately 200 mA but
of shorter duration than 30 ms will simply be clipped by
zener action and diverted from the load. Transients of very
high voltage will produce larger currents and, hence, will
melt the fuse more rapidly. In the limiting case where
transient power might eventually destroy the zener diode,
the fuse always melts first because of the slower thermal
time constant inherent in the zener diode’s larger geometry.
The curves in Figure 4 illustrate the change in zener
voltage as a function of changing current for a typical device
type.
http://onsemi.com
45
CHANGE, IN VZ
illustrated in Figure 5. If a short circuit occurs in the
capacitive section of the load (represented by C) the
resulting fault current is limited by the resistive section
(represented by R) to a value which may not be great enough
to melt the fuse. The fault current could be sufficient,
however, to damage the supply and other components in the
load.
The problem is resolved by employing a zener diode to
protect against supply surges as described in the previous
section, and by selecting a separate fuse to protect from load
faults. The load fuse in Figure 5 is chosen close to the normal
operating current. Abnormal supply surges do not affect it
and equipment operates reliably but with ample protection
for the supply against load changes.
VZ = V1
V2
V3
LOG IZ/IZK
Figure 4. Change in VZ for Changes in IZ
ZENER DIODES AND RECLOSING
DISCONNECT ELEMENTS
If an actual curve for the device being used is not
available, the zener voltage at a specific current above or
below the test current may be approximated by equation 4.
An interesting application of zener diodes as overvoltage
protectors, which offers the possibility of designing for both
long and short duration surges, is shown in Figure 6.
(4)
Where: V = VZ + ZZT (I−IZT)
VZ = zener voltage at test current IZT
ZZT = zener impedance at test current IZT
IZT = test current
V = zener voltage at current I
R
For a given design, the maximum zener voltage to expect
for the higher zener current should be determined to make
sure the limits of the circuit are met. If the maximum limit
is excessive for the original device selection, the next lower
voltage rating should be used.
The previous discussion on design consideration for
protective circuits incorporating fuses is applicable to any
protective element that permanently disconnects the supply
when actuated. Rather than a fuse, a non-resetting magnetic
circuit breaker could have been used, and the same
reasoning would have applied.
Figure 6. Zener Diode Reclosing Circuit Breaker
Protective Circuit
In the event of a voltage overload exceeding a chosen
zener voltage, a large current will be drawn through the
diode. The reclosing disconnect element opens after an
interval determined by its time constant, and the supply is
disconnected. After another interval, again depending on the
switch characteristics, the supply is reconnected and the
voltage “sampled” by the zener diode. This leads to an
“on-off” action which continues until the supply voltage
drops below the predetermined limit. At no time can the load
voltage or current exceed that set by the zener. The chief
advantage in this type of circuit is the elimination of fuse
replacement in similar fusing circuits, while providing
essentially the same load protection.
In many actual problems the designer must choose a
protective circuit to perform still another task. Not only must
the equipment be protected from the voltage surges in the
supply, but the supply itself often requires protection from
shorts or partial shorts in the load. A direct short in the load
is fairly easy to handle, as the drastic current change permits
the use of fuses with ratings high enough to avoid problems
with supply surges. More common is the partial short, as
SUPPLY
FUSE
LOAD
IZ
POWER
SUPPLY
LOAD CURRENT SURGES
R
RECLOSING
CIRCUIT
BREAKER
LOAD
FUSE
R
POWER
SUPPLY
C
LOAD
Figure 5. Supply and Load with Zener Diode; Fuse Circuitry
http://onsemi.com
46
VOLTS
SURGE VOLTAGE
SUPPLY
VOLTAGE
OVER VOLTAGE
NORMAL OPERATING VOLTAGE
TIME
°C
BREAK TEMPERATURE
MAKE
TEMPERATURE
THERMAL
BREAKER
TEMPERATURE
TIME
AMPS
ZENER
DIODE
CURRENT
TIME
°C
MAXIMUM TJ
ZENER
DIODE
JUNCTION
TEMPERATURE
TIME
Figure 7. (Typical) Voltage, Current and Temperature Waveforms
for a Thermal Breaker
causes the thermal breaker to open sooner. In effect, the
zener diode rating must be high enough to ensure that
maximum junction temperature is not reached during the
longest interval that the thermal switch will be closed.
Manufacturers of thermally operated circuit breakers
publish current-time curves for their devices similar to that
shown in Figure 8. By estimating the peak supply
overvoltage and determining the maximum overvoltage
tolerated by the load, an estimation of peak zener current can
be made. The maximum breaker trip time may then be read
from Figure 8. (After the initial current surge, the duration
of “of” time is determined entirely by the breaker
characteristics and will vary widely with manufacture.) The
zener diode junction temperature rise during conduction
may be calculated now from the thermal time constant of the
device and the heatsink used.
Because the reclosing circuit breaker is continually
cycling on and off, the zener current takes on the
characteristics of a repetitive surge, as can be seen in
Figure 7.
It is difficult to define a set design procedure in this case,
because of the wide variety of reclosing, magnetic and
thermal circuit breakers available. Care should be taken to
ensure that the power dissipated in the zener diode during the
conduction time of the disconnect element does not exceed
its rating. As an example, assume the disconnect element
was a thermal breaker switch. The waveforms for a typical
over-voltage situation are shown in Figure 7.
It is apparent that the highest zener diode junction
temperature is reached during the first conduction period. At
this time the thermal breaker is cold and requires the greatest
time to reach its break temperature. The breaker then cycles
thermally between the make and break temperatures as long
as the supply voltage is greater than the zener voltage, as
shown in Figure 7.
The zener diode current and junction temperature
variation are shown in the last two waveforms of Figure 7.
Overvoltage durations longer than the trip time of the
thermal breaker do not affect the diode as the supply is
disconnected. An overvoltage of much higher level simply
http://onsemi.com
47
of using a zener diode to protect an ignition transistor. These
are shown in Figures 9a and 9b. In Figure 9b the transistor
is protected by a zener diode connected between base and
collector and in Figure 9a, the zener is connected between
emitter and collector. In both cases the voltage level of the
zener must be selected carefully so that the voltage stress on
the transistor is in a region where the safe operating area is
adequate for reliable circuit operation.
Figure 10 illustrates “safe” and “unsafe” selection of a
zener diode for collector-base protection of a transistor in the
ignition coil circuit. It can be seen that the safe operating area
of a transistor must be known if an adequate protective zener
is to be selected.
The zener diode must be able to take the stress of peak
pulse current necessary to clamp the voltage rise across the
transistor to a safe value. In a typical case, a 5 watt, 100 volt
zener transient suppressor diode is required to operate with
an 80 μs peak pulse current of 8 amperes when connected
between the collector-emitter of the transistor. The
waveform of this pulse current approaches a sine wave in
shape (Figure 11). The voltage rise across a typical transient
suppressor diode due to this current pulse is shown in Figure
12. This voltage rise of approximately 8 volts indicates an
effective zener impedance of approximately 1 ohm.
However, a good share of this voltage rise is due to the
temperature coefficient and thermal time constant of the
zener. The temperature rise of the zener diode junction is
indicated by the voltage difference between the rise and fall
of the current pulse.
TRIP TIME (SECONDS)
30
20
10
0
0
1
2
3
CURRENT (AMPS)
Figure 8. Trip Time versus Current for
Thermal Breaker
TRANSISTOR OVERVOLTAGE PROTECTION
In many electronic circuits employing transistors, high
internal voltages can be developed and, if applied to the
transistors, will destroy them. This situation is quite
common in transistor circuits that are switching highly
inductive loads. A prime example of this would be in
transistorized electronic ignition systems such as shown in
Figures 9a and 9b.
The zener diode is an important component to assure solid
state ignition system reliability. There are two basic methods
+12 V
10 Ω
1/2 W
1Ω
10 Ω
2N6031
+12 V
2N5879
1N6295
560 pF
1N5374B
1Ω
100 W
10 Ω
10 W
200 V
PAPER
5Ω
H.V. TO
DIST.
H.V. TO
DIST.
MALLORY
COIL
28100
2 μF
PRESTO-LITE
201
(A)
(B)
Figure 9. Transistor Ignition Systems with Zener
Overvoltage Surge Protection
http://onsemi.com
48
TYPICAL TRANSISTOR SAFE AREA LIMIT
10
9
2 A/div ZENER CURRENT
8
7
6
IC 5
LOAD
LINE
4
SAFE
3
UNSAFE
2
SAFE
1
0
0
10
20
30
40
50
60
VCE
70
80
90
100 110 120
COLLECTOR-BASE
ZENER CLAMP
TIME 10 μs/div
Figure 11. Zener Diode Current Pulse
1 A/div ZENER CURRENT
Figure 10. Safe Zener Protection
COLLECTOREMITTER
ZENER
CLAMP
0
ZENER VOLTAGE 1 V/div
100 V
Figure 12. Voltage-Current Representation on
100 V Zener
result in additional zener voltage rise. It is not uncommon to
observe a 15-volt rise above the zener device voltage rating
due to temperature coefficient and impedance under these
pulse current conditions.
The zener diode should be connected as close as possible
to the terminals of the transistor the zener is intended to
protect. This insures that induced voltage transients, caused
by current changes in long lead lengths, are clamped by the
zener and do not appear across the transistor.
In order to assure safe operation, the change in zener
junction temperature for the peak pulse conditions must be
analyzed. In making the calculation, the method described
in Chapter 3 should be used, taking into account duty cycle,
pulse duration, and pulse magnitude.
When the zener diode is connected between the collector
and emitter of the transistor, additional power dissipation
will result from the clipping of the ringing voltage of the
ignition coil by the forward conduction of the zener diode.
This power dissipation by the forward diode current will
http://onsemi.com
49
+
Figure 13. DC-DC Converter with Surge Protecting Diodes
voltmeter to measure 25 V, approximately 249 thousand
ohms are required in series.
Another example of overvoltage protection of transistor
operating in an inductive load switch capacity is illustrated
in Figure 13. The DC-DC converter circuit shows a
connection from collector to emitter of two zener diodes as
collector overvoltage protectors. Without some type of
limiting device, large voltage spikes may appear at the
collectors, due to the switching transients produced with
normal circuit operation. When this spike exceeds the
collector breakdown rating of the transistor, transistor life is
considerably shortened. The zener diodes shown are chosen
with zener breakdowns slightly below transistor breakdown
voltage to provide the necessary clipping action. Since the
spikes are normally of short duration (0.5 to 5 μs) and duty
cycle is low, normal chassis mounting provides adequate
heatsinking.
+
70K
179K
1N4746
25 V
(18 VOLT
ZENER DIODE)
−
μA
Figure 14. Meter Protection with Zener Diode
The protection provided by the addition of an 18 volt zener
is illustrated in Figure 15. With an applied voltage of
25 volts, the 100 μAmps current in the circuit produces a
drop of 17.9 volts across the series resistance of 179
thousand ohms. A further increase in voltage causes the
zener diode to conduct, and the overload current is shunted
away from the meter. Since ON Semiconductor zener diodes
have zener voltages specified within 5 and 10%, a safe
design may always be made with little sacrifice in meter
linearity by assuming the lowest breakdown voltage within
the tolerance. The shunting effect on the meter of the reverse
biased diode is generally negligible below breakdown
voltage (on the order of 0.5° full scale). For very precise
work, the zener diode breakdown voltage must be accurately
known and the design equations solved for the correct
resistance values.
METER PROTECTION
The silicon zener diode can be employed to prevent
overloading sensitive meter movements used in low range
DC and AC voltmeters, without adversely affecting the
meter linearity. The zener diode has the advantage over
thermal protective devices of instantaneous action and, of
course, will function repeatedly for an indefinite time (as
compared to the reset time necessary with thermal devices).
While zener protection is presently available for voltages as
low as 2.4 volts, forward diode operation can be used for
meter protection where the voltage drop is much smaller. A
typical protective circuit is illustrated in Figure 14. Here the
meter movement requires 100 μAmps for full scale
deflection and has 940 ohms resistance. For use in a
http://onsemi.com
50
In a system that is handling large amounts of power, it may
become impractical to employ standard zener shunt
protection because of the large current it would be required
to carry. The SCR crowbar technique shown in Figure 16 can
be effectively used in these situations. The zener diode is still
the transient detection component, but it is only required to
carry the gate current for SCR turn on, and the SCR will
carry the bulk of the shunt current. Whenever the incoming
voltage exceeds the zener voltage, it avalanches, supplying
gate drive to the SCR which, when fired, causes a current
demand that will trip the circuit breaker. The resistors shown
are for current limiting so that the SCR and zener ratings are
not exceeded.
The circuit of Figure 17 is designed to disconnect the
supply in the event a specified load current is exceeded. This
is done by means of a series sense resistor and a compatible
zener to turn the shunt SCR on. When the voltage across the
series resistor, which is a function of the load current,
becomes sufficient to break over the zener, the SCR is fired,
causing the circuit breaker to trip.
METER CURRENT (μA)
200
150
100
WITH PROTECTIVE CIRCUIT
50
0
0
50
100
150
200
TOTAL CURRENT (μA)
Figure 15. Meter Protection with
Zener Diodes
ZENER DIODES USED WITH SCRS FOR
CIRCUIT PROTECTION
An interesting aspect of circuit protection incorporating
the reliable zener diode is the protective circuits shown in
Figures 16 and 17.
CIRCUIT
BREAKER
R6
R1
ZENER
AC
R2
SCR
SCR
R4
R3
R5
ZENER
Figure 16. SCR Crowbar Over-Voltage Protection Circuit for AC Circuit Operation
CIRCUIT
BREAKER
Figure 17. SCR Longterm Current Overload Protection
http://onsemi.com
51
ZENER TRANSIENT SUPPRESSORS
3. Large dice are used, or special tests are imposed on the
product to guarantee higher ratings than those shown
in Figure 18.
4. The specifications may be based on a JEDEC
registration or part number of another manufacturer.
The transient suppressor is used as a shunt element in
exactly the same manner as a conventional zener. It offers
the same advantages such as low insertion loss, immediate
recovery after operation, a clamping factor approaching
unity, protection against fast rising transients, and simple
circuitry. The primary difference is that the transient
suppressor extends these advantages to higher power levels.
Even in the event of transients with power contents far in
excess of the capacity of the zeners, protection is still
provided the load. When overloaded to failure, the zener will
approximate a short. The resulting heavy drain will aid in
opening the fuse or circuit breaker protecting the load
against excess current. Thus, even if the suppressor is
destroyed, it still protects the load.
The design of the suppressor-fuse combination for the
required level of protection follows the techniques for
conventional zeners discussed earlier in this chapter.
PPK(nom), NOMINAL PEAK POWER (kW)
The data of Figure 18 applies for non-repetitive conditions
and at a lead temperature of 25°C. If the duty cycle increases,
the peak power must be reduced as indicated by the curves
of Figure 19. Average power must be derated as the lead or
ambient temperature rises above 25°C. The average power
derating curve normally given on data sheets may be
normalized and used for this purpose.
TRANSIENT SUPPRESSION CHARACTERISTICS
Zener diodes, being nearly ideal clippers (that is, they
exhibit close to an infinite impedance below the clipping
level and close to a short circuit above the clipping level), are
often used to suppress transients. In this type of application,
it is important to know the power capability of the zener for
short pulse durations, since they are intolerant of excessive
stress.
Some ON Semiconductor data sheets such as the ones for
devices shown in Table 1 contain short pulse surge
capability. However, there are many data sheets that do not
contain this data and Figure 18 is presented here to
supplement this information.
Package
Description
1N4728A
1W
DO-41
Double Slug
Glass
1N6267A
5W
Case 41A
Axial Lead
Plastic
250 mW TO 1 W TYPES
0.02
GLASS DO-35 & GLASS DO-41
0.01
0.01 0.02
0.05 0.1 0.2
1
2
5
10
50
100
PULSE WIDTH
0.3
0.2
10 ms
0.1
0.07
0.05
1 ms
0.03
Case 102
Surmetic 40
0.02
1N746A/957B
/4370A
500 mW
DO-35
Double Slug
Glass
0.01
0.1
DO-35
0.5
1
0.7
0.5
5W
500 mW
5 WATT TYPES
0.2
0.1
0.05
1N5333B
1N5221B
1N6267 SERIES
2
1
0.5
Figure 18. Peak Power Ratings of
Zener Diodes
DERATING FACTOR
Steady State
Power
1 TO 3 W TYPES
PLASTIC DO-41
20
10
5
PULSE WIDTH (ms)
Table 1. Transient Suppressor Diodes
Series
Numbers
100
50
100 μs
10 μs
0.2
0.5
1
2
5
10
20
D, DUTY CYCLE (%)
Double Slug
Glass
Figure 19. Typical Derating Factor for
Duty Cycle
Some data sheets have surge information which differs
slightly from the data shown in Figure 18. A variety of
reasons exist for this:
When it is necessary to use a zener close to surge ratings,
and a standard part having guaranteed surge limits is not
suitable, a special part number may be created having a surge
limit as part of the specification. Contact your nearest
ON Semiconductor OEM sales office for capability, price,
delivery, and minimum order quantities.
1. The surge data may be presented in terms of actual
surge power instead of nominal power.
2. Product improvements have occurred since the data
sheet was published.
http://onsemi.com
52
MATHEMATICAL MODEL
with the zener is also effective in reducing the stress imposed
by very short duration transients.
To illustrate use of the data, a common application will be
analyzed. The transistor in Figure 20 drives a 50 mH
solenoid which requires 5 amperes of current. Without some
means of clamping the voltage from the inductor when the
transistor turns off, it could be destroyed.
The means most often used to solve the problem is to
connect an ordinary rectifier diode across the coil; however,
this technique may keep the current circulating through the
coil for too long a time. Faster switching is achieved by
allowing the voltage to rise to a level above the supply before
being clamped. The voltage rating of the transistor is 60 V,
indicating that approximately a 50 volt zener will be
required.
The peak current will equal the on-state transistor current
(5 amperes) and will decay exponentially as determined by
the coil L/R time constant (neglecting the zener impedance).
A rectangular pulse of width L/R (0.01 s) and amplitude of
IPK (5 A) contains the same energy and may be used to select
a zener diode. The nominal zener power rating therefore
must exceed (5 A × 50) = 250 watts at 10 ms and a duty cycle
of 0.01/2 = 0.5%. From Figure 19, the duty cycle factor is
0.62 making the single pulse power rating required equal to
250/0.62 = 403 watts. From Figure 18, one of the 1N6267
series zeners has the required capability. The 1N6287 is
specified nominally at 47 volts and should prove
satisfactory.
Although this series has specified maximum voltage
limits, equation 7 will be used to determine the maximum
zener voltage in order to demonstrate its use.
Since the power shown on the curves is not the actual
transient power measured, but is the product of the peak
current measured and the nominal zener voltage measured
at the current used for voltage classification, the peak current
can be calculated from:
IZ(PK) =
P(PK)
VZ(nom)
(5)
The peak voltage at peak current can be calculated from:
VZ(PK) = FC x VZ(nom)
(6)
where FC is the clamping factor. The clamping factor is
approximately 1.20 for all zener diodes when operated at
their pulse power limits. For example, a 5 watt, 20 volt
zener can be expected to show a peak voltage of 24 volts
regardless of whether it is handling 450 watts for 0.1 ms
or 50 watts for 10 ms. This occurs because the voltage is
a function of junction temperature and IR drop. Heating
of the junction is more severe at the longer pulse width,
causing a higher voltage component due to temperature
which is roughly offset by the smaller IR voltage
component.
For modeling purposes, an approximation of the zener
resistance is needed. It is obtained from:
RZ(nom) =
VZ(nom)(FC−1)
PPK(nom) / VZ(nom)
(7)
The value is approximate because both the clamping
factor and the actual resistance are a function of temperature.
RZ =
CIRCUIT CONSIDERATIONS
47(1.20 − 1)
9.4
=
= 0.9 Ω
500/47
10.64
At 5 amperes, the peak voltage will be 4.5 volts above
nominal or 51.5 volts total which is safely below the 60
volt transistor rating.
It is important that as much impedance as circuit
constraints allow be placed in series with the zener diode and
the components to be protected. The result will be a lower
clipping voltage and less zener stress. A capacitor in parallel
26 Vdc
50 mH, 5 Ω
10 ms
2s
USED TO SELECT A ZENER DIODE HAVING THE PROPER
VOLTAGE AND POWER CAPABILITY TO PROTECT THE TRANSISTOR
Figure 20. Circuit Example
http://onsemi.com
53
ZENER VOLTAGE SENSING CIRCUITS
AND APPLICATIONS
BASIC CONCEPTS OF VOLTAGE SENSING
TRANSISTOR-ZENER SENSING CIRCUITS
Numerous electronic circuits require a signal or voltage
level to be sensed for circuit actuation, function control, or
circuit protection. The circuit may alter its mode of
operation whenever an interdependent signal reaches a
particular magnitude (either higher or lower than a specified
value). These sensing functions may be accomplished by
incorporating a voltage dependent device in the system
creating a switching action that controls the overall
operation of the circuit.
The zener diode is ideally suited for most sensing
applications because of its voltage dependent
characteristics. The following sections are some of the more
common applications and techniques that utilize the zener in
a voltage sensing capacity.
The zener diode probably finds its greatest use in sensing
applications in conjunction with other semiconductor
devices. Two basic widely used techniques are illustrated in
Figures 1a and 1b.
In both of these circuits the output is a function of the input
voltage level. As the input goes from low to high, the output
will switch from either high to low (base sense circuit) or
low to high (emitter sense circuit), (see Figure 2).
The base sense circuit of Figure 1a operates as follows:
When the input voltage is low, the voltage dropped across R2
is not sufficient to bias the zener diode and base emitter
junction into conduction, therefore, the transistor will not
conduct. This causes a high voltage from collector to
emitter. When the input becomes high, the zener is biased
into conduction, the transistor turns on, and the collector to
emitter voltage, which is the output, drops to a low value.
VIN
SENSING LEVEL
R3
R1
VIN
VIN =
VIN BOTH CIRCUITS
Z1
Q1
TIME
VO1
VO1
R2
R2 + R1
VZ + VBE(SAT)
R2
(OUTPUT OF FIGURE 1a)
(a)
VOUT BASE SENSE
Z1
R2
TIME
VO2
VIN
(OUTPUT OF FIGURE 1b)
Q1
R1
R3
VO2
TIME
VOUT EMITTER SENSE
(b)
Figure 2. Outputs of Transistor-Zener Voltage
Sensing Circuits
Figure 1. Basic Transistor-Zener Diode
Sensing Circuits
http://onsemi.com
54
of these devices tend to change and have mechanical contact
problems. A solid state regulator that controls the charge rate
by sensing the battery voltage is inherently more accurate
and reliable. A schematic of a simplified solid state voltage
regulator is shown in Figure 3.
The purpose of an alternator regulator is to control the
battery charging current from the alternator. The charge
level of the battery is proportional to the battery voltage
level. Consequently, the regulator must monitor the battery
voltage level allowing charging current to pass when the
battery voltage is low. When the battery has attained the
proper charge the charging current is switched off. In the
case of the solid state regulator of Figure 3, the charging
current is controlled by switching the alternator field current
on and off with a series transistor switch, Q2. The switching
action of Q2 is controlled by a voltage sensing circuit that is
identical to the base sense circuit of Figure 1a. When
under-charged, the zener Z1 does not conduct keeping Q1
off. The collector-emitter voltage of Q1 supplies a forward
bias to the base-emitter of Q2, turning it on. With Q2 turned
on, the alternator field is energized allowing a charging
current to be delivered to the battery. When the battery
attains a proper charge level, the zener conducts causing Q1
to turn on, and effectively shorting out the base-emitter
junction of Q2. This short circuit cuts off Q2, turns off the
current flowing in the field coil which consequently, reduces
the output of the alternator. Diode D1 acts as a field
suppressor preventing the build up of a high induced voltage
across the coil when the coil current is interrupted.
The emitter sense circuit of Figure 1b operates as follows:
When the input is low the voltage drop across R3 (the output)
is negligible. As the input voltage increases the voltage drop
across R2 biases the zener into conduction and forward
biases the base-emitter junction. A large voltage drop across
R3 (the output voltage) is equal to the product of the collector
current times the resistance, R3. The following relationships
indicate the basic operating conditions for the circuits in
Figure 1.
Circuit
1a
1b
Output
High
VOUT = VIN − ICOR3 ≅ VIN
Low
VOUT = VIN − ICR3 = VCE(sat)
Low
VOUT = VIN − VZ − VCE(off) = ICOR3
High
VOUT = VIN − VCE(sat) = ICR3
In addition, the basic circuits of Figure 1 can be rearranged
to provide inverse output.
AUTOMOTIVE ALTERNATOR VOLTAGE
REGULATOR
Electromechanical devices have been employed for many
years as voltage regulators, however, the regulation setting
ALTERNATOR
OUTPUT
B+
R3
R1
Z1
Q1
R2
Q2
R4
D1
Figure 3. Simplified Solid State Voltage Regulator
http://onsemi.com
55
ALTERNATOR
FIELD
In this configuration, the sensing circuit is composed of Z1
and Q1 with biasing components. It is similar to the sensing
circuit shown in Figure 1b. The potentiometer R1 adjusts the
conduction point of Q1 establishing the proper charge level.
When the battery has reached the desired level, Q1 begins to
conduct. This draws Q2 into conduction, and therefore
shorts off Q3 which is supplying power to the alternator
field. This type of regulator offers greater sensitivity with an
increase in cost.
In actual operation, this switching action occurs many
times each second, depending upon the current drain from
the battery. The battery charge, therefore, remains
essentially constant and at the maximum value for optimum
operation.
A schematic of a complete alternator voltage regulator is
shown in Figure 4.
It is also possible to perform the alternator regulation
function with the sensing element in the emitter of the
control transistor as shown in Figure 5.
ALTERNATOR
OUTPUT
B+
0.05 Ω
30 Ω
100 Ω
2N4234
1N961B
SENSING
ZENER
DIODE
2N5879
15 Ω
RT*
THERMISTOR
30 Ω
1N3493
BIAS
DIODE
0.05 μF
FEEDBACK
CAPACITOR
70 Ω
TO ALTERNATOR
FIELD COIL
1N4001
FIELD
SUPPRESSION
DIODE
*THE VALUE OF RT DEPENDS ON THE SLOPE OF THE VOLTAGE REGULATION
VERSUS TEMPERATURE CURVE.
Figure 4. Complete Solid State Alternator Voltage Regulator
ALTERNATOR
OUTPUT
B+
R5
D1
R3
Q2
R4
Q3
R1
Q1
R6
R2
Z1
D2
Figure 5. Alternator Regulator with Emitter Sensor
http://onsemi.com
56
ALTERNATOR
FIELD
UNIJUNCTION-ZENER SENSE CIRCUITS
does not exceed the Zener voltage VZ, the circuit will
generate output pulses. At some given point, the required Vp
for triggering will exceed VZ. Since Vp is clamped at VZ, the
circuit will not oscillate. This, in essence, means the circuit
is GO as long as VIN is below a certain level, and NO GO
above the critical clamp point.
The circuit of Figure 7, is a clamped base UJT oscillator.
In this circuit VBB is clamped at a voltage VZ and the emitter
tied to a voltage dividing network by a diode D1. When the
input voltage is low, the voltage drop across R2 is less than
Vp. The forward biased diode holds the emitter below the
trigger level. As the input increases, the R2 voltage drop
approaches Vp. The diode D1 becomes reversed biased and,
the UJT triggers. This phenomenon establishes the
operating criterion that the circuit is NO GO at a low input
and GO at an input higher than the clamp voltage. Therefore,
the circuits in Figures 6 and 7 are both input voltage
sensitive, but have opposite input requirements for a GO
condition. To illustrate the usefulness of the clamped UJT
relaxation oscillators, the following two sections show them
being used in practical applications.
Unijunction transistor oscillator circuits can be made
GO-NO GO voltage sensitive by incorporating a zener diode
clamp. The UJT operates on the criterion: under proper
biasing conditions the emitter-base one junction will
breakover when the emitter voltage reaches a specific value
given by the equation:
Vp = ηVBB + VD
where:
Vp
η
VBB
VD
(1)
= peak point emitter voltage
= intrinsic stand-off ratio for the device
= interbase voltage, from base two to base one
= emitter to base one diode forward junction drop
Obviously, if we provide a voltage clamp in the circuit
such that the conditions of equation 1 are met only with
restriction on the input, the circuit becomes voltage
sensitive. There are two basic techniques used in clamping
UJT relaxation oscillators. They are shown in Figure 6 and
Figure 7.
The circuit in Figure 6 is that of a clamped emitter type.
As long as the input voltage VIN is low enough so that Vp
+
RB2
RT
VP = ηVBB + VD
VIN
VBB
C
VE
Z
UJT
RB1
VOUT
−
Figure 6. UJT Oscillator, GO — NO GO Output,
GO for Low VIN — NO GO for High VIN
+
RT
R1
VIN
RB2
D1
UJT
R2
CT
VE
RB1
Z
−
Figure 7. UJT — NO GO Output, NO GO for Low VIN — GO for High VIN
http://onsemi.com
57
VOUT
BATTERY VOLTAGE SENSITIVE SCR CHARGER
While operating, the oscillator will produce pulses in the
pulse transformer connected across the resistance, RGC
(RGC represents the gate-to-cathode resistance of the
controlled rectifier), at a frequency determined by the
resistance, capacitance, R.C. time delay circuit.
Since the base-to-base voltage on the unijunction
transistor is derived from the charging battery, it will
increase as the battery charges. The increase in base-to-base
voltage of the unijunction transistor causes its peak point
voltage (switching voltage) to increase. These waveforms
are sketched in Figure 9 (this voltage increase will tend to
change the pulse repetition rate, but this is not important).
A clamped emitter unijunction sensing circuit of the type
shown in Figure 6 makes a very good battery charger
(illustrated in Figure 8). This circuit will not operate until the
battery to be charged is properly connected to the charger.
The battery voltage controls the charger and will dictate its
operation. When the battery is properly charged, the charger
will cease operation.
The battery charging current is obtained through the
controlled rectifier. Triggering pulses for the controlled
rectifier are generated by unijunction transistor relaxation
oscillator (Figure 9). This oscillator is activated when the
battery voltage is low.
A
C
RECTIFIED A.C.
VOLTAGE FROM
CHARGER
SCR
G
+
A
R1
E
B2
R2
V
B1
12 V
UJT
C1
Z1
T1
R3
−
R1 3.9K, 1/2 W
R2 1K, POT.
R3 5.1K, 1/2 W
C1 .25 μf
Z1 1N753, 6.2 V
SCR MCR3813
UJT 2N2646
T1 PR1, 30T, no. 22
SEC, 45T, NO. 22
CORE: FERROX CUBE
203F181-303
Figure 8. 12 Volt Battery Charger Control
VB
B2
1
BATTERY
CHARGING
BATTERY
CHARGED
TIME
VC
1
ZENER
VOLTAGE
R1
UJT
UJT PEAK
POINT VOLTAGE
B2
B1
R2
Z1
C1
T1
RGC
TIME
VR
GC
+
SCR
CONDUCTS
SCR
NONCONDUCTING
TIME
Figure 9. UJT Relaxation Oscillator Operation
http://onsemi.com
58
VBATT.
Figure 10 shows a permanent magnet alternator regulator
designed to regulate a 15 ampere output. The two SCRs are
connected on the ac side of the bridge, and short out the
alternator when triggered by the unijunction voltage
sensitive triggering circuit. The sensing circuit is of the type
shown in Figure 7. The shorted output does not appreciably
increase the maximum output current level.
A single SCR could be designed into the dc side of the
bridge. However, the rapid turn-off requirement of this type
of circuit at high alternator speeds makes this circuit
impractical.
The unijunction circuit in Figure 10 will not oscillate until
the input voltage level reaches the voltage determined by the
intrinsic standoff ratio. The adjustable voltage divider will
calibrate the circuit. The series diode in the voltage divider
circuit will compensate for the emitter-base-one diode
temperature
change,
consequently,
temperature
compensation is necessary only for the zener diode
temperature changes.
Due to the delay in charging the unijunction capacitor,
when the battery is disconnected the alternator voltage will
produce high stress voltage on all components before the
SCRs will be fired. The 1N971B Zener was included in the
circuit to provide a trigger pulse to the SCRs as soon as the
alternator output voltage level approaches 30 volts.
When the peak point voltage (switching voltage) of the
unijunction transistor exceeds the breakdown voltage of the
Zener diode, Z1, connected across the delay circuit
capacitor, C1, the unijunction transistor ceases to oscillate.
If the relaxation oscillator does not operate, the controlled
rectifier will not receive trigger pulses and will not conduct.
This indicates that the battery has attained its desired charge
as set by R2.
The unijunction cannot oscillate unless a voltage
somewhere between 3 volts and the cutoff setting is present
at the output terminals with polarity as indicated. Therefore,
the SCR cannot conduct under conditions of a short circuit,
an open circuit, or a reverse polarity connection to the
battery.
ALTERNATOR REGULATOR FOR
PERMANENT MAGNET FIELD
SEC. 1
In alternator circuits such as those of an outboard engine,
the field may be composed of a permanent magnet. This
increases the problem of regulating the output by limiting
the control function to opening or shorting the output.
Because of the high reactance source of most alternators,
opening the output circuit will generally stress the bridge
rectifiers to a very high voltage level. It is, therefore,
apparent that the best control function would be shorting the
output of the alternator for regulation of the charge to the
battery.
200 Ω
5K Ω
27 Ω
1N971B
+
MCR
2304-2
1N4001
MDA2500
T1
PRI.
ALT.
OUT
SEC. 2
MCR
2304-2
1N960B
200 Ω
2N2646
0.1 μF
27 Ω
T1
CORE: ARNOLD no. 4T5340 D1 DD1
PRIMARY 125 TURNS AWG 36
SEC no. 1 125 TURNS AWG 36
SEC no. 2 125 TURNS AWG 36
TRIFILAR WOUND
+
BATTERY
Figure 10. Permanent Magnet Field Alternator Regulator
http://onsemi.com
59
+
Z
VIN
VOUT
R
−
VIN
BASE CIRCUIT
VOUT
VIN
VOUT
VZ
+
+
(LEVEL DETECTION)
(MAGNITUDE REDUCTION)
TYPICAL OUTPUTS
Figure 11. Zener-Resistor Voltage Sensitive Circuit
VZ = 20 V
+
Z
+
VIN = 24 V − 28 V
VOLTMETER
10 V FULL SCALE
R
−
−
Figure 12. Improving Meter Resolution
ZENER-RESISTOR VOLTAGE SENSING
which the meter can be read. By employing a 20 volt zener,
one can use a 10 voltmeter instead of the 30 volt unit, thereby
utilizing 40% of the meter movement instead of 13.3% with
a corresponding increase in accuracy and readability. For
ultimate accuracy a 24 volt zener could be combined with a
5 voltmeter. This combination would have the disadvantage
of providing little room for voltage fluctuations, however.
In Figure 13, a number of sequentially higher-voltage
Zener sense circuits are cascaded to actuate transistor
switches. As each goes into avalanche its respective
switching transistor is turned on, actuating the indicator
light for that particular voltage level. This technique can be
expanded and modified to use the zener sensors to actuate
some form of logic system.
A simple but useful sense circuit can be made from just a
Zener diode and resistor such as shown in Figure 11.
Whenever the applied signal exceeds the specific Zener
voltage VZ, the difference appears across the dropping
resistor R. This level dependent differential voltage can be
used for level detection, magnitude reduction, wave
shaping, etc. An illustrative application of the simple series
Zener sensor is shown in Figure 12, where the resistor drop
is monitored with a voltmeter.
If, for example, the input is variable from 24 to 28 volts,
a 30 voltmeter would normally be required. Unfortunately,
a 4 volt range of values on a 30 volt scale utilizes only 13.3%
of the meter movement — greatly limiting the accuracy with
Z1
LIGHT
(1)
Z2
Z3
LIGHT
(2)
LIGHT
(3)
INPUT
OUTPUT
Q1
R1
RE1
Q2
R2
RE2
Q3
R3
Figure 13. Sequential Voltage Level Indicator
http://onsemi.com
60
RE3
MISCELLANEOUS APPLICATIONS
OF ZENER TYPE DEVICES
INTRODUCTION
diode, the voltage across NB windings of the timing
transformer T1 is clamped to VBE of the ON device, giving
an inverter frequency of
Many of the commonly used applications of zener diodes
have been illustrated in some depth in the preceding
chapters. This chapter shows how a zener diode may be used
in some rarer applications such as voltage translators, to
provide constant current, wave shaping, frequency control
and synchronized SCR triggers.
The circuits used in this chapter are not intended as
finished designs since only a few component values are
given. The intent is to show some general broad ideas and
not specific designs aimed at a narrow use.
f=
where BS1A1 is the flux capacity of T1 transformer core.
The effect on output frequency of VBE variations due to
changing load or temperature can be reduced by using a
zener diode in series with VBE as shown in Figure 1. For
this circuit, the output frequency is given by
FREQUENCY REGULATION OF A
DC TO AC INVERTER
f=
A
NC
B1
Q1
N6
T2
Z1
N1
+
N6
B2
−
N1
N2
(VBE + VZ) x 108
4BS1A1NB
If VBE is small compared to the zener voltage VZ, good
frequency accuracy is possible. For example, with VZ =
9.1 volts, a 40 Watt inverter using 2N3791 transistors
(operating from a 12 volt supply), exhibited frequency
regulation of ±2% with ±25% load variation.
Care should be taken not to exceed V(BR)EBO of the
non-conducting transistor, since the reverse emitter-base
voltage will be twice the introduced series voltage, plus VBE
of the conducting device.
Transformer T2 should not saturate at the lowest inverter
frequency.
Inverter starting is facilitated by placing a resistor from
point A to B1 or a capacitor from A to B2.
Zener diodes are often used in control circuits, usually to
control the magnitude of the output voltage or current. In this
unusual application, however, the zener is used to control the
output frequency of a current feedback inverter. The circuit
is shown in Figure 1.
T1
VBE x 108
4BS1A1NB
LOAD
SIMPLE SQUARE WAVE GENERATOR
The zener diode is widely used in wave shaping circuits;
one of its best known applications is a simple square wave
generator. In this application, the zener clips sinusoidal
waves producing a square wave such as shown in Figure 2a.
In order to generate a wave with reasonably vertical sides,
the ac voltage must be considerably higher than the zener
voltage.
Clipper diodes with opposing junctions built into the
device are ideal for applications of the type shown in
Figure 2b.
Q2
NC
Figure 1. Frequency Controlled Current
Feedback Inverter
The transformer T1 functions as a current transformer
providing base current IB = (NC/NB)IC. Without the zener
http://onsemi.com
61
ZENER
VOLTAGE
R
A.C. INPUT
Z
OUTPUT
FORWARD
DROP
VOLTAGE
(a) Single Zener Diode Square Wave Generator
ZENER Z1
VOLTAGE
R
Z1
A.C. INPUT
OUTPUT
Z2
ZENER Z2
VOLTAGE
(b) Opposed Zener Diodes Square Wave Generator
Figure 2. Zener Diode Square Wave Generator
http://onsemi.com
62
TRANSIENT VOLTAGE SUPPRESSION
LIGHTNING
At any given time there are about 1800 thunderstorms in
progress around the world, with lightning striking about 100
times each second. In the U.S., lightning kills about 150
people each year and injures another 250. In flat terrain with
an average lightning frequency, each 300 foot structure will
be hit, on average, once per year. Each 1200 foot structure,
such as a radio or TV tower, will be hit 20 times each year,
with strikes typically generating 600 million volts.
Each cloud-to-ground lightning flash really contains from
three to five distinct strokes occurring at 60 ms intervals,
with a peak current of some 20,000 amps for the first stroke
and about half that for subsequent strokes. The final stroke
may be followed by a continuing current of around 150 amps
lasting for 100 ms.
The rise time of these strokes has been measured at around
200 nanoseconds or faster. It is easy to see that the
combination of 20,000 amps and 200 ns calculates to a value
of dI/dt of 1011 amps per second! This large value means that
transient protection circuits must use RF design techniques,
particularly considerate of parasitic inductance and
capacitance of conductors.
While this peak energy is certainly impressive, it is really
the longer-term continuing current which carries the bulk of
the charge transferred between the cloud and ground. From
various field measurements, a typical lightning model has
been constructed, as shown in Figure 1.
INTRODUCTION
Electrical transients in the form of voltage surges have
always existed in electrical distribution systems, and prior to
the implementation of semiconductor devices, they were of
minor concern. The vulnerability of semiconductors to
lightning strikes was first studied by Bell Laboratories in
1961.1 A later report tried to quantify the amount of energy
certain semiconductors could absorb before they suffered
latent or catastrophic damage from electrostatic discharge.2
Despite these early warnings, industry did not begin to
address the issue satisfactorily until the late 1970s. Listed
below are the seven major sources of overvoltage.
• Lightning
• Sunspots
• Switching of Loads in Power Circuits
• Electrostatic Discharge
• Nuclear Electromagnetic Pulses
• Microwave Radiation
• Power Cross
Most electrical and electronic devices can be damaged by
voltage transients. The difference between them is the
amount of energy they can absorb before damage occurs.
Because many modern semiconductor devices, such as
small signal transistors and integrated circuits can be
damaged by disturbances that exceed the voltage ratings at
only 20 volts or so, their survivability is poor in unprotected
environments.
In many cases, as semiconductors have evolved their
ruggedness has diminished. The trend to produce smaller
and faster devices, and the advent of MOSFET and gallium
arsenide FET technologies has led to an increased
vulnerability. High impedance inputs and small junction
sizes limit the ability of these devices to absorb energy and
to conduct large currents. It is necessary, therefore, to
supplement vulnerable electronic components with devices
specially designed to cope with these hazards. Listed below
are the four primary philosophies for protecting against
transients.
• Clamping, or “clipping” is a method of limiting the
amplitude of the transient.
• Shunting provides a harmless path for the transient,
usually to ground by way of an avalanche or a crowbar
mechanism.
• Interrupting opens the circuit for the duration of the
transient.
• Isolating provides a transient barrier between hostile
environments and vulnerable circuits through the use of
transformers or optoisolators.
Selection of the proper protective method should be made
based upon a thorough investigation of the potential sources
of the overvoltage hazard. Different applications and
environments present different sources of overvoltage.
40 μs
60
60
TIME (ms)
Flash with No Continuing Current
150 A
60
0.3
100
TIME (ms)
Flash with Continuing Current
Figure 1. Typical Lightning Model, with and without
Continuing Current
http://onsemi.com
63
Depending on various conditions, continuing current may
or may not be present in a lightning strike. A severe lightning
model has also been created, which gives an indication of the
strength which can be expected during worst case conditions
at a point very near the strike location. Figure 2 shows this
model. Note that continuing current is present at more than
one interval, greatly exacerbating the damage which can be
expected. A severe strike can be expected to ignite
combustible materials.
Burying cables does not provide appreciable protection as
the earth is almost transparent to lightning radiated fields. In
fact, underground wiring has a higher incidence of strikes
than aerial cables.3
SUNSPOTS
The sun generates electromagnetic waves which can
disrupt radio signals and increase disturbances on residential
and business power lines. Solar flares, which run in cycles
of 11 years (1989 was a peak year) send out electromagnetic
waves which disrupt sensitive equipment.
Although not quantified, the effects of sunspot activity
should be considered. Sunspots may be the cause of
sporadic, and otherwise unexplainable problems in such
sensitive areas.
400 A
200 A
10
60
110 160
460 520 580 640 700
SWITCHING OF LOADS IN POWER CIRCUITS
Inductive switching transients occur when a reactive load,
such as a motor or a solenoid coil, is switched off. The
rapidly collapsing magnetic field induces a voltage across
the load’s winding which can be expressed by the formula:
860
TIME (ms)
Figure 2. Severe Lightning Model
A direct hit by lightning is, of course, a dramatic event. In
fact, the electric field strength of a lightning strike nearby
may be enough to cause catastrophic or latent damage to
semiconductor equipment. It is a more realistic venture to try
to protect equipment from these nearby strikes than to
expect survival from a direct hit.
With this in mind, it is important to be able to quantify the
induced voltage as a function of distance from the strike.
Figure 3 shows that these induced voltages can be quite high,
explaining the destruction of equipment from relatively
distant lightning flashes.
V = −L (dI/dt)
where L is inductance in henrys and dI/dt is the rate of
change of current in amps per second.
Such transients can occur from a power failure or the
normal opening of a switch. The energy associated with the
transient is stored within the inductance at power
interruption and is equal to:
W = 1/2 Li 2
where W is energy in joules and i is instantaneous current in
amps at the time of interruption.
As an example, a 1.4 to 2.5 kV peak transient can be
injected into a 120 vac power line when the ignition system
of an oil furnace is fired. It has also been shown that there are
transients present on these lines which can reach as high as
6 kV. In locations without transient protection devices, the
maximum transient voltage is limited to about 6 kV by the
insulation breakdown of the wiring.
Inductive switching transients are the silent killers of
semiconductors as they often occur with no outward
indication. A graphic example is the report of a large
elevator company indicating the failure of 1000 volt
rectifiers during a power interruption. In another area, power
interruption to a 20 HP pump motor in a remote area was
directly related to failure of sensitive monitoring equipment
at that same site.4
INDUCED VOLTAGE IN 1 m OF WIRE (V)
10000
1000
100
10
1
0.1
1
10
DISTANCE FROM STRIKE (km)
Figure 3. Voltage Induced by Nearby
Lightning Strike
http://onsemi.com
64
The electrical waveform involved in ESD is a brief pulse,
with a rise time of about 1 ns, and a duration of 100−300 μs.
The peak voltage can be as large as 30 kV in dry weather, but
is more commonly 0.5−5.0 kV.6 The fastest rise times occur
from discharges originating at the tip of a hand-held tool,
while discharges from the finger tip and the side of the hand
are slightly slower.7 A typical human with a body
capacitance of 150 pF, charged to 3 microcoulombs, will
develop a voltage potential of 20 kV, according to the
formula:
28 Vdc
B A
WAVEFORM AT POINT A
V
V=Q/C
t1
where V is voltage, Q is charge and C is capacitance. The
energy delivered upon discharge is:
t2
TIME (μs)
V = 600 V pk-pk
No. of Repetitions = 5 to 100
W = 1/2 CV2
t1 = 0.2 to 10 μs
t2 = 50 to 1000 μs
where W is energy in joules, C is capacitance and V is
voltage.
It is interesting to note that most microcircuits can be
destroyed by a 2500 volt pulse, but a person cannot feel a
static spark of less than 3500 volts!
Figure 4. Switching Transient Definition for Aircraft
and Military Buses, per Boeing Document D6-16050
After characterizing electrical overstress on aircraft
power buses, Boeing published Document D6-16050 as
shown in Figure 4.
The military has developed switching transient
definitions within several specifications including:
DOD-STD-1399 for shipboard
MIL-STD-704 for aircraft
MIL-STD-1275 for ground vehicles
NUCLEAR ELECTROMAGNETIC PULSES (NEMP)
When a nuclear weapon is detonated, a very large flux of
photons (gamma rays) is produced. These rays act to
produce an electromagnetic field known as a nuclear
electromagnetic pulse or NEMP. When a nuclear detonation
occurs above the atmosphere, a particulary intense pulse
illuminates all objects on the surface of the earth, and all
objects in the lower atmosphere within line of sight of the
burst. A burst 300 −500 km above Kansas would illuminate
the entire continental U.S.
A typical NEMP waveform is a pulse with a rise time of
about 5 ns and a duration of about 1 μs. Its peak electric field
is 50 −100 kV/m at ground level. After such a pulse is
coupled into spacecraft, aircraft and ground support
equipment, it produces a waveform as described in
MIL-STD-461C. The insidious effect of NEMP is its broad
coverage and its potential for disabling military defense
systems.
The International Electrotechnical Commission (IEC) is
now promoting their specification IEC 801-4 throughout
the European community. This describes an inductive
switching transient voltage threat having 50 ns wide spikes
with amplitudes from 2 kV to 4 kV occurring in 300 ms wide
bursts.5
Besides these particular military specifications, many are
application specific and functional tests exist. A supplier of
transient voltage suppressor components will be expected to
perform to a wide variety of them.
ELECTROSTATIC DISCHARGE (ESD)
ESD is a widely recognized hazard during shipping and
handling of many semiconductor devices, especially those
that contain unprotected MOSFETs, semiconductors for use
at microwave frequencies and very high speed logic with
switching times of 2 ns or less. In response to this threat,
most semiconductors are routinely shipped in containers
made of conductive material.
In addition to various shipping precautions, electronic
assembly line workers should be grounded, use
grounded-tip soldering irons, ionized air blowers and other
techniques to prevent large voltage potentials to be
generated and possibly discharged into the semiconductors
they are handling.
Once the assembled device is in normal operation, ESD
damage can still occur. Any person shuffling his feet on a
carpet and then touching a computer keyboard can possibly
cause a software crash or, even worse, damage the keyboard
electronics.
MICROWAVE RADIATION
Microwaves can be generated with such high power that
they can disable electronic hardware upon which many
military systems depend. A single pulse flux of 10−8 MJ/cm2
burns out receiver diodes, and a flux of 10−4 MJ/cm2 causes
bit errors in unshielded computers.8 With automobiles
utilizing MPU controls in more applications, it is important
to protect against the effects of driving by a microwave
transmitter. Likewise, a nearby lightning strike could also
have detrimental effects to these systems.
POWER CROSS
Yet another source of electrical overstress is the accidental
connection of signal lines, such as telephone or cable
television, to an ac or dc power line. Strictly speaking, this
phenomenon, known as a power cross, is a continuous state,
not a transient. However, the techniques for ensuring the
http://onsemi.com
65
(TVS) devices. TVS devices handle high peak currents
while holding peak voltage below damaging levels, but have
relatively low energy capability and cannot protect against
a power cross fault. The first TVS used by telephone
companies is the carbon block, but its peak let-through
voltage was too high for modern equipment using
unprotected solid state circuitry. A number of other
components fill today’s needs.
The power cross condition causes a problem with
telephone lines. Fast acting fuses, high speed circuit
breakers and positive temperature coefficient thermistors
have been successfully used to limit or interrupt current
surges exceeding a millisecond.
Over the years, telecommunications switching equipment
has been transitioning from electromechanical relays to
integrated circuits and MOSFET technology. The newer
equipment operates at minimal electrical currents and
voltages, which make it very efficient. It is therefore quite
sensitive to electrical overloads caused by lightning strikes
and other transient voltage sources, and by power crosses.
Because of the deployment of new technology, both in
new installations and in the refurbishment of older systems,
the need for transient protection has grown rapidly. It is
widely recognized that any new equipment must include
protection devices for reasons of safety, reliability and long
term economy.
The major telecom companies, in their never ending quest
for the elimination of electromechanical technology have
been looking at a number of novel methods and
implementations of protection. These methods provide for
solutions to both the primary and secondary protection
categories.
A number of studies have been conducted to determine the
transient environment on telephone lines. Very little has
been done with data lines because a typical situation does not
exist. However, information gathered from telephone line
studies can serve as a guide for data lines.
Past studies on telephone lines coupled with the high
current capability of arc type arrestors and the conservative
nature of engineers seem to have produced specifications
which far exceed the real need. A recent study by Bell South
Services9 reported that the highest level of transient energy
encountered was well below standards and specifications in
common use. Now, solid state devices perform adequately
for many applications. However the stringent specifications
of some regulatory agencies promote arc-type arrestors,
though solid state devices would be a better choice.
survival of signal electronics after a power cross are similar
to techniques used for protection against transient
overvoltages.
STANDARDIZED WAVEFORMS
Fortunately, measurements of these hazards have been
studied and documented in several industry specifications.
For
example,
Bellcore
Technical
Advisory
TA-TSY-000974 defines the generic measurement
waveform for any double exponential waveform, which is
the basis for most of the specific applications norms.
The predominant waveform for induced lightning
transients, set down by Rural Electrification Administration
Document PE-60, is shown in Figure 5. This pulse test,
performed at the conditions of 100 V/μs rise, 10/1000 μs, Ip
= 1 kV, is one of the two most commonly specified in the
industry. The other is the 8/20 μs waveform, shown in
Figure 6.
Ip
.9 Ip
.5 Ip
.1 Ip
0
1
2
TIME, ms
3
Figure 5. Pulse Waveform (10/1000 μs)
lp
.9 lp
.5 lp
.1 lp
0
10
20
30
TIME, μs
Figure 6. Pulse Waveform (8/20 μs)
TRANSIENT VOLTAGE SUPPRESSION
AND TELECOM
TRANSIENT VOLTAGE SURGE SUPPRESSION
COMPONENTS ON DATA AND TELEPHONE LINES
PRIMARY PROTECTION
Primary protection is necessary to protect against high
voltage transients which occur in the outdoor environment.
These transients include induced lightning surges and ac
cross conditions.
As such, primary protection is located at the point where
wiring enters the building or terminal box. It is the first line
defense against outside hazards. TVS devices located where
lines enter a building are called primary protectors.
Lines carrying data and telephone signals are subject to a
number of unwanted and potentially damaging transients
primarily from two sources: lightning and “power crosses.”
A power cross is an accidental connection of a signal line to
a powerline. Transients from lightning can impress voltages
well above a kilovolt on the line but are of short duration,
usually under a millisecond. Lightning transients are
suppressed by using Transient Voltage Surge Suppressor
http://onsemi.com
66
Protectors connected to indoor lines are referred to as
secondary protectors. Both primary and secondary
protectors are required to provide complete equipment
protection.
Today, primary protection is most generally accomplished
through the use of surge protector modules. For telecom,
these are designed specifically for the environment and the
standards dictated by the telecom applications. They
typically contain a two or three element gas arrestor tube and
a mechanically-triggered heat coil. Some also include air
gap carbon block arrestors which break over at voltages
above about 1500 volts.
Some modules contain high speed diodes for clamp
response in the low nanoseconds. This provides protection
until the gas tube fires, generally in about one microsecond.
The diodes may be connected between the tip, ring and
ground conductors in various combinations. The 5ESS
electronic switching system norms dictate design and
performance requirements of TVS modules in use today.
Test methods are spelled out in REA PE-80, a publication
of the Rural Electrification Administration.
In the U.S. alone, 58 million primary protection modules
are sold annually, about 40 million for central offices and 18
million for station locations, such as building entrances.
Eighty percent of these use gas tubes, 16% use air-gap
carbon blocks, and only 2% (so far) are solid state.
P
0.9 P
WAVEFORM IS DEFINED AS tr/td
WHERE
tr : FRONT TIME = 1.25 (b - a)
= (T1-T0)
td : DURATION = (T2 -T0)
0.5 P
0.1 P
a b T1
T0
TIME
T2
Figure 7. Definition of Double Exponential
Impulse Waveform
The 10/1000 μs wave approximates the worst case
waveform observed on data and telecom lines. TVS devices
intended for this service are usually rated and characterized
using a 10/1000 waveform. The Bell South study revealed
that the worst transient energy handled by primary
protectors on lines entering a central office was equivalent
to only 27 A peak of a 10/1000 wave. This level is
considerably less than that required by secondary protectors
in most of the standards in use today. This finding is
particularly significant because the Bell South service area
includes Central Florida, the region experiencing the highest
lightning activity in the U.S.
The United States Federal Communications Commission
(FCC) has defined mandatory requirements for equipment
which is to be connected to the U.S. telephone network. In
some cases, U.S. equipment must meet standards developed
by the Rural Electrification Agency (REA). Many nations
demand compliance to standards imposed by the
Consultative Committee, International Telegraph and
Telephone (CCITT). In addition, most equipment users
demand safety certification from U.L., which has its own
standards.
The FCC Standards are based on a worst case residue from
a carbon block primary protector installed where the phone
line enters the building. The CCITT standard is applicable
for situations lacking primary protection, other than wiring
flashover. Companies entering the telephone equipment or
protector market will need to obtain and become familiar
with the appropriate governing standards.
SECONDARY PROTECTION
Secondary protection is necessary for the equipment
inputs, and as such, is located between the primary protector
and the equipment. Secondary protection is generally
accomplished with one or more TVS components, as
opposed to the modules used for primary. It is often placed
on a circuit board along with other components handling
other duties, such as switching. Secondary protection is
applied to lines associated with long branch circuits which
have primary protection a significant distance away, to
internal data lines, and to other locations requiring
additional local hazard-proofing.
While not as open to external transients as the primary,
secondary can still see peak open circuit voltages in excess
of 1000 volts and short circuit currents of hundreds of amps.
These transients may be locally generated, or they may be
residuals from the primary protectors upstream.
TRANSIENT VOLTAGE PROTECTION
COMPONENTS
STANDARDS
GENERAL TVS CHARACTERISTICS
Transient voltage waveforms are commonly described in
terms of a dual exponential wave as defined in Figure 7. The
standard chosen for power lines is a 1.2/50 μs voltage wave
which causes an 8/20 μs current wave. Although the source
of the most severe transients on telecom lines is the same as
for power lines and lightning, the higher impedance per unit
length of the telephone line stretches the waves as they
propagate through the lines.
A number of transient voltage suppressor (TVS) devices
are available. Each finds use in various applications based
upon performance and cost. All types are essentially
transparent to the line until a transient occurs; however,
some devices have significant capacitance which loads the
line for ac signals. A few of the these are described in
Table 1.
Based upon their response to an overvoltage, TVS devices
fit into two main categories, clamps and crowbars. A clamp
conducts when its breakdown voltage is exceeded and
reverts back to an open circuit when the applied voltage
http://onsemi.com
67
drops below breakdown. A crowbar switches into a low
voltage, low impedance state when its breakover voltage is
exceeded and restores only when the current flowing
through it drops below a “holding” level.
where K is a constant of proportionality and s is an exponent
which defines the “sharpness factor” of the knee. The
exponent s is 1 for a resistor and varies from 5 to over 100
for the clamping devices being used in TVS applications. A
high value of s i.e., a sharp knee, is beneficial. A TVS device
can be chosen whose breakdown voltage is just above the
worst case signal amplitude on the line without concern of
loading the line or causing excessive dissipation in the TVS.
As the current density in the clamp becomes high, the
incremental resistance as described by Equation 1 becomes
very small in comparison to the bulk resistance of the
material. The incremental resistance is therefore ohmic in
the high current region.
Unfortunately, a uniform terminology for all TVS devices
has not been developed; rather, the terms were developed in
conjunction with the appearance of each device in the
marketplace. The key characteristics normally specified
define operation at voltages below the knee and at currents
above the knee.
Leakage current is normally specified below the knee at
a voltage variously referred to as the stand-off voltage, peak
working voltage or rated dc voltage. Some devices are rated
in terms of an RMS voltage, if they are bidirectional. Normal
signal levels must not exceed this working voltage if the
device is to be transparent.
Breakdown voltage is normally specified at a fairly low
current, typically 1 mA, which places operation past the
knee region. Worst case signal levels should not exceed the
breakdown voltage to avoid the possibility of circuit
malfunction or TVS destruction.
The voltage in the high current region is called the
clamping voltage, VC. It is usually specified at the maximum
current rating for the device. To keep VC close to the
breakdown voltage, s must be high and the bulk resistance
low. A term called clamping factor, (FC) is sometimes used
to describe the sharpness of the breakdown characteristic.
FC is the ratio of clamping voltage to the breakdown voltage.
As the V-I characteristic curve of the TVS approaches a right
angle, the clamping factor approaches unity. Clamping
factor is not often specified, but it is useful to describe clamp
device behavior in general terms.
CLAMP DEVICES
All clamp devices exhibit the general V-I characteristic of
Figure 8. There are variations; however, some clamps are
asymmetric. In the non clamping direction, some devices
such as the zener TVS exhibit the forward characteristic of
a diode while others exhibit a very high breakdown voltage
and are not intended to handle energy of “reverse” polarity.
Under normal operating conditions, clamp devices appear
virtually as an open circuit, although a small amount of
leakage current is usually present. With increasing voltage
a point is reached where current increases rapidly with
voltage as shown by the curved portion of Figure 8. The
rapidly changing curved portion is called the “knee region.”
Further increases in current places operation in the
“breakdown” region.
I O
0
V
Figure 8. Static Characteristics of a Clamp Device
In the knee region the V-I characteristic of clamping
devices can be approximated by the equation:
I = K Vs
(1)
Table 1. Comparison of TVS Components
Protection
Time
Protection
Voltage
Power
Dissipation
Reliable
Performance
Expected
Life
> 1 μs
60 −100 V
Nil
No
Limited
10 −20 ns
> 300 V
Nil
No
Degrades
Fusing required. Degrades.
Voltage level too high.
AVALANCHE TVS
50 ps
3 −400 V
Low
Yes
Long
Low power dissipation.
Bidirectional requires dual.
THYRISTOR TVS
< 3 ns
30 −400 V
Nil
Yes
Long
High capacitance.
Temperature sensitive.
Type
GAS TUBE
MOV
http://onsemi.com
68
Other
Considerations
Only 50 −2500 surges.
Can short power line.
relatively low frequency applications. In a data line or
telecom application the turn-off delay causes a loss of
intelligence after the transient surge has subsided.
A telephone line has both ac and dc signals present.
Crowbars can be successfully used to protect telecom lines
from high current surges. They must be carefully chosen to
ensure that the minimum holding current is safely above the
maximum dc current available from the lines.
Clamp devices generally react with high speed and as a
result find applications over a wide frequency spectrum. No
delay is associated with restoration to the off state after
operation in the breakdown region.
CROWBAR DEVICES
Crowbar TVS devices have the general characteristics
shown in Figure 9. As with clamp devices, asymmetric
crowbars are available which may show a diode forward
characteristic or a high voltage breakdown in one direction.
TVS DEVICES
A description of the various types of TVS devices follows
in the chronological order in which they became available.
Used appropriately, sometimes in combination, any
transient protection problem can be suitably resolved. Their
symbols are shown in Figure 10.
VT
I2
3
2
VT
VD
1
I1
V3O1
V3O2
Air-Gap
Carbon Block
2- and 3-Element
Gas Tubes
Heat Coil
Switch
Metal Oxide
Varistor (MOV)
Figure 9. Static Characteristics of a
Bidirectional Crowbar Device
The major difference between a crowbar and a clamp is
that, at some current in the breakdown region, the device
switches to a low voltage on-state. In the clamping region
from I1 to I2, the slope of the curve may be positive as shown
by segment 1, negative (segment 2) or exhibit both
characteristics as shown by segment 3. A slightly positive
slope is more desirable than the other two curves because a
negative resistance usually causes a burst of high frequency
oscillation which may cause malfunction in associated
circuitry. However, a number of performance and
manufacturing trade-offs affect the shape of the slope in the
clamping region.
A crowbar TVS has an important advantage over a clamp
TVS in that it can handle much larger transient surge current
densities because the voltage during the surge is
considerably lower. In a telephone line application, for
example, the clamping level must exceed the ring voltage
peak and will typically be in the vicinity of 300 volts during
a high current surge. The on-state level of the crowbar may
be as low as 3 volts for some types which allows about two
orders of magnitude increase in current density for the same
peak power dissipation.
However, a crowbar TVS becomes “latched” in the
on-state. In order to turn off its current flow the driving
voltage must be reduced below a critical level called the
holding or extinguishing level. Consequently, in any
application where the on-state level is below the normal
system voltage, a follow-on current occurs. In a dc circuit
crowbars will not turn off unless some means is provided to
interrupt the current. In an ac application crowbars will turn
off near the zero crossing of the ac signal, but a time delay
is associated with turn-off which limits crowbars to
Zener
Regulator
Unidirectional
Avalanche TVS Device
Thyristor TVS Devices
Bidirectional
Avalanche TVS Device
Dual Thyristor TVS Devices
Figure 10. TVS Devices and Their Symbols
AIR GAP ARRESTORS
The air gap is formed by a pair of metal points rigidly fixed
at a precise distance. The air ionizes at a particular voltage
depending upon the gap width between the points. As the air
ionizes breakover occurs and the ionized air provides a low
impedance conductive path between the points.
The breakover threshold voltage is a function of the air’s
relative humidity; consequently, open air gaps are used
mainly on high voltage power lines where precise
performance is not necessary. For more predictable
behavior, air gaps sealed in glass and metal packages are also
available.
Because a finite time is required to ionize the air, the actual
breakover voltage of the gap depends upon the rate of rise of
the transient overvoltage. Typically, an arrestor designed for
a 120 V ac line breaks over at 2200 volts.
Air gaps handle high currents in the range of 10,000
amperes. Unfortunately, the arc current pits the tips which
causes the breakover voltage and on-state resistance to
increase with usage.
http://onsemi.com
69
CARBON BLOCK ARRESTORS
Key electrical specifications for this TVS type include
breakover voltage (dc & pulse), maximum holdover
voltage, arc voltage, and maximum surge current.
The breakover voltage is rated at a slow rate of rise,
5000 V/s, essentially dc to a gas arrestor. Typical dc voltage
ratings range from 75 V through 300 V to provide for most
communication systems protection requirements. The
maximum pulse voltage rating is that level at which the
device fires and goes into conduction when subjected to a
fairly rapid rate of voltage rise, (dv/dt) usually 100 V/μs.
Maximum rated pulse voltages typically range from 400 V
to 600 V, depending on device type.
A typical waveform of a gas surge arrestor responding to
a high voltage pulse is shown in Figure 12. From the
waveform, it can be seen that the dv/dt of the wave is 100
v/μs and the peak voltage (the breakover voltage) is 520 V.
The carbon block arrestor, developed around the turn of
the century to protect telephone circuits, is still in place in
many older installations. The arrestor consists of two carbon
block electrodes separated by a 3 to 4 mil air gap. The gap
breaks over at a fairly high level − approximately 1 kV − and
cannot be used as a sole protection element for modern
telecom equipment. The voltage breakdown level is
irregular. With use, the surface of the carbon block is burned
which increases the unit’s resistance. In addition, the burned
material forms carbon tracks between the blocks causing a
leakage current path which generates noise. Consequently,
many of the carbon blocks in service are being replaced by
gas tubes and are seldom used in new installations.
SILICON CARBIDE VARISTORS
The first non-linear resistor to be developed was called a
“varistor.” It was made from specially processed silicon
carbide and found wide use in high power, high voltage TVS
applications. It is not used on telecom lines because its
clamping factor is too high: s is only about 5.
ARRESTOR VOLTAGE (VOLTS)
600
GAS SURGE ARRESTORS
Gas surge arrestors are a sophisticated modification of the
air gap more suited to telecom circuit protection. Most often
used is the “communication” type gap which measures
about 3/8 inch in diameter and 1/4 inch thick. A cross section
is shown in Figure 11. They consist of a glass or ceramic
envelope filled with a low pressure inert gas with specialized
electrodes at each end. Most types contain a minute quantity
of radioactive material to stabilize breakover voltage.
Otherwise, breakover is sensitive to the level of ambient
light.
Because of their small size and fairly wide gap,
capacitance is very low, only a few picofarads. When not
activated, their off-state impedance or insulation resistance
is virtually infinite.
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇ
ÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
ÇÇÇÇÇÇÇÇÇ
IGNITION AID
400
200
0
0
1
2
3
TIME (500 ns/DIVISION)
Figure 12. Voltage Waveform of Gas Surge Arrestor
Responding to a Surge Voltage
Gas surge arrestors fire faster but firing voltage increases
as the transient wave fronts increase in slope as illustrated in
Figure 13. The near vertical lines represent the incident
transient rise time. Note that the response time is greater than
0.1s at slow rise times but decreases to less than 0.1 μs for
risetimes of 20 kV/μs. However, the firing voltage has
increased to greater than 1000 V for the gas tube which
breaks over at 250 Vdc.
The driving circuit voltage must be below the holdover
voltage for the gap to extinguish after the transient voltage
has passed. Holdover voltage levels are typically 60% to
70% of the rated dc breakdown voltage.
Arc voltage is the voltage across the device during
conduction. It is typically specified at 5 to 10 V under a low
current condition, but can exceed 30 V under maximum
rated pulse current.
The maximum surge current rating for a 8/20 μs
waveform is typically in the 10 kA to 20 kA range for
communication type devices. For repetitive surges with a
10/1000 wave, current ratings are typically 100 A,
comfortably above the typical exposure levels in a telephone
subscriber loop.
ACTIVATING
COMPOUND
DISCHARGE
REGION
ELECTRODES
INSULATOR
(GLASS OR
CERAMIC MATERIAL)
Figure 11. Gas Arrestor Cross Section
http://onsemi.com
70
100 V/SEC
100 V/ms
100 V/ μs
500 V/s
μ
1.0
0.6
10KV/s
μ
470 Vdc
0.8
20KV/s
μ
SPARKOVER VOLTAGE (kV)
1.2
1KV/s
μ
5KV/s
μ
1.4
350 Vdc
250 Vdc
0.4
150 Vdc
0.2
0
8
7
6
5
4
3
2
1
1
10
RESPONSE TIME (SECONDS)
Figure 13. Typical Response Time of a Gas Surge Arrestor
Gas tubes normally provide long life under typical
operating conditions, however; wear-out does occur.
Wear-out is characterized by increased leakage current and
firing voltage. An examination of gas tubes in service for six
to eight years revealed that 15% were firing outside of their
specified voltage limits.9 Because firing voltage increases
with use, protectors often use an air gap backup in parallel
with the gas tube. End-of-life is often specified by
manufacturers as an increase of greater than 50% of
breakover or firing voltage. Other limits include a decrease
in leakage resistance to less than 1 mW.
The features and limitations of gas tube surge arrestors are
listed below.
Advantages:
• High current capability
• Low capacitance
• Very high off-state impedance
Disadvantages:
• Slow response time
• Limited life
• High let-through voltage
• Open circuit failure mode
semiconductor diodes. The TVS cells are built by depositing
the polycrystalline material on a metal plate to increase their
thermal mass thereby raising energy dissipation. The cells
exhibit typical diode characteristics and a non-linear reverse
breakdown which is useful for transient suppression. Cells
can be made which are “self-healing”; that is, the damage
which occurs when subjecting them to excessive transient
current is repaired with time.
Selenium cells are still used in high power ac line
protection applications because of their self-healing
characteristic; however, their high capacitance and poor
clamping factor (s » 8) rule them out for data or telecom line
applications.
METAL OXIDE VARISTORS
The metal oxide varistor (MOV) is composed of zinc
oxide granules in a matrix of bismuth and other metal
oxides. The interface between the zinc oxide and the matrix
material exhibits characteristics similar to that of a p-n
junction having a voltage breakdown of about 2.6 V. With
this structure the electrical equivalent is that of groups of
diodes in parallel which are stacked in series with similar
parallel groups to provide the desired electrical parameters.
The taller the stack, the higher the breakdown and operating
voltage. Larger cross-sections provide higher current
capability. The structure of an MOV is shown in Figure 14.
Principally because of their high firing voltage, gas surge
arrestors are not suitable for use as the sole element to
protect modern equipment connected to a data or telecom
line. However, they are often part of a protection network
where they are used as the primary protector at the building
interface with the outside world.
MICROVARISTOR
ZINC OXIDE
SELENIUM CELLS
INTERGRANULAR
Polycrystalline diodes formed from a combination of
selenium and iron were the forerunners of monocrystalline
Figure 14. MOV Cross Section
http://onsemi.com
71
MOV (27 V)
Vert: 10 V/div
Horiz: 0.5 ms/div
Transient Source Impedance: 0.55 Ω
Vpeak = 62.5 V
Figure 15. MOV Clamping Voltage Waveform
for devices most often used on power lines. Leakage current
behavior is similar to that of a p-n junction. It roughly
doubles for every 10°C increase in temperature and also
shows an exponential dependence upon applied voltage. At
a voltage of 80% of breakdown, leakage currents are several
microamperes at a temperature of 50°C.
Although the theory of MOV operation is not fully
developed, behavior is similar to a bidirectional avalanche
diode. Consequently its response time is very fast.
Life expectancy is an important characteristic generated
under pulse conditions. A typical example is shown in
Figure 16. The data applies to 20 mm diameter disc types
having rated rms voltage from 130 V to 320 V. Lifetime
rating curves are usually given for each device family.
MOVs, formed from a ceramic-like material, are usually
produced in the shape of discs with most widely used MOVs
having diameters of 7 mm, 14 mm, and 20 mm. The disc
surfaces are coated with a highly conductive metal such as
silver to assure uniform conduction through the cross
sectional area of the device. After terminal attachment the
parts are coated with a durable plastic material.
The typical voltage spectrum of MOVs ranges from 8 V
through 1000 V for individual elements. Pulse current
capability (8/20 μs) ranges from a few amperes to several
thousands of amperes depending on the element’s size. The
V-I characteristic of MOVs is similar to Figure 8. Their
clamping factor is fairly good; s is in the vicinity of 25.
Key electrical specifications include: operating voltage,
breakdown voltage, peak current maximum clamping
voltage, and leakage current.
The maximum operating voltage specified is chosen to be
below the breakdown voltage by a margin sufficient to
produce negligible heating under normal operating
conditions. Breakdown voltage is the transition point at
which a small increase in voltage results in a significant
increase in current producing a clamping action. Maximum
limits for breakdown voltage are typically specified at
1 mA with upper end limits ranging from 20% to 40%
greater than the minimum breakdown voltage.
Maximum peak current is a function of element area and
ranges from tens of amperes to tens of thousands of amperes.
MOVs are typically pulse rated with an 8/20 μs waveform
since they are intended primarily for use across power lines.
The clamping characteristics of a 27 V ac rated MOV, with
a 4 joule maximum pulse capability is shown in Figure 15.
The transient energy is derived from an exponentially
decreasing pulse having a peak amplitude of 90 V. The pulse
generator source impedance is 0.55 W. Peak clamping
voltage is 62.5 V while the developed current is 50 A. The
clamping factor calculates to be 2.3.
Leakage currents are listed for MOVs intended for use in
sensitive protection applications but are not normally listed
RATED PULSE CURRENT − AMPERES
10,000
5000
1
2
10
2000
102
1000
500
200
100
50
20 INDEFINITE
10
5
2
1
106 105
104
103
20
100
1000
10,000
IMPULSE DURATION μs
Figure 16. MOV Pulse Life Curve
For a single 8/20 μs pulse, the device described in Figure
16 is rated at 6500 A; however, it must be derated by more
than two orders of magnitude for large numbers of pulses.
Longer duration pulses also require further derating. For
example, for a 10/1000 μs duration pulse, this family of
devices has a maximum pulse rating of about 100 A on a
single shot basis and devices must be derated to less than
10A for long lifetimes in excess of 100,000 pulses.
http://onsemi.com
72
End-of-life for an MOV is defined as the voltage
breakdown degrading beyond the limits of ± 10%. As MOVs
are pulsed, they degrade incrementally as granular
interfaces are overheated and changed to a highly
conductive state. Failure occurs in power line applications
when the breakdown voltage has degraded to the point
where the MOV attempts to clip the powerline peaks. In
telecom applications, their breakdown must be above the
peaks of the impressed ac line during a ring cycle or a power
cross; otherwise an immediate catastrophic failure will
occur.
When MOVs fail catastrophically they initially fail short.
However, if a source of high energy is present as might occur
with a power cross, the follow-on current may cause the part
to rupture resulting in an open circuit.
The advantages and shortcomings of using an MOV for
general purpose protection in microprocessor based
circuitry include the following:
Advantages:
• High current capability
• Broad voltage spectrum
• Broad current spectrum
• Fast response
• Short circuit failure mode
Disadvantages:
• Gradual decrease of breakdown voltage
• High capacitance
59-03
318-07
403
403A
Key electrical parameters include maximum operating
voltage, maximum reverse breakdown voltage, peak pulse
current, peak clamping voltage, peak pulse power, and
leakage current.
The normal operating or working voltage is usually called
the reverse standoff voltage in specification sheets. Devices
are generally available over the range of 5 V through 250 V.
Standoff voltage defines the maximum peak ac or dc voltage
which the device can handle. Standoff voltage is typically
10% to 15% below minimum reverse breakdown voltage. A
listing of TVS products available from ON Semiconductor
is shown in Table 2.
Table 2. ON Semiconductor Zener TVS Series
VZ
RANGE
PULSE POWER
RATING
(100/1000
PULSE)
PACKAGE
*SA5.0ASA170A
6.8-200
500 W
Axial
*P6KE6.8A P6KE200A
6.8-200
600 W
Axial
*1.5KE6.8A 1.5KE250A
6.8-250
1500 W
Axial
1SMB5.0AT3 1SMB170AT3
6.8-200
600 W
SMB
P6SMB6.8AT3 P6SMB200T3
6.8-200
600 W
SMB
1SMC5.0AT3 1SMC78AT3
6.8-91
1500 W
SMC
1.5SMC6.8AT3
1.5SMC91AT3
6.8-91
1500 W
SMC
MMBZ15VDLT1
15
ESD Protection
>15 kV
SOT-23
DEVICE
SERIES
ZENER TVS
Zener TVS devices are constructed with large area silicon
p-n junctions designed to operate in avalanche and handle
much higher currents than their cousins, zener voltage
regulator diodes. Some manufacturers use small area mesa
chips with metal heatsinks to achieve high peak power
capability. However, ON Semiconductor has determined
that large area planar die produce lower leakage current and
clamping factor. The planar construction cross section is
shown in Figure 17 and several packages are shown in
Figure 18.
PLANAR DIE
41A
Figure 18. Typical Insertion and Surface Mount Silicon
TVS Packages — Zeners and Thyristors
The capacitance of MOVs is fairly high because a large
device is required in order to achieve a low clamping factor;
consequently, they are seldom used across telecom lines.
PLASTIC
ENCAPSULATION
17-02
SOLDER
ÈÈ
ÈÈ
ÈÈ
ÈÈ
* Available in bidirectional configurations
Figure 17. Zener TVS Cross-Section
http://onsemi.com
73
6 kW, which is four times greater than the rating at 1ms. The
current handling capability is also increased roughly by this
same factor of four.
The reverse breakdown voltage is specified at a bias level
at which the device begins to conduct in the avalanche mode.
Test current levels typically are 1 mA for diodes which
breakdown above 10 V and 10 mA for lower voltage
devices. Softening of the breakdown knee, that is, lower s,
for lower voltage p-n junction devices requires a higher test
current for accurate measurements of reverse breakdown
voltage. Diodes that break down above 10 V display a very
sharp knee; s is over 100.
Peak pulse current is the maximum upper limit at which
the device is expected to survive. Silicon p-n junctions are
rated for constant power using a particular transient
waveform; consequently, current is a function of the peak
clamping voltage. For example, a 6.8 V device handles
about 28 times the pulse current that a 220 V device will
withstand; however, both the 6.8 V and 220 V types dissipate
the same peak power under the same pulse waveform
conditions. Most Zener TVS diodes are rated for 10/1000 μs
waveform pulses which are common in the telecom industry.
The clamping voltage waveform of a 27 V Zener TVS
having a 1.5 joule capability is illustrated in Figure 19. Its
peak voltage is 30.2 V. The transient energy source is the
same as applied to the MOV whose response is shown in
Figure 10. However, the current through the Zener TVS is
over 100 A, much higher than occurs with the MOV because
the clamping voltage is significantly lower. Despite the
higher pulse current, the Zener displays much better
clamping action; its clamping factor is 1.1.
PP , PEAK POWER (kW)
100
1
VC , CLAMPING VOLTAGE (VOLTS)
0.1μs
1μs
10μs
100μs
1 ms
10 ms
tP, PULSE WIDTH
Figure 20. Peak Pulse Power Rating
for a Popular Zener TVS Family
To increase power capability devices are stacked in series.
For example, doubling the power capability requirement for
a 100 V, 1.5 kW Zener TVS is easily done by placing two 50
V devices in series. Clamping factor is not significantly
affected by this arrangement.
Although leakage current limits are relatively high for the
industry low voltage types (500 μA to 1000 μA), dropping
off to 5 μA or less for voltages above 10 V, the planar die in
use by ON Semiconductor exhibit considerably less leakage
than the specified limits of the industry types.
Capacitance for the popular 1500 W family exceeds
10,000 pF at zero bias for a 6.8 V part, dropping
exponentially to less than 100 pF for a 200 V device.
Capacitance drops exponentially with a linear increase in
bias. The capacitance of a 6.8 V device is 7000 pF, while the
200 V part measures under 60 pF, at their respective standoff
voltages.
Capacitance loads the signal line at high frequencies. For
high speed data transmission circuits, low capacitance is
achieved by placing two diodes in a series stack as shown in
Figure 21. Under normal operation the top diode (DS)
operates at essentially zero bias current. Since its power
dissipation requirement is small, its area can be much
smaller than that of the TVS diode (DZ) in order to provide
low capacitance. The top diode normally is not intended to
be used in avalanche. Consequently if a negative voltage
exceeding the reverse rating of the stack could occur, the low
capacitance diode must be protected by another diode (DP)
shown connected by dotted lines on Figure 21. The
arrangement of Figure 21 is satisfactory for situations where
the signal on the line is always positive. When the signal is
ac, diode DP is replaced by another low capacitance stack,
connected in anti-parallel.
60
50
40
30
20
10
Figure 14
10
0
TIME (500 μs/DIVISION)
Figure 19. Zener TVS Clamping Voltage Waveform
Peak pulse power is the instantaneous power dissipated at
the rated pulse condition. Common peak pulse power ratings
are 500 W, 600 W, and 1500 W for 10/1000 μs waveforms.
As the pulse width decreases, the peak power capability
increases in a logarithmic relationship. An example of a
curve depicting peak pulse power versus pulse width is
shown in Figure 20. The graph applies to the 1.5 kW series
(10/1000 pulse) of TVS diodes and can be interpolated to
determine power ratings over a broad range of pulse widths.
At 50 μs, the maximum rated power shown in the curve is
http://onsemi.com
74
Advantages:
• High repetitive pulse power ratings
• Low clamping factor
• Sub-nanosecond turn-on
• No wearout
• Broad voltage spectrum
• Short circuit failure mode
Disadvantages:
• Low non-repetitive pulse current
• High capacitance for low voltage types
DS
DP
DZ
Because of their fast response and low clamping factor,
silicon devices are used extensively for protecting
microprocessor based equipment from voltage surges on dc
power buses and I/O ports.
Figure 21. A Series Stack to Achieve Low
Capacitance with Zener TVS Diodes
Switching speed is a prime attribute of the zener TVS.
Avalanche action occurs in picoseconds but performing tests
to substantiate the theory is extremely difficult. As a
practical matter, the device may be regarded as responding
instantaneously. Voltage overshoots which may appear on
protected lines are the result of poor layout and packaging
or faulty measurement techniques.
The p-n junction diode is a unidirectional device. For use
on ac signal lines, bidirectional devices are available which
are based upon stacking two diodes back to back. Most
manufacturers use monolithic NPN and PNP structures. The
center region is made relatively wide compared to a
transistor base to minimize transistor action which can cause
increased leakage current.
No wearout mechanism exists for properly manufactured
Zener diode chips. They are normally in one of two states;
good, or shorted out from over-stress. Long-term life studies
show no evidence of degradation of any electrical
parameters prior to failure. Failures result from stress which
causes separation of the metal heat sink from the silicon chip
with subsequent overheating and then failure. Like MOVs,
silicon chips quickly fail short under steady state or long
duration pulses which exceed their capabilities.
The strengths and weaknesses of Zener TVS devices are
listed below.
THYRISTOR DIODES
The most recent addition to the TVS arsenal is the
thyristor surge suppressor (TSS). The device has the low
clamping factor and virtually instantaneous response
characteristic of a silicon avalanche (Zener) diode but, in
addition, it switches to a low voltage on-state when
sufficient avalanche current flows. Because the on-state
voltage is only a few volts, the TSS can handle much higher
currents than a silicon diode TVS having the same chip area
and breakdown voltage. Furthermore, the TSS does not
exhibit the large overshoot voltage of the gas tube.
Thyristor TVS diodes are available with unidirectional or
bidirectional characteristics. The unidirectional type
behaves somewhat like an SCR with a Zener diode
connected from anode to gate. The bidirectional type
behaves similarly to a triac having a bidirectional diode
(Diac) from main terminal to gate.
Packaged TSS chips are shown in Figure 18. Figure 22a
shows a typical positive switching resistance bidirectional
TSS chip. Construction of the device starts with an n
material wafer into which the p-bases and n-emitters are
diffused. There are four layers from top to bottom on each
side of the chip, forming an equivalent SCR. Only half the
device conducts for a particular transient polarity.
http://onsemi.com
75
Figure 22a. Chip Construction
A
ÈÈÈÈ
ÈÈÈÈ
A
P EMITTER
P BASE
N EMITTER
N BASE
ÈÈÈÈ
ÈÈÈÈ
K
Figure 22b. Cross Section
of Left Side
K
Figure 22c. Circuit Model
of Left Side
MT1
MT2
Figure 22d. Circuit Symbol
of the NPN transistor drives the base of the PNP causing the
two devices to hold one another on. Both the p and n emitters
flood the chip with carriers resulting in high electrical
conductivity and surge current capability.
When driven with high voltage ac, which occurs during a
power cross, positive resistance TSS devices act like a Zener
diode until the ac voltage drives the load line through the
point where regeneration occurs. Then it abruptly switches
to a low voltage. When the peak ac current is just below the
current required for breakover, the device operates mainly
as a Zener and power dissipation is high although the current
is low. When the ac current peak is well above breakover,
(>10 A), the device operates mainly as an SCR, and the low
on-state voltage causes power dissipation to be relatively
low.
Negative resistance devices operate in a similar fashion.
However, their behavior is dependent upon the “load line,”
that is, the equivalent resistance which the device “sees.”
When the load resistance is high (>1000 W) behavior is
similar to that of a positive resistance TSS in that high
The “gate” does not trigger the SCR, instead, operation in
the Zener mode begins when the collector junction
avalanches. Note that the p-bases pass through the
n-emitters in a dot pattern and connect to the contact metal
covering both halves of the chip. This construction
technique provides a low resistance path for current flow
and prevents it from turning on the NPN transistor.
Therefore, at relatively low currents, the device acts like a
low gain PNP transistor in breakdown. The Zener diode is
the collector base junction of the PNP transistor. Negative
resistance TSS devices are similarly constructed but start
with p-type material wafer, allowing the fabrication of a
high-gain NPN transistor. The switchback in voltage with
increasing current is caused by the gain of the NPN.
Both device types switch on completely when the current
flow through the base emitter shunt resistance causes
enough voltage drop to turn on the emitter and begin four
layer action. Now the device acts like an SCR. The collector
current of the PNP transistor (Figure 22c) provides the base
drive for the NPN transistor. Likewise, the collector current
http://onsemi.com
76
on-state conduction mode. The transition stage to
conduction may have any of the slopes shown in Figure 9.
The important voltages which define the thyristor
operating characteristics are also shown in Figure 9. VD is
operating voltage, VC is the clamping voltage and VT is
on-state voltage.
On-state voltage for most devices is approximately 3 V.
Consequently, transient power dissipation is much lower for
the thyristor TVS than for other TVS devices because of its
low on-state voltage. For example, under power cross
conditions Bell South Services reported their tests showed
that the thyristor TVS devices handled short bursts of
commercial power with far less heating than arc type surge
arrestors.9
Capacitance is also a key parameter since in many cases
the TSS is a replacement for gas surge arrestors which have
low capacitance. Values for the TSS range from 100 pF to
200 pF at zero volts, but drop to about half of these values
at a 50 Vdc bias.
Holding current (IH) is defined as the current required to
maintain the on-state condition. Device thru-current must
drop below IH before it will restore to the non-conducting
state. Turn-off time is usually not specified but it can be
expected to be several milliseconds in a telecom application
where the dc follow-on current is just slightly below the
holding current.
The major advantages and limitations of the thyristor are:
Advantages:
• Fast response
• No wearout
• Produces no noise
• Short circuit failure mode
Disadvantages:
• Narrow voltage spectrum
• Non-restoring in dc circuits unless current is below IH
• Turn-off delay time
instantaneous power dissipation occurs as the load line is
driven along the high voltage region of the TSS prior to
switching.
When a TSS with a negative resistance characteristic is
driven with a low “load” resistance, switching occurs when
the load line is tangent to the peak of the negative resistance
curve. Thus, complete turn-on can occur at a very low
current if the load resistance is low and the device has a
“sharp” switchback characteristic.
Leakage and the Zener knee voltage increase with
temperature at eight percent and 0.11% per °C respectively
for 200 V positive resistance types. But the current required
to cause regeneration falls with temperature, causing less
Zener impedance contribution to the breakover voltage,
resulting in a large reduction in the breakover voltage
temperature coefficient to as little as 0.05%/°C. Negative
resistance types can show positive or negative breakover
voltage coefficients depending on temperature and the
sharpness of the negative switchback.
The response of both positive and negative switching
resistance units to fast transients involves a race between
their Zener and regenerative attributes. At first the device
conducts only in the small chip area where breakdown is
occurring. Time is required for conduction to spread across
the chip and to establish the currents and temperatures
leading to complete turn-on. The net result is that both types
exhibit increasing breakover voltage with fast transients.
However, this effect is very small compared to gas discharge
tubes, being less than 25% of the breakover voltage.
Negative resistance types are more sensitive to unwanted
turn-on by voltage rates (dv/dt) at peak voltages below the
avalanche value. The transient current that flows to charge
the self-capacitance of the device sets up an operating point
on the negative resistance slope leading to turn-on.
Reduction of dv/dt capability becomes significant when the
signal voltage exceeds 80% of the avalanche value.
Complete turn-off following a transient requires the load
line to intersect the device leakage characteristic at a point
below the avalanche knee. During turn-off the load line must
not meet an intermediate conducting state which can occur
with a negative resistance device. Positive resistance types
are free of states causing turn-off “sticking.” Both types
have temperature sensitive holding currents that lie between
1 and 4 mA/°C.
Recent product developments and published studies have
generated much interest. Based on a study sponsored by Bell
South,9 the authors concluded that these new devices offered
the highest level of surge protection available.
Key electrical parameters for the Thyristor TVS include
operating voltage, clamping voltage, pulse current, on-state
voltage, capacitance, and holding current.
Operating voltage is defined as the maximum normal
voltage which the device should experience. Operating
voltages from 60 V to 200 V are available.
Clamping voltage is the maximum voltage level attained
before thyristor turn-on and subsequent transition to the
The thyristor TVS is finding wide acceptance in telecom
applications because its characteristics uniquely match
telecom requirements. It handles the difficult “power cross”
requirements with less stress than other TVS devices while
providing the total protection needed.
SURGE PROTECTOR MODULES
TYPES OF SURGE PROTECTOR MODULES
Several component technologies have been implemented
either singly or in combination in surge protector modules
and devices. The simplest surge protectors contain nothing
more than a single transient voltage suppression (TVS)
component in a larger package. Others contain two or more
in a series, parallel or series-parallel arrangement. Still
others contain two or more varieties of TVS elements in
combination, providing multiple levels of protection.
Many surge protectors contain non-semiconductor
elements such as carbon blocks and varistors. If required,
http://onsemi.com
77
UTILITY
other modes of protection components may be incorporated,
such as circuit breakers or EMI noise filters.
Surge protector modules are one solution to the
overvoltage problem. Alternatives include:
• Uninterruptible power systems (UPS), whose main
duty is to provide power during a blackout, but
secondarily provide protection from surges, sags and
spikes.
• Power line conditioners, which are designed to isolate
equipment from raw utility power and regulated
voltages within narrow limits.
Both UPS and power line conditioners are far more
expensive than surge protector modules.
THE 6 MAJOR CATEGORIES OF
SURGE PROTECTION MODULES
The power transmission and distribution equipment
industry has an obvious need for heavy duty protection
against overvoltage transients. Many utility situations
require a combination of techniques to provide the necessary
solution to their particular problems. This industry utilizes
many forms of transient suppression outside the realm of
semiconductors.
DATACOM
Local area networks and other computer links require
protection against high energy transients originating on their
data lines. In addition, transients on adjacent power lines
produce electromagnetic fields that can be coupled onto
unprotected signal lines. Datacom protectors have a ground
terminal or pigtail which must be tied to the local equipment
ground with as short a lead as possible. Datacom protectors
should be installed on both ends of a data link, or at all nodes
in a network. This protection is in addition to the ac line
transient protection, which is served by the plug-in or
hardwired protection modules. Some datacom protector
modules contain multi-stage hybrid circuits, specially
tailored for specific applications, such as 4−20 mA analog
current loops.
Plug-in
Hardwired
Utility
Datacom
Telecom
RF and microwave
PLUG-IN MODULES
Plug-in modules come in a variety of sizes and shapes, and
are intended for general purpose use. They permit the
protection of vulnerable electronic equipment, such as home
computers, from overvoltage transients on the 115 vac line.
These products are sold in retail outlets, computer stores and
via mail order. Most models incorporate a circuit breaker or
fuse, and an on/off switch with a neon indicator. The module
may have any number of receptacles, with common models
having from two to six. These products comply with UL
1449, and are generally rated to withstand the application of
multiple transients, as specified in IEEE 587. Plug-in
modules generally provide their protection through the use
of these devices which are typically connected between line
and neutral, and between neutral and ground.
TELECOM
Included here are devices used to protect central office and
station telecommunications (telecom) equipment against
voltage surges. None of these devices are grounded through
an ac power receptacle. Those that are grounded through an
ac power receptacle are categorized as plug-in modules. Not
only can overvoltages cause disruptions of telecom service,
but they can destroy the sophisticated equipment connected
to the network. Also, users or technicians working on the
equipment can be injured should lightning strike nearby. It
is estimated that 10 to 15 people are killed in the U.S. each
year while talking on the telephone during lightning storms.
For these reasons, surge protectors are used both in central
offices and in customer premises.
There are three types of telecom surge protectors now in
service: air-gap carbon block, gas tube, and solid state. The
desire of the telecom market is to convert as many of the
non-solid state implementations into solid state as cost will
permit.
HARDWIRED
Hardwired modules take on a wide variety of styles,
depending upon their designed application. They provide
protection for instrumentation, computers, automatic test
equipment, industrial controls, motor controls, and for
certain telecom situations.
Many of these modules provide snubbing networks
employing resistors and capacitors to produce an RC time
constant. Snubbers provide common mode and differential
mode low-pass filtering to reduce interference from line to
equipment, and are effective in reducing equipment
generated noise from being propogated onto the line.
Snubbers leak current however, and many of these modules
are designed with heat sinks and require mounting to a
chassis. The surge protection is performed in a similar
manner to the plug-in modules mentioned earlier.
Hardwired products, therefore, present a prime opportunity
for avalanche TVS components.
SELECTING TVS COMPONENTS
From the foregoing discussion it should be clear that the
silicon junction avalanche diode offers more desirable
characteristics than any other TVS component. Its ability to
clamp fast rising transients without overshoot, low clamping
factor, non-latching behavior, and lack of a wearout
mechanism are the overriding considerations. Its one-shot
surge capability is lower than most other TVS devices but is
normally adequate for the application. Should an unusually
severe event occur, it will short yet still protect the
equipment.
http://onsemi.com
78
Telephone and RS-422 lines are called balanced lines
because the signal is placed between two lines which are
“floating” with respect to ground. A signal appears between
each signal line and ground but is rejected by the receiver;
only the difference in potential between the two signal lines
is recognized as the transmitted signal. This system has been
in use for decades as a means of providing improved noise
immunity, but protection from transient surge voltages is
more complex.
A cost-effective means of protecting a balanced line is
shown in Figure 23. The bridge diode arrangement allows
protection against both positive and negative transients to be
achieved, an essential requirement but the TVS devices Z1
and Z2 need only be unidirectional. The diodes are chosen
to have low capacitance to reduce loading on the line and
low turn-on impedance to avoid causing an overshoot on the
clamped voltage level. Although a zener TVS is shown, a
TSS is more appropriate when a telephone line having ring
voltages is to be protected.
For example, an RS-232 data line is specified to operate
with a maximum signal level of ± 25 V. Failure analysis
studies11 have shown that the transmitters and receivers
used on RS-232 links tolerate 40 V transients. A
1.5KE27CA diode will handle the maximum signal level
while holding the peak transient voltage to less than 40 V
with a 40 A 10/1000 pulse which is adequate for all indoor
and most outdoor data line runs. As a practical matter, few
data links use 25 V signals; 5 V is most common.
Consequently, much lower voltage silicon diodes may be
used which will allow a corresponding increase in surge
current capability. For example, a 10 V breakdown device
from the same 1500 W family will clamp to under 15 V
(typically 12 V) when subjected to a 100 A pulse.
Telecommunications lines which must accommodate the
ring voltage have much more severe requirements. For
example, one specification11 from Bellcore suggests that
leakage current be under 20 mA over the temperature range
from −40°C to +65°C with 265 V peak ac applied. To meet
this specification using Zener TVS parts, devices must be
stacked. Devices which breakdown at 160 V are chosen to
accommodate tolerances and the temperature coefficient. A
part number with a 10% tolerance on breakdown could
supply a unit which breaks down at 144 V. At − 40°C
breakdown could be 133 V. The breakdown of two devices
stacked just barely exceeds the worst case ring peak of
265 V. A 1500 W unit has a surge capability of 6.5 A
(10/1000) which is too low to be satisfactory while higher
power units are too expensive as a rule.
Another problem which telephone line service presents to
a surge suppressor is survival during a power cross. An
avalanche diode is impractical to use because the energy
delivered by a power cross will produce diode failure before
any overcurrent protective element can react.
As indicated by the Bell South studies, the thyristor TVS
is ideal for telephone line applications. Suppliers offer
unidirectional and bidirectional units which meet the FCC
impulse wave requirements as shown in Table 1. In addition,
the thyristor can handle several cycles of 50/60 Hz power
before failure. The ON Semiconductor MKT1V200 series,
for example, can handle 10 A for 4 cycles, which is enough
time for a low current fuse or other current activated
protective device to react.
D1
D3
Z1
D2
D4
Z2
Figure 23. A Method of Reducing Capacitance and
Protecting a Balanced Line
Since transients are usually common-mode, it is important
that the TVS circuit operate in a balanced fashion;
otherwise, common mode transients can cause differential
mode disturbances which can be devastating to the line
receiver. For example, suppose that an identical positive
common mode surge voltage appears from each
line-to-ground. Diodes D2 and D4 will conduct the transients
to Z2. However, if one of these diodes has a slower turn-on
or higher dynamic impedance than the other, the voltage
difference caused by the differing diode response appears
across the signal lines. Consequently, the bridge diodes must
be chosen to be as nearly identical as possible.
Should a differential mode transient appear on the signal
lines, it will be held to twice that of the line-to-ground
clamping level. In many cases a lower clamping level is
needed which can be achieved by placing another TVS
across the signal lines. It must be a bidirectional low
APPLICATION CONSIDERATIONS
In most cases, it is not advisable to place a zener TVS
directly across a data line because of its relatively high
capacitance. The arrangement previously discussed and
shown in Figure 21 works well for an unbalanced line such
as RS-232. When using discrete steering diodes, they should
have low capacitance and low turn-on impedance to avoid
causing an overshoot on the clamped voltage level.
Most noise and transient surge voltages occur on lines
with respect to ground. A signal line such as RS-232 which
uses ground as a signal reference is thus very vulnerable to
noise and transients. It is, however, easy to protect using a
single TVS at each end of the line.
http://onsemi.com
79
resulting current flow. To insure that this occurs,
semiconductor devices use heavy gauge clips or bonding
wires between the chip and terminals. In addition, parts are
available in plastic packages having a spring type shorting
bar which shorts the terminals when the package softens at
the very high temperatures generated during a severe
overload.
The shorted TVS protects the equipment, but the line
feeding it could be destroyed if the source of energy which
shorted the TVS is from a power cross. Therefore, it is wise
and necessary for a UL listing to provide a series element
such as a fuse or PTC device to either open the circuit or
restrict its value to a safe level.
The design of circuit boards is critical and layout must be
done to minimize any lead or wiring inductance in series
with the TVS. Significant voltage is developed in any loop
subject to transients because of their high current amplitudes
and fast risetimes.
capacitance device. With a line-to-line TVS in the circuit
diode matching is not required.
Other schemes appearing in the literature use two
bidirectional TVS devices from each line-to-ground as
shown in Figure 24 that of the line-to-ground voltage. To
avoid generating a differential mode signal, the TVS must
be closely matched or a third TVS must be placed
line-to-line. By using a third TVS differential mode
transients can be held to a low level.
TIP
T1
T3
T2
REFERENCES
1. D. W. Bodle and P. A. Gresh, “Lightning Surges in
Paired Telephone Cable Facilities,” The Bell System
Technical Journal, Vol. 4, March 1961, pp. 547−576.
2. D. G. Stroh, “Static Electricity Can Kill Transistors,”
Electronics, Vol. 35, 1962, pp. 90 −91.
3. J. D. Norgard and C. L. Chen, “Lightning-Induced
Transients on Buried Shielded Transmission Lines,”
IEEE Transactions on EMC, Vol. EMC28, No. 3,
August 1986, pp. 168 −171.
4. O. M. Clark, “Transient Voltage Suppression (TVS),”
1989, pp. 6 −7.
5. Clark, p. 7.
6. World Information Technologies, “U. S. Electrical and
Electronic Surge Protection Markets,” 1989, p. 5.
7. T. J. Tucker, “Spark Initiation Requirements of a
Secondary Explosive,” Annals of the New York
Academy of Sciences, Vol. 152, Article I, 1968, pp.
643−653.
8. H. K. Florig, “The Future Battlefield: A Blast of
Gigawatts?,” IEEE Spectrum, Vol. 25, No. 3, March
1988, pp. 50 −54.
9. Mel Thrasher, “A Solid State Solution,” Telephony,
June 1989, pp. 48 −52.
10. A. Urbieta, “Sensitivity Study to EOS/SSD of Bipolar
Integrated Circuits,” EOS-8, 1986.
11. M. Tetreault and F.D. Martzloff, “Characterization of
Disturbing Transient Waveforms on Computer Data
Communication Lines,” EMC Proceedings, Zurich,
March 1985, pp. 423 −428.
12. F. Martzloff, “Coupling, Propagation, and Side Effects
of Surges in an Industrial Building Wiring System,”
RING
Figure 24. Protecting a Balanced Line with
Bidirectional TSS Devices
The arrangement of Figure 25 offers the advantage of
lower capacitance when differential mode transient
protection is required. If all three TSS devices have the same
capacitance (C), the line-to-line and line-to-ground
capacitance of Figure 24 is 1.5C. However, the arrangement
of Figure 25 exhibits a capacitance of only C/2. To design the
circuit to handle the same simultaneous common mode
energy as the circuit of Figure 24, T3 must be twice as large
as T1 and T2. In this case the capacitance of T3 is doubled
which causes the line-to-ground capacitance to be 2C/3, still
a considerable improvement over the arrangement of
Figure 24.
TIP
T1
T3
T2
RING
Figure 25. Preferred Method of Protecting a
Balanced Line Using Bidirectional TSS Devices
Protectors are usually designed to be “fail safe” if their
energy ratings are exceeded, but the definition of “fail safe”
is often dependent upon the application. The most common
requirement is that the surge voltage protective element
should fail short and remain shorted regardless of the
http://onsemi.com
80
IEEE 472/ANSI C 37.90.1. Requirements for protective
relays, including 10/1000 nS waveform testing is described.
UL943. This requirement defines a 0.5 μs/100 kHz
waveform for ground fault and other switching applications.
VDE 0420. Industrial remote control receivers are
detailed, and test procedures defined.
VDE 0860/Part 1/II. This includes a description of
0.2/200 μs, 10kV test requirements.
Conference Record of IEEE IAS Meeting, 1988, pp.
1467−1475.
IMPORTANT REGULATORY REQUIREMENTS
AND GUIDELINES
GENERAL
DOD-STD-1399, MIL-STD-704, MIL-STD-1275,
MIL-STD-461C. These military specifications are
important, if we intend to target devices for military or
commercial aviation markets.
IEEE 587. This specification describes multiple transient
waveforms.
UL1449. This is a compulsory test which demonstrates
performance to criteria establishing the maximum voltage
that can pass through a device after clamping has taken
place. It is important that we comply, and say so on our data
sheets.
TELECOM
CCITT IX K.17, K.20, K.15. These documents relate to
repeaters.
EIA PM-1361. This document covers requirements for
telephone terminals and data processing equipment.
FTZ 4391 TV1. This is a general German specification
for telecom equipment.
FCC Part 68. The Federal Communications Commission
(FCC) requirements for communications equipment is
defined. Of special note is §68.302, dealing with telecom
power lines.
NT/DAS/PRL/003. Telephone instrument, subscriber
equipment and line requirements are documented.
PTT 692.01. This is a general Swiss specification for
telecom exchange equipment.
REA PE-60. The Rural Electrification Administration
(REA) has documented the predominant waveform for
induced lightning transients. This test is now commonly
known as the 10/1000 μs pulse test.
REA PE-80. The REA defines requirements for gas tubes
and similar devices for telecom applications.
TA-TSY-000974. This technical advisory by Bellcore
defines double exponential waveforms, which are the basis
for many telecom applications norms.
UL 1459 and UL 4978. These document detail tests for
standard telephone equipment and data transmission.
TA-TSY-000974. This technical advisory by Bellcore
defines double exponential waveforms, which are the basis
for many telecom applications norms.
UL 1459 and UL 4978. These document detail tests for
standard telephone equipment and data transmission.
INDUSTRIAL
ANSI/IEEE C62.41. Established by the American
National Standards Institute (ANSI) and the Institute of
Electrical and Electronic Engineers (IEEE), this guideline
tests the effectiveness of devices to typical power
disturbances. To meet the most rigorous category of this
spec, a device or module must be capable of withstanding a
maximum repetitive surge current pulse of 3000 amps with
a 8/20 μs waveform.
IEC TC 102 D. Requirements for remote control
receivers for industrial applications are detailed in this
International Electrotechnical
Commission
(IEC)
document.
IEC 255-4 and IEC TC 41. These documents describe
testing for static relays for industrial use.
IEC 801-1 thru -3. These are specifications for various
industrial control applications.
IEC 801-4. The IEC specifies transient voltage impulses
which occur from the switching of inductive loads. We must
be aware of the importance of this specification, especially
in Europe, and characterize our devices’ performance to it.
http://onsemi.com
81
AN784/D
Transient Power Capability
of Zener Diodes
Prepared by
Applications Engineering and
Jerry Wilhardt, Product Engineer —
Industrial and Hi-Rel Zener Diodes
http://onsemi.com
APPLICATION NOTE
INTRODUCTION
Because of the sensitivity of semiconductor components
to voltage transients in excess of their ratings, circuits are
often designed to inhibit voltage surges in order to protect
equipment from catastrophic failure. External voltage
transients are imposed on power lines as a result of lightning
strikes, motors, solenoids, relays or SCR switching circuits,
which share the same ac source with other equipment.
Internal transients can be generated within a piece of
equipment by rectifier reverse recovery transients,
switching of loads or transformer primaries, fuse blowing,
solenoids, etc. The basic relation, v = L di/dt, describes most
equipment developed transients.
P PK(NOM) , NOMINAL PEAK POWER (kW)
1. The surge data may be presented in terms of actual surge
power instead of nominal power.
2. Product improvements have occurred since the data
sheet was published.
Package
Description
1N4728
1W
DO-41
Double Slug Glass
1N6267
5W
Case 41A
Axial Lead Plastic
1N5333A
5W
Case 17
Surmetic 40
1N746/957
A/4371
400 mW
DO-35
Double Slug Glass
1N5221A
500 mW
DO-35
Double Slug Glass
March, 2001 − Rev.0
1N6267 SERIES
5
10
Figure 1. Peak Power Ratings of Zener Diodes
3. Larger dice are used, or special tests are imposed on the
product to guarantee higher ratings than those shown on
Figure 1.
4. The specifications may be based on a JEDEC
registration or part number of another manufacturer.
The data of Figure 1 applies for non-repetitive conditions
and at a lead temperature of 25°C. If the duty cycle increases,
the peak power must be reduced as indicated by the curves
of Figure 2. Average power must be derated as the lead or
ambient temperature rises above 25°C. The average power
derating curve normally given on data sheets may be
normalized and used for this purpose.
At first glance the derating curves of Figure 2 appear to be
in error as the 10 ms pulse has a higher derating factor than
the 10 μs pulse. However, when the derating factor for a
given pulse of Figure 2 is multiplied by the peak power value
of Figure 1 for the same pulse, the results follow the
expected trend.
Some data sheets have surge information which differs
slightly from the data shown in Figure 1. A variety of reasons
exist for this:
© Semiconductor Components Industries, LLC, 2001
2
1
0.5
Power is defined as VZ(NOM) x IZ(PK) where VZ(NOM) is the nominal
zener voltage measured at the low test current used for voltage
classification.
Table 1. Transient Suppressor Diodes
Steady
State Power
20
10
5
5 WATT TYPES
0.2 1 TO 3 W TYPES
0.1
PLASTIC DO-41
0.05
250 mW TO 1 W TYPES
0.02
GLASS DO-35 & GLASS DO-41
0.01
0.05 0.1
0.2
0.5
1
2
0.01 0.02
PULSE WIDTH (ms)
ZENER DIODE CHARACTERISTICS
Zener diodes, being nearly ideal clippers (that is, they
exhibit close to an infinite impedance below the clipping
level and close to a short circuit above the clipping level), are
often used to suppress transients. In this type of application,
it is important to know the power capability of the zener for
short pulse durations, since they are intolerant of excessive
stress.
Some ON Semiconductor data sheets such as the ones for
devices shown in Table 1 contain short pulse surge
capability. However, there are many data sheets that do not
contain this data and Figure 1 is presented here to
supplement this information.
Series
Numbers
100
50
82
Publication Order Number:
AN784/D
AN784/D
CIRCUIT CONSIDERATIONS
It is important that as much impedance as circuit
constraints allow be placed in series with the zener diode and
the components to be protected. The result will be a lower
clipping voltage and less zener stress. A capacitor in parallel
with the zener is also effective in reducing the stress imposed
by very short duration transients.
To illustrate use of the data, a common application will be
analyzed. The transistor in Figure 3 drives a 50 mH solenoid
which requires 5 amperes of current. Without some means
of clamping the voltage from the inductor when the
transistor turns off, it could be destroyed.
When it is necessary to use a zener close to surge ratings,
and a standard part having guaranteed surge limits is not
suitable, a special part number may be created having a surge
limit as part of the specification. Contact your nearest
ON Semiconductor OEM sales office for capability, price,
delivery, and minimum order criteria.
MATHEMATICAL MODEL
DERATING FACTOR
1
0.7
0.5
0.3
0.2
PULSE WIDTH
10 ms
0.1
0.07
0.05
1 ms
0.03
2s
10 μs
0.1
0.2
0.5
1
2
5
10
20
50
100
Used to select a zener diode having the proper voltage
and power capability to protect the transistor.
Figure 3. Circuit Example
D, DUTY CYCLE (%)
Figure 2. Typical Derating Factor for Duty Cycle
The means most often used to solve the problem is to
connect an ordinary rectifier diode across the coil; however,
this technique may keep the current circulating through the
coil for too long a time. Faster switching is achieved by
allowing the voltage to rise to a level above the supply before
being clamped. The voltage rating of the transistor is 60 V,
indicating that approximately a 50 volt zener will be
required.
The peak current will equal the on-state transistor current
(5 amperes) and will decay exponentially as determined by
the coil L/R time constant (neglecting the zener impedance).
A rectangular pulse of width L/R (0.01 sec) and amplitude
of IPK (5 A) contains the same energy and may be used to
select a zener diode. The nominal zener power rating
therefore must exceed (5 A × 50) = 250 watts at 10 ms and
a duty cycle of 0.01/2 = 0.5%. From Figure 2, the duty cycle
factor is 0.62 making the single pulse power rating required
equal to 250/0.62 = 403 watts. From Figure 1, one of the
1N6267 series zeners has the required capability. The
1N6287 is specified nominally at 47 volts and should prove
satisfactory.
Although this series has specified maximum voltage
limits, equation 3 will be used to determine the maximum
zener voltage in order to demonstrate its use.
Since the power shown on the curves is not the actual
transient power measured, but is the product of the peak
current measured and the nominal zener voltage measured
at the current used for voltage classification, the peak current
can be calculated from:
IZ(PK) =
P(PK)
VZ(NOM)
(1)
The peak voltage at peak current can be calculated from:
VZ(PK) = FC × VZ(NOM)
(2)
where FC is the clamping factor. The clamping factor is
approximately 1.20 for all zener diodes when operated at
their pulse power limits. For example, a 5 watt, 20 volt zener
can be expected to show a peak voltage of 24 volts regardless
of whether it is handling 450 watts for 0.1 ms or 50 watts for
10 ms. This occurs because the voltage is a function of
junction temperature and IR drop. Heating of the junction is
more severe at the longer pulse width, causing a higher
voltage component due to temperature which is roughly
offset by the smaller IR voltage component.
For modeling purposes, an approximation of the zener
resistance is needed. It is obtained from:
RZ(NOM) =
VZ(NOM)(FC-1)
PPK(NOM)/VZ(NOM)
26 Vdc
10 ms
100 μs
0.02
0.01
50 mH, 5 Ω
RZ =
(3)
47(1.20 − 1)
500/47
=
9.4
10.64
= 0.9Ω
At 5 amperes, the peak voltage will be 4.5 volts above
nominal or 51.5 volts total which is safely below the 60 volt
transistor rating.
The value is approximate because both the clamping
factor and the actual resistance are a function of temperature.
http://onsemi.com
83
AN843/D
A Review of Transients and
Their Means of Suppression
Prepared by
Steve Cherniak
Applications Engineering
http://onsemi.com
APPLICATION NOTE
INTRODUCTION
One problem that most, if not all electronic equipment
designers must deal with, is transient overvoltages.
Transients in electrical circuits result from the sudden
release of previously stored energy. Some transients may be
voluntary and created in the circuit due to inductive
switching, commutation voltage spikes, etc. and may be
easily suppressed since their energy content is known and
predictable. Other transients may be created outside the
circuit and then coupled into it. These can be caused by
lightning, substation problems, or other such phenomena.
These transients, unlike switching transients, are beyond the
control of the circuit designer and are more difficult to
identify, measure and suppress.
Effective transient suppression requires that the impulse
energy is dissipated in the added suppressor at a low enough
voltage so the capabilities of the circuit or device will not be
exceeded.
di
a voltage equal to L dt . The energy (J) in the transient is equal
to 1/2Li2 and usually exists as a high power impulse for a
relatively short time (J = Pt).
If load 2 is shorted (Figure 1), devices parallel to it may
be destroyed. When the fuse opens and interrupts the fault
current, the slightly inductive power supply produces a
transient voltage spike of V + L di with an energy content of
dt
J = 1/2Li2. This transient might be beyond the voltage
limitations of the rectifiers and/or load 1. Switching out a
high current load will have a similar effect.
TRANSFORMER PRIMARY
BEING ENERGIZED
If a transformer is energized at the peak of the line voltage
(Figure 2), this voltage step function can couple to the stray
capacitance and inductance of the secondary winding and
generate an oscillating transient voltage whose oscillations
depend on circuit inductance and capacitance. This
transient’s peak voltage can be up to twice the peak
amplitude of the normal secondary voltage.
In addition to the above phenomena the capacitively
coupled (CS) voltage spike has no direct relationship with
the turns ratio, so it is possible for the secondary circuit to see
the peak applied primary voltage.
REOCCURRING TRANSIENTS
Transients may be formed from energy stored in circuit
inductance and capacitance when electrical conditions in the
circuit are abruptly changed.
Switching induced transients are a good example of this;
ǒdtǓ
the change in current di in an inductor (L) will generate
FUSE
POWER
SUPPLY
+
+
A
LOAD
1
B
LOAD
2
SHORT
ACROSS
LOAD 2
0
VAB
−
−
Figure 1. Load Dump with Inductive Power Supply
© Semiconductor Components Industries, LLC, 2001
March, 2001 − Rev.0
84
Publication Order Number:
AN843/D
AN843/D
+
CS
VLine
PEAK SHOULD
BE 30% LIGHTER
−
SWITCH
A
LOAD
MAY HAVE STRAY
INDUCTANCE OR
CAPACITANCE
B
+
VAB
SWITCH
VAB − CLOSED
Figure 2. Situation Where Transformer Capacitance Causes a Transient
TRANSFORMER PRIMARY
BEING DE-ENERGIZED
If the transformer is driving a high impedance load,
transients of more than ten times normal voltage can be
created at the secondary when the primary circuit of the
transformer is opened during zero-voltage crossing of the ac
line. This is due to the interruption of the transformer
magnetizing current which causes a rapid collapse of the
magnetic flux in the core. This, in turn, causes a high voltage
transient to be coupled into the transformer’s secondary
winding (Figure 3).
Transients produced by interrupting transformers
magnetizing current can be severe. These transients can
destroy a rectifier diode or filter capacitor if a low
impedance discharge path is not provided.
SWITCH “ARCING”
When a contact type switch opens and tries to interrupt
current in an inductive circuit, the inductance tries to keep
current flowing by charging stray capacitances. (See
Figure 4.)
LINE
VOLTAGE
SWITCH
AC
LINE
MAGNETIZING
CURRENT AND
FLUX
Im
LOAD
SECONDARY
VOLTAGE
SWITCH
OPENED
Figure 3. Typical Situation Showing Possible Transient When Interrupting
Transformer Magnetizing Current
VCap
LINE
VOLTAGE
TRANSIENT
SENSITIVE
LOAD
VLine
Figure 4. Transients Caused by Switch Opening
http://onsemi.com
85
AN843/D
This can also happen when the switch contacts bounce
open after its initial closing. When the switch is opened (or
bounces open momentarily) the current that the inductance
wants to keep flowing will oscillate between the stray
capacitance and the inductance. When the voltage due to the
oscillation rises at the contacts, breakdown of the contact
gap is possible, since the switch opens (or bounces open)
relatively slowly compared to the oscillation frequency, and
the distance may be small enough to permit “arcing.” The
arc will discontinue at the zero current point of the
oscillation, but as the oscillatory voltage builds up again and
the contacts move further apart, each arc will occur at a
higher voltage until the contacts are far enough apart to
interrupt the current.
Vpk
0.9 Vpk
t = 10 μs (f = 100 kHz)
0.1 Vpk
0.5 μs
60% OF Vpk
Figure 5. 0.5 μs 100 kHz Ring Wave
WAVESHAPES OF SURGE VOLTAGES
The oscillating portion of the waveform produces voltage
polarity reversal effects. Some semiconductors are sensitive
to polarity changes or can be damaged when forced into or
out of conduction (i.e. reverse recovery of rectifier devices).
The sensitivity of some semiconductors to the timing and
polarity of a surge is one of the reasons for selecting this
oscillatory waveform to represent actual conditions.
Indoor Waveshapes
Measurements in the field, laboratory, and theoretical
calculations indicate that the majority of surge voltages in
indoor low-voltage power systems have an oscillatory
waveshape. This is because the voltage surge excites the
natural resonant frequency of the indoor wiring system. In
addition to being typically oscillatory, the surges can also
have different amplitudes and waveshapes in the various
places of the wiring system. The resonant frequency can
range from about 5 kHz to over 500 kHz. A 100 kHz
frequency is a realistic value for a typical surge voltage for
most residential and light industrial ac wire systems.
The waveshape shown in Figure 5 is known as an “0.5 μs
− 100 kHz ring wave.” This waveshape is reasonably
representative of indoor low-voltage (120 V − 240 V) wiring
system transients based on measurements conducted by
several independent organizations. The waveshape is
defined as rising from 10% to 90% of its final amplitude in
0.5 μs, then decays while oscillating at 100 kHz, each peak
being 60% of the preceding one.
The fast rise portion of the waveform can induce the
effects associated with non-linear voltage distribution in
windings or cause dv/dt problems in semiconductors.
Shorter rise times can be found in transients but they are
lengthened as they propagate into the wiring system or
reflected from wiring discontinuities.
Outdoor Locations
Both oscillating and unidirectional transients have been
recorded in outdoor environments (service entrances and
other places nearby). In these locations substantial energy is
still available in the transient, so the waveform used to
model transient conditions outside buildings must contain
greater energy than one used to model indoor transient
surges.
Properly selected surge suppressors have a good
reputation of successful performance when chosen in
conjunction with the waveforms described in Figure 6. The
recommended waveshape of 1.2 × 50 μs (1.2 μs is associated
with the transients rise time and the 50 μs is the time it takes
for the voltage to drop to 1/2Vpk) for the open circuit voltage
and 8 × 20 μs for the short circuit current are as defined in
IEEE standard 28-ANSI Standard C62.1 and can be
considered a realistic representation of an outdoor transients
waveshape.
I
V
Ipk
Vpk
0.9 Vpk
0.9 Ipk
0.5 Ipk
0.5 Vpk
0.3 Vpk
0.1 Ipk
0.1 Vpk
t1
50 μs
t2
t1 × 1.67 = 1.2 μs
(a) Open-Circuit Voltage Waveform
20 μs
t2 × 1.25 = 8 μs
(b) Discharge Current Waveform
Figure 6. Unidirectional Wave Shapes
http://onsemi.com
86
AN843/D
branch of a power distribution system and can also be caused
by lightning.
The type of device under test determines which
waveshape in Figure 6 is more appropriate. The voltage
waveform is normally used for insulation voltage withstand
tests and the current waveform is usually used for discharge
current tests.
AC POWER LINE TRANSIENTS
Transients on the ac power line range from just above
normal voltage to several kV. The rate of occurrence of
transients varies widely from one branch of a power
distribution system to the next, although low-level transients
occur more often than high-level surges.
Data from surge counters and other sources is the basis for
the curves shown in Figure 7. This data was taken from
unprotected (no voltage limiting devices) circuits meaning
that the transient voltage is limited only by the sparkover
distance of the wires in the distribution system.
RANDOM TRANSIENTS
The source powering the circuit or system is frequently
the cause of transient induced problems or failures. These
transients are difficult to deal with due to their nature; they
are totally random and it is difficult to define their
amplitude, duration and energy content. These transients are
generally caused by switching parallel loads on the same
20
HIGH EXPOSURE
10
9
8
7
6
5
P
E
A
K
S
U
R
G
E
V
O
L
T
A
G
E
4
MEDIUM EXPOSURE
3
2
LOW EXPOSURE
1
0.9
0.8
0.7
0.6
(kV) 0.5
0.4
0.3
0.2
0.1
0.01
0.1
1
10
SURGES PER YEAR
100
Figure 7. Peak Surge Voltage versus Surges per Year*
*EIA paper, P587.1/F, May, 1979, Page 10
http://onsemi.com
87
1000
AN843/D
open circuit voltages do not change very much throughout
the structure since the wiring does not provide much
attenuation. There are three categories of service locations
that can represent the majority of locations from the
electrical service entrance to the most remote wall outlet.
These are listed below. Table 1 is intended as an aid for the
preliminary selection of surge suppression devices, since it
is very difficult to select a specific value of source
impedance.
The low exposure portion of the set of curves is data
collected from systems with little load-switching activity
that are located in areas of light lightning activity.
Medium exposure systems are in areas of frequent
lightning activity with a severe switching transient problem.
High exposure systems are rare systems supplied by long
overhead lines which supply installations that have high
sparkover clearances and may be subject to reflections at
power line ends.
When using Figure 7 it is helpful to remember that peak
transient voltages will be limited to approximately 6 kV in
indoor locations due to the spacing between conductors
using standard wiring practices.
Category I: Outlets and circuits a “long distance” from
electrical service entrance. Outlets more than 10 meters
from Category II or more than 20 meters from Category
III (wire gauge #14 − #10)
TRANSIENT ENERGY LEVELS AND
SOURCE IMPEDANCE
The energy contained in a transient will be divided between
the transient suppressor and the source impedance of the
transient in a way that is determined by the two impedances.
With a spark-gap type suppressor, the low impedance of the
Arc after breakdown forces most of the transient’s energy to
be dissipated elsewhere, e.g. in a current limiting resistor in
series with the spark-gap and/or in the transient’s source
impedance. Voltage clamping suppressors (e.g. zeners,
mov’s, rectifiers operating in the breakdown region) on the
other hand absorb a large portion of the transient’s surge
energy. So it is necessary that a realistic assumption of the
transient’s source impedance be made in order to be able to
select a device with an adequate surge capability.
The 100 kHz “Ring Wave” shown in Figure 5 is intended
to represent a transient’s waveshape across an open circuit.
The waveshape will change when a load is connected and the
amount of change will depend on the transient’s source
impedance. The surge suppressor must be able to withstand
the current passed through it from the surge source. An
assumption of too high a surge impedance (when testing the
suppressor) will not subject the device under test to
sufficient stresses, while an assumption of too low a surge
impedance may subject it to an unrealistically large stress;
there is a trade-off between the size (cost) of the suppressor
and the amount of protection obtained.
In a building, the transient’s source impedance increases
with the distance from the electrical service entrance, but
Category II: Major bus lines and circuits a “short distance”
from electrical service entrance. Bus system in industrial
plants. Outlets for heavy duty appliances that are “close”
to the service entrance.
Distribution panel devices.
Commercial building lighting systems.
Category III. Electrical service entrance and outdoor
locations.
Power line between pole and electrical service entrance.
Power line between distribution panel and meter.
Power line connection to additional near-by buildings.
Underground power lines leading to pumps, filters, etc.
Categories I and II in Table 1 correspond to the extreme
range of the “medium exposure” curve in Figure 7. The
surge voltage is limited to approximately 6 kV due to the
sparkover spacing of indoor wiring.
The discharge currents of Category II were obtained from
simulated lightning tests and field experience with
suppressor performance.
The surge currents in Category I are less than in Category
II because of the increase in surge impedance due to the fact
that Category I is further away from the service entrance.
Category III can be compared to the “High Exposure”
situation in Figure 7. The limiting effect of sparkover is not
available here so the transient voltage can be quite large.
Table 1. Surge Voltages and Currents Deemed to Represent the Indoor Environment Depending Upon Location
Energy (Joules) Dissipated in a
Suppressor with a Clamping Voltage of (3)
Category
I
II
III
Waveform
0.5 μs 100 kHz
Ring Wave
0.5 μs 100 kHz
Ring Wave
1.2 × 50 μs
8 × 20 μs
1.2 × 50 μs
8 × 20 μs
Surge Voltage(1)
Surge Current(2)
250 V
500 V
1000 V
6 kV
200 A
0.4
0.8
1.6
6 kV
500 A
1
2
4
3 kA
20
40
80
6 kV
10 kV or more
10 kA or more
Notes: 1. Open Circuit voltage
Notes: 2. Discharge current of the surge (not the short circuit current of the power system)
Notes: 3. The energy a suppressor will dissipate varies in proportion with the suppressor’s clamping voltage, which can be different with different system voltages (assuming the same
Notes: 3. discharge current).
http://onsemi.com
88
AN843/D
LIGHTNING TRANSIENTS
There are several mechanisms in which lightning can
produce surge voltages on power distribution lines. One of
them is a direct lightning strike to a primary (before the
substation) circuit. When this high current, that is injected
into the power line, flows through ground resistance and the
surge impedance of the conductors, very large transient
voltages will be produced. If the lightning misses the
primary power line but hits a nearby object the lightning
discharge may also induce large voltage transients on the
line. When a primary circuit surge arrester operates and
limits the primary voltage the rapid dv/dt produced will
effectively couple transients to the secondary circuit through
the capacitance of the transformer (substation) windings in
addition to those coupled into the secondary circuit by
normal transformer action. If lightning struck the secondary
circuit directly, very high currents may be involved which
would exceed the capability of conventional surge
suppressors. Lightning ground current flow resulting from
nearby direct to ground discharges can couple onto the
common ground impedance paths of the grounding
networks also causing transients.
Dump” can cause the most destructive transients; it is when
the battery becomes disconnected from the charging system
during high charging rates. This is not unlikely when one
considers bad battery connections due to corrosion or other
wiring problems. Other problems can exist such as steady
state overvoltages caused by regulator failure or 24 V
battery jump starts. There is even the possibility of incorrect
battery connection (reverse polarity).
Capacitive and/or inductive coupling in wire harnesses as
well as conductive coupling (common ground) can transmit
these transients to the inputs and outputs of automotive
electronics.
The Society of Automotive Engineers (SAE) documented
a table describing automotive transients (see Table 2) which
is useful when trying to provide transient protection.
Considerable variation has been observed while gathering
data on automobile transients. All automobiles have their
electrical systems set up differently and it is not the intent of
this paper to suggest a protection level that is required. There
will always be a trade-off between cost of the suppressor and
the level of protection obtained. The concept of one master
suppressor placed on the main power lines is the most
cost-effective scheme possible since individual suppressors
at the various electronic devices will each have to suppress
the largest transient that is likely to appear (Load Dump),
hence each individual suppressor would have to be the same
size as the one master suppressor since it is unlikely for
several suppressors to share the transient discharge.
AUTOMOTIVE TRANSIENTS
Transients in the automotive environment can range from
the noise generated by the ignition system and the various
accessories (radio, power window, etc.) to the potentially
destructive high energy transients caused by the charging
(alternator/regulator) system. The automotive “Load
Table 2. Typical Transients Encountered in the Automotive Environment
Energy Capability
Length of
Transient
Steady State
5 Minutes
4.5−100 ms
Cause
Voltage Amplitude
∞
Failed Voltage Regulator
+18 V
∞
Booster starts with 24 V battery
±24 V
≥ 10 J
Load Dump — i.e., disconnection of battery during
high charging rates
≤ 0.32 s
Inductive Load Switching Transient
≤ 0.2 s
Alternator Field Decay
90 ms
Ignition Pulse
Disconnected Battery
1 ms
Mutual Coupling in Harness
15 μs
Ignition Pulse Normal
≤ 125 V
<1J
−300 V to +80 V
<1J
−100 V to −40 V
< 0.5 J
≤ 75 V
<1J
≤ 200 V
< 0.001 J
3V
Accessory Noise
≤ 1.5 V
Transceiver Feedback
≈ 20 mV
http://onsemi.com
89
Possible Frequency
of Application
Infrequent
Infrequent
Infrequent
Often
Each Turn-Off
≤ 500 Hz
Several Times in
vehicle life
Often
3 500 Hz
Continuous
50 Hz to 10 kHz
R.F.
AN843/D
intergranular boundaries. These boundaries are the source of
the MOV’s non-linear electrical behavior.
MOV electrical characteristics are mainly controlled by
the physical dimensions of the polycrystalline structure
since conduction occurs between the zinc oxide grains and
the intergranular boundaries which are distributed
throughout the bulk of the device.
The MOV polycrystalline body is usually formed into the
shape of a disc. The energy rating is determined by the
device’s volume, voltage rating by its thickness, and current
handling capability by its area. Since the energy dissipated
in the device is spread throughout its entire metal oxide
volume, MOV’s are well suited for single shot high power
transient suppression applications where good clamping
capability is not required.
The major disadvantages with using MOV’s are that they
can only dissipate relatively small amounts of average
power and are not suitable for many repetitive applications.
Another drawback with MOV’s is that their voltage
clamping capability is not as good as zeners, and is
insufficient in many applications.
Perhaps the major difficulty with MOV’s is that they have
a limited life time even when used below their maximum
ratings. For example, a particular MOV with a peak current
handling capability of 1000 A has a lifetime of about 1 surge
at 1000 Apk, 100 surges at 100 Apk and approximately 1000
surges at 65 Apk.
There will, of course, be instances where a need for
individual suppressors at the individual accessories will be
required, depending on the particular wiring system or
situation.
TRANSIENT SUPPRESSOR TYPES
Carbon Block Spark Gap
This is the oldest and most commonly used transient
suppressor in power distribution and telecommunication
systems. The device consists of two carbon block electrodes
separated by an air gap, usually 3 to 4 mils apart. One
electrode is connected to the system ground and the other to
the signal cable conductor. When a transient over-voltage
appears, its energy is dissipated in the arc that forms between
the two electrodes, a resistor in series with the gap, and also
in the transient’s source impedance, which depends on
conductor length, material and other parameters.
The carbon block gap is a fairly inexpensive suppressor
but it has some serious problems. One is that it has a
relatively short service life and the other is that there are
large variations in its arcing voltage. This is the major
problem since a nominal 3 mil gap will arc anywhere from
300 to 1000 volts. This arcing voltage variation limits its use
mainly to primary transient suppression with more accurate
suppressors to keep transient voltages below an acceptable
level.
Gas Tubes
The gas tube is another common transient suppressor,
especially in telecommunication systems. It is made of two
metallic conductors usually separated by 10 to 15 mils
encapsulated in a glass envelope which is filled with several
gases at low pressure. Gas tubes have a higher current
carrying capability and longer life than carbon block gaps.
The possibility of seal leakage and the resultant of loss
protection has limited the use of these devices.
TRANSIENT SUPPRESSION
USING ZENERS
Zener diodes exhibit a very high impedance below the
zener voltage (VZ), and a very low impedance above VZ.
Because of these excellent clipping characteristics, the zener
diode is often used to suppress transients. Zeners are
intolerant of excessive stress so it is important to know the
power handling capability for short pulse durations.
Most zeners handle less than their rated power during
normal applications and are designed to operate most
effectively at this low level. Zener transient suppressors
such as the ON Semiconductor 1N6267 Mosorb series are
designed to take large, short duration power pulses.
This is accomplished by enlarging the chip and the
effective junction area to withstand the high energy surges.
The package size is usually kept as small as possible to
provide space efficiency in the circuit layout, and since the
package does not differ greatly from other standard zener
packages, the steady state power dissipation does not differ
greatly.
Some data sheets contain information on short pulse surge
capability. When this information is not available for ON
Semiconductor devices, Figure 8 can be used. This data
applies for non-repetitive conditions with a lead temperature
of 25°C.
It is necessary to determine the pulse width and peak
power of the transient being suppressed when using
Figure 8. This can be done by taking whatever waveform
Selenium Rectifiers
Selenium transient suppressors are selenium rectifiers
used in the reverse breakdown mode to clamp voltage
transients. Some of these devices have self-healing
properties which allows the device to survive energy
discharges greater than their maximum capability for a
limited number of surges. Selenium rectifiers do not have
the voltage clamping capability of zener diodes. This is
causing their usage to become more and more limited.
METAL OXIDE VARISTORS (MOV’S)
An MOV is a non-linear resistor which is voltage
dependent and has electrical characteristics similar to
back-to-back zener diodes. As its name implies it is made up
of metal oxides, mostly zinc oxide with other oxides added
to control electrical characteristics. MOV characteristics are
compared to back-to-back zeners in Photos 2 through 7.
When constructing MOV’s the metal oxides are sintered
at high temperatures to produce a polycrystalline structure
of conductive zinc oxide separated by highly resistive
http://onsemi.com
90
AN843/D
P PK(NOM) , NOMINAL PEAK POWER (kW)
the transient is and approximating it with a rectangular pulse
with the same peak power. For example, an exponential
discharge with a 1 ms time constant can be approximated by
a rectangular pulse 1 ms wide that has the same peak power
as the transient. This would be a better approximation than
a rectangular pulse 10 ms wide with a correspondingly lower
amplitude. This is because the heating effects of different
pulse width lengths affect the power handling capability, as
can be seen by Figure 8. This also represents a conservative
approach because the exponential discharge will contain ≈
1/2 the energy of a rectangular pulse with the same pulse
width and amplitude.
Eqtn 1: VZ(pK) = FC (VZ(nom))
Unless otherwise specified FC is approximately 1.20 for
zener diodes when operated at their pulse power limits.
For example, a 5 watt, 20 volt zener can be expected to
show a peak voltage of 24 volts regardless of whether it is
handling 450 watts for 0.1 ms or 50 watts for 10 ms. (See
Figure 8.)
This occurs because the zener voltage is a function of both
junction temperature and IR drop. Longer pulse widths
cause a greater junction temperature rise than short ones; the
increase in junction temperature slightly increases the zener
voltage. This increase in zener voltage due to heating is
roughly offset by the fact that longer pulse widths of
identical energy content have lower peak currents. This
results in a lower IR drop (zener voltage drop) keeping the
clamping factor relatively constant with various pulse
widths of identical energy content.
An approximation of zener impedance is also helpful in
the design of transient protection circuits. The value of
RZ(nom) (Eqtn 2) is approximate because both the clamping
factor and the actual resistance is a function of temperature.
100
50
20
10
5
1N6267 SERIES
2
1
0.5
5 WATT TYPES
0.2 1 TO 3 W TYPES
0.1
PLASTIC DO-41
0.05
250 mW TO 1 W TYPES
0.02
GLASS DO-35 & GLASS DO-41
0.01
0.01 0.02
0.05 0.1 0.2
0.5
1
2
Eqtn 2: RZ(nom) =
5
Figure 8. Peak Power Ratings of Zener Diodes
When used in repetitive applications, the peak power must
be reduced as indicated by the curves of Figure 9. Average
power must be derated as the lead or ambient temperature
exceeds 25°C. The power derating curve normally given on
data sheets can be normalized and used for this purpose.
ZENER versus MOV TRADEOFFS
The clamping characteristics of Zeners and MOV’s are
best compared by measuring their voltages under transient
conditions. Photos 1 through 9 are the result of an
experiment that was done to compare the clamping
characteristics of a Zener (ON Semiconductor 1N6281,
approximately 1.5J capability) with those of an MOV (G.E.
V27ZA4, 4J capability); both are 27 V devices.
Photo 1 shows the pulse generator output voltage. This
generator synthesizes a transient pulse that is characteristic
of those that may appear in the real world.
Photos 2 and 3 are clamping voltages of the MOV and
Zener, respectively with a surge source impedance of
500 Ω.
Photos 4 and 5 are the clamping voltages with a surge
source impedance of 50 Ω .
Photos 6 and 7 simulate a condition where the surge
source impedance is 5 Ω .
Photos 8 and 9 show a surge source impedance of 0.55 Ω,
which is at the limits of the Zener suppressor’s capability.
DERATING FACTOR
1
0.7
0.5
0.3
0.2
PULSE WIDTH
10 ms
1 ms
0.03
100 μs
0.02
10 μs
0.01
0.1
0.2
0.5
1
2
5
10
20
PpK(nom)
VZ(nom) = Nominal Zener Voltage
PpK(nom) = Found from Figure 8 when device type and
pulse width are known. For example, from Figure 8 a
1N6267 zener suppressor has a PpK(nom) of 1.5 kW at a
pulse width of 1 ms.
As seen from equation 2, zeners with a larger PpK(nom)
capability will have a lower RZ(nom).
10
PULSE WIDTH (ms)
0.1
0.07
0.05
V2Z(nom) (FC −1)
50
D, DUTY CYCLE (%)
Figure 9. Typical Derating Factor for Duty Cycle
The peak zener voltage during the peak current of the
transient being suppressed can be related to the nominal
zener voltage (Eqtn 1) by the clamping factor (FC).
http://onsemi.com
91
AN843/D
PHOTO 1
Open Circuit Transient Pulse
Vert: 20 V/div
Horiz: 0.5 ms/div
Vpeak = 90 V
10
09
0
1
00
%
PHOTO 2
MOV (27 V)
Vert: 10 V/div
Horiz: 0.5 ms/div
Transient Source Impedance: 500 Ω
Vpeak = 39.9 V
10
09
0
1
00
%
http://onsemi.com
92
AN843/D
PHOTO 3
Zener (27 V)
Vert: 10 V/div
Horiz: 0.5 ms/div
Transient Source Impedance: 500 Ω
Vpeak = 27 V
10
09
0
1
00
%
PHOTO 4
MOV (27 V)
Vert: 10 V/div
Horiz: 0.5 ms/div
Transient Source Impedance: 50 Ω
Vpeak = 44.7 V
10
09
0
1
00
%
PHOTO 5
Zener (27 V)
Vert: 10 V/div
Horiz: 0.5 ms/div
Transient Source Impedance: 50 Ω
Vpeak = 27 V
10
09
0
1
00
%
http://onsemi.com
93
AN843/D
PHOTO 6
MOV (27 V)
Vert: 10 V/div
Horiz: 0.5 ms/div
Transient Source Impedance: 5 Ω
Vpeak = 52 V
10
09
0
1
00
%
PHOTO 7
Zener (27 V)
Vert: 10 V/div
Horiz: 0.5 ms/div
Transient Source Impedance: 5 Ω
Vpeak = 28 V
10
09
0
1
00
%
PHOTO 8
MOV (27 V)
Vert: 10 V/div
Horiz: 0.5 ms/div
Transient Source Impedance: 0.55 Ω
Vpeak = 62.5 V
http://onsemi.com
94
AN843/D
PHOTO 9
Zener (27 V)
Vert: 10 V/div
Horiz: 0.5 ms/div
Transient Source Impedance: 0.55 Ω
Vpeak: 30.2 V
Peak Power: Approx 2000 Wpeak
(The limit of this device’s capability)
be placed as close to the load as possible to minimize
overshoot due to wiring (or any inductive) effect. (See
Figure 11b.)
As can be seen by the photographs, the Zener suppressor
has significantly better voltage clamping characteristics
than the MOV even though that particular Zener has less
than one-fourth the energy capability of the MOV it was
compared with. However, the energy rating can be
misleading because it is based on the clamp voltage times the
surge current, and when using an MOV, the high impedance
results in a fairly high clamp voltage. The major tradeoff
with using a zener type suppressor is its cost versus power
handling capability, but since it would take an “oversized”
MOV to clamp voltages (suppress transients) as well as the
zener, the MOV begins to lose its cost advantage.
If a transient should come along that exceeds the
capabilities of the particular Zener, or MOV, suppressor that
was chosen, the load will still be protected, since they both
fail short.
The theoretical reaction time for Zeners is in the
picosecond range, but this is slowed down somewhat with
lead and package inductance. The 1N6267 Mosorb series of
transient suppressors have a typical response time of less
than one nanosecond. For very fast rising transients it is
important to minimize external inductances (due to wiring,
etc.) which will minimize overshoot.
Connecting Zeners in a back-to-back arrangement will
enable bidirectional voltage clamping characteristics. (See
Figure 10.)
If Zeners A and B are the same voltage, a transient of
either polarity will be clamped at approximately that voltage
since one Zener will be in reverse bias mode while the other
will be in the forward bias mode. When clamping low
voltage it may be necessary to consider the forward drop of
the forward biased Zener.
The typical protection circuit is shown in Figure 11a. In
almost every application, the transient suppression device is
placed in parallel with the load, or component to be
protected. Since the main purpose of the circuit is to clamp
the voltage appearing across the load, the suppressor should
A
A
OR
B
B
Figure 10. Zener Arrangement for
Bidirectional Clamping
ÎÎÎ
ÎÎÎ
Zin
LOAD
VL
Vin
Figure 11a. Using Zener to Protect Load
Against Transients
TRANSIENT
INPUT
ZENER
VOLTAGE
PEAK VOLTAGE
DUE TO OVERSHOOT
Figure 11b. Overshoot Due to Inductive Effect
http://onsemi.com
95
AN843/D
These transients may be generated by normal circuit
operations such as inductive switching circuits, energizing
and deenergizing transformer primaries, etc. They do not
present much of a problem since their energy content,
duration and effect may easily be obtained and dealt with.
Random transients found on power lines, or lightning
transients, present a greater threat to electronic components
since there is no way to be sure when or how severe they will
be. General guidelines were discussed to aid the circuit
designer in deciding what size (capability and cost)
suppressor to choose for a certain level of protection. There
will always be a tradeoff between suppressor price and
protection obtained.
Several different suppression devices were discussed with
emphasis on Zeners and MOV’s, since these are the most
popular devices to use in most applications.
Zener capacitance prior to breakdown is quite small (for
example, the 1N6281 27 Volt Mosorb has a typical
capacitance of 800 pF). Capacitance this small is desirable
in the off-state since it will not attenuate wide-band signals.
When the Zener is in the breakdown mode of operation
(e.g. when suppressing a transient) its effective capacitance
increases drastically from what it was in the off-state. This
makes the Zener ideal for parallel protection schemes since,
during transient suppression, its large effective capacitance
will tend to hold the voltage across the protected element
constant; while in the off-state (normal conditions, no
transient present), its low off-state capacitance will not
attenuate high frequency signals.
Input impedance (Zin) always exists due to wiring and
transient source impedance, but Zin should be increased as
much as possible with an external resistor, if circuit
constraints allow. This will minimize Zener stress.
REFERENCES
1. GE Transient Voltage Suppression Manual, 2nd
edition.
CONCLUSION
The reliable use of semiconductor devices requires that
the circuit designer consider the possibility of transient
overvoltages
destroying
these
transient-sensitive
components.
2. ON Semiconductor Zener Diode Manual.
http://onsemi.com
96
DESIGN CONSIDERATIONS AND PERFORMANCE
OF ON SEMICONDUCTOR TEMPERATURE-COMPENSATED
ZENER (REFERENCE) DIODES
Prepared by
Zener Diode Engineering
and
Ronald N. Racino
Reliability and Quality Assurance
INTRODUCTION
6.4
This application note defines ON Semiconductor
temperature-compensated zener (reference) diodes,
explains the device characteristics, describes electrical
testing, and discusses the advanced concepts of device
reliability and quality assurance. It is a valuable aid to those
who contemplate designing circuits requiring the use of
these devices.
Zener diodes fall into three general classifications:
Regulator diodes, reference diodes and transient voltage
suppressors. Regulator diodes are normally employed in
power supplies where a nearly constant dc output voltage is
required despite relatively large changes in input voltage or
load resistance. Such devices are available with a wide range
of voltage and power ratings, making them suitable for a
wide variety of electronic equipments.
Regulator diodes, however, have one limitation: They are
temperature-sensitive. Therefore, in applications in which
the output voltage must remain within narrow limits during
input-voltage, load-current, and temperature changes, a
temperature-compensated regulator diode, called a
reference diode, is required.
The reference diode is made possible by taking advantage
of the differing thermal characteristics of forward- and
reverse-biased silicon p-n junctions. A forward-biased
junction has a negative temperature coefficient of
approximately 2 mV/°C, while reverse-biased junctions
have positive temperature coefficients ranging from about
2 mV/°C at 5.5 V to 6 mV/°C at 10 V. Therefore it is
possible, by judicious combination of forward- and
reverse-biased junctions, to fabricate a device with a very
low overall temperature coefficient (Figure 1).
The principle of temperature compensation is further
illustrated in Figure 2, which shows the voltage-current
characteristics at two temperature points (25 and 100°C) for
both a forward- and a reverse-biased junction. The diagram
shows that, at the specified test current (IZT), the absolute
value of voltage change (ΔV) for the temperature change
between 25 and 100°C is the same for both junctions.
Therefore, the total voltage across the combination of these
two junctions is also the same at these temperature points,
since one ΔV is negative and the other is positive. However,
the rate of voltage change with temperature over the
6.2
6.2 − VOLT REFERENCE DIODE
(COMBINATION OF ZENER
AND FORWARD DICE)
6
5.8
DIODE VOLTAGE DROP (V)
(REFERENCED TO IZT = 7.5 mA)
ZENER DIE
5.6
5.4
0.8
0.6
FORWARD-BIASED COMPENSATING DIE
0.4
0.2
0
−75
−50
−25
0
+25
+50
TEMPERATURE (°C)
+75
+100
Figure 1. Temperature Compensation
of a 6.2 Volt Reference Diode (1N821 Series)
temperature range defined by these points is not necessarily
the same for both junctions, thus the temperature
compensation may not be linear over the entire range.
Figure 2 also indicates that the voltage changes of the two
junctions are equal and opposite only at the specified test
current. For any other value of current, the temperature
compensation may not be complete.
http://onsemi.com
97
2. Voltage-Temperature Stability. The temperature
stability of zener voltage is sometimes expressed by means
of the temperature coefficient. This parameter is usually
defined as the percent voltage change across the device per
degree centigrade. This method of indicating voltage
stability accurately reflects the voltage deviation at the test
temperature extremes but not necessarily at other points
within the specified temperature range. This fact is due to
variations in the rate of voltage change with temperature for
the forward- and reverse-biased dice of the reference diode.
Therefore, the temperature coefficient is given in
ON Semiconductor data sheets only as a quick reference,
for designers who are accustomed to this method of
specification.
A more meaningful way of defining temperature stability
is the “box method.” This method, used by
ON Semiconductor, guarantees that the zener voltage will
not vary by more than a specified amount over a specified
temperature range at the indicated test current, as verified by
tests at several temperatures within this range.
Some devices are accurately compensated over a wide
temperature range (−55°C to 100°C), others over a narrower
range (0 to 75°C). The wide-range devices are, as a rule,
more expensive. Therefore, it would be economically
wasteful for the designer to specify devices with a
temperature range much wider than actually required for the
specific device application.
During actual production of reference diodes, it is difficult
to predict the compensation accuracy. In the interest of
maximum economy, it is common practice to test all devices
coming off the production line, and to divide the production
lot into groups, each with a specified maximum ΔVZ. Each
group, then, is given a different device type number.
DIRECTION OF CURRENT FLOW
PACKAGE
OUTLINE
25°C
+
+ΔV
100°C
I
IZT
VR
VF
−ΔV
I
25°C
IZT
−
100°C
FORWARD-BIASED
P-N JUNCTION
REVERSE-BIASED
ZENER JUNCTION
Figure 2. Temperature Compensation of
P-N Junctions
IMPORTANT ELECTRICAL CHARACTERISTICS
OF REFERENCE DIODES
The three most important characteristics of reference
diodes are 1) reference voltage, 2) voltage-temperature
stability, and 3) voltage-time stability.
1. Reference Voltage. This characteristic is defined as the
voltage drop measured across the diode when the specified
test current passes through it in the zener direction. It is also
called the zener voltage (VZ, Figure 3). On the data sheets,
the reference voltage is given as a nominal voltage for each
family of reference diodes.
The nominal voltages are normally specified to a
tolerance of ±5%, but devices with tighter tolerances, such
as ±2% and ±1%, are available on special order.
I F (mA)
0.3
0.2
0.1
VR (V)
−4
VZ
−2
2
IZ
4
6
8
10
12
VF (V)
−0.1
−0.2
I R (mA)
−6
−0.3
Figure 3. Typical Voltage − Current Characteristic of Reference Diodes
http://onsemi.com
98
14
16
The reason that the device reference voltage may change
in either the negative or positive direction is that after
assembly, some of the devices within a lot may be
overcompensated while others may be undercompensated.
In any design, the “worst-case” condition must be
considered. Therefore, in the above example, it can be
assumed that the maximum voltage change will not exceed
31 mV.
It should be understood, however, that the above
calculations give the maximum possible voltage change for
the device type, and by no means the actual voltage change
for the individual unit.
3. Voltage-Time Stability. The voltage-time stability of
a reference diode is defined by the voltage change during
operating time at the standard test current (IZT) and test
temperature (TA). In general, the voltage stability of a
reference diode is better than 100 ppm per 1000 hours of
operation.
On the data sheet, the voltage-temperature characteristics
of the most widely used device types are illustrated in a
graph similar to the one shown in Figure 4. The particular
production line represented in this figure produces 6.2 volt
devices, but the line yields five different device type
numbers (1N821 through 1N829), each with a different
temperature coefficient. The 1N829, for example, has a
maximum voltage change of less than 5 mV over a
temperature range of −55 to +100°C, while the 1N821 may
have a voltage change of up to 96 mV over the same
temperature range.
100
1N821,A
IZT = 7.5 mA
1N823,A
50 ΔVZ = +31 mV
1N825,A
25
10
−55°C
+100°C
1N827,A
9
1N829,A
0
I Z , ZENER CURRENT (mA)
ΔVZ , MAXIMUM VOLTAGE CHANGE (mV)
(Referenced to −55°C)
75
1N827,A
−25
1N825,A
−50 ΔVZ = −31 mV
1N823,A
−75
1N821,A
−100
−55
0
50
+25°C
8
6
+25°C
+100°C
5
4
−75
100
IZT
7.5
7
−55°C
−50
−25
0
25
50
ΔVZ, MAXIMUM VOLTAGE CHANGE (mV)
(Referenced to IZT = 7.5 mA)
In the past, design data and characteristic curves on data sheets for
reference diodes have been somewhat limited: The devices have
been characterized principally at the recommended operating point.
ON Semiconductor has introduced a data sheet, providing device
data previously not available, and showing limit curves that permit
worst-case circuit design without the need for associated tests required in conjunction with the conventional data sheets.
Figure 5. Current Dependence of Zener Voltage
at Various Temperatures
(1N821 Series)
THE EFFECT OF CURRENT VARIATION ON
ZENER VOLTAGE
Figure 4. Temperature Dependence
of Zener Voltage (1N821 Series)
The nominal zener voltage of a reference diode is
specified at a particular value of current, called the zener test
current (IZT). All measurements of voltage change with
temperature are referenced to this test current. If the
operating current is varied, all these specifications will
change.
The effect of current variation on zener voltage, at various
temperatures, is graphically illustrated on the 1N821 data
sheet as “Zener Current versus Maximum Voltage Change.”
A typical example of such a graph is shown for the 1N821
series in Figure 5. The voltage change shown is due entirely
to the impedance of the device at the fixed temperature. It
does not reflect the change in reference voltage due to the
change in temperature since each curve is referenced to IZT
= 7.5 mA at the indicated temperature. As shown, the
Graphs such as these permit the selection of the
lowest-cost device that meets a particular requirement. They
also permit the designer to determine the maximum voltage
change of a particular reference diode for a relatively small
change in temperature. This is done by drawing vertical lines
from the desired temperature points at the abscissa of the
graph to intersect with each the positive- and negative-going
curves of the particular device of interest. Horizontal lines
are then drawn from these intersects to the ordinate of the
graph. The difference between the intersections of these
horizontal lines with the ordinate yields the maximum
voltage change over the temperature increment. For
example, for the 1N821, a change in ambient temperature
from 0 to 50°C results in a voltage change of no more than
about ±31 mV.
http://onsemi.com
99
OTHER CHARACTERISTICS
greatest voltage change occurs at the highest temperature
represented in the diagram. (See “Dynamic Impedance”
under the next section).
Figure 5 shows that, at 25°C, a change in zener current
from 4 to 10 mA causes a voltage shift of about 90 mV.
Comparing this value with the voltage-change example in
Figure 4 (31 mV), it is apparent that, in general, a greater
voltage variation may be due to current fluctuations than to
temperature change. Therefore, good current regulation of
the source should be a major consideration when using
reference diodes in critical applications.
It is not essential, however, that a reference diode be
operated at the specified test current. The new
voltage-temperature characteristics for a change in current
can be obtained by superimposing the data of Figure 5 on
that of Figure 4. A new set of characteristics, at a test current
of 4 mA, is shown for the 1N823 in Figure 6, together with
the original characteristics at 7.5 mA.
In addition to the three major characteristics discussed
earlier, the following parameters and ratings of reference
diodes may be considered in some applications.
Power Dissipation
The maximum dc power dissipation indicates the power
level which, if exceeded, may result in the destruction of the
device. Normally a device will be operated near the
specified test current for which the data-sheet specifications
are applicable. This test current is usually much below the
current level associated with the maximum power
dissipation.
Dynamic Impedance
Zener impedance may be construed as composed of a
current-dependent
resistance
shunted
by
a
voltage-dependent capacitance. Figure 7 indicates the
typical variations of dynamic zener impedance (ZZ) with
current and temperature for the 1N821 reference diode
series. These diagrams are given in the 1N821 data sheet. As
shown, the zener impedance decreases with current but
increases with ambient temperature.
+50
7.5 mA
0
1000
800
600
400
Z Z , MAXIMUM ZENER IMPEDANCE (OHMS)
Δ VZ (mV) (REFERENCED TO −55° C)
+100
−50
4 mA
−100
−150
−200
−50
0
50
100
TEMPERATURE (°C)
Figure 6. Voltage Change with Temperature
for 1N823 at Two Different Current Levels
200
100
80
60
40
20
10
8
6
4
2
100°C
−55°C
25°C
1
1
2
4
6
8 10
20
40
60 80 100
IZ, ZENER CURRENT (mA)
From these characteristics, it is evident that the voltage
change with temperature for the new curves is different from
that for the original ones. It is also apparent that if the test
current varies between 7.5 and 4 mA, the voltage changes
would lie along the dashed lines belonging to the given
temperature points. This clearly shows the need for a
well-regulated current source.
It should be noted, however, that even when a
well-regulated current supply is available, other factors
might influence the current flowing through a reference
diode. For example, to minimize the effects of
temperature-sensitive passive elements in the load circuit on
current regulation, it is desirable that the load in parallel with
the reference diode have an impedance much higher than the
dynamic impedance of the reference diode.
Figure 7. Variation of Zener Impedance
With Current and Temperature (1N821 Series)
The impedance of a reference diode is normally specified
at the test current (IZT). It is determined by measuring the ac
voltage drop across the device when a 60 Hz ac current with
an rms value equal to 10% of the dc zener current is
superimposed on the zener current (IZT). Figure 8 shows the
block diagram of a circuit used for testing zener impedance.
http://onsemi.com
100
DC POWER
SUPPLY
1k
A
HP
ǒ88i8.8
Ǔ
712A
8888
600
READ
(A)
10M
SET
E1
Rx=
SIGNAL
GENERATOR
0.1μF
1k
E1−E2
E2
S1
READ
Rx=Zz
E2
SET
METER
DUMMY
LOAD
100pF
(B)
R2
1
HP
ǒ88i8.8
Ǔ
650A
8888
AC
VTVM
HP
ǒ88i8.8
Ǔ
400H
8888
DC
VTVM
HP
ǒ88i8.8
Ǔ
412A
8888
Figure 8. Block Diagram of Test Circuit for Measuring Dynamic Zener Impedance
ELECTRICAL TESTING
device design and process development stages to establish
a firm foundation for a comprehensive reliability program.
The design is then evaluated by thorough reliability testing,
and the results are supplied to the Design Engineering
department. This closed-loop feedback procedure provides
valuable information necessary to improve important design
features such as electrical instability due to surface effects,
mechanical strength, and uniformly low thermal resistance
between the die and ambient environment.
All devices are tested electrically as a last step in the
manufacturing process.
The subsequent final test procedures represent an
automated and accurate method of electrically classifying
reference diodes. First, an electrical test is performed on all
devices to insure the correct voltage-breakdown and
stability characteristics. Next, the breakdown voltage and
dynamic impedance are measured. Finally, the devices are
placed in an automatic data acquisition system that
automatically cycles them through the complete
temperature range specified. The actual voltage
measurements at the various temperature points are retained
in the system computer memory until completion of the full
temperature excursion. The computer then calculates the
changes in voltage for each device at each test temperature
and classifies all units on test into the proper category. The
system provides a printed readout for every device,
including the voltage changes to five digits during
temperature cycling, and the corresponding EIA type
number, as well as the data referring to test conditions such
as device position, lot number, and date.
Process Control
There are more than 2000 variables that must be kept
under control to fabricate a reliable reference diode. The
in-process quality control group controls most of these
variables. It places a strict controls on all aspects of
manufacturing from materials procurement to the finished
product. Included in this broad spectrum of controls are:
• Materials Control. All materials purchased or
fabricated in-plant are checked against rigid specifications.
A quality check on vendors’ products is kept up to date to
insure that only materials of a proven quality level will be
purchased.
• In-Process Inspection and Control. Numerous on-line
inspection stations maintain a statistical process control
program on specific manufacturing processes. If any of
these processes are found to be out of control, the discrepant
material is diverted from the normal production flow and the
cognizant design engineer notified. Corrective action is
initiated to remedy the cause of the discrepancy.
DEVICE RELIABILITY AND
QUALITY ASSURANCE
Insuring a very low failure rate requires maximum
performance in all areas effecting device reliability: Device
design, manufacturing processes, quality control, and
reliability testing. ON Semiconductor’s basic reliability
concept is based on the belief that reference diode reliability
is a complex yet controllable function of all these variables.
Under this “total reliability” concept, ON Semiconductor
can mass-produce high-reliability reference diodes.
The reliability of a reference diode fundamentally
depends upon the device design, regardless of the degree of
effort put into device screening and circuit designing.
Therefore, reliability measures must be incorporated at the
Reliability Testing
The Reliability Engineering group evaluates all new
products and gives final conclusions and recommendations
to the device design engineer. The Reliability Engineering
group also performs independent testing of all products and
includes, as part of this testing program,
step-stress-to-failure testing to determine the maximum
capabilities of the product.
http://onsemi.com
101
SOME STRAIGHT TALK ABOUT MOSORBS
TRANSIENT VOLTAGE SUPPRESSORS
IMPORTANT SPECIFICATIONS FOR
MOSORB PROTECTIVE DEVICES
INTRODUCTION
Distinction is sometimes made between devices
trademarked Mosorb (by ON Semiconductor Inc.), and
standard zener/avalanche diodes used for reference,
low-level regulation and low-level protection purposes. It
must be emphasized from the beginning that Mosorb
devices are, in fact, zener diodes. The basic semiconductor
technology and processing are identical. The primary
difference is in the applications for which they are designed.
Mosorb devices are intended specifically for transient
protection purposes and are designed, therefore, with a large
effective junction area that provides high pulse power
capability while minimizing the total silicon use. Thus,
Mosorb pulse power ratings begin at 500 watts − well in
excess of low power conventional zener diodes which in
many cases do not even include pulse power ratings among
their specifications.
MOVs, like Mosorbs, do have the pulse power
capabilities for transient suppression. They are metal oxide
varistors (not semiconductors) that exhibit bidirectional
avalanche characteristics, similar to those of back-to-back
connected zeners. The main attributes of such devices are
low manufacturing cost, the ability to absorb high energy
surges (up to 600 joules) and symmetrical bidirectional
“breakdown” characteristics. Major disadvantages are: high
clamping factor, an internal wear-out mechanism and an
absence of low-end voltage capability. These limitations
restrict the use of MOVs primarily to the protection of
insensitive electronic components against high energy
transients in application above 20 volts, whereas, Mosorbs
are best suited for precise protection of sensitive equipment
even in the low voltage range − the same range covered by
conventional zener diodes. The relative features of the two
device types are covered in Table 1.
Typically, a Mosorb suppressor is used in parallel with the
components or circuits being protected (Figure 1), in order
to shunt the destructive energy spike, or surge, around the
more sensitive components. It does this by avalanching at its
“breakdown” level, ideally representing an infinite
impedance at voltages below its rated breakdown voltage,
and essentially zero impedance at voltages above this level.
In the more practical case, there are three voltage
specifications of significance, as shown in Figure 1a.
a) VRWM is the maximum reverse stand-off voltage at
which the Mosorb is cut off and its impedance is at its
highest value − that is, the current through the device is
essentially the leakage current of a back-biased diode.
b) V(BR) is the breakdown voltage − a voltage at which the
device is entering the avalanche region, as indicated by
a slight (specified) rise in current beyond the leakage
current.
c) VRSM is the maximum reverse voltage (clamping
voltage) which is defined and specified in conjunction
with the maximum reverse surge current so as not to
exceed the maximum peak power rating at a pulse
width (tp) of 1 ms (industry std time for measuring
surge capability).
RELATIVE FEATURES OF MOVs and MOSORBS
Table 1.
MOV
Mosorb/Zener Transient Suppressor
• High clamping factor.
• Symmetrically bidirectional.
• Energy capability per dollar usually higher than a silicon device.
However, if good clamping is required the energy capability would
have to be grossly overspecified resulting in higher cost.
• Inherent wear out mechanism clamp voltage degrades after every
pulse, even when pulsed below rated value.
• Ideally suited for crude ac line protection.
• High single-pulse current capability.
• Degrades with overstress.
• Good high voltage capability.
• Limited low voltage capability.
• Very good clamping close to the operating voltage.
• Standard parts perform like standard zeners. Symmetrical bidirectional devices available for many voltages.
• Good clamping characteristic could reduce overall system cost.
• No inherent wear out mechanism.
•
•
•
•
•
Ideally suited for precise DC protection.
Medium multiple-pulse current capability.
Fails short with overstress.
Limited high voltage capability unless series devices are used.
Good low voltage capability.
http://onsemi.com
102
In the data sheets of some protective devices, the
parameter for response time is emphasized. Response time
on these data sheets is defined as the time required for the
voltage across the protective device to rise from 0 to V(BR),
and relates primarily to the effective series impedance
associated with the device. This effective impedance is
somewhat complex and changes drastically from the
blocking mode to the avalanche mode. In most applications
(where the protective device shunts the load) this response
time parameter becomes virtually meaningless as indicated
by the waveforms in Figures 1b and 1c. If the response time
as defined is very long, it still would not affect the
performance of the surge suppressor.
However, if the series inductance becomes appreciable, it
could result in “overshoot” as shown in Figure 1d that would
be detrimental to circuit protection. In Mosorb devices,
series inductance is negligible compared to the inductive
effects of the external circuitry (primarily lead lengths).
Hence, Mosorbs contribute little or nothing to overshoot
and, in essence, the parameter of response time has very little
significance. However, care must be exercised in the design
of the external circuitry to minimize overshoot.
In practice, the Mosorb is selected so that its VRWM is
equal to or somewhat higher than the highest operating
voltage required by the load (the circuits or components to
be protected). Under normal conditions, the Mosorb is
inoperative and dissipates very little power. When a
transient occurs, the Mosorb converts to a very low dynamic
impedance and the voltage across the Mosorb becomes the
clamping voltage at some level above V(BR). The actual
clamping level will depend on the surge current through the
Mosorb. The maximum reverse surge current (IRSM) is
specified on the Mosorb data sheets at 1 ms and for a
logrithmically decaying pulse waveform. The data sheet
also contains curves to determine the maximum surge
current rating at other time intervals.
Typically, Mosorb devices have a built-in safety margin at
the maximum rated surge current because the clamp voltage,
VRSM, is itself, guardbanded. Thus, the parts will be
operating below their maximum pulse-power (Ppk) rating
even when operated at maximum reverse surge current).
If the transients are random in nature (and in many cases
they are), determining the surge-current level can be a
problem. The circuit designer must make a reasonable
estimate of the proper device to be used, based on his
knowledge of the system and the possible transients to be
encountered. (e.g., transient voltage, source impedance and
time, or transient energy and time are some characteristics
that must be estimated). Because of the very low dynamic
impedance of Mosorb devices in the region between V(BR)
and VRSM, the maximum surge current is dependent on, and
limited by the external circuitry.
In cases where this surge current is relatively low, a
conventional zener diode could be used in place of a Mosorb
or other dedicated protective device with some possible
savings in cost. The surge capabilities most of
ON Semiconductor’s zener diode lines are discussed in
ON Semiconductor’s Application Note AN784.
SUMMARY
In selecting a protective device, it is important to know as
much as possible about the transient conditions to be
encountered. The most important device parameters are
reverse working voltage (VRWM), surge current (IRSM), and
clamp voltage (VRSM). the product of VRSM and IRSM yields
the peak power dissipation, which is one of the prime
categories for device selection.
The selector guide, in this book, gives a broad overview
of the Mosorb transient suppressors now available from ON
Semiconductor. For more detailed information, please
contact your ON Semiconductor sales representative or
distributor.
http://onsemi.com
103
Z
Vout
Vin
MOSORB
PROTECTED
LOAD
Figure 1.
VOLTAGE
VOLTAGE
Vin
Vin
Vout
Vout
VRSM
VRSM
V(BR)
VRWM
V(BR)
VRWM
TIME
TIME
tp
tclamping, VERY SHORT
tp = PULSE WIDTH OF INCOMING TRANSIENT
Figure 1a.
Figure 1b.
VOLTAGE
VOLTAGE
Vin
Vin
OVERSHOOT
Vout
Vout
VRSM
VRSM
V(BR)
VRWM
V(BR)
VRWM
TIME
TIME
tclamping, VERY LONG
tclamping, WITH OVERSHOOT
Figure 1c.
Figure 1d.
http://onsemi.com
104
TYPICAL MOSORB APPLICATIONS
DC Power Supplies
Input/Output Regulator Protection
AC
INPUT
MOSORBS
DC
POWER
SUPPLY
IC
VOLTAGE
REGULATOR
+
+
Vout
MOSORB
MOSORB
−
DC Motors − Reduces EMI
+B
MOSORB
−
−
Op Amp Protection
+
Memory Protection
+
+5 V
DC
MOTOR
MOSORB
IC
OP AMP
MOSORB
MOSORB
−
MOS
MEMORY
MOSORB
−8 V
MOSORB
−10 V
MOSORB
MOSORB
−B
Inductive Switching Transistor Protection
VDD
VGG
ADDRESS BUS
RAM
ROM
DATA BUS
CPU
MOSORB MOSORB
I/O
CLOCK
CONTROL BUS
Gnd
MOSORBS
Microprocessor Protection
KEYBOARD
TERMINAL
PRINTER
ETC.
A
B FUNCTIONAL
C DECODER
D
I/O
Gnd
MOSORBS
Computer Interface Protection
http://onsemi.com
105
AR450 − CHARACTERIZING OVERVOLTAGE TRANSIENT SUPPRESSORS
Prepared by
Al Pshaenich
ON Semiconductor Power Products Division
The use of overvoltage transient suppressors for
protecting electronic equipment is prudent and
economically justified. For relatively low cost, expensive
circuits can be safely protected by one or even several of the
transient suppressors on the market today. Dictated by the
type and energy of the transient, these suppressors can take
on several forms.
For example, in the telecommunication field, where
lightning induced transients are a problem, such primary
suppressors as gas tubes are often used followed by
secondary, lower energy suppressors. In an industrial or
automotive
environment,
where
transients
are
systematically generated by inductive switching, the
transient energy is more well-defined and thus adequately
suppressed by relatively low energy suppressors. These
lower energy suppressors can be zener diodes, rectifiers with
defined reverse voltage ratings, metal oxide varistors
(MOVs), thyristors, and trigger devices, among others. Each
device has its own niche: some offer better clamping factors
than others, some have tighter voltage tolerances, some are
higher voltage devices, others can sustain more energy and
still others, like the thyristor family, have low on-voltages.
The designer’s problem is selecting the best device for the
application.
Thus, the intent of this article is twofold:
R1
R2
S1
VZ
S2
VIN
IZ
IZ
C
DUT
10%
tW
Figure 1a. Simplified Exponential Tester
VEE
IZ
RL
IZ
tW
DUT
Figure 1b. Simplified Rectangular Tester
Using PNP Switch
Figure 1. Basic Surge Current Testers
automate this circuit, the series switches S1 and S2 can be
replaced with appropriately controlled transistors or SCRs.
One method of easily implementing a rectangular surge
current pulse is shown in the simplified schematic of
Figure 1b. A PNP transistor switch connected to the
positive supply VEE applies power to the DUT. The current
is obviously set by varying either VEE and/or RL. If however,
the transistor switch were replaced with a variable, constant
current source, measurement procedures are simplified as
how the limiting resistor need not be selected for various
current conditions.
As in most surge current evaluations, the DUT is
ultimately subjected to destructive energy (current, voltage,
pulse width), the failure points noted, and the derated points
plotted to produce the energy limitation curve. Of particular
interest is the junction temperature at which the DUTs are
operated, be it near failure or at the specified derated point.
This measurement relates to the overall reliability of the
suppressor, i.e., can the suppressor sustain one surge current
pulse or a thousand, and will it be degraded when operated
above the specified maximum operating temperature?
The Rectangular Current Surge Suppressor Test Circuit to
be described addresses these questions by implementing and
measuring the rectangular current capability of the
1. To describe the operation of the surge current test
circuits used in characterizing lower energy transient
suppressors.
2. To define the attributes of the various suppressors,
allowing the circuit designer to make the
cost/performance tradeoffs.
Surge suppressors are generally specified with
exponentially decaying and/or rectangular current pulses.
The exponential surge more nearly simulates actual surge
current conditions − capacitor discharges, line and switching
transients, lightning induced transients, etc., whereas
rectangular surge currents are usually easier to implement
and control.
To generate an exponential rating, a charged capacitor is
simply dumped into the device under test (DUT) and the
energy of each successive pulse increased until the device
ultimately fails. The simplified circuit of Figure 1a describes
the circuit. By varying the size of the capacitor C, the
limiting resistor R2, and the voltage to which C is charged
to, various peak currents and pulse widths (defined to the
10% discharge point in this paper) can be obtained. To
http://onsemi.com
106
CMOS gate ICs. The remaining blocks are the Sample and
Hold (S/H) circuit and two detectors for determining the
status of failed DUTs, either shorted devices or open.
Shown in Figures 3 and 4 respectively, are the complete
circuit and significant waveforms. Clocking for the system
is derived from a CMOS, two inverters, astable MV (gates
1A and 1B) whose output triggers the two input NOR gate
configured monostable MV (gates 1C and 1D) to produce
the Pulse Generator output pulse (Figure 4b). Alternatively,
a single pulse can be obtained by setting switch S2 to the One
Shot position and depressing the pushbutton Start switch S1.
Contact bounce is suppressed by the 100 ms MV (gates 4C
and D). Frequency of the astable MV, set by potentiometer
R1, can vary from about 200 Hz to 0.9 Hz and the pulse
width, controlled by R2 and the capacitor timing selector
switch S3, from about 300 μs to 1.3 s.
The positive going Pulse Generator output feeds the
Constant Current Amplifier IZ and turns on, in order, NPN
transistor Q1, PNP transistor Q3, NPN Darlington Q4, PNP
Power Darlington Q6 and parallel connected PNP Power
Transistors Q8 and Q9. Transistor Q4 is configured as a
constant current source whose current is set by the variable
base voltage potentiometer R3. Thus, the voltage to the
bases of Q6, Q8 and Q9 are also accordingly varied.
Transistors Q8 and Q9 (MJ14003, IC continuous of 60 A),
also connected as constant current sources with their 0.1 Ω
emitter ballasting resistors, consequently can produce a
rectangular current pulse from a minimum of about 0.5 A
and still have adequate gain for 1 ms pulses of 150A peak.
Due to propagation delays of this amplifier, the IZ current
waveform is as shown in Figure 4f. Since Q8 and Q9 must
be in the linear region for constant current operation, these
transistors are power dissipation limited at high currents to
the externally connected power supply V+ of 60 V. Thus the
maximum DUT voltage, taking into account the clamping
factor of the device, should be limited to about 50 V. At
wider pulse widths and consequently lower currents before
the DUT fails, the V+ supply should be proportionally
reduced to minimize Q8, Q9 dissipation. As an example, a
28 V surge suppressor operating at 100 ms pulse widths can
be tested to destructive limits with V+ of about 40 V.
Although a zener diode is shown as the DUT in the
schematic, the test devices can be any rectifier with defined
reverse voltage, e.g., surge suppressors.
Immediately after the power pulse is applied to the DUT,
the negative going sense pulse from the 300 μs MV (Gate
2A, Figure 4e) turns on series connected PNP transistor Q10
and NPN transistor Q11 of the Diode Forward Current
Switch IF. Sense current, set by current limiting resistor
selector switch S4, thus flows up from ground through the
forward biased DUT, the limiting resistor, and Q11 to the
−15 V supply. The result, by monitoring the cathode of the
DUT, is a 300 μs wide, approximately −0.6 V pulse.
suppressor and determining the device junction temperature
TJ shortly after the end of the surge current pulse. Knowing
TJ, the energy to the DUT can be limited just short of failure
and thus a complete surge curve generated with only one, or
a few DUTs (Figure 6). Second, with the junction
temperature known, a reliability factor can be determined
for a practical application.
CIRCUIT OPERATION FOR THE
RECTANGULAR CURRENT TESTER
The Surge Suppressor Test Circuit block diagram is
shown in Figure 2 with the main blocks being the Constant
Current Amplifier supplying IZ to the DUT (a zener diode
in this instance) during the power pulse and the Diode
Forward Current Switch supplying IF during the
temperature sensing time. These two pulses are applied
sequentially, first the much larger IZ, and then the very small
sense current IF. During the IF time, the forward voltage VF
of the diode is measured from which the junction
temperature of the zener diode can be determined. This is
simply done by calibrating the forward biased DUT with a
specified low value of IF in a temperature chamber, one point
at 25°C and a second point at some elevated temperature.
The result is the familiar diode forward voltage versus
temperature linear plot with a slope of about −2 mV/°C for
typical diodes (Figure 7a). Comparing the plot with the test
circuit measured VF yields the DUT junction temperature
for that particular pulse width and IZ (Figure 7b).
25 μS
BLANKING
MV
CLOCK
PULSE
GEN
GATE
SAMPLE
PULSE
SHORT
DETECTOR
300 μS
SENSE
MV
OPEN
DETECTOR
VF
S/H
IZ
DUT
IZ
IF
IF
Figure 2. Surge Suppressor Test
Circuit Block Diagram
The System Clock, Pulse Generator, the several
monostable multivibrators (25 μs Blanking, Sample Pulse
and 300 μs Sense MVs) and Gate are fashioned from three
http://onsemi.com
107
Figure 3. Surge Suppressors Surge Current Fixture
http://onsemi.com
108
47 k
100 k
0.001 μF
22 k
5.6 k
3.9 k
Q1
2N3904
0.1 μF
+15 V
47 k
4
SW
S6
8
9
10 0.1 μF
47 k
300
R3
1k
SWS2
12
13
1.8M
+15 V
V2
3.3 k
12 k
+15 V
10
2N6287
Q6
22
1/2 W
390
6
7
1C
W SW 0.4 μF
S3
N
P.W. M.V.
2.7 k
0.0033 μF
R2
MR821
Q7
−15 V
Q11
MPS8099
1.3 k
1N914
I2
1k
0.1 μF
1D 9
100 k
DUT
5 pF
Q9
≤ 60 V
CI R4
Q8
10 mΩ
560
3D
(2)
0.1 Ω
25 W
V+
12
13
39 k
270 pF
47 k
1k
7
−15 V
−
4
+
6
0.1 μF
U6 LF355
2
3
+15 V 0.1 μF
+15 V
150 k
Q12
47 k 2N3906
+15 V
10 k
6
5
1
2
3B
3
4 8
9
22 k
10 k
U4A,B
U7 LF355
S/H
D
3
+15 V
Q14
MPS8099
3C
10
3
−15 V
1k
10 k
2B
TRIAD
F90X
4
10
390 pF
VF
0.1 μF
0.1 μF
4.7 k
10 k
−15 V
1N914
0.1 μF
4.7 k
12
13
7
2D
+15 V
6
5
14
7
4.7 k
1
RESET SW S5
OPEN DETECTOR
1k
2N5060
Q17
470
1W
OPEN LED
RED
Q18
2N3904
10 k
−0.7
DUT
GATE
SAMPLE
DELAY
SENSE
+15 V
RED
3
3
200 μF
25 V
μC
7915
(2) 200 μF
25 V
2
MC
7815
220 pF
1N914
100 k
U9
SHORT LED #1
0.1 μF 0.1 μF
8
−15 V
−
4
+
0.1 μF
+15 V
U5B
(1/2) MC1458
1k
Q16
2N5060
470
1W
+
U8
PWR ON
LED
2
2
RED
1.2 k
1/2 W
100 μS SAMPLE DELAY M.V.
2C
390 k
+15 V
200 μF
25 V
MDA101A
+
−
120 V
SHORT
DETECTOR #1
8
9
300 μs
SENSE M.V.
5
6
100 k
270 pF
390 pF
U2
MC14001
2A
0.1 μF
+15 V
1N914
1 5
+ 7 6
2
−
0.1 μF
4
25 k
+15 V
47 k
4A
Q13
2N4858
S
100 k
100 k
2
1
3 5
6
25 μs M.V.
1F 11
(1/2) MC14001
4 10 k
4B
3A
100 k
12
1
2
+15 V
1.2M
2N3906
Q15
U5A
(1/2)
MC1458
2
1k
−
1
1/2 W
3
SHORT
+
0.1 μF
LED #2
10 k
RED
47 k
1k
10 k
0.001
100 k
330 pF
0.001 μF
100
k
0.001 μF
10 k
2W
+15 V
SHORT
DET #2
1N914
(2)
8
1E
16
13
270 pF
+15 V
U3
MC14001
14
15
+15 V
10.000 μF
75 V
5M
PW CONT
+15 V
60 μs < T2N < 8.5 mS
7 ms < T2W < 1.1 S
5
1k
SYNC
SW S4
4.7 k
25 mA
5 mA
10 mA
43 k
2N6668
1k
4,7 k
1W
U1
MC14572
100 k
IF SWITCH 0.1 μF
Q10
2N3906
10 k
180
Q4
11
270 pF
4D
U4C,D
(1/2) MC14001
MPSA29
Q5 8599
MPS
CLOCK
FREE 47 k
RUN
2N3906
Q3
1/2 W
1.2 k 820
Q2
2N3906
10 k
0.01 μF
0.1
1 SHOT
1B 3
4C
680
1/2 W
10 k
15 k
1/2 W
I2 SWITCH
FREQ CONT
5 ms < T1 < 1.2S
5M
100 mS MV
CONTACT BOUNCE 1A 1
2
22 k
1M
R1
+15 V
S1 START SW
300 μS
+3 V
8 μS
100 μS
25 μs
V2
I2
P.W.
8 μs
SAMPLE
PULSE
+15 V
0.1 μF
−15 V
0.1 μF
+15 V
(Figure 4D) is required to blank out turn-on switching
transients to produce the waveform shown in Figure 4I.
During the power pulse, U5A is normally high for a good
DUT (Figure 4J). This waveform is NOR’d with gate 3B
(inverted waveform of Figure 4I) to produce a low level
(0 V) gate 3C output (Figure 4K).
If, however, the DUT is shorted, U5A output switches low
resulting in a positive pulse output from G3C. This pulse
triggers the SCR on, lighting the LED in its anode circuit and
turning on the PNP transistor Q2 across the emitter-base of
Q3, thus clamping off the IZ power pulse. The circuit (Q16)
can be reset by opening switch S5.
By and large, this Short Detector circuit was found
adequate to protect transistors Q8 and Q9. However, for
some wide pulse widths, relatively high current conditions,
the propagation delay through the Short Detector was too
great, resulting in excessive FBSOA (Forward Bias Safe
Operating Area) stress on Q8 and Q9. Consequently, a faster
Short Detector #2 was implemented.
For accurate measurements of this pulse amplitude,
sample and hold circuitry is employed. This consists of unity
gain buffer amp U6, series FET switch Q13 and capacitor
hold buffer amp U7. The sample pulse (Figure 4H) to the
gate of the FET is delayed about 100 μs (by monostable MV
G-2C and G-2D) to allow for switching and thermal
transients to settle down. This pulse is derived from the
negative going, trailing edge output pulse of Gate 2D cutting
off transistor Q18 for the RC time constant in its base circuit.
The result is an approximate 8 μs wide sample pulse.
Consequently, the DC output voltage from hold amplifier
U7 is a measure of the DUT junction temperature.
Invariably, most DUTs will fail short. When the surge
suppressor tester is in the Free-run Mode, the power pulse
subsequent to the DUT shorting could excessively stress the
constant current drivers Q8 and Q9. To prevent this
occurrence, the Short Detector circuit was implemented.
This circuit consists of comparator U5A, 2 input NOR gate
configured 25 μs monostable MV (G1E and G1F), Gate
Circuit G3A, 3B and 3C, and SCR Q16. The 25 μs MV
(4A)
CLOCK
G1B
(4B)
PULSE GEN
G1D
(4C)
PULSE GEN
G1C
(4D)
25 μS MV
G1E
(4E)
300 μS
SENSE MV
G2A
(4F)
IZ
(4G)
100 μS SAMPLE
DELAY MV G2D
8 μS
(4H) SAMPLE PULSE Q8
(4I)
GATE
G3C
(4J)
SHORT
COMPARATOR U6A
(4K)
SHORT SCR
TRIGGER G3C
(4L)
VF
U6
+15 V
−14 V
GOOD
(4M)
OPEN SCR
TRIGGER U5B
SHORT
SHORT
GOOD
−0.6 V
−14 V
GOOD
OPEN
OPEN
GOOD
Figure 4. Surge Suppressor Test Circuit Waveforms
http://onsemi.com
109
high voltage (V+ = 200 to 350 V) and large capacitors
(>3000 μF), the power dissipated in the current limiting
resistor R1 can be substantial, thus necessitating the
illustrated 20 W rating. For longer charging times, switch S3
is closed, doubling the timing capacitor and the astable MV
on time.
To discharge capacitor C1 and thereby generate the
exponential surge current, the SCR must be fired. This
trigger is generated by the positive going one second pulse
from gate 1A being integrated by the R2C2 network, and
then shaped by gates 1C and 1D. The net result of about 100
μs time delay from gate 1D ensures noncoincident timing
conditions. This pulse output is then differentiated by C3-R3
with the positive going leading edge turning on Q3, Q4 and
finally the SCR with about a 4 ms wide, 15 mA gate pulse.
Consequently, the DUT is subjected to a surge current pulse
whose magnitude is dictated by the voltage on the capacitor
C1 and value of resistor RS, and also whose pulse width to
the 10% point is 2.3 RSC1. For a fixed pulse width, the DUT
is then stressed with increasing charge (by increasing V+)
until failure occurs, usually a shorted device.
If the DUT is the SCR (or MOS SCR), the failed condition
is obvious as the capacitor C1 will not be allowed to charge
for subsequent timing cycles. However, when the DUT is the
zener, rectifier, SIDAC or even an MOV, and the SCR is an
adequately rated switch, the circuit will still discharge
through the shorted DUT, but now the SCR alone will be
stressed by the surge current. A shorted DUT can be detected
by noting the voltage across the device during testing.
One problem encountered when stressing SCRs with high
voltage is when the DUT fails short. The limiting resistor
R1, which is only rated for 20 W, would now experience
continuous power dissipation for the full On time − as much
as 123 W ([350 V]2/1K). To prevent this occurrence, the PR1
Short Protection Circuit is incorporated. Since this is only a
problem when high V+’s (>100 V) are used, the circuit can
be switched in or out by means of switch S2. When
activated, this circuit monitors the voltage on capacitor C1
some time after the charging cycle begins. If the capacitor is
charging, normal operation occurs. However, if the SCR
DUT is shorted, the absence of voltage on the capacitor is
detected and the system is disabled.
The circuit consists of one CMOS IC with NAND gates
2A and 2B comprising a one second monostable time delay
MV and gates 2C and 2D forming a comparator and NAND
gate, respectively.
The negative going, trailing edge of gate 2A is
differentiated by R4-C4, and amplified by Q5 to form a
positive, 10 ms wide pulse (delayed by 1 sec) to gate 2D
input. If the capacitor C1 is shorted, gate 2C output is high,
allowing the now negative pulse from gate 2D to turn on
PNP transistor Q6 and SCR Q7. This latches the input to the
astable MV gate 1A low, disabling the timing and
consequently removing the power from R1. Resetting the
tester for a new device is accomplished by depressing the
pushbutton switch S1.
This circuit, connected to the collectors of Q8 and Q9, uses
a differentiating network (R4, C1) to discriminate between
the normally relatively slow fall time of the voltage pulse on
the DUT, and the exceedingly fast fall time when the device
fails. Thus, the R4-C1 time constant (5 ns) will only generate
a negative going trigger to PNP transistor Q12 when the DUT
voltage collapses during device failure. The positive going
output from Q12 resets the flip-flop (gates 4A and 4B), which
turns on the NPN transistor Q14. This transistor supplies
drive to the two PNP clamp transistors (Q5 and Q7) placed
respectively across the emitter-bases of the high, constant
current stages Q6 and Q8 and Q9. Propagation delay is thus
minimized, providing greater protection to the power stages
of the tester. As an added safety feature, the positive going
output from Gate 3C when Short Detector #1 is activated is
also used to trigger the flip-flop.
On the few occasions when the DUT fails open, then the
Open Detector consisting of comparator U5B and SCR Q17
comes into play. This circuit measures the DUT integrity
during the sense time. For a good DUT (VF < −1 V), U5B
output remains low (see Figures 4L and 4M). However for
an open DUT, VF switches to the negative rail and U5B goes
high, turning on Q17. As in the Short Detector, Q2 clamps
off the IZ power amplifier.
All of the circuitry including the +15 V and −15 V
regulated power supplies are self-contained, with the
exception of the V+ supply. For high current, narrow pulse
width testing, this external supply should have 10 to 15 A
capability. If not, additional energy storing capacitors across
the supply output may be required.
CIRCUIT OPERATION FOR THE
EXPONENTIAL SURGE CURRENT TESTER
To generate the surge current curve of peak current versus
exponential discharge pulse width, the test circuit of
Figure 5 was designed. This tester is an implementation of
the simplified capacitor discharge circuit shown in
Figure 1A, with the PNP high voltage transistor Q2
allowing the capacitor C to charge through limiting resistor
R1 and a triggered SCR discharging the capacitor. As shown
in Figure 5, the DUTs can be of any technology, although the
device connected to the capacitor and discharge limiting
resistor RS is shown as an MOS SCR. It could just as well
have been an SCR as the DUT or as the switch for the zener
diode, rectifier, SIDAC, etc., DUTs.
System timing for this Exponential Surge Current Tester
is derived from a CMOS quad 2 input NOR gate with gates
1A and 1B comprising a non-symmetrical astable MV of
about 13 seconds on and about one second off (switch S3
open). The positive On pulse from gate 1B turns on the 500
V power MOSFET Q1 and the following PNP transistor Q2.
The extremely high current gain FET allows for the large
base current variation of Q2 with varying supply voltage
(V+). This capacitor charging circuit has a 400 V blocking
capability (limited by the VCEO of Q2) and thus the capacitor
C1 used should be comparably rated. When operating with
http://onsemi.com
110
5
+15 V
1
+15 V
4
1D
6
3
+15 V
1C
2
1.5K
1/2 W
47K
C2
0.001 μF
+15 V
GRN
LED
2N3904
+15 V
R2
100 k
8
9
12
10
13
1A
14
1B
1N914
22M
Q3
2N3904
CAP
DISCHARGE
SW
S4
ON TIME
0.47 μF
SW
S3
ZENER DUT
R1
RS
RECTIFIER DUT
1N4005
C1
13S
0.47 μF
1N914
22K
1K
20 W
1.8M
10K
Q4
2N3906
Q2
MJE5852
7
18M
R3
10K
1K
Q1
MTP2N50
1K
11
10K
V+ ≤350 V
15K
20 W
22K
MC14011
1M
+15 V
C3
0.1 μF
SIDAC DUT
10
2W
25S
DUT SHORT
INDICATOR
+15 V
1K
MOS SCR
DUT
SCR
DUT/SW
150K
2W
10K
RED
LED
1K
1/2 W
+15 V
+15 V
1N914
0.1 μF
1N914
22K
C4
R4
0.1 μF 47K
100K
+15 V
9
2
3
0.01 μF
2A
10
2C
13
SW S1
11 47K
2D
SW
S2
22K
Q5
2N3906
5
6
RESET
LV
1M
+15 V
1
12
8
+15 V
4
HV
Q6
2N3906
2N5060
Q7
2B
6.8K
10K
15M
1 SEC DELAY MV
MC14011
1K
0.1 μF
Figure 5. Exponential Surge Current Tester
Exponential surge current curves, as well as rectangular,
are generated by destructive testing of at least several DUTs
at various pulse widths and derating the final curve by
perhaps 20−30%. These tests were conducted at low duty
cycles (<2%). To ensure multicycle operation, the DUTs are
then tested for about 1000 surges at a derated point on the
curve.
voltages generally greater than 30 V. Also, since energy
capability is related to die size, this parameter is also listed.
For several devices, both rectangular and exponential surge
current pulses are listed. Other devices were tested with only
rectangular pulses (where the junction temperature can be
determined) and still others, whose applications include
crowbars, with exponential current only.
TEST RESULTS
In trying to make a comparison of the several different
technologies of transient suppressors, some common
denominator has to be chosen, otherwise, the amount of
testing and data reduction becomes unwieldy. For this
exercise, voltage was used, generally in the 20 V to 30 V
range, although some of the more unique suppressors
(SIDACs, MOS SCRs, SCRs) were tested at their operating
voltage. As an example, the SIDAC trigger families of
devices were tested with voltages greater than their
breakover voltage (104 V to 280 V) and the SCRs were
subjected to exponential surge currents derived from
AVALANCHE RECTIFIER
The Rectangular Surge Current Tester was originally
designed for characterizing rectifier surge suppressors used
in automotive applications. For this operation, where
temperatures under the hood can reach well over 125°C, it
is important to know the device junction temperature at
elevated ambient temperature. Figures 6 and 7 describe the
results of such testing on a typical suppressor, the 24 V−32 V
MR2520L. It should be noted that these axial lead
suppressors, as well as all other axial lead devices tested,
were mounted between two spring loaded clips spaced 1
inch apart.
http://onsemi.com
111
As shown in Figure 6 of the actual current failure points
of the DUTs, at least four devices were tested at the various
pulse widths, tw (in this example from 0.5 ms to 100 ms).
Also shown in Figure 6 is the curve derived with a single
DUT at an energy level just short of failure. This
measurement was obtained by maintaining a constant
rectifier forward voltage drop, VF (0.25 V) for all pulse
widths (junction temperature, TJ of 230°C) by varying the
avalanche current. Thus, one device can be used,
non-destructively, to generate the complete rectangular
surge current curve.
It should also be pointed out that the definition for the
exponential tw in this article is the current discharge point to
the 10% value of the peak test current IZM. Expressed in time
constant τ, this would be 2.3 RC. Some data sheets describe
tw to the 50% point of IZM (0.69 τ) and others to 5 τ. To
normalized these time scales (abscissa of curves) simply
change the scales accordingly; i.e., IZM/2 pulse widths would
be multiplied by 2.3/0.69 = 3.33 for tw at 10% current pulses.
Figure 7a describes the actual temperature calibration
curve (measured in a temperature chamber) of the
MR2520L and Figure 7b, the junction temperature of the
DUT at various 10 ms rectangular pulse current amplitudes.
These temperatures are taken from the calibration curve (in
actuality, an extremely linear curve), knowing the rectifier
forward voltage drop immediately (within 100 μs) after
cessation of power. Note that the junction temperature just
prior to device failure is about 290°C.
V F , FORWARD VOLTAGE (VOLTS)
1
10
20
50
1
2
5
tW, RECTANGULAR PULSE WIDTH (ms)
I Z , PEAK SURGE CURRENT (AMPS)
0.2
50
0
100
150
200
250
TJ, JUNCTION TEMPERATURE (°C)
Figure 7b. Measured
Forward Voltage
300
350
IZ
(A)
VF
(V)
TJ
(°C)
11
0.64
225
10
0.57
75
20
0.48
120
30
0.36
180
40
0.25
230
50
0.15
290
0.10
DUT
FAILED
55
Figure 7. Calculated Junction Temperature
Of The MR2520L Surge Suppressor
At Various Avalanche Currents
apparent that the device is very conservatively specified.
Also, the relative magnitudes of the two curves reflect the
differences in the rms values of the two respective pulses.
ACTUAL DUTS FAILURE POINTS
ONE DUT WITH VF = 0.25 V
TJ = 230°C
0.5
0.4
RECTANGULAR
PULSE
tW = 10 ms
I = 10 mA
F
50
10
0.6
Figure 7a. Temperature Calibration Curve
Of The MR2520L
MR2520L RECTIFIER
SURGE SUPPRESSOR,
RECTANGULAR PULSE
VZ = 28 V TYP
TA = 25°C
20
0.8
0
200
100
MR2520L AVALANCHE RECTIFIER
SURGE SUPPRESSOR
IF = 10 mA
SIDAC
SIDACs are increasingly being used as overvoltage
transient suppressors, particularly in telephone applications.
Being a high voltage bilateral trigger device with relatively
high current capabilities, they serve as a costeffective
overvoltage protection device. As in other trigger devices,
when the SIDACs breakover voltage is exceeded, the device
switches to a low voltage conduction state, allowing an
inordinate amount of surge current to be passed. This is well
illustrated by the surge current curves of Figure 9 which
describe the small die size ([37]2mil) axial lead, Case 59-04,
MKP9V240 SIDAC. The curves show that this 240 V device
was able to handle, to failure, as much as 31 A and 15 A,
respectively, for 1 ms exponential and rectangular current
pulses. Under the same pulse conditions, the large die
([78]2mil) MK1V270 SIDAC handled 170 A and 60 A,
respectively, as shown in Table 2.
100
Figure 6. Experimental Rectangular Surge
Current Capability Of The MR2520L Rectifier
Surge Suppressor
ZENER OVERVOLTAGE TRANSIENT
SUPPRESSOR
Illustrated in Figure 8 are the actual rectangular and
exponential surge current curves of the P6KE30 overvoltage
transient suppressor, an axial lead, Case 17, 30 V zener
diode characterized and specified for surge currents. This
device is specified for 600 W peak for a 1 ms exponential
pulse measured at IZM/2. From the exponential curve, it is
http://onsemi.com
112
OVERALL RATINGS
I Z , PEAK SURGE CURRENT (AMPS)
100
P6KE30 OVERVOLTAGE
TRANSIENT SUPPRESSOR
VZ = 30 V
EXPONENTIAL
PSPEC = 600 W pk
IZ
@
2
50
The compilation of all of the testing to date on the various
transient suppressors is shown in Tables 1 and 2. Table 1
describes the zener suppressors, avalanche rectifiers and
MOVs, comparing the die size and normalized costs
(referenced to the MOV V39MA2A). From this data, the
designer can make a cost/performance judgment.
Of interest is that the small pellet MOV is not the least
expensive device. The P6KE30 overvoltage transient
suppressor costs about 85% of the MOV, yet it can handle
about three times the current (2.5 A to 0.7 A) for a 100 ms
rectangular pulse. Under these conditions, the resultant
clamping voltages for the zener and MOV were 32 V and
60 V respectively.
Also shown in the table is a 1.5 W zener diode specified
for zener applications. This low surge current device costs
three times the MOV, illustrating that tight tolerance zener
diodes are not cost effective and that the user should use
devices designed and priced specifically for the suppressor
application.
Thyristor type surge suppressors are shown in Table 2.
They include four SIDAC series, two SCRs designed and
characterized specifically for crowbar applications and also
the MOS SCR MCR1000. The MOS SCR, a process
variation of the vertical structure power MOSFET,
combines the input characteristics of the FET with the
latching action of an SCR.
All devices were surge current tested with the resultant
peak currents being impressively high. The TO-220 (150)2
mil SCR MCR69 for example, reached peak current levels
approaching 700 A for a 1 ms exponential pulse. The
guaranteed, derated, time base translated curves for the
crowbar SCR family of devices are shown in Figure 10, as
is the MK1V SIDAC in Figure 11.
Figures 12A−C describe the guaranteed, reverse surge
design limits for the avalanche rectifier devices. These three
figures illustrate, respectively, the peak current, power and
energy capabilities of these overvoltage transient
suppressors derived from exponential testing. The peak
power, Ppk, ordinate of the curve is simply the product of the
derated IZ and VZ and the energy curve, the product of Ppk
and tw.
10
RECTANGULAR
5
IZ
10
%
tW
1
0.1
0.5 1
50 100
5 10
tW, PULSE WIDTH (ms)
500 1000
Figure 8. Surge Current Capability Of The P6KE30
Overvoltage Transient Suppressor As A Function Of
Exponential & Rectangular Pulse Widths
I Z , PEAK SURGE CURRENT (AMPS)
100
SIDAC MKP9V240
240 V
CASE 59-04
372 MILS
50
20
EXPONENTIAL
10
RECTANGULAR
5
IZ
tW
10
%
0.5
1
2
1
0.2
2
5
10
20
50
100
tW, PULSE WIDTH (ms)
Figure 9. Measured Surge Current To Failure
Of A SIDAC MKP9V240
http://onsemi.com
113
3000
1
Ipk
1000
5 TC
tW
MCR70
NORMALIZED PEAK SURGE CURRENT
Ipk , PEAK CURRENT (AMPS)
C = 8400 μF
TA = 25°C
ESR ≈ 25 mΩ
N = 2000 PULSES
VC ≤ 60 V
f = 3 PULSES/MIN.
MCR71
MCR69
MCR68
300
MCR67
100
30
0.1
0.5
1
5
10
50
N = 2000 PULSES
0.8
0.6
0.4
100
25
0
50
75
100
TA, AMBIENT TEMPERATURE (°C)
tW, BASE PULSE WIDTH (ms)
a. Peak Surge Current versus Pulse Width
125
b. Peak Surge Current versus
Ambient Temperature
Figure 10. SCR Crowbar Derating Curves
I RSM , PEAK REVERSE CURRENT (AMPS)
I Z , PEAK SURGE CURRENT (AMPS)
100
SIDAC MK1V115
V(BO) = 115 V MAX
TA = 25°C
50
30
10
IZ
5
10%
3
tW
1
0.3 0.5
10
30
3
5
tW, PULSE WIDTH (ms)
1
50
100
300
MR2530L
TC = 25°C, DUTY CYCLE ≤ 1%
200
MR2525L
100
MR2520L
70
50
30
SEE NOTE FOR TIME
CONSTANT DEFINITION
20
300
1
2
5
10000
TC = 25°C, DUTY CYCLE ≤ 1%
7000 MR2530L
MR2525L
5000
3000
MR2520L
2000
1000
SEE NOTE FOR TIME
CONSTANT DEFINITION
700
500
1
2
5
10
20
50 100
τ, TIME CONSTANT (ms)
200
200
500 1000
Figure 12a. Peak Current
WRSM , PEAK REVERSE ENERGY (JOULES)
PRSM , PEAK REVERSE POWER (WATTS)
Figure 11. Exponential Surge Current
Capability Of The MK1V SIDAC, Pulse Width
versus Peak Current
10
20
50 100
τ, TIME CONSTANT (ms)
300
200
TC = 25°C, DUTY CYCLE ≤ 1%
100
50
MR2530L
30
MR2525L
SEE NOTE FOR TIME
CONSTANT DEFINITION
20
500 100
1
0
NOTE: τ = RC
MR2520L
2
5
10
20
50
100
τ, TIME CONSTANT (ms)
Figure 12c. Energy
Figure 12b. Peak Power
Figure 12. Guaranteed Reverse Surge Design Limits for the
MR2525L & MR2530L Overload Transient Suppressors
http://onsemi.com
114
200
500 1000
Table 1. Measured Surge Current Capability of Transient Suppressors
Spec.
TitleP
a
r
t
N
o
.
Surge Supp.,
Overvoltage
Transient
Suppressor
1.5 W Zener
Diode
Zener
C
l
a
mpi
n
g
Fa
c
t
o
r
V
Powe Die
m
1
ms
V1
Rect.
Exp.
Rect.
Exp.
Rect.
Exp.
Rect.
0
0
ms
Cas Volt(Ene Siz Exp.
1
m
s
Dev
Type
Avalanche
Rectifier
P
e
a
C
k
u
r
r
e
n
a
t
P
t
u
l
s
e
W
idths,Ipk(Amps)
Overvoltage
Transient
Suppressor
1
0
0
m
s
2.5KW
Peak
1502
mil
85 A
40
30
18
27V + 1.2
22V
MR2525L
24−32 V
10KW
Peak
1962
mil
150 A
70
54
37
31
+ 1.3
23
1N5936A
30 V
1.5 W
Cont.
372
mil
600 W
Peak
602
mil
1500 W
Peak
1042
mil
194 05
194−05
DO 41
DO−41
1N5932A
20 V
P6KE30
30 V
17
P6KE10
10 V
1.5KE30
30 V
41A 02
41A−02
1.5KE24
MOV**
2
0
m
s
24−32 V
MR2520L
MOSORB
Metal
Oxide
Varistor
1
0
m
s
24 V
V39MA2A
Axial
Lead
28 V
0.16
ǒJoules
Ǔ
3 mm
V33ZA1
Radial
Lead
26 V
1.0
ǒJoules
Ǔ
7 mm
12 A
5
6
2.5
5
2
3
1.3
41 + 1.4
30
23 A
6
10
2.8
7
2.3
5
1.4
28
+ 1.2
23
43 A
14
14
5
10
4.5
5
2.5
41 + 1.3
32
5.5
16
+ 1.2
13
9
Nor
Cos
*
4.0
32
3.2
0 85
0.85
24 A
12
35 A
10
4
35
+ 1.1
33
45 A
14
6
30V
+ 1.1
28V
9A
5
0.7
80V
60V
6A
0.7A
1.0
35
4A
105V
80V
35A
4A
1.4
18
1.8
**G.E.
Table 2. Measured Surge Current Of Thyristor Type Devices
Ipk @ tW
Technology
SIDAC
SCR
Device
Voltage
Ratings
MKP9V130 Series
104 V−135 V
MKP9V240 Series
220 V−280 V
MK1V135 Series
120 V−135 V
MK1V270 Series
220 V−280 V
MCR68 Series
MCR69 Series
Case
Die
Size
59 04
59−04
372 mil
267 01
267−01
782 mil
25 V−400
V 400 V
O
TO−220
MOS SCR
MCR1000 Series
1 ms
200 V−600 V
10 ms
Exponent.
Rectang.
Exponent.
Rectang.
40 A
13 A
16 A
8A
31 A
15 A
20 A
8A
140 A
80 A
55 A
30 A
170 A
60 A
90 A
28 A
Norm
Cost
*
0 87
0.87
11
1.1
922 mil
300 A
170 A
1.2
1502
mil
700 A
400 A
1.9
127 mil
x
183 mil
250 A
170 A
9.3
*Normalized to G.E. MOV V39MA2A, Qty 1-99, 1984 Price
zener operating at a 1.0% duty cycle, the respective derating
factors for 10 μs and 10 ms pulses are 0.08 and 0.47. The
non-repetitive peak power capabilities for these two pulses
(10 μs and 10 ms) are about 1300 W and 50 W respectively,
resulting in repetitive power and current capabilities of
about 104 W and 24 W and consequently 5.2 A and 1.2 A.
Additionally, the published non-repetitive peak power
ratings of the various zener diode packages are illustrated in
Figure 13. Figure 14 describes the typical derating factor for
repetitive conditions of duty cycles up to 20%. Using these
two empirically derived curves, the designer can then
determine the proper zener for the repetitive peak current
conditions.
At first glance the derating of curves of Figure 14 appear
to be in error as the 10 ms pulse has a higher derating factor
than the 10 μs pulse. However, when the mathematics of
multiplying the derating factor of Figure 14 by the peak
power value of Figure 13 is performed, the resultant
respective power and current capability of the device
follows the expected trend. For example, for a 5 W, 20 V
MOV
All of the surge suppressors tested with the exception of
the MOV are semiconductors. The MOV is fabricated from
a ceramic (Zn0), non-linear resistor. This device has wide
acceptance for a number of reasons, but for many
applications, particularly those requiring good clamping
http://onsemi.com
115
PPK(NOM), NOMINAL PEAK POWER (kW)
SOURCE IMPEDANCE RS = 500 Ω
100
50
20
10
5
1N6267 SERIES
2
1
0.5
0.2
0.1
0.05
5 WATT TYPES
1 TO 3 W TYPES
PLASTIC DO-41
250 mW TO 1 W TYPES
0.02
GLASS DO-35 & GLASS DO-41
0.01
0.05 0.1 0.2
0.5
1
0.01 0.02
PULSE WIDTH (ms)
2
5
10
Power is defined as VZ(NOM) x IZ(PK) where VZ(NOM) is the
nominal zener voltage measured at the low test current used for
voltage classification.
Figure 13. Peak Power Ratings of Zener Diodes
27 V MOV
G.E. V27ZA4, 4 JOULES CAPABILITY
Figure 15a.
DERATING FACTOR
1
0.7
0.5
SOURCE IMPEDANCE RS = 500 Ω
0.3
0.2
PULSE WIDTH
10 ms
0.1
0.07
0.05
1 ms
0.03
100 μs
0.02
0.01
0.1
10 μs
0.2
0.5
1
2
5
10
D, DUTY CYCLE (%)
20
50
100
Figure 14. Typical Derating Factor for Duty Cycle
factors, the MOV is found lacking; (clamping factor is
defined as the ratio of VZ at the test current to that at 1.0 mA).
This is photographically illustrated in Figure 15 which
compares a 27 V zener (1N6281) with a 27 V MOV
(V27ZA4). The input waveform, through a source
impedance resistance to the DUTs, was an exponentially
decaying voltage waveform of 90 V peak. Figures 15A and
B compare the output waveforms (across the DUTs) when
the source impedance was 500 Ω and Figures 15C and D for
a 50 Ω condition. The zener clamped at about 27 V for both
impedances whereas the MOV was about 40 V and 45 V
respectively.
Surge current capabilities of a comparably powered MOV
were also determined, as shown in the curve of Figure 16.
Although the MOV, a V39MA2A, is specified as a 28 V
27 V ZENER DIODE
ON Semiconductor 1N6281, APPROX. 1.5 JOULES
Figure 15b.
continuous device (39 V ±10% at 1 mA) at the pulse widths
and currents tested, the resultant voltage VZ across the MOV
− 80 V at about 6 A − necessitated a high voltage fixture. This
was accomplished with a circuit similar to that of Figure 1B.
But MOVs do have their own niche in the marketplace, as
described in Table 3, the Relative Features of MOVs and
MOSORBs.
http://onsemi.com
116
10
I Z , PEAK SURGE CURRENT (AMPS)
SOURCE IMPEDANCE RS = 50 Ω
5
3
1
G.E. V39MA2A MOV
VDCM = 28 V
VNOM = 39 V @ 1 mA
TA = 25°C
0.5
0.3
0.1
1
3
10
5
tW, PULSE WIDTH (ms)
30
50
100
Figure 16. Rectangular Surge Current Capability
Of The V39MA2A MOV
27 V MOV
Table 3. Relative Features of MOVs
and MOSORBs
Figure 15c.
SOURCE IMPEDANCE RS = 50 Ω
MOV
27 V ZENER DIODE
Figure 15d.
Figure 15. Clamping Characteristics of a
27 V Zener Diode and 27 V MOV
High Clamping Factor
Very good clamping close to
the operating voltage.
Symmetrically bidirectional
Standard parts perform like
standard zeners. Symmetrical
bidirectional devices available
for many voltages.
Energy capability per dollar
usually much greater than a
silicon device. However, if
good clamping is required a
higher energy device would be
needed, resulting in higher
cost.
Good clamping
characteristics could reduce
overall cost.
Inherent wear out mechanism,
clamp voltage degrades after
every pulse, even when
pulsed below rated value.
No inherent wear out
mechanism.
Ideally suited for crude AC line
protection.
Ideally suited for precise DC
protection.
High single-pulse current
capability.
Medium multiple-pulse
current capability.
Degrades with overstress.
Fails short with overstress.
Good high voltage capability.
Limited high voltage capability
unless series devices are
used.
Limited low voltage capability.
Good low voltage capability.
http://onsemi.com
117
MOSORB/Zener Transient
Suppressor
REFERENCES
SUMMARY
The surge current capabilities of low energy overvoltage
transient suppressors have been demonstrated, including
cost/performance comparison of rectifiers, zeners, thyristor
type suppressors, and MOVs. Both rectangular and
exponential testing have been performed with the described
testers. Additionally, the Rectangular Current Surge Tester
has the capability of measuring the diode junction
temperature of zeners and rectifiers at various power levels,
thus establishing safe operating limits.
1. Cherniak, S., A Review of Transients and Their Means
of Suppression, ON Semiconductor Application Note
AN843.
2. Wilhardt, J., Transient Power Capability of Zener
Diodes, ON Semiconductor Application Note AN784.
3. Pshanenich, A., Characterizing the SCR for Crowbar
Applications, ON Semiconductor Application Note
AN879.
4. Pshaenich, A., The SIDAC, A New High Voltage
Trigger that Replaces Circuit Complexity and Cost,
ON Semiconductor Engineering Bulletin EB-106.
5. General Electric, Transient Voltage Suppression
Manual, Second Edition.
http://onsemi.com
118
MEASUREMENT OF ZENER VOLTAGE TO THERMAL EQUILIBRIUM
WITH PULSED TEST CURRENT
Prepared by
Herb Saladin
Discrete Power Application Engineering
INTRODUCTION
current (IZT) pulse width other than the pulse width used by
the manufacturer.
Thermal equilibrium (TE) is reached when diode junction
temperature has stabilized and no further change will occur
in VZ if the IZT time is increased.2 This absolute value can
vary depending on the mounting method and amount of
heatsinking. Therefore, thermal equilibrium conditions
have to be defined before meaningful correlation can exist.
Normalized VZ curves are shown for four package styles
and for three to five voltage ratings per package. Pulse
widths from 1 ms up to 100 seconds were used to arrive at
or near thermal equilibrium for all packages with a given
method of mounting.
This paper discusses the zener voltage correlation
problem which sometimes exists between the manufacturer
and the customer’s incoming inspection. A method is shown
to aid in the correlation of zener voltage between thermal
equilibrium and pulse testing. A unique double pulsed
sample and hold test circuit is presented which improves the
accuracy of correlation.1
Several zener voltages versus zener pulsed test current
curves are shown for four package styles. An appendix is
attached for incoming inspection groups giving detailed
information on tolerances involved in correlation.
For many years the major difficulty with zener diode
testing seemed to be correlation of tight tolerance voltage
specifications where accuracy between different test setups
was the main problem. The industry standard and the EIA
Registration system adopted thermal equilibrium testing of
zener diodes as the basic test condition unless otherwise
specified. Thermal equilibrium was chosen because it was
the most common condition in the final circuit design and it
was the condition that the design engineers needed for their
circuit design and device selection. Thermal equilibrium
testing was also fairly simple to set-up for sample testing at
incoming inspection of standard tolerance zeners.
In recent years with the advent of economical
computerized test systems many incoming inspection areas
have implemented computer testing of zener diodes which
has been generating a new wave of correlation problems
between customers and suppliers of zener diodes.
The computerized test system uses short duration pulse
test techniques for testing zener diodes which does not
directly match the industry standard thermal equilibrium
test specifications.
This paper was prepared in an attempt to clarify the
differences between thermal equilibrium and short duration
pulse testing of zener diodes, to provide a test circuit that
allows evaluation at various pulse widths and a suggested
procedure for incoming inspection areas that will allow
meaningful correlation between thermal equilibrium and
pulse testing.
In the measurement of zener voltage (VZ), the
temperature coefficient effect combined with test current
heating can present a problem if one is attempting to
correlate VZ measurements made by another party (Final
Test, Quality Assurance or Incoming Inspection).2 This
paper is intended as an aid in determining VZ at some test
Mounting
There are five conditions that can affect the correlation of
VZ measurements and are: 1) instrumentation, 2) TA, 3) IZT
time, 4) PD and 5) mounting. The importance of the first
four conditions is obvious but the last one, mounting, can
make the difference between good and poor correlation. The
mounting can have a very important part in VZ correlation
as it controls the amount of heat and rate of heat removal
from the diode by the mass and material in contact with the
diode package.
Two glass axial lead packages (DO-35 and DO-41),
curves (Figures 5 and 6) were measured with standard
Grayhill clips and a modified version of the Grayhill clips to
permit lead length adjustment.
Test Circuit
The test circuit (Figure 8) consists of standard CMOS
logic for pulse generation, inverting and delaying. The logic
drives three bipolar transistors for generation of the power
pulse for IZT. VZ is fed into an unique sample and hold (S/H)
circuit consisting of two high input impedance operational
amplifiers and a field effect transistor switch.
For greater accuracy in VZ measurements using a single
pulse test current, the FET switch is double pulsed. Double
pulsing the FET switch for charging the S/H capacitor
increases accuracy of the charge on the capacitor as the
second pulse permits charging the capacitor closer to the
final value of VZ.
The timing required for the two pulse system is shown in
waveform G-3C whereby the initial sample pulse is delayed
from time zero by a fixed 100 μs to allow settling time and
the second pulse is variable in time to measure the analog
input at that particular point. The power pulse (waveform
G-2D) must also encompass the second sample pulse.
http://onsemi.com
119
To generate these waveforms, four time delay monostable
multivibrators (MV) are required. Also, an astable MV, is
required for free-running operation; single pulsing is simply
initiated by a push-button switch S1. All of the pulse
generators are fashioned from two input, CMOS NOR gates;
thus three quad gate packages (MC14001) are required.
Gates 1A and 1B form a classical CMOS astable MV clock
and the other gates (with the exception of Gate 2D) comprise
the two input NOR gate configured monostable MV’s. The
Pulse Width variable delay output (Gate 1D) positions the
second sample pulse and also triggers the 100 μs Delay MV
and the 200 μs Extended Power Pulse MV, The respective
positive going outputs from gates 3A and 2C are diode
NOR’ed to trigger the Sample Gate MV whose output will
consequently be the two sample pulses. These pulses then
turn on the PNP transistor Q1 level translator and the
following S/H N-channel FET series switch Q2. Op amps
U4 and U5, configured as voltage followers, respectively
provide the buffered low output impedance drive for the
input and output of the S/H. Finally, the pulse extended
Power Gate is derived by NORing (Gate 2D) the Pulse
Width Output (Gate 1D) with the 200 μs MV output (Gate
2C). This negative aging gate then drives the Power
Amplifier, which, in turn, powers the D.U.T. The power
amplifier configuration consists of cascaded transistors
Q3−Q5, scaled for test currents up to 2 A.
Push button switch (S4) is used to discharge the S/H
capacitor. To adjust the zero control potentiometer, ground
the non-inverting input (Pin 3) of U4 and discharge the S/H
capacitor.
Table 1. S3 — Pulse Width
Switch
Position
*C(μF)
t(ms)
1
2
3
4
5
6
7
8
9
10
11
12
13
0.001
0.004
0.006
0.01
0.04
0.06
0.1
0.4
0.6
1.0
1.2
6.0
10
1
3
5
10
30
50
100
300
500
1K
3K
5K
10K
*Approximate Values
Using Curves
Normalized VZ versus IZT pulse width curves are shown
in Figure 1 through 6. The type of heatsink used is shown or
specified for each device package type. Obviously, it is
beyond the scope of this paper to show curves for every
voltage rating available for each package type. The object
was to have a representative showing of voltages including
when available, one diode with a negative temperature
coefficient (TC).
These curves are actually a plot of thermal response
versus time at one quarter of the rated power dissipation.
With a given heatsink mounting, VZ can be calculated at
some pulse width other than the pulse width used to specify
VZ.
For example, refer to Figure 5 which shows normalized
VZ curves for the axial lead DO-35 glass package. Three
mounting methods are shown to show how the mounting
effects device heating and thus VZ. Curves are shown for a
3.9 V diode (1N5228B) which has a negative TC and a 12 V
diode (1N5242B) having a positive TC.
In Figure 5, the two curves generated using the Grayhill
mountings are normalized to VZ at TE using the
ON Semiconductor fixture. There is very little difference in
VZ at pulse widths up to about 10 seconds and mounting only
causes a very small error in VZ. The maximum error occurs
at TE between mountings and can be excessive if VZ is
specified at TE and a customer measures VZ at some narrow
pulse width and does not use a correction factor.
Using the curves of Figure 5, VZ can be calculated at any
pulse width based upon the value of VZ at TE which is
represented by 1 on the normalized VZ scale. If the 1N5242B
diode is specified at 12 V ± 1.0% at 90 seconds which is at
TE, VZ at 100 ms using either of the Grayhill clips curves
would be 0.984 of the VZ value at TE or 1 using the
ON Semiconductor fixture curve. If the negative TC diode
is specified at 3.9 V ± 1.0% at TE (90 seconds), VZ at 100
ms would be 1.011 of VZ at TE (using ON Semiconductor
fixture curve) when using the Grayhill Clips curves.
Testing
The voltage VCC, should be about 50 volts higher than the
D.U.T. and with RC selected to limit the IZT pulse to a value
making VZT IZT = 1/4 PD (max), thus insuring a good current
source. All testing was performed at a normal room
temperature of 25°C. A single pulse (manual) was used and
at a low enough rate that very little heat remained from the
previous pulse.
The pulse width MV (1C and 1D) controls the width of the
test pulse with a selector switch S3 (see Table 1 for capacitor
values). Fixed widths in steps of 1, 3 and 5 from 1 ms to 10
seconds in either a repetitive mode or single pulse is
available. For pulse widths greater than 10 seconds, a stop
watch was used with push button switch (S1) and with the
mode switch (S2) in the > 10 seconds position.
For all diodes with VZ greater than about 6 volts a resistor
voltage divider is used to maintain an input of about 6 V to
the first op amp (U4) so as not to overload or saturate this
device. The divider consists of R5 and R6 with R6 being
10 kΩ and R5 is selected for about a 6 V input to U4.
Precision resistors or accurate known values are required for
accurate voltage readout.
http://onsemi.com
120
In using the curves of Figure 5 and 6, it should be kept in
mind that VZ can be different at TE for the three mountings
because diode junction temperature can be different for each
mounting at TE which is represented by 1 on the VZ
normalized scale. Therefore, when the correlation of VZ
between parties is attempted, they must use the same type of
mounting or know what the delta VZ is between the two
mountings involved.
The Grayhill clips curves in Figure 6 are normalized to the
ON Semiconductor fixture at TE as in Figure 5. Figures 1
through 4 are normalized to VZ at TE for each diode and
would be used as Figures 5 and 6.
Measurement accuracy can be affected by test equipment,
power dissipation of the D.U.T., ambient temperature and
accuracy of the voltage divider if used on the input of the first
op-amp (U4). The curves of Figures 1 through 6 are for an
ambient temperature of 25°C, at other ambients, θVZ has to
be considered and is shown on the data sheet for the
1N5221B series of diodes. θVZ is expressed in mV/°C and
for the 1N5228B diode is about −2 mV/°C and for the
1N5242B, about 1.6 mV/°C. These values are multiplied by
the difference in TA from the 25°C value and either
subtracted or added to the calculated VZ depending upon
whether the diode has a negative or positive TC.
General Discussion
The TC of zener diodes can be either negative or positive,
depending upon die processing. Generally, devices with a
breakdown voltage greater than about 5 V have a positive
TC and diodes under about 5 V have a negative TC.
Conclusion
Curves showing VZ versus IZT pulse width can be used to
calculate VZ at a pulse width other than the one used to
specify VZ. A test circuit and method is presented to obtain
VZ with a single pulse of test current to generate VZ curves
of interest.
References
1. Al Pshaenich, “Double Pulsing S/H Increases System
Accuracy”; Electronics, June 16, 1983.
2. ON Semiconductor Zener Diode Manual, Series A,
1980.
http://onsemi.com
121
FIGURES 1 thru 8 — Conditions: Single Pulse, TA = 25°C, VZ IZT = 1/4 PD (Max) Each device normalized to VZ at TE.
AXIAL LEAD PACKAGES: MOUNTING STANDARD GRAYHILL CLIPS
1.06
VZ , ZENER VOLTAGE (NORMALIZED)
VZ , ZENER VOLTAGE (NORMALIZED)
1.06
1.04
VZ = 3.9 V
1.02
1
6.2 V
0.98
12 V
0.96
75 V
0.94
0.92
1
4
10
40 100
400 1K
4K 10K
PW, PULSE WIDTH (ms)
1.04
1.02
1
6.2 V
0.98
12 V
0.96
0.94
0.92
40K 100K
VZ = 3.9 V
1
Figure 1. DO-35 (Glass) 500 mW Device
40K100K
1.06
VZ , ZENER VOLTAGE (NORMALIZED)
VZ , ZENER VOLTAGE (NORMALIZED)
40 100
400 1K
4K 10K
PW, PULSE WIDTH (ms)
Figure 2. DO-41 (Glass) 1 Watt Device
1.06
1.04
VZ = 3.3 V
1.02
1
6.2 V
0.98
12 V
0.96
0.94
0.92
10
4
1
4
10
40 100
400 1K
4K 10K
PW, PULSE WIDTH (ms)
40K 100K
VZ = 3.9 V
1.04
1.02
1
0.98
6V
0.96
13 V
0.94
0.92
150 V
68 V
1
Figure 3. DO-41 (Plastic) 1.5 Watt Device
4
10
40 100
400 1K
4K 10K
PW, PULSE WIDTH (ms)
40K100K
Figure 4. Case 17 (Plastic) 5 Watt Device
http://onsemi.com
122
THREE MOUNTING METHODS: DO-35 AND DO-41
1.004
1.018
1.014
MOUNTINGS:
GRAYHILL CLIPS
MODIFIED, L = 3/8″
1N5228B (VZ = 3.9 V)
VZ , ZENER VOLTAGE (NORMALIZED)
1.012
STANDARD,
L = 11/16″
1.008
1.004
ON SEMICONDUCTOR
FIXTURE L = 1/2″
1
VZ , ZENER VOLTAGE (NORMALIZED)
1.022
VZ = 12 V
0.996
0.992
GRAYHILL CLIPS
STANDARD, L = 11/16″
0.992
ON SEMICONDUCTOR
FIXTURE
L = 1/2″
0.988
0.984
0.98
0.976
0.996
MOUNTINGS:
GRAYHILL CLIPS
STANDARD, L = 11/16″
MODIFIED, L = 3/8″
1N4742A
1
1
MODIFIED
L = 3/8″
4
10
40 100
400 1K
4K 10K
PW, PULSE WIDTH (ms)
Figure 6. DO-41 (Glass) 1 Watt Device
0.988
0.984
0.98
1N5242B (VZ = 12 V)
0.976
MOUNTING FIXTURE
0.972
0.968
1
4
10
40 100
400 1K
4K 10K
PW, PULSE WIDTH (ms)
40K 100K
40K 100K
1.69
Figure 5. DO-35 (Glass) 500 mW Device
.75
1.41
2.31
.78
Figure 7. Standard Grayhill Clips
http://onsemi.com
123
Figure 8. Zener Voltage Double Pulsing S/H Test Circuit
http://onsemi.com
124
27 k
S2B
100
100 k
3A
VDD
+12 V
1N914
**Tek Current Probe
**P6302/AM503
27 k
27 k
R1
22 pF
1/2 MC14001
U3
3B
330 k
100 μs
DELAY MV
2C
1B
680 k
+12 V
+12 V
510 pF
C1
10 k
MC14001
510 pF
200 μs
2B
T1 ≈ 2.2R1C1
1A
VDD
+12 V
U3
3/4 MC14001
0.001 μF
0.001 μF
2A
VDD
1N914
≈2 R1
+12 V
0.1
47 k
1N914
47 k
1N914
2D
47 k
12 k
1/4 MC14001
U2
0.001
μF
0.001 μF
2N
3906
22 k
+12 V
4.7 k
Q3
10 k
+12 V
3C
1k
MPSA42
Q4
510 pF
12 k
1
+12 V
C15
C5
C4
C3
C2
330 k
U3
1/2 MC14001
1C
EXTENDED POWER-PULSE MV
100 k
0.001 μF
MODE
SEL SW
S2A
FREE
RUN
ONE
SHOT
>10 SEC
START SW
S1
47 k
S3
100 k
DUT
R5
PULSE
WIDTH
MV
R6
10 k
Q1
2N3906
10 k
−12 V
−
+
U4
MC1741
VIN
+12 V
−12 V
1N914
22 k
+12 V
SAMPLE GATE MV
SEE
TABLE 1
1D
T2 ≈ 0.6C2 (R2 + R3)
P.W.
CONTROL
POWER AMPLIFER
**
RC
68 k
2W
Q5A*
MJE350
VCC ≤ 250 V
3D
13
3
2
R2
5M
R3
+12 V
VIN =
0.1 μF
0.1 μF
33 pF
2N4856
R5 + 10 k
VIN (10 k)
S
=
D
Q2
K
U5
LF155J
0.1 μF
S4
−
+
25 k
VCC ≤ 250 V
−12 V
+12 V
Q5A
MJE35
100 μs
100 μs
100 μs
DUT
68 k
2W
Q5B
MJE
5850
0.1
μF
0.1 μF
VO =
VIN
K
200 μs
VCC ≤ 250 V
ZERO
CONTROL
+ 12 V
RC
TIMING WAVEFORMS
*FOR DUT
CURRENTS:
200 mA ≤ IZT ≤ 2 A
VIN
1N914
12 k
1k
G-3C
G-3A
Q4
G-2C
G-2D
G-2A
GATE
G-1D
APPENDIX A
Recommended Incoming Inspection Procedures
Zener Voltage Testing
Pulsed versus Thermal Equilibrium
This section is primarily for use of incoming inspection
groups. The subject covered is the measurement of zener
voltage (VZ) and the inherent difficulty of establishing
correlation between supplier and buyer when using pulsed
test techniques. This difficulty, in part, is due to the
interpretation of the data taken from the variety of available
testers and in some cases even from the same model types.
It is therefore, our intent to define and reestablish a
standardized method of measurement to achieve correlation
no matter what test techniques are being used. This
standardization will guarantee your acceptance of good
product while maintaining reliable correlation.
stable junction temperatures). The device should also be
powered from a constant current source to limit changes of
power dissipated and impedance.
All of the above leads us to an understanding of why
various pulse testers will give differing VZ readings; these
differences are, in part, due to the time duration of test (pulse
width), duty cycle when data logging, contact resistance,
tolerance, temperature, etc. To resolve all of this, one only
needs a reference standard to compare their pulsed results
against and then adjust their limits to reflect those
differences. It should be noted that in a large percentage of
applications the zener diode is used in thermal equilibrium.
ON Semiconductor guarantees all of it’s axial leaded
zener products (unless otherwise specified) to be within
specification ninety (90) seconds after the application of
power while holding the lead temperatures at 30 ± 1°C, 3/8
of an inch from the device body, any fixture that will meet
that criteria will correlate. 30°C was selected over the
normally specified 25°C because of its ease of maintenance
(no environmental chambers required) in a normal room
ambient. A few degrees variation should have negligible
effect in most cases. Hence, a moderate to large heatsink in
most room ambients should suffice.
Also, it is advisable to limit extraneous air movements
across the device under test as this could change thermal
equilibrium enough to affect correlation.
DEFINITION OF TERMS
Temperature Coefficient (TC):
The temperature stability of zener voltages is sometimes
expressed by means of the temperature coefficient (TC).
This parameter is usually defined as the percent voltage
change across the device per degree centigrade, or as a
specific voltage change per degree centigrade. Temperature
changes during test are due to the self heating effects caused
by the dissipation of power in the zener junction. The VZ will
change due to this temperature change and will exhibit a
positive or negative TC, depending on the zener voltage.
Generally, devices with a zener voltage below five volts will
have a negative TC and devices above five volts will exhibit
a positive TC.
SETTING PULSED TESTER LIMITS
Pulsed test techniques do not allow a sufficient time for
zener junctions to reach TE. Hence, the limits need to be set
at different values to reflect the VZ at lower junction
temperatures. Since there are many varieties of test systems
and possible heatsinks, the way to establish these limits is to
actually measure both TE and pulsed VZ on a serialized
sample for correlation.
The following examples show typical delta changes in
pulsed versus TE readings. The actual values you use for
pulsed conditions will depend on your tester. Note, that there
are examples for both positive and negative temperature
coefficients. When setting the computer limits for a positive
TC device, the largest difference is subtracted from the
upper limit and the smallest difference is subtracted from the
lower limit. In the negative coefficient example the largest
change is added to the lower limit and the smallest change
is added to the upper limit.
Thermal Equilibrium (TE)
Thermal equilibrium (TE) is reached when the diode
junction temperature has stabilized and no further change
will occur. In thermal equilibrium, the heat generated at the
junction is removed as rapidly as it is created, hence, no
further temperature changes.
MEASURING ZENER VOLTAGE
The zener voltage, being a temperature dependent
parameter, needs to be controlled for valid VZ correlation.
Therefore, so that a common base of comparison can be
established, a reliable measure of VZ can only occur when
all possible variables are held constant. This common base
is achieved when the device under test has had sufficient
time to reach thermal equilibrium (heatsinking is required to
stabilize the lead or case temperature to a specified value for
http://onsemi.com
125
• Thermal equilibrium specifications:
VZ at 10 mA, 2.7 V minimum, 3.3 V maximum:
(Negative TC)
ON Semiconductor Zeners
• Thermal equilibrium specifications:
VZ at 10 mA, 9 V minimum, 11 V maximum:
(Positive TC)
TE
Pulsed
Difference
9.53 V
9.35 V
9.46 V
9.56 V
9.50 V
9.45 V
9.38 V
9.83 V
9.49 V
9.40 V
−0.08 V
−0.07 V
−0.08 V
−0.07 V
−0.10 V
TE
Pulsed
Difference
2.78 V
2.84 V
2.78 V
2.86 V
2.82 V
2.83 V
2.91 V
2.84 V
2.93 V
2.87 V
+0.05 V
+0.07 V
+0.05 V
+0.07 V
+0.05 V
Computer test limits:
Set VZ min. limit at 2.7 V + 0.07 V = 2.77 V
Set VZ max. limit at 3.3 V + 0.05 V = 3.35 V
Computer test limits:
Set VZ max. limit at 11 V − 0.10 V = 10.9 V
Set VZ min. limit at 9 V − 0.07 V = 8.93 V
http://onsemi.com
126
ON Semiconductor and
are registered trademarks of Semiconductor Components Industries, LLC (SCILLC). SCILLC reserves the right to make changes without further notice
to any products herein. SCILLC makes no warranty, representation or guarantee regarding the suitability of its products for any particular purpose, nor does SCILLC assume any
liability arising out of the application or use of any product or circuit, and specifically disclaims any and all liability, including without limitation special, consequential or incidental
damages. “Typical” parameters which may be provided in SCILLC data sheets and/or specifications can and do vary in different applications and actual performance may vary over
time. All operating parameters, including “Typicals” must be validated for each customer application by customer’s technical experts. SCILLC does not convey any license under
its patent rights nor the rights of others. SCILLC products are not designed, intended, or authorized for use as components in systems intended for surgical implant into the body,
or other applications intended to support or sustain life, or for any other application in which the failure of the SCILLC product could create a situation where personal injury or death
may occur. Should Buyer purchase or use SCILLC products for any such unintended or unauthorized application, Buyer shall indemnify and hold SCILLC and its officers, employees,
subsidiaries, affiliates, and distributors harmless against all claims, costs, damages, and expenses, and reasonable attorney fees arising out of, directly or indirectly, any claim of
personal injury or death associated with such unintended or unauthorized use, even if such claim alleges that SCILLC was negligent regarding the design or manufacture of the part.
SCILLC is an Equal Opportunity/Affirmative Action Employer. This literature is subject to all applicable copyright laws and is not for resale in any manner.
PUBLICATION ORDERING INFORMATION
LITERATURE FULFILLMENT:
N. American Technical Support: 800−282−9855 Toll Free
Literature Distribution Center for ON Semiconductor
USA/Canada
P.O. Box 61312, Phoenix, Arizona 85082−1312 USA
Phone: 480−829−7710 or 800−344−3860 Toll Free USA/Canada Japan: ON Semiconductor, Japan Customer Focus Center
2−9−1 Kamimeguro, Meguro−ku, Tokyo, Japan 153−0051
Fax: 480−829−7709 or 800−344−3867 Toll Free USA/Canada
Phone: 81−3−5773−3850
Email: [email protected]
© Semiconductor Components Industries, LLC, 2005
April, 2005 − Rev. 1
127
ON Semiconductor Website: http://onsemi.com
Order Literature: http://www.onsemi.com/litorder
For additional information, please contact your
local Sales Representative.
HBD854/D
Publication Order
Number:
NLAS3158/D