Arc'd/fused cap lead in power supply - Need Help!

[Gaz]

Hi guys,

I've had this happen twice now in different amps, and I'm totally baffled.
What's happening is, for some unknown reason, the lead on an electrolytic cap in the power supply is fusing open and arc'ing. Not sure how this would even happen.

The caps leads are bent with a small screwdriver so as to not stress where they are exiting the cap, and also stuck down with 100% silicone, which you can see in the photo.

The other time was also in the power supply, but the main B+ cap.

Any ideas why this would happen?!

Thanks.

[xtian]

Never had a failure like that. Is that a Nichicon cap?

What's the operating voltage across that cap?

Did it fail during operation, or during startup?

Happened twice in two different amps? Bad batch of caps?

[Gaz]

Thanks for the thoughts. Not sure about when exactly it happened because it was intermittent until I found it. The silicone was holding the cap from flopping around so it was still making contact sometimes. It was only when I could see it arc'ing with the lights off I noticed where it was coming from.
I thought this time it was a different problem.

It's a Nichicon cap (CY Series) rated at 350V. They are stacked with 220K/2W balancing resistors for a total rating of 700V. The voltage there is only about 445V.

I actually just dismantled the cap further, but inside it looks normal, meaning no stress or anything where it enters the actual cap inside the casing. Both times it happened right where the lead enters the rubber, and I'm not sure if that a hint or not.

I would love to chalk it up to bad parts, but the other one that failed was a different type and from a different order.

[trobbins]

Is the failed location on the lead where someone introduced a bend in that lead?

Is the silicone near or on the failed location?

Perhaps you have cracking in the lead where it was bent, followed by some corrosion due to the acetic acid of the silicone when it was applied, followed by vibration of the large capacitor stressing the cracks.

[strelok]

That would be my guess. Failure due to chemical or mechanical degradation. You'd have to be putting some serious current through there to cause the lead to get hot enough to fuse open. Something else would give long before that ever happened.

[Firestorm]

That's awfully strange. And twice! When you first power up, current is virtually infinite for about one thirtieth of a second. Are the leads well separated? Some silicones are conductive until fully cured. Did you start up too soon?

[Gaz]

Thanks for all the ideas. I have considered most of these.

The leads are as separated as they would be as if mounted normally into PCB, and only a 'soft' bend using a small screw driver. The lead is a couple mm away from the cap's can, and so sign of arcing through the insulation. It could've been my bending because this has only happened on these two caps, but I am always pretty gentle when bending them. This amp was operating fine for a couple months before this came up...

The fused/severed spot was right where the lead enters the cap, not quite at the bend. This is also typically where the silicone is, and so I am worried that may be the cause.

The silicone I use is GE Silicone II and is 100% silicone. No acetic acid, and is not supposed to be conductive in any state, according to Andy Fuch from another thread. I have also tried taking a dried strip of silicone and connecting to the B+ to see if it would act as a conductor, but 0V. Also, I cannot measure any voltage on the dried, aged silicone under the caps either.

I have made other amps using this method, some up to 4 years ago, and these are the only ones I have had a problem with. That said, I'm not naive, and think it happening to two both in the power supply cannot be a mere coincidence.

Is there any electrical problem that could cause this without blowing the fuse? The cap is itself also seems to be good, and the only noise/problem is the broken/arc'ing lead itself. Again, does that just point to the f*ck'ing
silicone?...

Okay, I just ripped up all the caps in the PS, and tried to carefully to remove the silicone, and I cannot find any look of "corrosion," however, I'm not sure if I would be able to see it or not. Here's a couple photos of the silicone removed from another cap in the power supply, before and after. Sorry they are not the best - so hard to get good macro shots.

[trobbins]

Are the amps subject to long term vibration - perhaps from transportation?

[Gaz]

Are the amps subject to long term vibration - perhaps from transportation?

Nothing out of the ordinary, just travel in cars to and from practice/gigs. Only other vibration would be on top of a 4x12 blaring at 100W.

[Gaz]

And not sure if this matters, but check out where it happened before, at the junction of the two B+ caps.

[sluckey]

I don't think this failure was electrical. I think it's purely mechanical. Looks like the lead simply detached from the capacitor plate. That could be due to a manufacturing defect or too much stress when bending the lead. Once the lead became detached the arcing between the plate and lead caused the burning on the lead.

Not much to do if it's a manufacturing defect. But you can minimize stress on the lead by grasping the lead with a pair of round nose pliers (use needle nose if you don't have round nose)BETWEEN the body of the cap and the point you want to make the bend. If you need to make another bend to go around a turret lug, terminal, eyelet, etc., always hold the lead with pliers between the cap body and the point of the lead.

[Gaz]

I don't think this failure was electrical. I think it's purely mechanical. Looks like the lead simply detached from the capacitor plate. That could be due to a manufacturing defect or too much stress when bending the lead. Once the lead became detached the arcing between the plate and lead caused the burning on the lead.

Not much to do if it's a manufacturing defect. But you can minimize stress on the lead by grasping the lead with a pair of round nose pliers (use needle nose if you don't have round nose)BETWEEN the body of the cap and the point you want to make the bend. If you need to make another bend to go around a turret lug, terminal, eyelet, etc., always hold the lead with pliers between the cap body and the point of the lead.

I totally agree that there's no way logically this could be electrical without something else frying (or fuse popping), or some sign of arc'ing to ground.
The PS is so simple too...

But, I did do exactly as you instructed, and have been doing it that way for a while to keep from stressing the internal connections. Also, the lead broke above the bend, where the lead enters the rubber. I dismantled the cap as well, and past the rubber (about an 1/8") all looks well. The stress did not happen inside the cap, but right where it enters and where the silicone's edge is, which you can see in the photo.

Any other ideas?

[sluckey]

When you disassembled the cap did you find a short stub of the broken lead? If so, was it still attached to the foil plate?

[Structo]

Maybe when you wrapped the lead around the turret, it strained the connection to the innards and with heat and use it failed.

Double check the adhesive to make sure it isn't conductive.

[Firestorm]

I keep coming back to this because it's so odd. Any chance the plus side is shorting to the outer case? Radials weren't designed to have the leads bent sideways.