Power Supply Troubleshooting - Flyback keeps killing clamp diodes

[Snee_A]

I have an off line flyback converter power supply that keeps killing its clamp diodes and I'm not entirely sure what to attribute the issue to.

Initial symptom was a "ticking" sound and highly unstable output voltages. The ticking turned out to be the switching controller repeatedly turning on and off again. It's got a stable output to the fet when its on, but it quickly recognizes an issue and shuts off.

The FET checks out ok though I haven't taken it out of circuit to be sure. Feedback resistors are ok. Oscillator freq is about 48Khz and based on R12 and C3 that's about right.

The only problem component I could find was D6, a 100V zener diode. It's paired with a 400V avalanche rectifier D5, but that one seems to survive the issue. This happens even with D3, D4, and D7 out of circuit.

All measurements and observations taken under no load, I don't have a suitable synthetic load unfortunately.

Attached is the schematic and scope measured from the anode of D6 to the anode of D5, right before D5 dies. About 288Vpp, coming from a bridge rectifier fed by 120V. It starts around 200Vpp and quickly ramps to ~300 and that's when D5 dies and it settles there.

Any and all help is greatly appreciated.

[doktor pyta]

Maybe operation without the load is the problem?
You could try adding a loading resistor at one output.

[bdunham7]

First things I would check would be C14, C19 and C22.  Maybe C17 and C21 as well. Then D1, D2, C1 and C8. I wouldn't try to run it with parts missing. 

[capt bullshot]

+1 to bdunham7,
but I'd check C1 and C8 first

[Snee_A]

We are possibly on to something; everything in bdunham's list was nominal (tested out of circuit) except C8, which is 22uF instead of 47uf. I'll look for a replacement and let you know what results.

I realize that no-load operation is less than ideal, but as I said I don't even have sufficient power resistors to load this at all, and the device I took it out of is somewhat precious so I can't use it as a load. I'd hope that a well designed supply would not kill diodes simply because it was turned on with no load...

[bdunham7]

C8 is the one which will affect the stability of the feedback the most and works the hardest, so no surprise there.  But if you are ordering parts, I'd suggest just replacing at least C14, C19 and C22 as well unless you can properly test them for ESR.   

[Snee_A]

We have a winner! Didn't have a 63V 47uF on hand but I did have 35V 100uF that I could pair in series, and the supply works just fine. Haven't tested it in the device, but I'm pretty certain that was the problem.

I will heed your advice and swap the electrolytics just to be safe, I just wanted to make sure I wasn't putting lipstick on a dead pig before I bought them.

So, being relatively new to flyback topology, I take it that D2/C8 are responsible for conditioning the output of the feedback winding somehow? I'm very curious to know how you suspected C8. I feel like I just missed it because I did notice a pretty substantial sawtooth ripple at VCC of the switching controller but didn't know what to think of it given that the input cap was fine.

Much thanks to everyone for the assistance, I hope to be able to pay it forward.

[bdunham7]

In simple terms, if the secondary side is 'full' and can't absorb the energy from the transformer, it will go back into the primary and feedback windings.  D2/C8 integrate that and supply both the V_cc and feedback to the controller IC.  D1 and C1 further filter and isolate V_cc.  The controller responds to the feedback by reducing the drive pulses to the FET.  If  the feedback doesn't work or if the output capacitors are bad, it may keep driving until something else breaks, like your poor zener.  It's hard to know what the chain of failures looks like, but I think high ESR on the output sides causes all sorts of knock-on issues in these supplies.