Understanding the Noise Issue Out of Inductor Based DC-DC Converter

AND8172/D
Understanding the Noise
Issue Out of Inductor Based
DC-DC Converter
Prepared by: Michael Bairanzade
ON Semiconductor
http://onsemi.com
APPLICATION NOTE
Abstract
NCP5007/NCP5005 Basic Analysis
Portable applications, such as cellular phones and
hand-held computers, have sensitive electronic functions,
making any electromagnetic noise a real issue. Even a low
power converter can generate noise during transients since
the converters operate in the megahertz range. The high
frequency switching can interact with the parasitic circuit
elements to create uncontrolled noise. In addition, the
associated voltage spikes might be difficult to filter out, a
detailed analysis of the converter operation being necessary
to properly design the system. Of course, the mechanical
layout is a key parameter to minimize the EMI perturbation.
The main purpose of this document is to clarify the source of
such a noise, and assess the risk of system's EMI pollution.
The NCP5007 device, together with a PWM based structure,
will be used as vehicles to perform the associated
engineering test in a real application, but the same may apply
to other Boost converters, such as the NCP5005.
The NCP5007/NCP5005 are dc-dc converters dedicated
to the supply of High Efficiency White LEDs used as an
LCD back light or photo flash light source. Generally
speaking, with a typical 3.8 V Vf drop during normal
operation, these LEDs are not capable to operate from a
standard 3.6 V battery pack. A boost circuit must be
provided to raise the voltage up to the bias called by such a
diode. Moreover, these LEDs are commonly connected in
series to get a constant current through the diodes, assuring
smooth and consistent light across the LCD panel.
The basic application, depicted in Figure 1, drives the
back light of a standard LCD display, the command bit EN
being provided by an external CPU or other digital
controller.
Vbat
U1
Vbat
GND
2
1
C1
5
4.7 mF
L1
NCP5007
EN
GND
22 mH
3
Pulse
D1
GND
Vout
4
FB
MBR0530
C2
1.0 mF
R1
D5
D4
D3
D2
GND
GND
47 W
LWT67C
LWT67C
LWT67C
LWT67C
Figure 1. Typical Back Light Application
© Semiconductor Components Industries, LLC, 2008
February, 2008 - Rev. 1
1
Publication Order Number:
AND8172/D
AND8172/D
With a 10 W sense resistor, the dc current through the
LEDs is 20 mA with a "10% tolerance over the
temperature range. The output voltage will be 19 V ( typical)
and the 1.0 mF capacitor must be sized accordingly. Using
ceramic type X5R is mandatory for both C1 and C2.
The EN signal (Enable, pin 3 ) is useful to power ON/OFF
the diodes, the chip providing a constant current to the LEDs
when EN = Vbat. On the other hand, the EN signal can also
be used to control the brightness of the back light by means
of a PWM signal applied to pin 3.
Basically, the chip operates with two cycles in a PFM
mode:
Cycle #1: the energy is stored into the inductor
Cycle #2: the energy is dumped to the load
Of course, from a practical standpoint, the inductor must
be sized to cope with the peak current present in the circuit
to avoid saturation of the core. On top of that, the ferrite
material shall be capable to operate at high frequency
(>1.0 MHz) to minimize the Foucault's losses developed
during the cycles.
The waveforms captured on the NCP5007 Demo Board,
V3.10, show two voltage spikes VSPK#1 and VSPK#2
across the output voltage (Figure 2). The VSPK#1 signal is
synchronized with the positive going slope of the output
voltage, the VSPK#2 signal being synchronized with the
negative slope of the same voltage. Both of these signals
come on top of the voltage ripple. Generally speaking, such
spikes will be named “noise'', but we do not know what is
the energy content and what could be the influence of the
electrical environment. At this point, let us consider the
switching operation only since there are no spikes outside
the transients as demonstrated by the waveforms.
Figure 2. NCP5007 Normal Operation
http://onsemi.com
2
AND8172/D
The silicon chip is connected to the external world by
means of bonding wires attached to the internal pads. The
bonding wires have finite inductance and stray capacitance.
In addition, the internal overvoltage protection circuits and
PCB trace leads contribute to such parasitic elements.
The simplified PSPICE model (Figure 3), includes the
parasitic inductances associated with the NCP5007 internal
L1
1
bonding, and the stray capacitance between the output pin
and ground. The load is built with the White LED LWT67C
model (courtesy: OSRAM GmbH) with a 1.0 mF reservoir
capacitor in parallel. The inductor includes the ESR and
stray capacitance developed by the wiring. The ESR of the
printed circuit tracks and wiring is neglected as it is not a first
order importance, unless the layout is really poor.
R1
2
22 mH
1.2 R
C4
Stray Capacitance
Schottky Stray Capacitance
C1
5 pF
140 pF
V1 +
3.6 V -
NCP5007 Chip
L2
1
D4
2
10 nH
Cdg
3
NMOS
2
Cgs
Pad
M1
1
MBR030
D1
LW_5413-TYP
Cds
C3
15 pF
C2
1 mF
2
D2
LW_5413-TYP
D3
LW_5413-TYP
L3
10 nH
Pad
Parasitic Inductance
Stray Capacitance
Figure 3. PSPICE Output Circuit Model
http://onsemi.com
3
AND8172/D
SPIKE VSPK#1
The period T is derived from:
Assuming the system operates in steady state, let us
consider the noise VSPK#1. The current flowing into the
NMOS is abruptly switched off and the voltage rises
(according to the Lenz's law) until the Schottky diode is
forward biased. The inductor current will be diverted to the
load when the output voltage will be higher than Vout + Vf.
During the positive going of the output voltage slope, the
stray capacitances and the parasitic inductances create an
LC network, yielding an oscillation as the NMOS device is
switched off. According to the MBR0530 data sheet, the
intrinsic stray capacitance varies from 170 pF (when
Vr = 0 V) to 40 pF when Vr = 12 V.
Let us consider Vr = 10 V since Vout = S Vf*LED, then:
F+
1
2 * P * ǸLC
F+
1
+ 178MHz
2 * P * Ǹ20 * 10-9 * 40 * 10-12
1
T+1+
+ 5.6ns.
F
178 * 106
(eq. 2)
This value is well within the signal given Figure 4,
captured during the test performed with the NCP5007 demo
board. Of course, the frequency varies along the output
voltage slope since all the semiconductor based capacitance
varies as the voltage varies across the junctions.
Beside the voltage effect, one must consider the level of
energy radiated by such a spike. In this application, the
energy is transferred from the parasitic inductance to
the stray capacitance, as depicted by the waveforms
(oscillations). Consequently, we can calculate the amount of
energy stored into the parasitic inductor:
Ej + 1 * L * I2
2
(eq. 1)
(eq. 3)
Ej + 1 * 20 * 10-9 * 0.062 + 36pJ
2
Such level of energy is extremely low and will not pollute
the environment.
Note: The dI/dt of the NCP5007 internal NMOS is 10 mA/ns, a very fast transient.
Figure 4. NCP5007 Output Voltage Noise at NMOS Turn Off
http://onsemi.com
4
AND8172/D
SPIKE VSPK#2
This energy is partially dissipated into the ohmic
resistance, partially into the RDS(on) of the internal NMOS
device (1.7 W typical for the NCP5007), yielding a
1000 mVpp spike as depicted in Figure 5.
Although visible, a 5.7 nJ amount of energy is likely too
low to pollute the environment of the power device.
Moreover, the energy is not radiated by an external
inductance element, but is limited to the parasitic structures
only, thus less prone to trouble the rest of the application.
The spike is developed during the negative going slope of
the output voltage, when the current transfer to the load is
completed and the NMOS switch turns on (Figure 5).
At this point, the voltage across the Schottky diode
reverses and the same parasitic structures are activated,
yielding the oscillation depicted by the top trace (Figure 5).
The energy is now transferred from the charged stray
capacitance to the parasitic inductor, yielding the
oscillations. Assuming the total stray capacitance is 80 pF
(NCP5007 pad, PCB, Schottky), we get:
Ej + 1 * C * V2
2
(V + sumofthewhiteLEDVf)
(C + totalstraycapacitance, firstorderonly)
Ej + 1 * 80 * 10-12 * 122 + 5.7nJ
(eq. 4)
2
Figure 5. NCP5007 Output Voltage Noise at NMOS Turn On
http://onsemi.com
5
AND8172/D
Discontinuous Mode of Operation Analysis
diode as a minimum parasitic capacitance since the reverse
voltage is maximum), the frequency is 6.2 MHz, yielding a
161 ns period. The calculated values are well within the
evaluation tests carried out with a PWM based white LED
driver used as a reference (Figure 6, top trace).
At this point, one can derive the level of energy radiated
during this period:
When a chip operates in a fully discontinuous mode, a
large peak current takes place in the inductor as depicted in
Figure 6. The PWM mode is prone to such a discontinuous
cycle. The output voltage is largely different, in comparison
with the NCP5007, as one expects from a discontinuous
mode of operation. The main consequence is the large
oscillations developed when the transfer of the current
between the inductor and the load is completed. Of course,
internal circuit can be implemented to avoid such oscillation
(so-called ring killer), but all ICs might not have such
features. The oscillations are no longer the only
consequence of the Schottky diode, but come from the main
inductor (22 mH) associated with the stray capacitances.
Assuming we have a 30 pF stray capacitance (the Schottky
Ej + 1 * L * I2 (I = level of the current during the oscillation
2
first order only)
1
Ej + * 22 * 10-6 * 0.042 + 4.4nJ
(eq. 5)
2
Although such a level is low, is it generated by the main
inductor and this element is prone to radiate most of this
energy out of the core, unless a shielding magnetic core is
used.
Figure 6. Typical PWM Normal Operation
http://onsemi.com
6
AND8172/D
1.7 W, compared to 0.5 W for the PWM chip. The
consequence is twofold:
1. With a lower RDS(on), the PWM based chip has
probably larger Cdg, Cgs and Cds capacitances,
thus slower dI/dt. Consequently, slower dI/dt
minimizes the effect of the parasitic inductance.
2. Similarly, lower RDS(on) generates lower voltage
spike for the same amount of energy transferred
from the stray capacitance.
A detailed view of the PWM based chip output voltage
shows a non-negligible spike when the inductor current hits
the maximum peak value (Figure 7).
Basically, this is the same mechanism as the one depicted
for the NCP5007 application, and a similar voltage spike is
developed with 800 mVpp amplitude.
Leaving aside the structure, the main difference between
the PWM mode and the NCP5007 from a power device
standpoint, is the RDS(on) of the internal NMOS. We have
Figure 7. PWM Chip Output Voltage Noise as the NMOS is Switched Off
http://onsemi.com
7
AND8172/D
SPECTRUM ANALYSIS
It is clear that, although the apparent voltage spikes are
higher for the NCP5007, the net result, in terms of real noise,
is much better for the NCP5007 in comparison with the
PWM circuit as depicted by the records.
In order to better analyze the spikes coming from the
chips, the noise return to battery has been measured and
recorded (Figure 10). The spectral density, mathematical
integration of the two curves, yields 0.826 mVrms for the
NCP5007, and 238 mVrms for the PWM based chip, both
over the 100 kHz to 30 MHz bandwidth.
Once again, the behavior of the NCP5007 is far better than
the PWM based white LED driver, particularly in the
100 kHz to 1.0 MHz frequency band.
The waveforms captured by an oscilloscope are very
useful to evaluate the behavior of a given product, but they
don't give information about the EMI performance. At this
point, a spectrum analyzer is necessary to characterize the
device over the considered bandwidth.
The spectrum analysis has been performed with the same
demo board, the antenna being located 10 mm above the
inductor. Although the tests have not been performed in a
shielded cage, the results provide a good understanding of
the situation in terms of EMI. The waveforms are provided
in Figures 8 and 9.
Figure 8. NCP5007 Spectrum Analysis (Unit: dB mV): Relative EMI
http://onsemi.com
8
AND8172/D
Figure 9. PWM Circuit Spectrum Analysis
http://onsemi.com
9
AND8172/D
Figure 10. Reflected Noise to Battery
CONCLUSION
Figures 8 and 9, demonstrates the excellent EMI behavior of
the NCP5007 with much lower noise, in comparison with
the PWM based white LED driver operation. The same
comment applies for the reflected noise to battery.
The waveforms are very useful to fully characterize the
system, but it is of prime importance to identify and quantify
the risks, in terms of noise and reliability, of the voltage
spikes when a detailed analysis is requested.
Since it is not possible to reduce the parasitic inductance
to zero (using the micro bump technique could improve but
not reduce the inductance to zero nH), and since fast dI/dt is
necessary to lower the switching losses into the chip, spikes
voltage might exist.
However, it is mandatory to consider the behavior of the
power device in the real environment, keeping into account
all the parasitic effects. The spectrum analysis, provided in
http://onsemi.com
10
AND8172/D
PSPICE SIMULATIONS
Using a lower RDS(on) yields the expected lower spike and
a low stray capacitance diode provides a much lower noise
during the transients, as expected.
The circuit has been simulated prior to run the analysis in
the laboratory. The same schematic (Figure 11) has been
used to evaluate the configurations.
Vbat
2
Vbat
L1
22 mH
+
3.6 V
1
-
C1
5 pF
R1
1.5 R
1
L5
1 nH
D1
R2
2
0.1 R
MBR030
2
2
L2
10 nH
L4
20 nH
1
1
2
3
M1
NMOS
IC = 0 V
1 mF
C4
C5
20 pF
D6
LW_5413-TYP
1
V1 = 0 V
V2 = 5.0 V
TD = 10 ns
TR = 10 ns
TF = 10 ns
PW = 1.0 ms
PER = 3.0 ms
+
-
D7
LW_5413-TYP
V2
2
D5
LW_5413-TYP
L3
10 nH
1
Figure 11. PSPICE Schematic Diagram
http://onsemi.com
11
AND8172/D
The parasitic elements are based on the existing package and silicon designs. The value of the stray capacitance ( C5)
depends, besides the internal structure, upon the PCB layout.
Test Conditions
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Temperature
Oscilloscope
Voltage Probes
Current Probe
Input Power Supply
Input Power Supply Bias
Output Load
Output DC Current
Inductor
Spectrum Analyzer
Spectrum Analyzer
EMI probe
HF probe
:Room, +20°C
:Tektronix, TDA754/D, operating in the DPO mode
:Tektronix P6139A
:Tektronix TCP202
:Tektronix PS2520G
:3.60 V
:3 White LED
:NCP5007 = 10 mA, PWM based chip = 18 mA (Rsense = 22 W)
:22 mH
:ROHDE & SCHWARZ, FSIQ 7
:HP4195A
:HP11941A
:HP41800A
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:
Literature Distribution Center for ON Semiconductor
P
.O. Box 61312, Phoenix, Arizona 85082-1312 USA
Phone: 480-829-7710 or 800-344-3860 Toll Free USA/Canada
Fax: 480-829-7709 or 800-344-3867 Toll Free USA/Canada
Email: [email protected]
N. American Technical Support: 800-282-9855 Toll Free
USA/Canada
ON Semiconductor Website: http://onsemi.com
Order Literature: http://www.onsemi.com/litorder
Japan: ON Semiconductor, Japan Customer Focus Center
2-9-1 Kamimeguro, Meguro-ku, Tokyo, Japan 153-0051
Phone: 81-3-5773-3850
http://onsemi.com
12
For additional information, please contact your
local Sales Representative.
AND8172/D