failed diode / PT question

[studiodunn]

After about 20 hours of play,tune,play,tune with this 100w amp I've built. The mains fuse blows. I pull the tubes and it blows again, so now I go through the debug process.

AC line only - fuse holds
Wire in HV and fuse blows. So I find a failed diode in the rectifier section. I replace all 4 and C120 & 121. I check the PT as it's unhooked and all voltages are good, so I hook things back up one step at a time and I'm back to the play & tune routine. I take a break come back and flip the switch - fuse pops. I go straight to the rectifier and find a failed diode.

I unhook all leads, replace the rectifier diodes and start over with the debug. AC line fine / PT voltages good. I hook up the HV secondary and fuse holds.

Here are my voltages with HV only hooked up. Something isn't right if my LV winding is showing 256v, right?

Screenshot 2023-09-28 at 10.20.11 AM.png

PT is APD 8053H. I'm running 4x6L6 off of the heater winding 6.3v@7A

Screenshot 2023-09-28 at 10.43.59 AM.png

[bepone]

remove the snubber caps.. put 2 diodes insted on one, so in total 8 pcs..move standby to the AC line, DPDT, sw.off both lines of transformer instead this DC

[johnnyreece]

Just checking; do you have the PT center tap *and* the diode bridge connected to ground? I believe that would do it.

[R.G.]

bepone has a good guess: failing snubber cap.

In situations like these, the dynamic characteristics matter. If (for instance) C110 C111 don't have reasonably close leakages, the lower-leakage cap takes more of the voltage, and can be pumped to a higher peak voltage than a rectifier can take. I would guess something like this for why the rectifiers blow. Swamping resistors of 220K to a meg across the snubber caps might help a lot. This is a common issue with stacked series filter caps, and can come up with series rectifiers too.

On the other hand, the snubber caps may well not be needed. the "UF" designator indicates that these are fast/soft recovery types, and they don't have the slam-off character of ordinary diodes, so they may not cause the haze of 120hz noise that is the reason for snubbing.

I might choose to use a sledgehammer: replace the UF5308s with Microchip APT15DQ120KG FREDs in a TO-220 package and leave off the snubber caps. These are rated for 1200V and 15A(!) with a surge rating of 110A and only cost $0.98 each at Mouser, versus the $0.56 each for UF5308s. I might also soften the blows of standby by including a small resistor in series with the snubber cap across the standby switch.

johnny also has a good point: check that it's running as a doubler, not a shorted FWR.

[studiodunn]

Just checking; do you have the PT center tap *and* the diode bridge connected to ground? I believe that would do it.

center tap is tied off and not grounded

[studiodunn]

remove the snubber caps.. put 2 diodes insted on one, so in total 8 pcs..move standby to the AC line, DPDT, sw.off both lines of transformer instead this DC

Thanks bepone, I'll ditch the snubber caps first see if this stabilizes things

bepone has a good guess: failing snubber cap.

In situations like these, the dynamic characteristics matter. If (for instance) C110 C111 don't have reasonably close leakages, the lower-leakage cap takes more of the voltage, and can be pumped to a higher peak voltage than a rectifier can take. I would guess something like this for why the rectifiers blow. Swamping resistors of 220K to a meg across the snubber caps might help a lot. This is a common issue with stacked series filter caps, and can come up with series rectifiers too.

On the other hand, the snubber caps may well not be needed. the "UF" designator indicates that these are fast/soft recovery types, and they don't have the slam-off character of ordinary diodes, so they may not cause the haze of 120hz noise that is the reason for snubbing.

I might choose to use a sledgehammer: replace the UF5308s with Microchip APT15DQ120KG FREDs in a TO-220 package and leave off the snubber caps. These are rated for 1200V and 15A(!) with a surge rating of 110A and only cost $0.98 each at Mouser, versus the $0.56 each for UF5308s. I might also soften the blows of standby by including a small resistor in series with the snubber cap across the standby switch.

johnny also has a good point: check that it's running as a doubler, not a shorted FWR.

Thanks for the suggestions RG. I'll try one route at a time and put the "Microchip APT15DQ120KG FREDs" in queue if things don't get sorted by doubling the diodes.

[studiodunn]

Just wanted to follow up and say that replacing the snubber caps with diodes has cured the amp. I've go probably 20 hours and 100 cycles on and off with no issues. Thanks for the quick help!

[R.G.]

So it was a shorted snubber cap??

[studiodunn]

So it was a shorted snubber cap??

I didn't test any of the caps when pulled. I just replaced all the diodes and moved on. So far so good. I played a show with it yesterday and amp worked perfectly.

[SoulFetish]

just curious, where did you buy you UF5408s?
A few weeks back a customer came in with a newer model Sovtek MIG 50 with a blown mains transformer. Even though it was out of warranty, they replaced the PT, no cost to customer. It turns out that a shorted 1N5408 probably took the transformer out (usually the mains fuse will blow, but we know that they their job isn't to prevent failures due to secondary faults). Anyway, we replaced the diode; installed the new PT; tested everything; all is good so we sent the amp back into the world.
Customer called I think a day later (maybe two). "Amp is blowing fuses after playing for an hour or so". We tell him to bring it back in, and wouldn't you know, New Power transformer is shorted. Crazy. The probable cause?:
another shorted 1N5408.
Again, contacted EHX and reported what had happened, sharing the date codes for the failed transformers-etc. This time they sent a whole mod kit along with a New PT. Included was a set of 8 new 5408s to double them up.
So, 1N5408s are pretty robustly rated rectifier diodes for our purposes and the same goes for the Ultra fast version. I can't say with any reasonable certainty, but I wonder if that they ran into a batch of sketchy diodes. When we rebuilt the rectifier section, I used series pairs of our HER208s (ultra fast recovery 1000V/2A) sourced from mouser to be safe.
Do you remember where you got them?
Im curious if the snubber caps were the issue, and removing them solved the problem, or if doubling up the diodes was the actual fix.

cheers.

[maxkracht]

I had an older MIG50 in with a blown soviet era bridge rectifier just last month... Not sure sure why they would be more prone to failure than anything else, the B+ is high, but not that high. Probably just a coincidence.

I watched a video recently about someone getting counterfeit voltage regulators from Mouser. I think Mouser is starting to allow 3rd party sellers like Digikey has been. In this instance, it was Mouser selling parts from Adafruit. The voltage regulator had ST branding and I believe the listing said ST in the title but the stated manufacturer was Adafruit. So, check listings carefully, supply chains are weird.

[SoulFetish]

I had an older MIG50 in with a blown soviet era bridge rectifier just last month... Not sure sure why they would be more prone to failure than anything else, the B+ is high, but not that high. Probably just a coincidence.

I watched a video recently about someone getting counterfeit voltage regulators from Mouser. I think Mouser is starting to allow 3rd party sellers like Digikey has been. In this instance, it was Mouser selling parts from Adafruit. The voltage regulator had ST branding and I believe the listing said ST in the title but the stated manufacturer was Adafruit. So, check listings carefully, supply chains are weird.

Dude, the crazy thing about the counterfeit parts market is the amount of effort put into making something look like an OEM part, as well as operate at least somewhat as expected. Why not just make the real product?

[maxkracht]

Dude, the crazy thing about the counterfeit parts market is the amount of effort put into making something look like an OEM part, as well as operate at least somewhat as expected. Why not just make the real product?

More graphic designers than engineers in the world? Some are quite impressive. I think the usual trick is to rebadge a cheaper part or to sell the bits that didn't pass qc. I have also watched some horrifying videos of people dipping old boards in a wok full of solder then smacking it on a rock so the chips fall out. Sort, clean, straighten pins, rebadge, on to Ebay or Ali express...