reducing RFI in half bridge forward convertor

Try winding the powerlead through a big ferrite ringcore. I did this with high voltage power supply and it worked perfectly. Keep the core as close as possible to the supply

about

My bet would be that your freewheeling diodes are ringing - test by placing an RC network across the diode. A few tens of a nF and 10-47 Ohm should be fine for a quick test (since you do not have an oscilloscope).

First, 4cm high-current traces are bad. Looking at design notes for theses things, the critical routes should be about 1cm or less.

We had a half-bridge that sucked despite good layout. It turned out our controller was turning on the one FET before the other was completely turned off; this "shoot through" caused horrible EMI. Look for the shoot trhough and - if suspected - try to increase the time before the "other" FET kicks in.

Hi there. I do have a scope , so I'll check it out. When you say freewheeling diode , I assume you are refering to the diode on the secondary that is connected between GND and the cathode of the rectifying diode. Cheers Rob

I'm running half bridge , so there is only 1 FET. The lead length I'll have to look at. Thanks Rob

by

cathode

Yes - that is often the worst offender.

The diodes in the half bridge itself can also ring but it is not very likely since they should be off by the time the switches turn on - unless you got a really bad transformer with lots of leakage.

1. You can use foil tape to accomplish some of what multi-layer PCBs allow.
2. Learn the concept of 'loop area' and apply it to minimize loop area for high current loops. This applies even to snubbers.
3. You must have high quality bypass capacitors immediately in the area of your main switching loop, so that these capacitors supply most of the high frequency currents. Otherwise, high frequency currents will flow on your power leads.
4. Create a short capacitive path between the primary and secondary sides of your converter that allows parasitic currents to flow through your deliberate path. Otherwise, such currents will flow wherever they will, potentially in a large loop. Paul Mathews

Thanks for all the tips guys. Putting a snubber across the diodes as suggested earlier did the trick. Lots of noise has been killed this way. I still thing a new layout is in order , but things are looking much better already. I have come across another problem though. The load transient response is truly dreadfull. A step in load from 0 to about 14A causes the output to drop (fast) about 6V to about 7V and then climb back up to 13.6 v over about

Thanks for all the help so far!! Cheers Rob

Here is the circuit:

Version 4 SHEET 1 1916 688 WIRE -720 176 -720 -144 WIRE -720 624 -720 256 WIRE -352 320 -368 320 WIRE -352 480 -352 384 WIRE -288 192 -320 192 WIRE -288 288 -288 272 WIRE -288 320 -352 320 WIRE -288 480 -352 480 WIRE -288 480 -288 384 WIRE -288 624 -720 624 WIRE -288 624 -288 480 WIRE -256 192 -288 192 WIRE -176 288 -176 272 WIRE -160 320 -288 320 WIRE -160 448 -160 400 WIRE -160 464 -160 448 WIRE -160 624 -288 624 WIRE -160 624 -160 528 WIRE -96 192 -176 192 WIRE -96 288 -176 288 WIRE -96 288 -96 256 WIRE -80 448 -160 448 WIRE -80 448 -80 384 WIRE -32 192 -96 192 WIRE -32 256 -96 256 WIRE -32 320 -160 320 WIRE -32 384 -80 384 WIRE -16 320 -32 320 WIRE 0 -144 -720 -144 WIRE 0 -128 0 -144 WIRE 0 -48 0 -64 WIRE 112 624 -160 624 WIRE 112 640 112 624 WIRE 128 128 128 32 WIRE 128 448 128 432 WIRE 128 624 112 624 WIRE 128 624 128 448 WIRE 144 -144 0 -144 WIRE 144 -32 144 -64 WIRE 144 32 128 32 WIRE 144 32 144 -32 WIRE 208 -32 144 -32 WIRE 208 48 208 32 WIRE 208 80 208 48 WIRE 224 48 208 48 WIRE 224 48 224 32 WIRE 272 -32 208 -32 WIRE 320 224 288 224 WIRE 320 224 320 144 WIRE 352 -32 336 -32 WIRE 352 48 224 48 WIRE 352 352 288 352 WIRE 352 400 352 352 WIRE 352 496 352 400 WIRE 352 624 128 624 WIRE 352 624 352 560 WIRE 416 400 352 400 WIRE 432 144 384 144 WIRE 432 224 400 224 WIRE 432 224 432 144 WIRE 496 224 432 224 WIRE 496 224 496 160 WIRE 496 336 496 224 WIRE 560 -144 144 -144 WIRE 560 112 560 -144 WIRE 560 224 560 208 WIRE 560 288 560 224 WIRE 560 624 352 624 WIRE 560 624 560 384 WIRE 592 224 560 224 WIRE 624 -144 560 -144 WIRE 624 -128 624 -144 WIRE 624 -32 624 -64 WIRE 624 64 624 48 WIRE 688 224 672 224 WIRE 736 -144 624 -144 WIRE 736 -112 736 -144 WIRE 736 144 736 -32 WIRE 736 400 496 400 WIRE 736 400 736 240 WIRE 736 464 736 400 WIRE 736 624 560 624 WIRE 736 624 736 544 WIRE 848 432 848 400 WIRE 848 480 848 432 WIRE 848 528 848 480 WIRE 848 624 848 608 WIRE 880 432 848 432 WIRE 896 -32 864 -32 WIRE 960 48 864 48 WIRE 960 80 960 48 WIRE 992 432 944 432 WIRE 1024 -32 960 -32 WIRE 1024 -16 1024 -32 WIRE 1024 48 960 48 WIRE 1056 -32 1024 -32 WIRE 1104 480 848 480 WIRE 1136 288 1136 272 WIRE 1136 432 1072 432 WIRE 1136 432 1136 384 WIRE 1136 448 1136 432 WIRE 1136 624 848 624 WIRE 1136 624 1136 512 WIRE 1136 656 1136 624 WIRE 1152 288 1136 288 WIRE 1168 -32 1136 -32 WIRE 1168 -16 1168 -32 WIRE 1168 48 1024 48 WIRE 1168 48 1168 32 WIRE 1216 -32 1168 -32 WIRE 1312 -32 1296 -32 WIRE 1312 -16 1312 -32 WIRE 1312 48 1168 48 WIRE 1376 -32 1312 -32 WIRE 1376 -32 1376 -112 WIRE 1376 48 1312 48 WIRE 1376 288 1328 288 WIRE 1376 288 1376 144 WIRE 1408 352 1328 352 WIRE 1472 -112 1376 -112 FLAG 112 640 0 FLAG 0 -48 0 FLAG 208 80 0 FLAG 960 80 0 FLAG 1408 352 Vfb FLAG -320 192 Vfb FLAG 1136 656 0 FLAG 1376 -112 Vout FLAG -368 320 Vref FLAG 1376 144 Vref FLAG -288 288 0 FLAG 1168 -32 Vsense FLAG 1136 192 Vsense FLAG 848 320 Vout FLAG 624 64 0 SYMBOL UC3845B 128 288 R0 SYMATTR InstName U1 SYMATTR Value UC3842B SYMBOL res -176 304 R0 SYMATTR InstName R1 SYMATTR Value 15k SYMBOL res 512 384 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R4 SYMATTR Value 1k SYMBOL cap -176 464 R0 SYMATTR InstName C1 SYMATTR Value 1n SYMBOL cap -112 192 R0 SYMATTR InstName C2 SYMATTR Value 100p SYMBOL res -192 176 R0 SYMATTR InstName R5 SYMATTR Value 150k SYMBOL res 720 448 R0 SYMATTR InstName R6 SYMATTR Value 0.375 SYMBOL cap 336 496 R0 SYMATTR InstName C3 SYMATTR Value 220p SYMBOL res 416 208 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R7 SYMATTR Value 1k SYMBOL res 128 -160 R0 SYMATTR InstName R8 SYMATTR Value 100k SYMBOL cap -16 -128 R0 SYMATTR InstName C4 SYMATTR Value 100=B5 SYMATTR SpiceLine V=3D400 Irms=3D900m Rser=3D0.2 MTBF=3D20000 Lser=3D0 ppPk= g=3D1 SYMBOL ind2 720 -128 R0 SYMATTR InstName L1 SYMATTR Value 2.3m SYMATTR Type ind SYMATTR SpiceLine Rser=3D0.15 SYMBOL ind2 336 -48 R0 WINDOW 0 30 92 Left 0 WINDOW 3 36 40 Left 0 SYMATTR InstName L2 SYMATTR Value 6=B5 SYMATTR Type ind SYMATTR SpiceLine Rser=3D0.1 SYMBOL schottky 336 -48 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName D3 SYMATTR Value 1N4937 SYMATTR Description Diode SYMATTR Type diode SYMBOL cap 192 -32 R0 SYMATTR InstName C5 SYMATTR Value 47=B5 SYMATTR SpiceLine V=3D25 Irms=3D2.4 Rser=3D0.05 MTBF=3D0 Lser=3D2.1p mfg=3D= "KEMET" pn=3D"T510E107K025AS" type=3D"Tantalum" ppPkg=3D1 SYMBOL ind2 880 -48 M0 WINDOW 0 36 80 Left 0 WINDOW 3 49 40 Left 0 SYMATTR InstName L3 SYMATTR Value 35=B5 SYMATTR Type ind SYMATTR SpiceLine Rser=3D0.1 SYMBOL schottky 896 -16 R270 WINDOW 0 32 32 VTop 0 WINDOW 3 0 32 VBottom 0 SYMATTR InstName D4 SYMATTR Value byv32-200 SYMATTR Description Diode SYMATTR Type diode SYMBOL voltage -720 160 R0 WINDOW 123 0 0 Left 0 WINDOW 39 24 132 Left 0 SYMATTR InstName V1 SYMATTR Value 311 SYMBOL ind2 1312 -48 R90 WINDOW 0 5 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName L4 SYMATTR Value 10=B5 SYMATTR SpiceLine Rser=3D0.1 SYMATTR Type ind SYMBOL Optos\\\\4N25 1232 352 R0 SYMATTR InstName U4 SYMBOL mystuff\\\\my\\ nmos 688 144 R0 WINDOW 38 57 51 Left 0 SYMATTR SpiceModel irf840_modified_to_800v SYMATTR InstName U5 SYMBOL mystuff\\\\tl431 1120 448 R0 SYMATTR InstName U7 SYMBOL res 832 512 R0 SYMATTR InstName R11 SYMATTR Value 2k2 SYMBOL res 1120 176 R0 SYMATTR InstName R13 SYMATTR Value 2k SYMBOL res 832 304 R0 SYMATTR InstName R14 SYMATTR Value 10k SYMBOL cap 944 416 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C9 SYMATTR Value 100n SYMBOL res 1088 416 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R17 SYMATTR Value 300 SYMBOL res -304 176 R0 SYMATTR InstName R2 SYMATTR Value 1k SYMBOL res -160 176 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R3 SYMATTR Value 10k SYMBOL cap -304 320 R0 SYMATTR InstName C8 SYMATTR Value 100n SYMBOL res 1360 -48 R0 SYMATTR InstName R9 SYMATTR Value 0.75 SYMBOL schottky 1040 48 R180 WINDOW 0 24 72 Left 0 WINDOW 3 24 0 Left 0 SYMATTR InstName D2 SYMATTR Value byv32-200 SYMATTR Description Diode SYMATTR Type diode SYMBOL cap 1296 -16 R0 SYMATTR InstName C6 SYMATTR Value 220=B5 SYMATTR SpiceLine V=3D25 Irms=3D355m Rser=3D0.22 MTBF=3D2000 Lser=3D0 mfg=3D"Nichicon" pn=3D"UPR1E331MPH" type=3D"Al electrolytic" ppPkg=3D1 SYMBOL ind 1152 -48 R90 WINDOW 0 5 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName L5 SYMATTR Value 27=B5 SYMBOL cap 1152 -32 R0 SYMATTR InstName C7 SYMATTR Value 1000=B5 SYMATTR SpiceLine V=3D25 Irms=3D600m Rser=3D0.13 MTBF=3D20000 Lser=3D0 mfg=3D"Nichicon" pn=3D"UPH1E331MRH" type=3D"Al electrolytic" ppPkg=3D1 SYMBOL ind2 640 64 R180 SYMATTR InstName L6 SYMATTR Value 2.3m SYMATTR Type ind SYMATTR SpiceLine Rser=3D0.2 SYMBOL diode 640 -64 R180 WINDOW 0 24 72 Left 0 WINDOW 3 24 0 Left 0 SYMATTR InstName D5 SYMATTR Value 1N4937 SYMBOL cap -368 320 R0 SYMATTR InstName C10 SYMATTR Value 1=B5 SYMBOL mystuff\\\\mynpn 496 112 R0 SYMATTR InstName U2 SYMATTR SpiceModel 2n2222a SYMBOL mystuff\\\\mypnp 496 384 M180 SYMATTR InstName U3 SYMATTR SpiceModel 2n2907a SYMBOL res 688 208 R90 WINDOW 0 0 56 VBottom 0 WINDOW 3 32 56 VTop 0 SYMATTR InstName R10 SYMATTR Value 27 SYMBOL cap 384 128 R90 WINDOW 0 0 32 VBottom 0 WINDOW 3 32 32 VTop 0 SYMATTR InstName C11 SYMATTR Value 10=B5 TEXT -98 560 Left 0 !.tran 2050m startup TEXT -48 480 Left 0 !.include uc384x.lib TEXT 816 -96 Left 0 !k l1 l2 l3 l6 1

Here is the uc384x:

Version 4 SymbolType CELL RECTANGLE Normal -160 -160 160 160 WINDOW 0 8 -176 Left 0 WINDOW 3 8 176 Left 0 SYMATTR Value UC3845B SYMATTR Prefix X SYMATTR Description High Speed Current Mode Pulse Width Modulator PIN -160 -32 LEFT 8 PINATTR PinName COMP PINATTR SpiceOrder 1 PIN -160 -96 LEFT 8 PINATTR PinName FB PINATTR SpiceOrder 2 PIN 160 64 RIGHT 8 PINATTR PinName Isense PINATTR SpiceOrder 3 PIN -160 96 LEFT 8 PINATTR PinName Rt/Ct PINATTR SpiceOrder 4 PIN 0 160 BOTTOM 8 PINATTR PinName GND PINATTR SpiceOrder 5 PIN 160 -64 RIGHT 8 PINATTR PinName Output PINATTR SpiceOrder 6 PIN 0 -160 TOP 8 PINATTR PinName Vcc PINATTR SpiceOrder 7 PIN -160 32 LEFT 8 PINATTR PinName Vref PINATTR SpiceOrder 8

The slow transient could be caused by a too-large inductor. What is the size of your inductor and what does the current ramp look like? (For a non-isolated DC-DC buck converter, I always have a series resistor from the rail for current sense, for instance) What frequency are you running the switcher?

(top posted because my mail might not quote properly)

Thanks for all the help so far!! Cheers Rob

Here is the circuit:

That's one hell of an inductor.

Graham Holloway

Things that make a switcher respond slowly:

Feedback loop too slow. Feedback gain too low. Converter frequency too low. Too much output capacitance. Too much inductance.

I don't recall if you said whether you are using any peak current feedback, but properly stabilized current feedback can be faster than voltage feedback alone. However, even a pure voltage feedback system can be much faster than 100ms. Paul Mathews

Hi there. The inductor in my application is 27uh. I don't think that this is to big. If I make it much smaller the inductor current becomes discontinous. I put a loop of wire through the inductor(toroid) conected to a 330 ohm resistor to see what the inductor current looked like. It is decidedly square in nature :0( I was expecting to see a triangle current waveform. I assume this means that my inductor may be saturating. I took this toroid out of a used pc power supply , so I assumed it would work for my application. I seem to have missed the boat somewhere. Do my deductions seem reasonable? Cheers Rob

This technique does not show you what the current is doing, but what the voltage across the inductor is doing.

It shows you that during part of the cycle, there is one voltage across the inductor, and at another part of the cycle, there is another voltage across the inductor. Hence the two level output.

Sorry, no. You still have a lot of basic things to learn about inductors.

I suggest you read through this tutorial for some leads.

Sorry , bad measurement method used there. Putting a low value resistor in series with the coil shows a triangular waveform , which I expect. dI is about 2.6A for a 9A load , which is about 35%. This is about right I think. If I am wrong here please let me know. Cheers Rob

Hi all. I'm being my own worst enemy here. Nothing like being off for a few days to see the error of one ways. I have been using 12v 50W light bulbs as a load. When cold there resistance is a fraction of what it is when hot. My circuit is in fact working fine. The current limit is kicking in and dropping the voltage. As the bulbs heat up the output rises to the correct voltage. Feel a bit of a fool :0( Using a 1 ohm 100W resistor the output behaves more like I would expect. Load change fom 0 - 13A the output drops about 0.8V for about 400uS. Good enough for me. Any ideas / opinions still welcome. Thanks for all the advice so far. cheers Rob

What is dt?

Of what?

I have no idea if I agree or not.

That's di , change in inductor current , sometimes refered to as ripple current I think. The 35% is the percentage of ripple current to load current (9A). Probably should be a little less , but I think it's ok. Supply seems to running ok. Still feeling my way a bit , but am getting there slowly. Cheers. Rob

Sorry , misunderstood your question at first. dt is about 3us.At home now , so am going from memory. The convertor is running at about 120Khz or so.