Is exceeding the PIV of a diode by 3 x tolerable in any instance?

[tom66]

I've been repairing a lot of 17PW25-4 power supplies. I've repaired 4 so far in the space of about 6 months. They are utter junk. They have the same failure mode: a shorted diode on the secondary and damage to the primary side, blown mains fuse, shorted FET.

The original power supply uses UF5402 (3A, 100V) diodes on the output (8 in parallel!) I have used HER308G (3A, 1kV) to replace them. This is for the 24V, 6A output.

I scoped the anode of the diodes on power up - the original UF5402 are 100V rated... there are repeated transients of up to 280V on the cathode when the power supply is turned ON (this is not AC plug in, this is turn on from standby) occuring over about a 2ms period.

I suspect this is why these power supplies are failing. Could this kill diodes? I haven't been able to measure the current, so I can't be sure this is  dumping much energy into them. But it cannot be good, right?

[edavid]

The UF5402 is actually a 200V rated part.   Are they breaking down at 280V?  It doesn't look like it from your scope trace.

[tom66]

Good point, I will make sure but I think they might have been UF5401s. I do recall them being 100V rated parts.

This scope shot was taken with 1kV diodes installed (HER308G) after the PSU was repaired.
 
Forgot to mention, the transients only occur when switched on. Once operating, the inverse voltage across the diodes is about 60~70V maximum.

[T3sl4co1l]

Try a ferrite bead in series with each diode (which will help with transient current balance as well), and an R+C across the diodes (you'll have to find correct values, but I would vaguely guess 10 ohm and 2nF would be a starting point).

It doesn't look like you even have enough bandwidth to observe the phenomenon.  It may be faster, crunchier or squarer than you can see.

Or even better, see if you can fit a proper TO-220 style device somewhere.

Tim