Application Notes

Subject
:
MHN-TD 70W Driver with UBA2030
Pages: 39
Summary
: This report describes a lamp driver concept intended for the MHL-TD 70W
lamp (Metal Halide Lamp) used in general lighting applications. The circuit
is a cascade of a buck converter and commutator. The latter is driven by
the UBA2030. This concept is principally used to investigate the lamp
behaviour, so it does not contain a preconditioner and is not EMC tested.
Keywords
: MHN lamp; UBA2030; Buck Converter; Commutator; Igniter
Philips Semiconductors
2
TABLE OF CONTENTS:
1. INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2. LAMP DATA.. . . . . .
2.1
Ignition Phase . .
2.2
Take-over phase
2.3
Run-up Phase. .
2.4
Burn Phase . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
7
7
7
7
7
3. GENERAL CIRCUIT DESCRIPTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.1
Power Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2
Commutator & Igniter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4. POWER CONTROL. . . . . . . . .
4.1
Current Sensing. . . . . . .
4.2
Power Control Management
4.3
Watch-dog Timer . . . . . .
4.4
Ignition Voltage Control . . .
4.5
Level Shift Driver . . . . . .
4.6
Power Components . . . . .
4.6.1
Buck Coil . . . . .
4.6.2
Power Mosfet . . .
4.6.3
Diode . . . . . . .
4.6.4
Buck Capacitor . .
4.7
Control Loop . . . . . . . .
4.7.1
Power Conversion
4.7.2
Measuring Circuit .
4.7.3
OTA in IC1 . . . .
4.7.4
Modulator in IC1 .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
11
11
11
13
14
14
15
15
15
16
16
17
17
17
17
18
5. COMMUTATOR & IGNITER. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.1
Commutator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.2
Igniter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
6. WAVEFORMS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
7. CONCLUSIONS & RECOMMENDATIONS. . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
APPENDIX 1 CALCULATIONS. . . . . . . . . .
A.1.1 Current Sensing. . . . . . . . . . . . .
A.1.2 Power Control Management . . . . . .
A.1.3 Watch-dog Timer . . . . . . . . . . . .
A.1.4 General Characteristics Buck Converter
A.1.5 Power Mosfet Losses . . . . . . . . . .
A.1.6 Diode Losses . . . . . . . . . . . . . .
APPENDIX 2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
29
29
29
32
32
33
36
COMPLETE CIRCUIT DIAGRAM. . . . . . . . . . . . . . . . . . . . . . . . . . 37
Philips Semiconductors
3
8. REVISION HISTORY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Philips Semiconductors
4
1.
INTRODUCTION.
This report describes a lamp driver circuit intended for the MHL-TD 70W lamp (Metal Halide Lamp)
belonging to the group of compact HID lamps (High Intensity Discharge lamps).
In contrast with TL and CFL lamps, HID lamps are high pressure discharge lamps. The presence of
iodine will make starting more difficult compared to the low pressure discharge lamps mentioned
before. An ignition voltage up to 25 kV can be needed to ignite an HID lamp in case of a hot re-strike.
For more details about lamp behaviour, see chapter 2.
The HID lamps are applied in the automotive area and the general lighting area. Both application areas
use the same lamp type but the operating requirements are rather different, so circuit topologies
depend strongly on the application area. The system described in this report gives a driver concept for
a general lighting application. This concept is principally used to investigate the lamp behaviour, so it
does not contain a preconditioner and EMC is not tested and measured.
Philips Semiconductors
5
Philips Semiconductors
6
2.
LAMP DATA.
The available lamp data is given in the next sub sections. Each sub section gives the characteristics of
the lamp valid for the corresponding operating phase.
2.1
Ignition Phase
During the ignition phase, especially in case of a hot re-strike when the lamp is still warm from previous
operation, a 25 kV voltage pulse should be supplied to the lamp with a fast rise-time. The total
breakdown occurs within 10 ns. The voltage level is more important than the rise-time. The ignition
voltage in case of a cold lamp amounts about 4-5 kV. So the ignition voltage increases with increasing
lamp temperature.
2.2
Take-over phase
After ignition, the lamp voltage breaks down to a quarter of the nominal lamp voltage (20 V) and a
conducting channel is produced. A take-over current should be applied in less than 1 ms preventing
cooling down of the lamp. In most cases, the take-over current is derived from a capacitor which is
loaded to a take-over voltage during the ignition phase and discharges to the lamp break-down voltage
immediately after break down (20 V). The take-over current depends on the lamp resistance at break
down. If the take-over current (or voltage) is too low, the lamp will extinguish and re-ignition is
necessary to fire the lamp once more. Here again, the take-over voltage increases with increasing
lamp temperature.
2.3
Run-up Phase
The lamp is characterised by the absence of a perceptible glow to arc transition. The lamp desires a
relative high run-up current preventing extinguishing of the lamp. When the run-up current is too large,
the electrode may be melting. The lamp voltage swells gradually to the nominal lamp voltage (85 V).
The required run-up time is about 2 minutes and depends on the used luminaire.
2.4
Burn Phase
The lamp is designed to be driven with an AC current to avoid difference in electrode temperature
which can cause acoustic resonance of the arc. However, re-ignition voltage pulses may be present
when the lamp current is interrupted for a relative long time i.d. at the moment that the lamp current
commutates. For example, the interruption time of a lamp driven with a sinusoidal voltage at 50 Hz will
be relative large. The interruption time of a commutation circuit can be very short (≤ 2 µs)
Acoustic resonance occurs in the frequency domain: 10 kHz - 1 MHz, however some free-of-acousticresonance windows do exists. The commutating frequency of the full-bridge should be limited to the
domain: 50 Hz - 10 kHz to avoid any risk on acoustic resonance. In practise, the commutating
frequency range is limited to 100 Hz - 400 Hz.
The nominal lamp voltage (mean value 85 V) depends on the lamp type and the lamp aging effect,
65 V - 105 V. The differential resistance of the lamp is rather small and negative. To obtain a stable
operating point, an impedance in series with the lamp is needed so that an ideal voltage source is
Philips Semiconductors
7
forbidden. The best lamp performance is achieved when the lamp is driven by a current source with
power regulation. Lamp power control is important for the life-time of the lamp.
Philips Semiconductors
8
3.
GENERAL CIRCUIT DESCRIPTION.
The complete circuit exists of two converters, namely:
• the buck converter which controls the lamp power and behaves like a current source.
• the full bridge converter which commutates the DC current from the buck converter into an AC
current. Furthermore it supplies the igniter to generate ignition pulses.
The block diagram is given in figure 1 and the complete circuit diagram in figure 20 on page 37.
Power Control
Commutator
+ 380 Vdc
PHP10N40E
L1
PHP10N40E
500 uH
PHP10N40E
T2
T4
Igniter
T1
C1
820 nF
D1
PHP10N40E
BYV29F/500
T3
Rs
IC2
IR2101
IC1
Level
Lamp
Control
MHN-TD 70W
PHP10N40E
T5
IC3
MC34262
Shifter
Lamp
UBA2030
Fig.1 block diagram
3.1
Power Control
The power control is achieved by a buck converter in SOPS mode running at a relative high frequency
of ± 70 kHz @ 85 V output voltage (lamp voltage). The SOPS mode operation is chosen to minimize
the switch-on losses in the power mosfet. The lamp power is stabilised by a feedback signal formed by
the addition of lamp voltage and lamp current to get a power controlled current source. The control IC1
is MC34262 which normally is used as power factor correction IC. The lamp power as function of lamp
voltage is parabolical. The top of the parabola, the nominal lamp power Pn and nominal lamp voltage
Un, are set to (Pn, Un) = (70 W, 85 V). The maximum lamp current is limited to 1 A, so the parabolical
relation is not valid for lamp currents smaller than 1 A.
The maximum output voltage of the buck converter would be 2*Un (170 V), due to the parabolical
function, see figure 2 on page 13. So an ignition voltage control circuit is added to replace the power
control by a voltage control when the voltage is larger than 140 V and forces the buck converter to act
like a voltage source of 340 V. Now, the output voltage range is >0 - 340 V, where the 340 V is used
for the igniter to generate ignition pulses.
Philips Semiconductors
9
Immediately after break-down, the output voltage breaks down to zero and the buck converter is not
able to operate due to the missing restart pulses. A Watch-dog Timer function is added to keep the
buck converter running, see section 4.3.
Finally, a level-shift IC2 (IR2101) is used to drive the power mosfet of the buck converter.
3.2
Commutator & Igniter
The commutation is achieved by a full bridge converter driven by the UBA2030 with a commutation
frequency of ± 100 Hz.
The igniter is connected between the two mid-points of the full bridge and generates ignition pulses of
4 kV peak with a duration of 100 µs and a repetition frequency of 100 Hz (commutator frequency). The
breakdown device is a sidac (break-over diode) with a typical breakdown voltage of 340 V.
Philips Semiconductors
10
4.
4.1
POWER CONTROL.
Current Sensing
The current sensing block measures the buck converter coil current by three in parallel connected
resistors R1, R2 and R3 called RS (1.57 Ω) followed by a voltage divider network R4 and R5 (both
100 Ω) to obtain information about the actual peak current in the buck coil, see figure 20 on page 37.
The peak current is controlled by a feedback signal at pin 1 of the IC1. The maximum peak current is
set by an internal voltage limiter inside the IC, named Uth(max) = 1.5 V. So the maximum lamp current
Ila,max is set to 1 A, see equation 1. The complete derivation is given in equation 13 on page 29.
R5
k U th, max
I la, max = --- ⋅ --------------------- …with …k = 1 + -----Rs
2
R4
(1)
Data: Uth,max = 1.5 V, Rs = 1.57 Ω, R5 = 100 Ω and R4 = 100 Ω.
Calculations: Ila,max = 0.96 A.
Another sensing signal is evident for proper operation of the buck converter namely, the zero current
detection at pin 5 of IC1. This signal senses indirectly the inductor current by monitoring the voltage
across the auxiliary winding on the buck coil. When it falls below the level of 1.4 V, it initiates the turnon of the buck mosfet.
During the run-up phase, the output voltage of the buck converter oscillates at every commutation for
some milliseconds down to zero, see section 5.1 on page 19. The zero current detection circuit is not
able to generate restart pulses because the secondary voltage stays for a relative long time lower than
1.4 V. To overcome this problem a Watch-dog Timer is added to keep the buck converter going on,
see section 4.3 on page 13.
4.2
Power Control Management
The feedback signal should be a multiplication of the lamp voltage and -current but for simplicity an
addition is used to obtain power control. But the floating lamp makes sensing difficult, so the lamp
voltage is sensed across the buck converter capacitor C1 and the lamp current is represented by the
voltage across Rs (1.57 Ω). The indirect lamp current sensing is valid because the average buck coil
current is equal to the lamp current (SOPS mode) and the average voltage across C1 is equal to the
lamp voltage. Addition of both lamp voltage and -current results in a parabolical relation between lamp
power and lamp voltage. Lamp power variation due to lamp voltage variation are the smallest when the
nominal lamp power is set to the top of the parabola.
The lamp voltage is measured by Ru (R6 in series with R7 → 550 kΩ) and the lamp current by Ri
(8.2 kΩ) which is the sum of Ri1 (R10 in parallel with R11 → 4 kΩ) and Ri2 (R8 in parallel with R9 →
4.2 kΩ). Equation 2 gives the relation between the nominal lamp power Pn, nominal lamp voltage Un,
the internal reference voltage Uref and sensing resistors Ru, Ri and Rs.
Un
Ru
α n = ------- = 2 ⋅ ---------- – 1
U ref
Ri
Philips Semiconductors
11
2
Un
R s = -----------------Pn ⋅ αn
(2)
Substituting the lamp data and IC data gives the values for the used components.
Data: Uref = 2.5 V, Pn = 70 W and Un = 85 V.
Calculations: αn = 67, Rs = 1.54 Ω.
Rs is formed by three PR02 resistors of 4.7 Ω connected in parallel, so the actual value of the sense
resistor Rs is 1.57 Ω.
The complete derivation of Pla(Ula) and Ila(Ula) is given in equation 14 through 20 in appendix A.1.2 on
page 29 and the end result is given in equation 3 and 4. Figure 2 gives the corresponding curves.
Equation 3 gives the relation between Pla(Ula) and Ila(Ula). Two regions are distinguished, namely:
• region 1, the region where Pla(Ula) is linear and Ila(Ula) constant.
• region 2, the region where Pla(Ula) is parabolical and Ila(Ula) linear.
Both are caused by the lamp current limitation Ila,max described in section 4.1.
Equation 4 gives the boundary values between region 1 and region 2, namely:
• Ula,min, the minimum lamp voltage where the parabolical relation of Pla(Ula) is valid.
• Ula,max, the maximum lamp voltage due to the parabolical relation of Pla(Ula).
• Pbr, the lamp power at the breakpoint. Note that P br ≤ P n .
Pn
U la

 I la = ------- ⋅  2 – --------
Un
Un


 I la = I la, max
U la ≥ U la, min
U la ≤ U la, min
U la
U la

 P la = P n ⋅ -------- ⋅  2 – --------
Un
Un


 P la = U la ⋅ I la, max
I la, max ⋅ U n

U la, min = U n ⋅  2 – -----------------------------
Pn


(3)
U la ≥ U la, min
U la ≤ U la, min
I la, max ⋅ U n

P br = I la, max ⋅ U n ⋅  2 – -----------------------------
Pn


(4)
From calculations in section 4.1, we know that the maximum lamp current Ila,max = 0.96 A. Substitution
of Ila,max in equation 4 gives a value for Ula,min and Pbr.
Data: Ila,max = 0.96 A, Pn = 70 W and Un = 85 V.
Calculations: Ula,min = 71 V and Pbr = 68 W.
Note that the maximum output voltage is limited to 2.Un = 170 V due to the position of the parabola.
This means that the igniter can not generate ignition pulses. An Ignition Voltage Control circuit solves
this problem, see section 4.4.
Philips Semiconductors
12
Pla[W]
Pn
Pbr
2.Un
Ula[V]
Ila[A]
Region 1
Ila,max
Region 2
In
Ula,min
Un
Ula[V]
Fig.2 Pla(Ula) and Ila(Ula)
4.3
Watch-dog Timer
The buck converter output voltage range is rather large >0 - 340 V. After ignition, the output voltage
breaks down to zero and the zero current detection signal, which initiates the turn-on of the buck
mosfet, is not able to generate pulses anymore see section 4.1. So a watch-dog timer is applied to
generate a restart pulse when IC1 does not generate pulses for a time longer than twd seconds. The
delay time twd is given by equation 5 derived in equation 21 on page 32 where E is the low voltage
supply, τwd = R12.C5 and Uref(s) the system reference voltage.
E
t wd = τ wd ⋅ ln  ---------------------------
 E – U ref ( s )
(5)
Data: E = 15 V, Uref(s) = 5 V, R12=12 kΩ and C5 = 10 nF. Calculations: τwd = 120 µs and twd = 49 µs.
The watch-dog timer key components are R12, C5, TR6 and IC4-1.
Philips Semiconductors
13
4.4
Ignition Voltage Control
Due to power control method described in section 4.2 on page 11, the maximum output voltage of the
buck converter is limited to 2.Un = 170 V. The Igniter circuit needs 360 V, so an Ignition Voltage
Control circuit is added to solve this problem. The Ignition Voltage Control circuit modifies the feedback
loop twofold. The current sensing information across Rs is inhibited so as a result the buck converter
behaves like a voltage source. Simultaneously, the feedback resistor divider ratio is adapted by the
open collector output of IC4-2 to generate the correct voltage on C1 needed for ignition UC1 = UC1,ign.
The Ignition Voltage Control circuit is activated when the output voltage of the buck converter UC1 is
larger than the trip-level Utrip. Utrip is fixed by R20, R21, R22 and Uref(s), see equation 6.
Ru
U C1, ign =  1 + -------- ⋅ U ref

R 
i1
U trip
R 20 + R 21
=  1 + ------------------------- ⋅ U ref ( s )


R
(6)
22
Data: Uref = 2.5 V, Ru (R6 in series with R7) = 550 kΩ, Ri1 (R10 in parallel with R11) = 4 kΩ.
Uref(s) = 5 V, R20 = 150 kΩ, R21 = 120 kΩ and R22 = 10 kΩ.
Calculation: UC1,ign = 346 V.
Utrip = 140 V.
Note: UC1,ign is an averaged value. Just after ignition, UC1 is smaller than 360 V due to the large
ignition currents. So, just before ignition, UC1 must larger than 360 V to retain the averaged value of
346 V.
4.5
Level Shift Driver
The Level Shift Driver function is necessary to drive the floating power N-mos TR1. The most important
characteristic of the driver IC2 is that the propagation delay between input -and output signal is small
and the maximum gate drive current is large. Both take care of fast turn-on and turn-off switching of the
power mosfet to minimize switching losses. However, the gate drive current of the applied Level Shift
Driver is rather small so that the mosfet is turned on with a small delay and optimal switching is not
achieved.
The driver sink current and the total parasitic capacitance parallel with the mosfet determine the switch
off losses of the mosfet, see section 4.6.2.
The floating power supply is formed by bootstrap diode D3 and bootstrap capacitor C6 which is
charged to 15 V when the buck diode D1 is conducting. During ignition, the current through the buck
diode is zero and the voltage on the source of the power mosfet stays larger than 15 V so that the
bootstrap diode is not loaded anymore and the mosfet can not be turned on. Now, an additional
winding on the buck coil L1 provides the floating power supply. The polarity is in phase with the buck
output voltage.
Philips Semiconductors
14
4.6
4.6.1
Power Components
Buck Coil
The buck coil L1 is an EF32 -core with 3C85 as core material and 63 turns 10*0.2 mm litzwire. Two
auxiliary windings of 6 and 3 turns are used for successively zero current detection and additional
floating supply for the level shift driver IC1, see respectively section 4.1 and 4.5. The value of the
inductor is ± 490 µH and the wire resistance 240 mΩ.
The operating frequency fop depends on the value of the buck coil as you can see in equation 7.
Equation 7 is derived from equation 22 through 24 on page 33 and is only valid when T osc « 2 ⁄ f op . Tosc
is the freewheel oscillation period time caused by the buck coil L1 and Cpar the total parasitic
capacitance from source to ground, Cpar = Cgd +Cds + CL1 + CD1. The oscillation starts when the
current through the buck diode D1 drops to zero and stops conducting.
f op
4.6.2
U la
1 – -------U la ⋅ U n
U in
= ---------------------- ⋅ ----------------2 ⋅ L ⋅ Pn
U la
2 – -------Un
 U la ≥ U la, min

 I la ≤ I la, max
(7)
Power Mosfet
The power mosfet TR1 is of the Philips type PHP10N40E (BUK457/400) with a typical
Rds(on) = 0.55 Ω. The conduction losses Pm,c are given by equation 8. The complete derivation is given
in equation 26 on page 33.
2
P m, c
P la
4
= --- ⋅ R ds ( on ) ⋅ --------------------3
U la ⋅ U in
(8)
Data: Pn = 70 W, Un = 85 V, Uin = 380 V and Rds(on) = 0.55 Ω.
Calculations: Pm,c = 111 mW.
The switch-on losses Pm,son are given by equation 9. The complete derivation is given in equation 27
and 28 on page 34. The operating frequency fop of the buck converter is determined by the value of the
buck coil L1, the input voltage Uin, the output voltage Ula, the output power Pla and the total parasitic
capacitance at the source of the mosfet Cpar. Cpar and L1 determine the freewheel oscillation period
time Tosc when the current through the buck diode drops to zero.
1
2
P m, son = f op ⋅ --- ⋅ C par ⋅ ( U in – 2 ⋅ U la )
2
T osc 2 1
C par =  ------------ ⋅ -- 2 ⋅ π L
2
f op
U la
U la
= ----------------------- ⋅  1 – --------
2 ⋅ L ⋅ P la 
U in
Philips Semiconductors
15
(9)
Data: Pn = 70 W, Un = 85 V, Uin = 380 V, Tosc = 1.5 µs, L = 500 µH
Calculations: fop = 80 kHz, Cpar = 115 pF, Pm,son = 200 mW
The switch-off losses are described in equation 29 through 31 on page 36 and the end result is given in
equation 10. Equation 10 gives an approximation of the real switch-off losses where some non-linear
quantities like Cds, Cdg and gm are supposed to be linear. So, equation 10 gives merely the switch-off
behaviour and not the quantitative switch-off losses, with Cp = Cds + CL1 + CD1.
˙

1
2  Î L
P m, soff = f ⋅ --- ⋅ C gd ⋅ U in ⋅  ---- – k
2
 Ig 
(10)
Cp
k = 1 + ---------C gd
From equation 10, we can see that the switch-off losses can be reduced by increasing the gate sink
current Ig or increasing the capacitance Cp. The disadvantage of increasing Cp is that the switch-on
losses will increase due to the increase of capacitance Cpar in equation 9, Cpar = Cgd +Cp.
Calculations are not given because the difference between the approximated relation in equation 10
and reality is too large. From measurements it could be derived that an extra capacitance Cext of
270 pF in parallel with diode D1 reduces the overall switching losses substantially, see figure 9 and 10
on page 23. The figures 11 through 14 give a more detailed picture of the switch-on and switch-off
behaviour depending on Cext.
4.6.3
Diode
The buck diode D1 is of the Philips type BYV29F/500V. The diode conduction losses Pd,c will be a first
order approximation described in equation 32 through 34 on page 36. The end result, given in equation
11, needs two input arguments; Uf as diode forward voltage and Rdio as diode resistance. Both values
are measured on a curve tracer.
2
P d, c
P la
P la
4 U in
= U f ⋅ -------- + R dio ⋅ --- ⋅  -------- – 1 ⋅ --------------------U la
3  U la  U la ⋅ U in
(11)
Data: Pn = 70 W, Un = 85 V, Uin = 380 V, Uf = 0.7 V and Rdio = 100 mΩ.
Calculations: Pd,c = 650 mW.
The reverse recovery losses are negligible because the diode current drops to zero at the end of the
period time.
The forward recovery losses are negligible too because the diode current at switch-on is rather low for
the applied diode.
4.6.4
Buck Capacitor
The output capacitor of the buck converter C1 is a film capacitor of the 379 MKP series of Philips. The
differential resistance of the lamp is rather small and negative. To obtain a stable operating point, an
Philips Semiconductors
16
impedance in series with the lamp is needed, so an ideal voltage source is forbidden. As a result the
output capacitor should be small. The film capacitor is preferred to the usual electrolytic capacitor for
life-time reasons.
4.7
Control Loop
The overall control loop consists of four blocks, namely:
• The power conversion transfer (buck converter and commutator)
• The measuring circuit transfer function (Ru, Ri and C11)
• The OTA transfer (IC1 and C9)
• The modulator transfer (IC1, Rs, R4 and R5)
The power conversion transfer function is rather complex, so a quantitative description is not given at
this moment. As a consequence, it does not make sense to describe the other transfer functions in a
quantitative way either. So all control blocks will not be described in a quantitative way but in a general
way.
Furthermore, instantaneous values are regarded to be averaged values in one HF period and the
corner frequency in the feedback loop will be much lower than the operating frequency of the buck
converter. So the given description is an averaged model description and is valid for rather slow
variations in the circuit.
4.7.1
Power Conversion
The power conversion circuit behaves like a controlled current source due to the SOPS operation
mode. The peak value of the buck coil current is controlled by the feedback loop and the minimum
value will always be zero, so the average current will be equal to half the peak value, see figure 17 on
page 29. Normally, the transfer function of these kind of systems are first order functions formed by
C1, the lamp resistance and the current source through L1. The igniter coil L3 changes this system into
a second order transfer function.
4.7.2
Measuring Circuit
The measuring circuit as described in section 4.2 is extended with an extra capacitor C11 at pin 1 of
IC1. The capacitor is added to suppress the oscillations on C1 caused by the commutator circuit during
the run-up phase, see section 5.1. C11 gives an extra pole in the feedback circuit and makes obtaining
stability more difficult.
4.7.3
OTA in IC1
The OTA is used to set the static error to zero by integrating capacitance C9 (disregarding the
tolerances on components). The corner frequency must be much lower than the operating frequency of
the buck converter.
Philips Semiconductors
17
4.7.4
Modulator in IC1
The modulator in IC1 is a comparator which makes the inductor peak value, measured by Rs and
tapped by R4 and R5, equal to the control voltage of the OTA. This function has no pole.
Finally, figure 3 gives the Nyquist diagram of the open loop transfer function simulated with PSTAR.
The parameters are the lamp resistance and the integration capacitance C9 of the OTA at pin 2 of IC1.
Jan 9, 1998
12:00:52
- y1-axis -
2.0
(LIN)
IMAG(I_INDUCTOR)
- Subvar C_9: 100.0n
R_LAMP: 2.0
2
C_9: 100.0n
R_LAMP: 20.0
3
C_9: 100.0n
R_LAMP: 200.0
4
C_9: 470.0n
R_LAMP: 2.0
5
C_9: 470.0n
R_LAMP: 20.0
6
C_9: 470.0n
R_LAMP: 200.0
1.0
3 62 5433
66
25426543
54
0.0
3
2
6
54
-1.0
-2.0
-2.0
0.0
-1.0
Analysis: AC
User:
2.0
1.0
(LIN)
REAL(I_INDUCTOR)
Simulation date: 09-01-98, 11:31:43
File: /user/optveld/simulation/pn711_uba2030_hid/HID_ipg/co@optveld/stability_42/NlDmCur_42.0_0
Fig.3 open loop transfer with parameters C9 and the lamp resistance
The system is stable for all parameter combinations because the point of oscillation (-1,0) is not
enclosed. But the phase margin in situation 3 is rather small and will produce an overshoot in the
power response.
.
Fig.4 the PSTAR simulation circuit
Philips Semiconductors
18
5.
5.1
COMMUTATOR & IGNITER.
Commutator
The commutator is a full bridge converter driven by the driver IC3 UBA2030. The diagonally transistor
pair TR2 and TR5 switch simultaneously and the diagonally transistor pair TR3 and TR4 switch
simultaneously either. Both transistor pairs switch alternately at a commutation frequency of ± 100 Hz,
fixed by R26 (100 kΩ) and C16 (33 nF) connected to the UBA2030. The commutation frequency is
asynchronous to the buck converter frequency. Locking both frequencies together is difficult and
makes no difference. The dead-time is fixed at 1.7 µs by resistor R27 (330 kΩ). The desired dead-time
tdt needed for proper switching is given in equation 12, with the preassumption that the maximum lamp
voltage Ula,max is 105 V. The nominal lamp voltage Un is 85 V, the nominal lamp power Pn is 70 W and
the dvdt limiting capacitor Cdvdt is 10 nF (C3 and C4).
C dvdt ⋅ U la 
t dt = --------------------------- 
I la
C dvdt ⋅ U la ⋅ U n

 ⇒ t dt = -------------------------------------U la
Pn
U

la

P n ⋅ 2 – -------I la = ------- ⋅  2 – -------- 

Un 
Un 
Un  
(12)
The dvdt limiting capacitor Cdvdt (C3 and C4) is determined by the maximum permissible dvdt at the
pin of the IC and the maximum current through the dvdt limiting capacitor. The maximum current is
about 30 A during ignition and the maximum permissible dvdt at the driver IC is 4 V/ns so that a
minimal Cdvdt = 7.5 nF is desired. The applied value for C3 and C4 is 10 nF.
The voltage supply for the UBA2030 is realised by connecting the HV-pin, pin 13, directly to the buck
output voltage.
The igniter coil L3 and buck output capacitor C1 will oscillate at each commutation due to the relative
low value of C1, see figure 5 on page 21. Just before commutation when the system is in steady-state,
the current through the lamp and L3 is equal to the averaged current delivered by the buck converter
hence, the resulting current through C1 is zero and the voltage across C1 is equal to the lamp voltage.
Just after commutation, the commutator switches the load in reverse so that initially twice the steadystate lamp current will flow through C1 which initiates an oscillating current through C1 and
corresponding voltage across C1. The oscillation is damped by the lamp resistance. Especially during
the run-up phase when the lamp resistance is very low, the oscillation is very large compared to the
averaged lamp voltage which would force the buck output voltage to negative values. The negative
output voltage is clamped by the body diode of the fullbridge power mosfets. The oscillation during the
run-up phase causes a small gap in the voltage across C1 after each commutation, see for a detailed
description section 4.1. Figure 8 on page 22 gives the voltage across C1 during the run-up phase and
figure 6 on page 21 during the burn phase.
Philips Semiconductors
19
5.2
Igniter
The igniter is formed by C2, D2, L2, L3 and Z4 through Z7. The igniter generates pulse voltages of
about 5000 V peak with a repetition frequency equal to the commutation frequency, ± 100 Hz. C2, L2
and L3 take care of the pulsating character of the igniter. The key components are;
• D2 is a break-over diode (sidac) type K1V36W manufactured by Shindengen with a typical breakover voltage Ubo = 360 V. This device is symmetrical in both directions.
• L2 is a fixed inductor of 270 µH which limits the current through the break-over diode and saturates
at 2 A. After saturation the inductor value is reduced to 20 µH.
• The igniter coil L3 is an EF32 core with 3C85 as core material, primary 7 turns 0.2 mm and
secondary 210 turns 0.45 mm solid copper wire. The secondary winding, connected in series with
the lamp, has an inductance of 6 mH. The maximum applied voltage between the windings mutual
and between the windings and core is 5600 V.
• Z4 through Z7 are zener diodes of the philips type BZD23C47C who limit the voltage across the
igniter coil to prevent corona.
Philips Semiconductors
20
6.
WAVEFORMS.
Ula
Ila
1) Ula
2) Ila
50 V/div
1 A/div
2 ms/div
2 ms/div
lamp voltage
lamp current
Fig.5 lamp voltage and lamp current, burn phase
UC1
Ila
1) UC1
2) Ila
50 V/div
1 A/div
2 ms/div
2 ms/div
output voltage buck converter
lamp current
Fig.6 buck output voltage and lamp current, burn phase
Philips Semiconductors
21
Ila
Ula
1) Ula
2) Ila
20 V/div
2 A/div
2 ms/div
2 ms/div
lamp voltage
lamp current
Fig.7 lamp voltage and lamp current, run-up phase
UC1
UC1
1) UC1
2) UC1
50 V/div
2 ms/div
50 V/div 100 µs/div
buck output voltage
buck output voltage, zoomed
Fig.8 buck output voltage, run-up phase
Philips Semiconductors
22
Us
Id
Ig
Ugs
1) Us
2) Id
3) Ig
4) Ugs
100 V/div
1 A/div
200 mA/div
10 V/div
2 µs/div
2 µs/div
2 µs/div
2 µs/div
source voltage T1
drain current T1
gate current T1
gate voltage T1
Fig.9 switch behaviour of T1
Us
Id
Ig
Ugs
1) Us
2) Id
3) Ig
4) Ugs
100 V/div
1 A/div
200 mA/div
10 V/div
2 µs/div
2 µs/div
2 µs/div
2 µs/div
source voltage T1
drain current T1
gate current T1
gate voltage T1
Fig.10 switch behaviour of T1, with extra Cext = 270 pF
Philips Semiconductors
23
Us
Id
Ig
Ugs
1) Us
2) Id
3) Ig
4) Ugs
100 V/div
1 A/div
200 mA/div
10 V/div
200 ns/div
200 ns/div
200 ns/div
200 ns/div
source voltage T1
drain current T1
gate current T1
gate source voltage T1
Fig.11 switch-on behaviour of T1
Us
Id
Ig
Ugs
1) Us
2) Id
3) Ig
4) Ugs
100 V/div
1 A/div
200 mA/div
10 V/div
200 ns/div
200 ns/div
200 ns/div
200 ns/div
source voltage T1
drain current T1
gate current T1
gate source voltage T1
Fig.12 switch-on behaviour of T1, with extra Cext = 270 pF
Philips Semiconductors
24
Us
Id
Ig
Ugs
1) Us
2) Id
3) Ig
4) Ugs
100 V/div
1 A/div
200 mA/div
10 V/div
100 ns/div
100 ns/div
100 ns/div
100 ns/div
source voltage T1
drain current T1
gate current T1
gate source voltage T1
Fig.13 switch-off behaviour of T1
Us
Id
Ig
Ugs
1) Us
2) Id
3) Ig
4) Ugs
100 V/div
1 A/div
200 mA/div
10 V/div
100 ns/div
100 ns/div
100 ns/div
100 ns/div
source voltage T1
drain current T1
gate current T1
gate source voltage T1
Fig.14 switch-off behaviour of T1, with extra Cext = 270 pF
Philips Semiconductors
25
UC1
IFB
1) UC1
2) IFB
100 V/div
10 A/div
2 ms/div
2 ms/div
buck output voltage on C1
current to full bridge converter
Fig.15 the voltage UC1 during ignition
Uign
Uign
1) Uign
2) Uign
2.5 kV/div
2.5 kV/div
2 ms/div
10 µs/div
ignition voltage
ignition voltage, zoomed
Fig.16 the secondary ignition voltage at the igniter coil L3
Philips Semiconductors
26
7.
CONCLUSIONS & RECOMMENDATIONS.
Conclusions
• The ignition coil takes care for a second order transfer of the power circuit and an extra pole needed
in the control circuit. So, the overall transfer is a fourth order one and makes loop stability difficult.
• An auxiliary winding on the buck coil is needed for the floating supply of the level shift IC in unloaded
situations.
• The control IC1 is not completely suited for the application.
Recommendations
• The igniter circuit should be redesigned to get a smaller value of the inductance in series with the
lamp.
• A lamp with ignition electrode makes the ignition coil in series with the lamp obsolete and ignition
could be done by a piezoelectric transformer.
Philips Semiconductors
27
Philips Semiconductors
28
APPENDIX 1
CALCULATIONS.
In this appendix, some formulas are derived to give a mathematical description of the buck converter.
The switching frequency always will be very high in relation to the commutation frequency of the full
bridge converter and the applied corner frequency in the feedback circuit. The buck converter operates
in a SOPS mode what means that the current through the buck coil is continues and alternates
1
between zero and Î L , so the average coil current I L = --- ⋅ Î L see Fig.17.
2
IL(t)
Î L
IL
t
Us(t)
Uin
2.Ulamp
Ulamp
t
ton
toff
Tosc/2
Fig.17 SOPS mode
The negative part drawn in the current in figure 17 is not according the weight scale.
A.1.1 Current Sensing
The maximum current Î L through the buck coil (L1) and the maximum lamp current Ila,max are given by
equation 13. Uth,max is the internal voltage limiter of the control IC.
U th, max 
Î L = k ⋅ --------------------- 
Rs 
Î L k U th, max
- = --- ⋅ -------------------- ⇒ …I la, max = I L = --Rs
2 2
R5 
k = 1 + ------ 
R4 
(13)
A.1.2 Power Control Management
The feedback signal is obtained by adding the lamp voltage Ula and lamp current Ila. The error
amplifier of control IC1 is an OTA. The output is connected to a relative large capacitor C9 so that the
corner frequency fc of the control circuit is very low compared to the switching frequency.
The inputs of the error amplifier must be equal, assuming that the control system is in steady state
condition. So the voltage at the negative input of the error amplifier must be equal to the internal
reference voltage Uref = 2.5 V at the positive input of the amplifier. Now, a simple expression can be
Philips Semiconductors
29
found between the lamp power and lamp voltage which is given in equation 14 through 16, where Ru is
the voltage feedback resistor, Ri the current feedback resistor and Rs the measuring resistor for the
current feedback.
U –U

I ⋅R –U
P la
-------- ⋅ R s – U ref
U la – U ref U la
- + ------------------------------------- = 0
 ⇒ … -----------------------Ru
Ri



la
ref
la
s
ref
- + ---------------------------------- = 0 
∑k Ik = 0 ⇒ … -----------------------Ru
Ri

I la
P la
= -------U la
(14)
U la R i
Ru

P la = -------- ⋅ ------- ⋅  1 + ------- ⋅ U ref – U la 
Rs Ru 
Ri 
U la

- ⋅ [ ( 1 + α ) ⋅ U ref – U la ]
 ⇒ …P la = -------------R
Ru
s⋅α

Define, …α = ------
Ri

(15)
U
ref
⋅ (1 + α) .
The roots of the parabola are U la = 0 ∩ U la = U ref ⋅ ( 1 + α ) and the top is given by U la = ----------
2
The best power control is achieved when both the nominal lamp voltage Un and nominal lamp power Pn are set
1
to the top of the parabola, hence U n = --- ⋅ U ref ⋅ ( 1 + α n ) ⇒ α n = 2 ⋅ U n ⁄ U ref – 1 . Substitution of Un into equation
2
2
Un
Un
15 gives the expression P n = ------------------ ⋅ ( 2U n – U n ) = ------------------ . Resuming, we get some simple expressions to
αn ⋅ Rs
αn ⋅ Rs
determine αn and Rs which are given in equation 16.
Un
Ru
α n = ------- = 2 ⋅ ---------- – 1
U ref
Ri
2
Un
R s = -----------------Pn ⋅ αn
U la
U la
P la = P n ⋅ -------- ⋅  2 – --------
Un 
Un 
(16)
The lamp current Ila as function of lamp voltage Ula is linear:
Pn
U la
I la = ------- ⋅  2 – --------
Un 
Un 
(17)
Observing that the maximum lamp current, limited to Ila,max (equation 13) will restrict the interval where
the linear relation Ila(Ula) is valid, see Fig.18. The corresponding lamp voltage at breakpoint Ila,max is
called Ula,min, see equation 18.
Pn
U la 
I la = ------- ⋅  2 – -------- 
Un 
Un  
I la, max ⋅ U n

⇒ U la, min = U n ⋅  2 – -----------------------------

I la = I la, max
Pn




U la = U la, min

Now, the parabolical relation Pla(Ula) is not valid anymore for lamp voltages lower than Ula,min.
Substitution of Ula,min in equation 16 gives the corresponding lamp power Pbr at the breakpoint:
Philips Semiconductors
30
(18)

U la
U la 

P la = P n ⋅ -------- ⋅  2 – -------- 

Un 
Un  
U la, min
U
la, min 
 2 – -----------------⇒
=
P
⋅
⋅
P
-----------------
br
n

Un
P la = P br
Un  



⇒…
U la = U la, min



I la, max ⋅ U n 

U la, min = U n ⋅  2 – ----------------------------- 
Pn



(19)
I la, max ⋅ U n

P br = I la, max ⋅ U n ⋅  2 – -----------------------------
Pn


Equation 20 gives an overview of the lamp current and lamp power formulas, where two regions are
distinguished. Region 1 is valid for U la ≤ U la, min , the linear region. Region 2 is valid for U la ≥ U la, min ,
the parabolical region, see figure 18.
Pla[W]
Pn
Pbr
2.Un
Ula[V]
Ila[A]
Region 1
Ila,max
Region 2
In
Ula,min
Un
Fig.18 Pla(Ula) and Ila(Ula)
Philips Semiconductors
31
Ula[V]
Pn
U la

 I la = ------- ⋅  2 – --------
Un
Un


 I la = I la, max
U la ≥ U la, min
U la ≤ U la, min
U la
U la

 P la = P n ⋅ -------- ⋅  2 – --------
Un
Un


 P la = U la ⋅ I la, max
(20)
U la ≥ U la, min
U la ≤ U la, min
A.1.3 Watch-dog Timer
The watch-dog time delay twd is an ordinary first order network with time constant τwd = Rwd.Cwd. The
unloaded capacitor is charged by a resistor to the system reference voltage Uref(s) and generates a
high to low transition to the control IC which turns the buck mosfet on. Equation 21 gives the watch-dog
delay time twd.
t
U ref ( s )
wd
– -------

τ wd
E

 ⇒ …t = τ ⋅ ln  -------------------------
= E⋅ 1–e
wd
wd


 E – U ref ( s-)


(21)
A.1.4 General Characteristics Buck Converter
The operating frequency fop of the buck converter is determined by the value of the buck coil L, the
input voltage Uin, the output voltage Ula, the output power Pla and the total parasitic capacitance at the
source of the mosfet Cpar. The latter determines the freewheel oscillation period time Tosc of the buck
coil L with the parasitic capacitance Cpar when the current through the buck diode drops to zero. The
derivation of the operating period time Top is given in equation 22.




2 ⋅ P la
U in
T osc
T osc
L ⋅ Î L
L ⋅ Î L
L ⋅ Î L 
t on = -----------------------  ⇒ T op = ------------ + ----------------------- + ------------ = L ⋅ --------------- ⋅ ----------------------------------------- + -----------U la U la ⋅ ( U in – U la )
2
2
U la U in – U la
U in – U la 

2 ⋅ P la 
Î L = --------------- 
U la

L ⋅ Î L
t off = -----------U la
(22)
The operating frequency fop is given in equation 23 where Tosc is assumed to be small in relation to
(ton + toff). So, during the ignition phase equation 23 is not valid (Ula is large and Pla is small).
2
f op
U la
U la
= ----------------------- ⋅  1 – --------
2 ⋅ L ⋅ P la 
U in
Philips Semiconductors
32
(23)
Combining equation 23 and 16 eliminates the lamp power Pla so that fop is merely a function of Ula.
Equation 24 gives fop as function of Ula and is only valid when the lamp current is not limited.
2
f op
P la
U la
U la
U la 
1 – -------= ----------------------- ⋅  1 – -------- 


U la ⋅ U n
U in
2 ⋅ L ⋅ P la
U in 
- ⋅ ---------------- ⇒ f op = --------------------⋅
⋅
U la
2
L
P
n
U la
U la 
2 – -------= P n ⋅ -------- ⋅  2 – -------- 
Un
Un 
Un  
(24)
Furthermore, we can derive the relation between the output voltage Ula and the input voltage Uin.
Assuming that the freewheel oscillation period time Tosc is small compared to (ton + toff) again, an
expression of Ula(Uin) is given in equation 25.

t off ⋅ U la

Î L = --------------------t on
L

- ⋅ U in = δ ⋅ U in
 ⇒ U la = -------T op
t on ⋅ ( U in – U la ) 
Î L = ---------------------------------------- 
L

(25)
A.1.5 Power Mosfet Losses
The power mosfet losses are divided into three parts:
• Conduction losses Pm,c caused by the RMS current through the mosfet Im.
• Switch-on losses Pm,son caused by the voltage across the mosfet at turn-on.
• Switch-off losses Pm,soff causes by the current through the mosfet at turn-off
power mosfet conduction losses
The power mosfet RMS current Im and the mosfet conduction losses Pm,c are:
2
2
2
t on Î L
 Î L 
Î L 
 ------- ⋅ t ⋅ dt = --------- ⋅ ---- = δ ⋅ ---- 
T op 3
3
 t on 
0

2
P la

4
2
 ⇒ P m, c = R ds ( on ) ⋅ I m = --3- ⋅ R ds ( on ) ⋅ --------------------U la ⋅ U in



P la

= 2 ⋅ -------
U la

t on
2
Im
1
= --------- ⋅
T op
U la
δ = -------U in
Î L = 2 ⋅ I la
∫
(26)
power mosfet switch-on losses
The switch-on losses Pm,son are given in equation 28. Tosc is the freewheel oscillation at the moment
that the diode current is zero. The buck coil L and the total parasitic capacitance Cpar from source
terminal to ground determine the oscillation period time Tosc. The oscillation period time Tosc is
Philips Semiconductors
33
measured so that Cpar can be calculated. The circuit is designed to switch-on at the moment that the
voltage across the mosfet is minimal, i.d. U ds, son = U in – 2 ⋅ U la and the voltage across the capacitor is
maximal, U s, son = 2 ⋅ U la . The energy dissipated in the mosfet Em,son is given in equation 27.
E m, son
∞

∂U s

= U ds ⋅ I ds ⋅ dt = ( U in – U s ) ⋅ C par ⋅ --------- ⋅ dt = C par ⋅ ( U in – U s ) ⋅ dU s 
∂t
⇒…
0
0
0


U s(0) = U s, so ∩ U s(∞) = U in

∞
∞
∫
∫
∫
(27)
1
2
E m, son = --- ⋅ C par ⋅ ( U in – U s, so )
2
Equation 28 gives the switch-on losses in the mosfet.
1
2
P m, son = f op ⋅ E m, son = f op ⋅ --- ⋅ C par ⋅ ( U in – 2 ⋅ U la )
2
T osc = 2 ⋅ π ⋅ L ⋅ C par ⇒ C par
T osc 2 1
=  ------------ ⋅ -- 2 ⋅ π L
(28)
power mosfet switch-off losses
The switch-off behaviour of the mosfet is determined by:
• the peak current through the conduction channel just before switch-off Ich which equals the buck
inductor peak current Î L .
• the gate sink current during turn-off Ig.
• the transconductance of the mosfet gm.
• the capacitors Cgs, Cdg and Cds of the mosfet and some external capacitors formed by the diode
capacitance, the buck coil capacitance and the parasitic capacitance of the source terminal to
ground.
Some quantities are non-linear like Cds, Cdg and gm. For simplicity, all quantities are supposed to be
linear. Now, the switch-off model description becomes an approximation of the real switch-off
behaviour. Furthermore, gm is supposed to be very large which means that a large channel current
variation ∆Ich is obtained by a small change in gate-source voltage ∆Ugs. So, Ugs is assumed to be
constant during switch-off and will be a little bit larger than the threshold voltage of the mosfet Uth.
Figure 19 gives the gate-source voltage Ugs, the drain source-voltage Uds and the current through the
drain-source channel Ich in case of previous assumptions. The current through the channel during
switch-off is denoted by Iso.The gate-source capacitance is not drawn in figure 19 because it does not
affect the switch-off losses (Ugs is almost constant). Equation 29 gives the relation between Ip and Ig.
Philips Semiconductors
34
Ugs
Î L
Uth
Uin
Id
Uds
Uin
Cdg
Ich
d
Ich
Î L
Ig
Iso
Ip
Cp
g
s
tso
Fig.19 switch-off model description




Ip
Cp
Ig
∂U ds
∂( U dg + U gs )
∂U dg 
- = ------- ⇒ I p = ---------- ⋅ I g
------------ = ---------------------------------- = -------------  ⇒ --------C gd
∂t
∂t
∂t  C gd C p

∂U ds
Ip

------------ = ---------------------------
∂t
C ds + C ext

∂U dg
Ig
------------- = ---------∂t
C gd
(29)
The drain-source channel current during switch-off Iso, the drain-source voltage Uds and the switch-off
time tso are given by equation 30, where k is used to reduce the formula length.
Cp
I so = Î L – ( I g + I p ) = Î L –  1 + ---------- ⋅ I g = Î L – k ⋅ I g

C gd
Ig
∂U ds ∂U dg
Ig
------------ = ------------- = ---------- ⇒ U ds(t) = ---------- ⋅ t
C gd
∂t
∂t
C gd
(30)
Ig
U in ⋅ C gd
U ds(t so) = ---------- ⋅ t so = U in ⇒ t so = ----------------------C gd
Ig
The switch-off losses are given by multiplication of the drain-source channel current Iso and the drainsource voltage Uds during the switch-off time tso:
Philips Semiconductors
35
t so
P m, soff
1
= --- ⋅
T
∫
t so
1
U ds(t) ⋅ I so ⋅ dt = --- ⋅
T
0
∫
0
Ig
---------- ⋅ t ⋅ ( Î L – k ⋅ I g ) ⋅ dt
C gd
Ig 1 2
1
P m, soff = --- ⋅ ( Î L – k ⋅ I g ) ⋅ ---------- ⋅ --- ⋅ t so
C gd 2
T
P m, soff
(31)
˙
I g 1 U in ⋅ C gd 2

1
1
2  Î L


= --- ⋅ ( Î L – k ⋅ I g ) ⋅ ---------- ⋅ --- ⋅ ----------------------- = f ⋅ --- ⋅ C gd ⋅ U in ⋅  ---- – k

T
C gd 2 
Ig
2
 Ig 
Be careful: equation 31 is only valid when Î L > k ⋅ I g is fulfilled, otherwise the switch-off losses are zero.
A.1.6 Diode Losses
The diode conduction losses Pd,c will be a first order approximation, see equation 32, with Uf as diode
forward voltage and Rdio as diode resistance. Both values are measured on a curve tracer. The
conduction losses are expressed in the average diode current and RMS diode current.
P d, c
T
T
T
T
∫
∫
∫
∫
0
0
0
0
1
1
1
1
2
= --- ⋅ u(t) ⋅ i(t) ⋅ dt = --- ⋅ [ U f + R dio ⋅ i(t) ] ⋅ i(t) ⋅ dt = --- ⋅ U f ⋅ i(t) ⋅ dt + --- ⋅ R dio ⋅ i(t) ⋅ dt
T
T
T
T
(32)
2
P d, c = U f ⋅ I d + R dio ⋅ I d
The RMS current, see equation 33, is similar to the derivation of the mosfet RMS current shown in
equation 26.
t off
2
Id
1
= --------- ⋅
T op
∫
0

2
2
t off Î 2
Î
Î
 ------- ⋅ t ⋅ dt = --------- ⋅ ---- = ( 1 – δ ) ⋅ ---- 
 t off 
T op 3
3

2
P la

4  U in 
2
U la
- – 1 ⋅ -------------------- ⇒ I d = --3- ⋅  ------δ = -------U la  U la ⋅ U in

U in


P la

Î = 2 ⋅ I la = 2 ⋅ -------
U la

(33)
The average diode current is equal to the lamp current, so the approximated conduction losses Pd,c
are:
2
P la
P la
4 U in
P d, c = U f ⋅ -------- + R dio ⋅ --- ⋅  -------- – 1 ⋅ --------------------

U la
3 U la
U la ⋅ U in
Philips Semiconductors
36
(34)
+15 Vdc
+380 Vdc
Philips Semiconductors
37
10nF
C5
Z1
BZX79C
15V
TR1
Fig.20 complete circuit diagram
5
4.7k
R13
GND
4
GND
4
4.7k
R14
3
2
8
C8
47pF
LIN
HIN
VB
NC
470nF
C7
3
2
IC4-1
LM393N
1
220
R15
Watch Dog Timer
LO
VS
HO
LM393N.sup
V+
IC4-3
8
R12
12k
NC
6
7
1
VCC
IC2
Buck Converter
C6
470nF
IR2101
PHP10N40E
R16
100k
*
*
R4
100
R5
100
4
5
7
R19
39k
VFB
MULT
3
2
1
MC34262P
COMP
GND
6
R18
82k
CS
ZCD
OUT
4.7
4.7
4.7
8
VCC
PR02
PR02
PR02
Rs
IC1
Z2
BZX79C
2.4V
R17
15k
R3
R2
R1
*
Ru
C11
220nF
470nF
R11
R10
Ri1
R7
330k
R6
220k
C9
470nF
C10
820nF
C1
4.7k
27k
R8
39k
Ri2
1
Ignition
Voltage
Control
LM393N
IC4-2
R9
4.7k
Provisional Circuit: MHN-TD 70W
3
2
R22
10k
R21
120k
R20
150k
TR3
PHP10N40E
10nF
C3
TR2
PHP10N40E
5.6nF
C12
17
13
6
5
8
9
7
3
Z3
TL431C
R23
10k
R25
10k
270k
100nF
R26
100k
47V
47V
47V
47V
TR5
PHP10N40E
10nF
C4
TR4
PHP10N40E
Commutator
C16 33nF
Uref=2.5V
R24
10k
C15
R27
UBA2030
Z7
BZD23C
Z4
BZD23C
Z5
BZD23C
Z6
BZD23C
2
1
GLL PGND GLR
12
FSL
FSR
10
SHR
SHL
11
GHL
GHR
15
SDH
SDL
16
DTC
SDHR
14
HV
EXO
4
VDD SGND RC
18
IC3
270 uH
L2
System
Reference
Voltage
C13 470nF
C14
K1V36W
Igniter
100nF
C2
100nF
Lamp
APPENDIX 2
COMPLETE CIRCUIT DIAGRAM.
Philips Semiconductor
38
Philips Semiconductors
Preliminary specification
Full bridge driver IC
UBA2030T
BRIEF APPLICATION INFORMATION
The UBA2030T is the commutator part in the complete system to drive a HID lamp. The lamp-life of the HID lamp can
be dependant of the amount of sodium that migrates through the quartz wall of the lamp. To minimize this migration, the
lamp has to operate negative with respect to the ground.
An application with a control circuit referenced to the high voltage in a full bridge with a HID lamp in an automotive
environment, see figure below. The BER pin is connected to the system ground. With the BE pin the bridge can be hold.
The HV pin delivers the supply current to the internal low voltage circuit. This pin can be connected to the system ground
or to e.g. a present LV DC supply (battery) as indicated in the figure by the dotted lines. To avoid malfunctioning of the
full-bridge the voltage at the HV pin must be higher than the voltage at the VDD pin (also during starting up the system).
The diode in series with the supply to the HV pin avoids discharging of C8 if the lamp is shorted during the ignition phase.
The EXO pin and the SD pin should be (logically) grounded during the start-up phase.
The dV/dT of the voltage at the EXO pin and the SD pin should be > 5V/ms.
The DTC pin is sensitive for capacitive coupling. For that reason C7 has to be added as close as possible to DTC pin.
The power transistors are relatively hard driven by the control IC. To avoid radiation problem due to hard switching of the
powers transistors a resistor in series with the gate can be added.
From LV DC supply
Bridge
Control
Circuit
SYSTEM GND
HMR
HML
C2
C8
SHR
GHR
FSR
HV
EXO
SD
DTC
VDD
SGND
SHL
GHL
FSL
BER
BE
RC
GLL
PGND
GLR
C1
IGNITOR
LAMP
LML
LMR
C5
C6
C8
C3
C7
R2
C4
R1
HV -570V max
Automotive configuration
1999 Febr 4
15
C1=150nF
C2=150nF
C3=220nF
C4=10nF
C7=100pF
R1=147kΩ
R2=50 to 1000kΩ
for a dead time of1µs R2=220kΩ