Working switch mode power supply with overheating power resistor

[geekyoyd]

I have a Crest X-Rack audio mixing rack that sort of works but there is a power resistor in the switch mode power supply that runs way too hot and is charring the PCB.  Sometimes, the power supply refuses to run but other times it runs and regulates well but the 5 Watt replacement resistor R18 (circled in red on the schematic, originally 2 Watt) I put in heats up to about 120 degs C (248 deg F) in open air, probably higher in the case.  Could anyone give me a tip on why this could be happening?  I have tested all the surrounding HV capacitors for leakage at 1200 Volts and all suspect electrolytics have been replaced.  Could the feedback circuit be faulty and causing the circuit to run at the wrong frequency?  Has anyone experienced PWM switches like the TOP227Y going bad without failing completely? Any help greatly appreciated.

[amyk]

Check if C25 is leaky.

[geekyoyd]

Check if C25 is leaky.

Thanks, I did check C25 with a uA meter at 1200 volts and it tested good.  I will try replacing it anyway at some point.

[timenutgoblin]

Maybe check D1 and D4? D1 is a 400V rated TVS diode and D4 is a 200V rated TVS rated diode. D5 is a rectifier diode.

Link to datasheet below for TVS diodes:

https://pdf1.alldatasheet.com/datasheet-pdf/view/1236105/YFWDIODE/P4KE400A.html

[geekyoyd]

Maybe check D1 and D4? D1 is a 400V rated TVS diode and D4 is a 200V rated TVS rated diode. D5 is a rectifier diode.

Link to datasheet below for TVS diodes:

https://pdf1.alldatasheet.com/datasheet-pdf/view/1236105/YFWDIODE/P4KE400A.html

I just checked the breakdown voltage on both and they were spot on 400V and 200V as spec.  D5 hit reverse avalanche at 850 Volts (spec 700V minimum)

[geekyoyd]

Looking at the example circuit shown in the TOP227Y datasheet, the transient suppressor circuit is just a TVR and a reverse blocking diode.  Most other circuits I have seen don't have the transient filter across the switch like this circuit does.  I wonder why it is needed.

[timenutgoblin]

Looking at the datasheet for the TOP227Y device, it gets it's power from the DRAIN terminal to produce it's internal supply rail(s). The CONTROL terminal should measure 5.7V (according to the datasheet). Check C14.

TL431 regulator REG2 looks like it's responsible for the +18VDC rail?

You could always replace the TOP227Y if it's starting to fail.

I have no further suggestions or ideas!

[Circlotron]

Looking at the example circuit shown in the TOP227Y datasheet, the transient suppressor circuit is just a TVR and a reverse blocking diode.  Most other circuits I have seen don't have the transient filter across the switch like this circuit does.  I wonder why it is needed.

Maybe keeping a lid on EMI?

[timenutgoblin]

Looking at the example circuit shown in the TOP227Y datasheet, the transient suppressor circuit is just a TVR and a reverse blocking diode.  Most other circuits I have seen don't have the transient filter across the switch like this circuit does.  I wonder why it is needed.

Maybe keeping a lid on EMI?

I wonder if there are any ferrite beads on the DRAIN and SOURCE terminals of the TOP227Y device? If not, probably not required.

[geekyoyd]

I suppose it might be to cut down EMI, especially since it is powering a professional audio mixer with very sensitive microphone inputs that could pick up interference.

There are were no ferrite beads on the PWM switch legs.

I think I will solder everything back in the board and see what frequency it is switching at.  I assume that the transient snubbing resistors will overheat if it is running at too high a frequency.

[thinkfat]

C25 and the rest of the snubber network with D4, D5, C39 and R20 are IMHO the best candidates. R18 burning up and charring the PCB means there's considerable current going through it, at a high voltage. C25 would be my first candidate, but if it checks out fine, you may be right about too high switching frequency. But it has to be considerably higher than rated to make a difference. Another possibility to take into account is that C25 is maybe not correctly dimensioned to provide sufficient drop at 100kHz.

[xavier60]

The frequency can be more safely checked on  the transformer secondary pins.
Another possible cause is high ESR output capacitors.

[geekyoyd]

Below are a pair of scope photos.  The first is the SMPS operating with a 240 VAC supply (yes, I do live in the UK) and the second is with a 120 VAC supply.  With a 120 VAC supply, R18, the snubber resistor that is normally running too hot, rises to 82 degC.  With a 240 VAC supply it rises to 153 degC (a bit too hot for a PCB and solder joints in my opinion).  I measured the voltage accross R18 (330 Ohm 2 Watt in the schematic but I have put a 5 Watt in there and it still overheats) with a high voltage differential probe on my Hameg HM604-2 scope.  The PWM is operating at near enough 100 kHz which is normal for the TOP227Y. 

From the datasheet:
• Minimize peak voltage and ringing on the DRAIN voltage
at turn-off. Use a Zener or TVS Zener diode to clamp the
drain voltage below the breakdown voltage rating of
TOPSwitch under all conditions, including start-up and
overload. The maximum recommended clamp Zener
voltage for the TOP2XX series is 200 V and the
corresponding maximum reflected output voltage on the
primary is 135 V. Please see Step 4: AN-16 in the 1996-97
Data Book and Design Guide or on our Web site.

So the maximum allowed reverse voltage on the switch is 200V.  I don't know, there doesn't seem to be enough area under that R18 voltage curve to dissipate more than 5 watts in a 330 Ohm resistor but I guess there must be.  I'm totally flumoxed by this one!  {:(

Taa for all your help so far.

[thinkfat]

This is where a DSO can be real handy for measuring the RMS voltage across R18, but maybe a good multimeter could do it just as well, provided it is spec'd for 100kHz. But quite obviously it is high enough to heat up the resistor to 150°C or thereabouts and depending on the PCBs temperature grade that might be OK (well, 150°C _is_ hot). Anyway, you cannot expect a 5W resistor to stay cold at 5W. The rating just states that it's able to get hot enough without burning up.

[cbutlera]

The maximum power that could be dissipated in R18 is if all of the stored energy in C25 were dissipated in it at every transition, which would be enough to account for the observed overheating. This would occur if the transition times for the voltage across the primary of T2 were very small.  So assuming that the operating frequency is as it should be, then I think that the rise and/or fall times are less than they should be.  Two possible causes for this could be a fault in the driving circuit for U2, or damage to the core of T2.

[fzabkar]

Maybe the supply would behave if you were to eliminate the ringing? That's where most of the energy in R18 is being dissipated, if I understand correctly. It might be worth seeing what happens if C25 were increased from 220pF to 330pF, say.

[xavier60]

Ringing is due to underloading and likely normal. I'd suggest reducing the value of R18.

[Circlotron]

Disconnect the unit from the mains and apply enough DC voltage across the 330 ohm resistor until it gets to the same temperature as it would normally. 41 VDC = 5 watts, so no more than that. The it should be easy to calculate how much it is dissipating when operating.

[geekyoyd]

Ringing is due to underloading and likely normal. I'd suggest reducing the value of R18.

I was running the PSU at only 2.5 Watts into a dummy load to keep the PWM switch cool.  For access, I just removed it from it's heatsink.

I was considering reducing the value of C25, not increasing it.  Wouldn't that dump more energy into R18?  I can see that lowing the value of R18 would cool it down.  I did consider possible damage in the transformer but there are no signs of overheating from shorted windings there.

The maximum power that could be dissipated in R18 is if all of the stored energy in C25 were dissipated in it at every transition, which would be enough to account for the observed overheating. This would occur if the transition times for the voltage across the primary of T2 were very small.  So assuming that the operating frequency is as it should be, then I think that the rise and/or fall times are less than they should be.  Two possible causes for this could be a fault in the driving circuit for U2, or damage to the core of T2.

Are you saying that the switching times are too fast creating excessive transients?

[fzabkar]

Ringing is due to underloading and likely normal. I'd suggest reducing the value of R18.

I was thinking that if the resonant frequency of the series circuit (T2 + C25 + R18) were less than the switching frequency (100kHz), then there would be no ringing.

f = 1/(2 x pi x sqrt( L x C))

Increasing C would reduce f.

Does that make any sense at all?

[Circlotron]

You can actually make the ringing work for you if you tune it's resonant frequency so the instantaneous voltage is lowest right at the point of switching device turn on.

[xavier60]

Some controllers do that, time the turn on to the bottom of the ring. Not this one though.

[geekyoyd]

Ringing is due to underloading and likely normal. I'd suggest reducing the value of R18.

I was thinking that if the resonant frequency of the series circuit (T2 + C25 + R18) were less than the switching frequency (100kHz), then there would be no ringing.

f = 1/(2 x pi x sqrt( L x C))

Increasing C would reduce f.

Does that make any sense at all?

Yes, it does make sense, I'll give all your suggestions a try.  It will be an interesting experiment and I may not get this power supply working properly in the end (say because of a dud transformer! {:(  ) but I will learn something useful in the process.

[cbutlera]

The maximum power that could be dissipated in R18 is if all of the stored energy in C25 were dissipated in it at every transition, which would be enough to account for the observed overheating. This would occur if the transition times for the voltage across the primary of T2 were very small.  So assuming that the operating frequency is as it should be, then I think that the rise and/or fall times are less than they should be.  Two possible causes for this could be a fault in the driving circuit for U2, or damage to the core of T2.

Are you saying that the switching times are too fast creating excessive transients?

Yes, when the transition occurs significantly faster than the RC time constant, then the majority of the energy stored in C25 1/2*C*V², where V is the voltage change, will be dissipated in the resistor.  My foremost thought is that there could be a fault with T2,  not a burnt winding, but a damaged core.  This could have the effect of reducing the coefficient of coupling between the windings, which would be the equivalent of adding an inductor in series with the winding.  Another possibility is that T2 was poorly wound from manufacture.  If it was intended to have an even distribution of the winding turns, but in fact doesn't.

[jonpaul]

Hello: Just saw this post now.

Snubber power is a very old and well known question. It should be perhaps a few % of the max output power, so 5W is a lot for a small supply.

The ringing is a natural result of energy stored in the leakage inductance of the magnetics after switch turn off.

Minimum load is a worst case, as network damping is reduced.

The PRF does not affect the ring freq, only determined by the Llkg, and parasitic inductances of the magnetics, combined with the snubber R-C.

Many flyback and forward converters use active snubbers, diodes, TAZ or extra windings to reduce the lost power of the required snubber.

Old Vicor patents are an excellent example.

To learn more the OP could search sites about converter  forward and flyback design, and snubbers.

Enjoy,

Jon