Lesser of 3 evils?

[slerner]

Hi, my project is nearly on the right track https://www.badcaps.net/forum/showthread.php?t=115083 However, there's one last electrolytic cap that I need to replace and it's just not available, even at higher voltage. So, this question is a bit more general & maybe off-topic from my previous thread.
I'm recapping a powered speaker that contains a mains transformer & all e-caps are 105c-rated general-purpose radial cans. The one I can't match is 12000uF 50V 10mm-spaced snap-in, 30mm diameter, ±20% tolerance. And here are my options:
A) iffy quality: I've already bought some 63V JCCON off Amazon. I don't have the knowledge or gear to truly evaluate them. And even if they're ok now, how long will they work?
B) under-spec temp: digikey has in-stock quality caps that match in all respects, except they're 85c rated
C) under-spec capacitance: they also have quality 10000uF caps that match in all other respects [and they have 15000uF but 35mm diameter. won't fit on the board]
I'm leaning towards C, on the basis that 10000uF is within the 20% tolerance. And the amp worked for 15 years with Chinese TWBOR caps.
Which corner would you cut?

[momaka]

Re: Lesser of 3 evils?

If I really had to make such choice... definitely B), unless the device really is known for running super-hot. Under-specing on the temp when you're dealing with 105C GP original caps is rarely an issue. Also, I'd always take a known reliable name brand like Panasonic, Rubycon, Nichicon, and United Chemicon with a lower temp spec over any no-name / cheapo cap brand. The former were used in CRT TVs (sometimes hot-running ones too) and they've proven themselves to last for many years. With the cheapo caps, you never know. That being said, since these caps will be connected to the output of a line-type transformer, chances are even cheap caps will survive for many years. But again, I would trust a name brand more than a cheap one.

You can also go with option C) but... nah, why loose capacitance when you don't have to.

BTW, if the board has the diameter space to take larger caps, then you can go with larger caps. If the only reason you're not doing that is because the holes on the PCB don't match with what's on the caps', you can always re-drill the board for new / wider holes, provided it's a single-sided (or double-sided at worst) PCB and doing so won't destroy other traces. I had to do that when replacing some through-hole CapXon caps in a JBL plate amp with Nichicon caps that had snap-in leads.

[slerner]

Re: Lesser of 3 evils?

I don't recall the back plate on these speakers getting above lukewarm, even after a full day of mixing. And that plate serves as the heatsink. The only time it got worryingly hot was when it failed [snap, crackle, POP]. I put the plate in my bench vise and plugged it back in. I think the laser thermometer only found heat where the transformer plugs into the woofer's amp and on one of the bloated 12000uF caps. I should've written it down, but I think the highest reading I got was 170F, about 77c.
So, I'm liking your reasoning. They've got 12000uF in Cornell 380LX and Rubycon USG, both 3000hr/85c.
My ideal choice would've been 105c 15000uF, except the two 30mm cans are already just a couple mm apart. Since the 35mm are also 10mm snap-in, I'd have to move all 4 holes. Someone could probably extend the traces. There's room. But, I'd rather recap annually than monkey with the PCB layout. Especially, as I'm looking to recap both speakers' boards plus a pair of new-old-stock replacement boards. They came with dry caps, since they're about the same age as my originals.
I guess my concern about the 85c rating was based on not knowing what I don't know, so I'm trying to keep everything to-spec. Unfortunately, for a perfect match at 105c Cornell requires a minimum purchase of 3000 caps. Anyone want to go in on this with me? I only need 8 caps, so you'd have to buy the other 2992 at like $6 each. Aww, come on.
Nobody? Well, then I'll probably go with the Rubycon 85c. They've got a beefy thick crimp that somehow fills me with confidence.
Many thanks!

[eccerr0r]

Re: Lesser of 3 evils?

This thing has a huge, so probably line frequency transformer - would the caps actually be overspecced at 105°C?

Is there a reason they need to be replaced?

I was wondering, 15mF is huge for switching PSUs... would imply somewhere near 100A or something, and may even be lower ESR using multiple smaller caps...

Since it's a line frequency capacitor it's probably not a good idea to go lower in capacitance unlike switch mode supplies.

[slerner]

Re: Lesser of 3 evils?

It does have a large transformer that plugs directly into mains power. I don't know why it was built with 105c caps. But, since I don't know, that's why I'm trying to stick with the original specs.
The reason for recapping isn't necessarily the amp's initial failure. Though it occurred after the thing spent a few months in storage. Until then, it had been plugged in almost continuously & working fine for 15 years. That's a cap failure scenario.
I first tried to fix it with new-old-stock replacement boards. A real bargain. Probably the last ones on the market. But, these pristine boards were dead on arrival. They have full factory inspection marks & packaging and were assembled about the same time as my originals. So, if they worked in 2007 and sat in a warehouse until now, the electrolyte must have evaporated. I pulled one of the original 12000uF caps (2007 Chinese TWBOR, not quality) and it weighed a full gram less than a new equivalent JCCON. Resistance topped out at 500ohm. I pulled another and it tested similarly weak.
Recapping seems necessary. If that doesn't suffice, I guess I'll have to go to engineering school because I'll be thoroughly stumped.
Anyway, I placed my digikey order today and it's all Japanese or German 105c, except for the 12000uF which simply isn't available. I went with Rubycon USG 12000uF 3000hr/85c. It's not ideal. But, all in all, new boards & quality new caps, it's a deluxe repair.

[jons]

Re: Lesser of 3 evils?

You should measure those old caps first as large capacitances rarely fail, and not in 15 years, I've seen 40 years old soviet 10000uF caps in fine condition, and they usually are even worse than chinese no-brands.

[slerner]

Re: Lesser of 3 evils?

You should measure those old caps first as large capacitances rarely fail, and not in 15 years, I've seen 40 years old soviet 10000uF caps in fine condition, and they usually are even worse than chinese no-brands.

I take your point and if I had the knowledge and equipment to properly test caps I would've tested the large ones.

[jons]

Re: Lesser of 3 evils?

These and similar clones are cheap (and good enough) even for novice.

[slerner]

Re: Lesser of 3 evils?

Thank you. I ought to get that.