Efficient switching in SMPS - Mitja Stachowiak

Efficient switching in SMPSProjectseminarTutor: M.Sc. Muhammad Bilal SaifBy Mitja Stachowiak, July 2016Integrated Electronic Systems | Univ.-Prof. Dr.-Ing. Klaus Hofmann

AbstractThis work is a preparation for bachelorthesis on switching mode power suplies. Its aim is to prepare thetasks for bachelorthesis, to decide for a certain topology, to find required materials, and to get a simpleprototype to work on.

Contents1 Situation 1

1.1 Measurement conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Measurements on existing SMPS 22.1 Topology and Frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.2 Voltage regulation and power saving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.3 Transformer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3 Construction of a halfbridge prototype 53.1 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53.2 Switching condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63.3 Zero voltage switching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73.4 Frequency stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1 SituationWhen designing a new SMPS, a lot of parameters have to be chosen, for example the topology, switchingfrequency, voltage control strategy or the type of semiconductor switches. For corresponding practical, alot of different power supplies have been analyzed. The 180W models all have too large sizes and slightlyworse efficiency, than the small 90W Hama power supply. There is also a defective HP Ultra Slim powersupply. Because the Hama-model is pasted with thermal conductivity paste and the transformer’s inputconnectors are unreachable, a new HP Ultra Slim power supply was bought and opened. This powersupply is meant to be the archetype for the experimental prototype. Some special elements, required forthe prototype, could be taken from the defective HP power supply. The first prototype will be built on acircuit board, later on, a PCB board should be designed.

1.1 Measurement conditions

The waveform measurements where done using digital oscilloscopes. For instrument protection, highvoltage probes (factor 1:100) are used for all primary side measurements. This makes the measurementunprecise, especially if the probe is not descrewed perfectly. For current waveforms, simple (unprecise)resistors without temperature compensation are used. Furthermore, if there are more channels to bemeasured, than the oscilloscope has inputs, the measurements are done separatly using the same trigger.Sometimes, even if two channels are connected to the oscilloscope at once, there seems to be a delay inthe signals. Maybe the oscilloscopes cannot take all channels synchronously.

1

2 Measurements on existing SMPS2.1 Topology and Frequency

First, the device topology of the device has to be determined. There are two switching transistors at theHama and HP power supply, which are inversly clocked with a frequency of about 120 KHz. Therefore itmust be a halfbridge topology.

HP Gate signals

V gat

e lo

w -

P5| V

gate

hig

h -

P5

t-0.000004 s -0.000002 s 0 s 0.000002 s 0.000004 s

0 V

-100 V

100 V

200 V

300 V

400 V

500 V

Transformer input signals

-100 V

0 V

100 V

200 V

300 V

400 V

500 V

V P2

- P5

| VP3

- P

5

-0.000004 s -0.000002 s 0 s 0.000002 s 0.000004 st

2

2.2 Voltage regulation and power saving

One important question is, how the output voltage is regulated. From corresponding practical, abreakdown in the efficiency of the Hama power supply for low powers is noted. For this work, theefficiency measurement was repeated for the HP power supply, which shows no breakdown:

HP_slim: EfficiencyHama: Efficiency

Effic

ienc

y

0 %

20 %

40 %

60 %

80 %

100 %

0 W 10 W 20 W 30 W 40 W 50 W 60 W 70 W 80 W 90 W

Pout

The power supplies use on-off-control, to hold the efficiency high for low powers. Hama stopps thiscontrol at 0.3 A output current, HP continues the control until the on to off percentage reaches 100 %(0.8 A). When the output current becomes larger than 0.8 A, there is no significant change at the signalsarround the transformer.

On-Off-Control

-300 V

-200 V

-100 V

0 V

100 V

200 V

300 V

400 V

V pri

-0.00015 s -0.0001 s -0.00005 s 0 s 0.00005 s 0.0001 s 0.00015 s 0.0002 s 0.00025 s 0.0003 st

Both power supplies have a wide range input from 100 V - 250 V AC. Modulating this voltage using acontrollable transformer has no influence on the signals arround the transformer, so the rough voltagecontrol happens in the PFC.

2.3 Transformer

From the defective HP ultra slim power supply, the transformer was cut out. It’s a Delta Tec MV-NBS9049. There is no technical datasheet or support for this transformer, so a Hameg LCR-Meter was usedto determine the replacement circuit. The transformer has 5 input connectors and 6 output connectorswhich results in the following symbol and replacement circuit:

2.2 Voltage regulation and power saving 3

SC1

MV-NBS9049

P3S3

S2P2

P1

P4

P5

S1

S4

S5

S6

P3

S3

S2

P2

P1

P4

P5

S1

S4

S5

S6

Coupling

Arbitrary SPICE Block

Lp23743.47µH

Rp23

2.031Ω

Rp15

113.84Ω

Rp45

259.1mΩ

Rs16

88mΩ

Rs25

51.9mΩ

Rs34

137.4mΩ

Lp157.671µH

Lp4529.7666µH

Ls34

4.56µH

Ls257.822µH

Ls16

7.852µH

K1 LLp15 LLp23 LLp45 LLs16 LLs25 LLs34 0.99

The coupling factor was not measured, the instru-mentation was too unprecise. Some tests using thistransformer in Flyback mode with open or shortedoutput resulted in unrealistic coupling factors any-where between 50 % and 130 %.The output windings S1-S6 and S2-S5 are connectedin center tap mode and there must be an active rec-tifier on the secondary side.The optocouplers just transfer DC-voltages from sec-ondary to primary side, so the switching signals forthe active rectifier are generated on the secondaryside.

2.3 Transformer 4

3 Construction of a halfbridge prototypeWith the identified parameters, a simple half-bridge prototype was built on a circuit board, just existing ofthe transformer, the switching transistors and a passive two-diode rectifier.

3.1 Simulation

SC1

MV-NBS9049

P3S3

S2P2

P1

P4

P5

S1

S4

S5

S6

D5

D6

V1

250Vrms 50Hz 0°

+

-

Q2V2125kHz15V

Q1

C2100nF

C3100nF

V315V

+

-

C4

D7

C54.7µF

R220kΩ

R35kΩ

R420kΩ

SC2

NCP5181

VBOOTVcc

DRV_HIIN_HI

BridgeIN_LO

DRV_LOGND

R1

0.2Ω

D1

D2

C1

12µFIC=13V

0.5Ω

12Ω

12Ω

0.5Ω

Rload4Ω

D3 C6

IC=352V

C71µF

0.5Ω

C8

0.47µF

Parasitics

One objective of this work was, to get not just the prototype but also a working simulation of it. The gatedriver was only available as an encrypted PSPICE model, that works since Orcad 16. The free versionof Orcad has a limit of 75 components but the driver consists of more than 80 sub-circuits and cannot

5

be simulated. Therefore this circuit plan was made with Multisim using ideal relais for the driver. Thestudent version of Multisim have a component limit, too, so this plan has to be ported to LTSpice or another free software to continue the work.The resistors R2, R3, R4 and the opamps generate the signals for the driver, having a certain death time,which can be modulated by changing the rise and falltime of the clock.

3.2 Switching condition

begin: VP3 - GNDbegin: Vgate high - GND begin: Vgate low - GND

-20 V

0 V

20 V

40 V

60 V

80 V

V gat

e lo

w -

GND

| Vga

te h

igh

- GN

D| V

P3 -

GND

0.0000012 s 0.0000014 s 0.0000016 s 0.0000018 s 0.000002 s 0.0000022 s

t

low gate turn-off

high gate turn-on

Starting up the raw halfbridge circuit with full in-put voltage alwais resulted in destruction of the gatedriver. When using a controlable source, it can beseen, that the ringing pushed the driver out of saveoperating area, even with voltages less then hundredvolt. The ringing influences neighbour connectionsand the driver’s supply voltage due to parasitic effects.If there is too much ringing, the gate driver doesn’tdo, what it is bounded for. The signals overshootstwice of the destination voltage.An other problem in this first experiment was, thatthe duty cycle wasn’t 50 % (Ton = 5µ s, f = 10 KHz).When using Halfbridges, the duty cycle shoult be 50 %, because otherwise the capacitive voltage divisionwon’t work.

First optimization

firstOpt: IsecfirstOpt: VP3 - GND firstOpt: Vgate low - GND

Isec

-0.5 A

-0.25 A

0 A

0.25 A

0.5 A

0.75 A

1 A

1.25 A

-10 V

0 V

10 V

20 V

30 V

40 V

50 V

V gat

e lo

w -

GND

| VP3

- G

ND

0 s 0.000002 s 0.000004 s 0.000006 s 0.000008 s

t

As a first optimizatrion, the duty cycle was set to 50 %(Ton = 5µ s, f = 100 KHz). The switching speedwas reduced by adding a resistor (27Ω) betweengate and driver and the rectified input voltage wasstabilized near the flyback diodes. There often is alarge (and expensive) ceramic capacitor on switch-ing mode power supplies. It seems, that electrolytecapacitors cannot buffer strong and high frequentringing.This optimizations significantly reduced the ringingon the bridge voltage, but the ringing on the gatesignals and the transformer output current was stillhigh.

Second optimizationFor next step, a ceramic capacitor (0.47µF) was added to the input voltage near the gate driver. Thisreduced the ringing on the gate signals and allowed to reduce the gate-driver-resistor to 12Ω. Since now,the ringing reduced, when the input voltage or the output current increased. Under this conditions, themost important goal of efficient switching regarding halfbridges is fullfilled: Zero voltage switching.

3.2 Switching condition 6

secondOpt: VP3 - GNDsecondOpt: Vgate low - GND

secondOpt: Vgate high - GND

secondOpt: Isec

Isec

-1.5 A

-1 A

-0.5 A

0 A

0.5 A

1 A

1.5 A

2 A

-25 V

0 V

25 V

50 V

75 V

100 V

V gat

e lo

w -

GND

| Vga

te h

igh

- GN

D| V

P3 -

GND

0.0000014 s 0.0000016 s 0.0000018 s 0.000002 s 0.0000022 s

t

high gate turn-on

secondOpt: VP3 - GNDsecondOpt: Vgate low - GND secondOpt: Vgate high - GND

secondOpt: Isec

Isec

0 V

50 V

100 V

-50 V

150 V

200 V

V gat

e lo

w -

GND

| Vga

te h

igh

- GN

D| V

P3 -

GND

0.000002 s0.00000125 s 0.0000015 s 0.00000175 s 0.00000225 s

t

0 A

-1 A

1 A

2 A

-1.5 A

-0.5 A

0.5 A

1.5 A

high gate turn-on

3.3 Zero voltage switching

Q2D2

Q1D1

off

on

Lσ1

Lh

Lσ2*

RLoad*

400V400V

0V

Q2D2

Q1D1

off

off

Lσ1

Lh

Lσ2*

RLoad*

400V ↑↑

Q2D2

Q1D1

ZVS

off

Lσ1

Lh

Lσ2*

RLoad*

400V

400V

400V

Zero voltage Switching (ZVS) means, that the voltage on the transistor is nearly zero, when the switchingaction occurs. After low gate turn-off, the transformer’s inductances drive the current for a certaintime, that shoult be long enough, to load the transistor’s and wiring capacitances until the flyback diodebecomes conductive. The high gate then should turn on, before this flyback current stopps.Zero voltage switching is a powerfull soft switching technique, that comes for free on halfbridges, if theparasitic capacitances and - inductances are well coordinated with the dead-time. Regarding the previousmeasurements, it can be seen, that ZVS is not just a nice feature for efficiency maximization, but that it isabsolutely necessary for high power applications.

3.4 Frequency stability

At least, the prototype was driven with the full rectified grid voltage and the frequency was modulatedbetween 80 and 160 KHz, but there were no abnormalities.

Halfbridge: VpriHalfbridge: Vs6-s1 Halfbridge: Isec

Isec

V s6-

s1| V

pri

t

-0.000005 s -8.470329472543e-… 0.000005 s 0.00001 s 0.000015 s

0 V

-300 V

-200 V

-100 V

100 V

200 V

300 V

-7.5 A

-5 A

-2.5 A

0 A

2.5 A

5 A

7.5 A

10 A

Halfbridge: VpriHalfbridge: Vs6-s1 Halfbridge: Isec

Isec

V s6-

s1| V

pri

t

-0.000005 s -8.470329472543e-… 0.000005 s 0.00001 s 0.000015 s

0 V

-300 V

-200 V

-100 V

100 V

200 V

-10 A

-7.5 A

-5 A

-2.5 A

0 A

2.5 A

5 A

7.5 A

10 A

3.3 Zero voltage switching 7