Okay to use 6.3V caps on 3.3V line of PC PSU?

[em132]

Hi,

I'm currently in the process of re-capping a NOS (new old stock) Sparkle ATX-400PN (from the late-2000s), which was chock-full of Teapos and CapXons . Am looking to put a yummy mix of NCCs, Panos and Rubys in their place.

Here's the problem I ran into: The 3.3V rail is filtered by a single 4,700uF/10V Teapo... The problem? This thing is only 10mm wide, and a wider (12mm) cap will not fit! I've searched the "big three" shops (Mouser, Digikey, Newark/Farnell) and all the 4,700uF/10V I found were at least 12mm wide. The closests matches I could find in 10mm were 3,300uF/10V and 4,700uF/6.3V.

So my question is, as the title says: Is it okay to use 6.3V caps on the 3.3V line of a PC PSU? (In fact, I don't recall ever seeing a 6.3V cap on a 3.3V rail; it's always a 10V, just like on the 5V rail.) If not, then would dropping from 4,700uF down to 3,300uF cause issues?

Thanks in advance.

[stj]

Re: Okay to use 6.3V caps on 3.3V line of PC PSU?

3900uf 6.3v FS series will fit

[momaka]

Re: Okay to use 6.3V caps on 3.3V line of PC PSU?

So my question is, as the title says: Is it okay to use 6.3V caps on the 3.3V line of a PC PSU?

Yes.

I do it almost all the time, especially in cases like this. I even do it on the 5V rail, and so far it hasn't been an issue with any of my recapped PSU for many years now (for some of them, they are approaching the 10 year mark.)

So yes, it's usually not a problem to drop from a 10V cap to a 6.3V cap in a PSU. 5V is as high as any of these will see... so that's why it's OK. 12V rail is clearly too much even for a 10V-rated cap, so that's why it always gets at least 16V -rated caps.

[redwire]

Re: Okay to use 6.3V caps on 3.3V line of PC PSU?

Only one 4,700uF 6.3V 10mm Dia part at Digi-Key: Rubycon ZLQ for 6.3ZLQ4700MEFC10X25 rated 2.5A 100kHz ripple current.

As a guideline, higher voltage-rated and larger size, larger value capacitors can take more ripple current. I would not downsize the part to 3,300uF.

I have shamefully stuffed in a 12.5mm dia. part and let it sit up and crooked. Otherwise, look at using/adding a polymer cap.

[em132]

Re: Okay to use 6.3V caps on 3.3V line of PC PSU?

Thanks momaka. I had a hunch a 6.3V would work. I just wanted some reassurance from the experts here because, like I said, I don't recall ever seeing a 6.3V cap used in a PSU before. Makes one wonder how Teapo managed to stuff 4,700uF@10V in a can the same size as a 3,300. Then again, it did test a bit high when I pulled it (around 4,900), which from what I understand, is not a good sign.

Also, why is it that these supposedly "top-tier" OEMs (FSP in this case) insist on using low/bottom-tier crap caps when quality ones really don't cost that much more? To wit, I'm also in the process of re-capping a Seasonic SS-350ES. Damn thing is filled to the brim with Su'scons!... AND a sleeve-bearing fan to boot!

Only one 4,700uF 6.3V 10mm Dia part at Digi-Key: Rubycon ZLQ for 6.3ZLQ4700MEFC10X25 rated 2.5A 100kHz ripple current.

Precisely the part I ordered (from Mouser), in anticipation of this thread's responses.

[momaka]

Re: Okay to use 6.3V caps on 3.3V line of PC PSU?

Thanks momaka. I had a hunch a 6.3V would work. I just wanted some reassurance from the experts here because, like I said, I don't recall ever seeing a 6.3V cap used in a PSU before.

I recapped a ThermalTake TR2-430W (XP550-NP) not too long ago, and both the 3.3V and 5V rail had 1000 uF 2200 uF caps rated for 6.3V only. But on a lot of cheap PSUs back in the early 2000's, I also regularly saw only 10V caps like you mentioned. So I think it's just down to manufacturers.

Generally, when considering two caps from the same brand and series, the cap with the larger can will have higher ripple current rating and lower ESR. So sometimes, manufacturers use a higher voltage -rated cap just to get lower ESR and higher RC specs out a particular manufacturer's brand and series. It's also a good way to save on costs, since they can just order a larger number of, say, 10V 2200 uF caps to do both the 3.3V and 5V rails, rather than get 6.3V caps for the 3.3V rail and 10V caps for the 5V rail (i.e. shop in bulk quantities, and get more savings.) Hence why it's such a common practice.

Makes one wonder how Teapo managed to stuff 4,700uF@10V in a can the same size as a 3,300. Then again, it did test a bit high when I pulled it (around 4,900), which from what I understand, is not a good sign.

Custom order... but yeah, whenever I see that kind of questionably-high-capacity-for-the-can-size, I tend deem the cap unreliable. It's not always the case, but seems to be more often than not.

As far as the capacity reading higher - that's OK, up to a certain point. Modern electrolytic caps are usually rated 20% on the capacity spec (unless specified otherwise). So a 4700 uF could read as low as 3760 uF or as high as 5640 uF and still be considered "in order". Of course, when the capacity starts going higher (~10-15% or more) and especially if the capacitor is a 2nd tier brand like Teapo, OST, and similar, you can bet the cap is starting to develop high internal electrical leakage, which simply means the electrolyte has probably began to break down and the cap would likely get worse over time (and soon!) So caps like that, I replace whenever possible.

4900 uF for a 4700 uF cap is not bad, though. I still wouldn't trust those old Teapo caps, of course (especially at this age.) But it might be a good idea to hang onto it for any temporary or emergency recaps when you run out of good parts.

Also, why is it that these supposedly "top-tier" OEMs (FSP in this case) insist on using low/bottom-tier crap caps when quality ones really don't cost that much more?

Question of the century, probably.

Well, "don't cost that much more" can become "we get significant savings" when you add a few cents per cap and multiply that by hundreds of thousands of units produced. It still makes no sense to me, of course, since the cheaper caps will not last anywhere near as long. But it does make sense to manufacturers. After all, they know their PSUs likely will end up in the great e-waste after about 5 years on average - not because the PSU failed, but because people (and especially businesses) often replace their fleet every 3-5 years. In fact, top-tier OEMs DID actually used to put high-quality Japanese caps in their PSUs. I still have a lot of LiteOn and Astec PSUs from the late 90's / ealry 2000's (mostly the Pentium II/3 era) that were build with Japanese caps. These PSUs are still kicking around and work fine. But the machines they are in (PII/P3/AMD Athlon) are way beyond obsolete at this point. So PSU manufacturers probably figured they really don't need to make this kind of quality PSUs when the rest of the machine will become obsolete way before then (and in the case of the early 2000's, this was even moreso true than now.)

To wit, I'm also in the process of re-capping a Seasonic SS-350ES. Damn thing is filled to the brim with Su'scons!... AND a sleeve-bearing fan to boot!

Su'scons aren't that bad, IME. About the same as Teapo and OST... which aren't that great either, I suppose. But that's still a million miles better than trash brands like Canicon, GL, CS, and JEE... or worse, the "no-name" brands that no one has ever heard of... or Cheng/Chang/Chong +/-X/ing.

As for sleeve bearing fans - I actually prefer them. They are more quiet.
My guess is, you have an Adda fan in there? Either way, many sleeve bearing fans are stuffed with cheap grease from the factory, which tends to gunk up and dry after some years of use... and then the fan seizes.
A proper cleaning of the sleeve bearing and a generous amount of quality machine oil will keep one running for many years. I have sleeve bearing fans that have been in well over 10 years of operation now without issues (mind you, not 24/7 operation, as I don't run my computers when I don't need them... but nevertheless, 10 years is 10 years. )

As a guideline, higher voltage-rated and larger size, larger value capacitors can take more ripple current. I would not downsize the part to 3,300uF.

Yes, downsizing capacity in a PSU is indeed not advisable.
However, I have run into instances where I could not stuff a 12.5 mm cap and really had to go down in capacity. It's not end of the world if you do, though - the PSU will just have higher ripple and noise output on that rail, particularly when loaded near its maximum current rating. If not planning to load near the maximum rating, downsizing a notch or two in capacity can be OK. For 4700 uF, 3900 uF would be fine (it's within the 20% spec lower capacity for 4700 uF.) Even 3300 uF would work... but again, that's where you might have to downgrade the output capability on the PSU on that rail. Moreover, one cannot just keep pushing smaller and smaller capacities and then downgrade the output. At some point when the capacity is too low, the PSU may not work right. This has to do with the way the filter is designed on the output (particularly, the size of the common-mode output toroid.)

I have shamefully stuffed in a 12.5mm dia. part and let it sit up and crooked.

Apevia did that on some of their PSUs to stuff in extra caps where there was no space for them on the PCB (my guess would be someone goofed up on the PCB silkscreen layout. ) If it worked there...

I haven't done this myself, but done something similar: in a PSU where I needed to do a full recap, I left about 5 mm extra length on the leads of all of the caps I was installing. This allowed me to bend them / move them slightly from their original silkscreen positions on the PCB, which is what allowed me to stuff a 12.5 mm cap in a spot that would otherwise not have allowed it at all (tightly surrounded by caps near by.) It was a in a Bestec PSU, IIRC... that is still somewhere in service in one of my PCs today.

Otherwise, look at using/adding a polymer cap.

I'd caution against using polymers in a PSU. If their ESR is way too low, you run the risk of output filter ringing and creating more R+N than going with a cap that has lower capacity and higher ESR (or ESR closer to the original.) Of course, it also does depend a lot on the PSU, particularly on the feedback compensation and output filter design. Some PSUs will be fine with polymers. Others will oscillate wildly. I've ran into both cases. In general, too low of ESR is not always a good thing. When considering replacements, I do try to stay closer to the original ESR whenever possible, especially in old half-bridge units. The more modern STF topologies from OEMs like Delta and HiPro seem to tolerate very low ESR caps OK.

[PeteS in CA]

Re: Okay to use 6.3V caps on 3.3V line of PC PSU?

Companies try to minimize the number of different parts in their production, so the 10V parts used for 5V outputs will be used for 3.3V outputs as well. But for repairs or upgrades, 6.3V caps for the 3.3V output is OK.

[ReeceyBurger123]

Re: Okay to use 6.3V caps on 3.3V line of PC PSU?

Ive used 6.3v on a circuit I made for a led supply. Its 5v been running fine for years not got round to sorting it yet Ill know if the psu voltage gets too high

[em132]

Re: Okay to use 6.3V caps on 3.3V line of PC PSU?

Thanks again for the replies.

A couple more related questions, if I may, this time regarding the Seasonic SS-350ES. I now have all the equivalent spec caps on hand needed for the transplant surgery. I also have a few more that are above spec in terms of capacitance, and I was wondering how safe (or unsafe) it would be use these instead as "upgrades" in a switching supply.

Specifically: Is it OK to replace the primary (hi Volt) cap, a 400V/220uF, with a 330uF? (Note that the SS-350ES is a "350W Active PFC F3" unit.) Also, OK to replace 2x 10V/1,000uF (both are filtering the 3.3V and 5V outputs) with 2x 1,500uF?

(If the mods don't mind, I'd rather keep this going in here instead of hogging the forum with separate threads.)

My guess is, you have an Adda fan in there?

Not for long; my Mouser order came in this morning. A shiny (shiny??) new Sanyo will take its place real soon. Ditto the cheap Yate-Loon in the Sparkle. Love those Sanyo Denki's. A tad pricey, but worth it.

[momaka]

Re: Okay to use 6.3V caps on 3.3V line of PC PSU?

Specifically: Is it OK to replace the primary (hi Volt) cap, a 400V/220uF, with a 330uF? (Note that the SS-350ES is a "350W Active PFC F3" unit.)

Yes, should be fine.

Those big caps are expensive, so PSU manufacturers often use the smallest possible size that would get the job done, particularly in more budget PSUs. In servers or where reliability is more important, one will see PSUs with equivalent power ratings have much higher capacity and better quality input caps.

With that said, the 330 uF primary cap would actually be a nice upgrade in terms of capacity.

I also like to "upgrade" the voltage rating on PSUs with APFC: 420V better, or 450V best. Again, higher voltage rating means bigger cap size... and that means higher cost. Most APFC circuits don't boost above 400V, but may have a spike or two very close to that, particularly at startup. So IME, 400V caps seem to get worn out faster in PSUs with APFC, hence my suggestion to go with 420V or 450V.

Lastly, I also like to add a polypropylene (metal film) cap in parallel with the primary electrolytic cap. Anything from 0.1 uF to 1 uF and rated for 630V would be nice to have, as the PP film cap will help filter out some of the HF ripple that the electrolytic cap then doesn't have to deal with... so again, I believe that translates to longer life for the primary electrolytic cap. Reason I say "I believe" is because I saw workstation and server PSUs do this, but not something I've seen in cheap(er) consumer units.

Also, OK to replace 2x 10V/1,000uF (both are filtering the 3.3V and 5V outputs) with 2x 1,500uF?

Yes, that should be fine too. If anything, you'll get better filtering on those rails.

Not for long; my Mouser order came in this morning. A shiny (shiny??) new Sanyo will take its place real soon. Ditto the cheap Yate-Loon in the Sparkle. Love those Sanyo Denki's. A tad pricey, but worth it.

Yeah, they are nice. I can't remember if I bought a Sanyo Denki or Panaflo off of Digikey some years back. It was for a recap for an old "retro" Sparkle PSU that had a failed BB fan (Nidec?) Anyways, the new fan from DK worked great. At close to $9, it wasn't exactly cheap... but for half of that or thereabouts, you can only get a no-name cheap sleeve bearing LED fan from eBay or Ali. So it doesn't make too much sense, especially when long-term reliability is key.

That said, I do still like YL and Adda fans. Again, with proper lubrication on the sleeve bearing, I find these last a very long time and push a good amount of air for the not much noise they make.