Looking for help in identifying Kuang Jin capacitor from PSU

[chunk69]

Hi friends,

I recently acquired an old PC power supply (along with several vintage motherboards from late 90s/early 2000s). The PSU was dead upon first test, but quickly identified the internal fuse had blown. I fixed that, re-tested and i'm pleased to report it seemed to work great after that - no obvious issues or reasons for the fuse to blow.

Afterwards, I inspected the capacitors more closely and noticed one pair required replacement. They weren't in bad shape by any stretch of imagination, but one was slightly bulged, and I wanted to swap them out for good measure.

The caps appear to be 'Kuang Jin', branded with a 'CS' logo, and has '2200uF 10V +105*C' printed on the side. The only other code shown on the cap is '0303M' - I haven't quite figured out what this means yet.

Now, without being able to track down a datasheet, i'm not entirely sure what to replace them with. I found some datasheets here - https://paullinebarger.net/DS/CS%20%...uang%20Jin%5d/ - but i'm struggling to identify which type they are.

Appreciate any tips here, i'm fairly new to the capacitor replacement game and keen to learn how to identify appropriate replacements, as I have many others to do.

K.

[stj]

what diameter?
i would just use panasonic FR if it will fit.

[edunlikely]

Based on the information you provided, the capacitors are likely electrolytic capacitors, as they are commonly used in power supply circuits. When selecting replacements, look for electrolytic capacitors with similar specifications and quality from reputable manufacturers.

[momaka]

I recently acquired an old PC power supply (along with several vintage motherboards from late 90s/early 2000s). The PSU was dead upon first test, but quickly identified the internal fuse had blown. I fixed that, re-tested and i'm pleased to report it seemed to work great after that - no obvious issues or reasons for the fuse to blow.

If the fuse really was blown, I'd really like to know why. Only way I've seen fuses blow in PSUs is if something is wrong or if someone shorted the primary heatsink to the case by accident. Otherwise, no. Only other exception I can think of is PSUs with APFC circuits, in which case sometimes the fuse can blow when the primary cap is bad or going bad (low capacity) but might not necessarily blow the fuse again the next time... at least until it sees some heavier load.

Afterwards, I inspected the capacitors more closely and noticed one pair required replacement. They weren't in bad shape by any stretch of imagination, but one was slightly bulged, and I wanted to swap them out for good measure.

Oh yes, they would be.
Slightly bulged = BAD already (means electrolyte has started to break down and is producing gases inside the case... so the cap is no good anymore.)

The caps appear to be 'Kuang Jin', branded with a 'CS' logo, and has '2200uF 10V +105*C' printed on the side. The only other code shown on the cap is '0303M' - I haven't quite figured out what this means yet.

"M" = tolerance code = 20% (virtually all caps in ATX PSUs are 20% tolerance, so MEH, nothing new here.)
"0303" likely means the cap was produced either in week 03 of year 2003 or March of 2003 (depends on how the manufacturer likes to denote the date code, so no standard there.) Either way, this is not essential info either.

IIRC, "CS-logo" caps don't have the series printed on their sleeves, so no way of really knowing their specs. But from personal experience repairing PSUs with these caps, I can tell you they are usually used on cheap(er) PSUs with oldschool half-bridge design - i.e. nothing crazy in terms of switching frequency or ripple current & ESR/impedance requirements. As such, the likes of Nichicon PW, Panasonic FC, United Chemicon LXZ, and Rubycon YXJ should be fine to use as replacements. This is not a full list of caps that would be compatible, though. You can go even lower in ESR, such as with Nichicon HE, United Chemicon KY/KYB, or Panasonic FK. And depending on how old the PSU is, sometimes you can go with even general purpose (not low impedance / low ESR -rated) caps like Rubycon PX, Nichicon VZ, and United Chemicon KMG if you are in a pinch and have nothing else.

Probably best to post some pictures of the PSU so we can see what it looks like inside and make suggestions based on that. You can also search badcaps.net to see if someone has already posted about such PSU before (either as a stand-alone thread or in the PSU Quality Pictorial thread.) Sometimes, as is often the case with generic PSUs, the brand might not match anything, but the model number (either on the PCB or on the case) might lead you to find a similar or the same PSU. So you can try that too, if you want.

[chunk69]

Hi,

Thanks for your replies, I really appreciate each and every one of them - very informative. Regarding the popped fuse, in my original post I forgot to mention. On one of the 'floppy' style power connectors, two of the conductors have been cut (it looks like something else had been hard wired in and snipped on removal). I can only imagine this was cut when the power supply was connected/on, causing a short.

Power supply model is: Advance MPT-300 'Thermo-Control and Noise Killer' - 300W max.

The caps appear to be 10x20mm. I guess I was just trying to figure out the ESR/imp to make sure I was replacing them with something equivalent. momaka - thanks for shedding some light on this, i'll take a look for some of the brands you recommended.

Finally, grabbed some snaps. I know it's nothing special, probably a cheap PSU, however it's of the correct vintage to have -5 & -12Vdc rails and the 20 + 6 pin power plugs for old motherboards, so worth repairing IMO.

[ChaosLegionnaire]

Oh yes, they would be.
Slightly bulged = BAD already (means electrolyte has started to break down and is producing gases inside the case... so the cap is no good anymore.)

^this^ i would draw a parallel with this and the women folk and say that u can only be pregnant or not. u cannot be slightly pregnant or heavily pregnant. that just doesnt make sense...

so the cap either has a flat top - not pregnant - in good working condition or its bulged - pregnant - failed and not working.

[PeteS in CA]

Based on the 330uF input caps and the heatsinks, the P/S might be good for 200W. It does look to be half-bridge topology, and based on the yellow inductor core (probably Micrometals 26 material), the switch frequency is probably be between 20 KHz and 50 KHz, and the switch devices may be bipolar transistors.

[momaka]

Regarding the popped fuse, in my original post I forgot to mention. On one of the 'floppy' style power connectors, two of the conductors have been cut (it looks like something else had been hard wired in and snipped on removal). I can only imagine this was cut when the power supply was connected/on, causing a short.

That still shouldn't cause the PSU to blow its fuse. Just about any competent ATX PSU will have short-circuit protection, and I'm pretty sure this one does too, as I'm actually familiar with these Macron Power PSUs.

But I suppose if the PSU has been tampered with before by another "technician", then anything is possible at this point.

Power supply model is: Advance MPT-300 'Thermo-Control and Noise Killer' - 300W max.

Ah, you have a Macron Power unit, and the pictures confirm it (I'm rather fond of these PSUs.)

Actually, when you mentioned those CS-logo caps, I immediately though in the back of my head that this could be a Macron PSU, since they used CS-logo caps quite frequently.

The caps appear to be 10x20mm. I guess I was just trying to figure out the ESR/imp to make sure I was replacing them with something equivalent. momaka - thanks for shedding some light on this, i'll take a look for some of the brands you recommended.

No probs! 👍
Since you're in the UK, I guess RS and Farnell would be where you can find genuine Japanese capacitor brands like Nichicon, United Chemicon, Rubycon, and Panasonic (Matsushita). Nichicon PW or HE and the equivalents from the other brands listed should be a good fit for this PSU.

Finally, grabbed some snaps. I know it's nothing special, probably a cheap PSU, however it's of the correct vintage to have -5 & -12Vdc rails and the 20 + 6 pin power plugs for old motherboards, so worth repairing IMO.

Macron Power are actually pretty decent units.
Yours is a slightly cheaper variant, as it has only 330 uF input caps, which as PeteS suggested, might only allow you to go up to 200W max peak (and more like 120-150W continuous at the most.) But for most retro / classic (5V heavy?) rigs, that's usually more than enough. Only a dual s462 AMD rig might be cutting it a little close with that limit. Most other retro PCs should be fine, especially the likes of a Pentium II or Pentium 3, which are pretty light on power consumption. On that note, I can see a TO-247 part on the secondary heatsink from the pictures you provided, which suggests (and is often the case with these Macron PSUs) that the 5V rail has a 30 Amp schottky diode... which means the 5V rail will be good for at least 20 Amps (100 Watts.)

Yes, definitely worth fixing!
Despite being a cheaper Macron variant, your unit still has proper input filtering components - something that we take for granted today, but many cheaper and even not-so-cheap units lacked back in the day.

At this point, my suggestion for you would be to change ALL of the electrolytic capacitors in this PSUs, (sans the 330 uF 200V input caps, unless you already have some spare parts.) As I found out with one Macron PSU I recapped a few years ago, nearly ALL of the CS-logo caps were starting to fail. So to save yourself some time, just do the recap once, but do it completely and you will have a pretty reliable PSU.
As for the Macron I recapped a few years back, here it is:
https://www.badcaps.net/forum/troubl...9806b-p-rev-a3
(and looks like my inline images are broken on that thread after the forum update... so uhm, sorry if the thread reads a little strange.)

Based on the 330uF input caps and the heatsinks, the P/S might be good for 200W. It does look to be half-bridge topology, and based on the yellow inductor core (probably Micrometals 26 material), the switch frequency is probably be between 20 KHz and 50 KHz, and the switch devices may be bipolar transistors.

Yup, it's a classic H-bridge design. BJTs are usually 13007 on the "lower-end" Macrons. PWM IC is TL494 and there's also a LM339 quad-comparator for helping with handling of protections. And yup, Micrometals -26 material. Main core is likely a T106-26... unless the photos show it smaller than it is... in which case it might be a T130-26. But I doubt it, judging by the 330 uF input caps. Macron only started putting the bigger toroid cores on units past 350W from what I recall seeing.

^this^ i would draw a parallel with this and the women folk and say that u can only be pregnant or not. u cannot be slightly pregnant or heavily pregnant. that just doesnt make sense...

so the cap either has a flat top - not pregnant - in good working condition or its bulged - pregnant - failed and not working.

That's an interesting analogy.

In the case of women, though, pregnant usually does not mean failed

[chunk69]

Thanks all,

Hopefully, ~200W will be enough for the rig I intend to build with this. I have a Supermicro P6DBE board with two Slot 1 350Mhz PII's (https://theretroweb.com/motherboards/s/supermicro-p6dbe). I already own a 3dfx Voodoo 1, and a 3dfx Voodoo 3 3000 AGP, so I think picking up a couple of Voodoo 2's and running them in SLI on this dual CPU board would be the ultimate retro build 😎

I'm torn on whether to replace the rest of the caps, on visual inspection, I can't see any evidence of bulging or leakage and I don't have the means to test them. Looking at RS Components, it looks like I have to buy caps in packs of 20 or more, so it could get expensive depending on how many diffferent types there are.

Glad to hear the PSU is not a complete POS, and glad I saved it from being thrown into a skip 😊

[PeteS in CA]

Depending on how much you want to repair/upgrade, replacing those 330uF input caps with 470uF or 560uF parts and the MJE13007 clones with MJE13009 clones might bump up the power capability and/or improve the output noise. The output inductor, output rectifiers, and heatsinks will still make the practical limit 235W or 250W. Half bridge topology and use of bipolar transistors will keep the efficiency in the ~70% range. Is the fan decent or a POS?

[momaka]

Hopefully, ~200W will be enough for the rig I intend to build with this. I have a Supermicro P6DBE board with two Slot 1 350Mhz PII's (https://theretroweb.com/motherboards/s/supermicro-p6dbe). I already own a 3dfx Voodoo 1, and a 3dfx Voodoo 3 3000 AGP, so I think picking up a couple of Voodoo 2's and running them in SLI on this dual CPU board would be the ultimate retro build 😎

The Pentium II CPUs would use around 30-40W max each... so you're looking at 80W there for the pair. And the VooDoo cards... I don't think much of the cards from that era really went past 10-15W. If they have passive coolers, put them at 10W each or maybe 15W just to be on the safe side. Then add 10W for the chipsets and another 10 for inefficiencies and other gremlins inside the machine. With an HDD or two, add another 20W... So the grand total would likely peak at no more than 140 Watts. Maybe 150W if you really want to be on the safe side (though I personally doubt you would see that high of a number, ever on such an old rig.)

So yeah, even a 200W PSU would do the job just fine.

I'm torn on whether to replace the rest of the caps, on visual inspection, I can't see any evidence of bulging or leakage and I don't have the means to test them. Looking at RS Components, it looks like I have to buy caps in packs of 20 or more, so it could get expensive depending on how many diffferent types there are.

The thing is, visual inspection is not really a good method to tell if caps are still good. It's only a good method to tell when caps are certainly failed (i.e. when they are bulging/leaking.) But if their tops look flat, they can still be failed (dry) or on their way out soon.

I find it that once a cap from a known crap brand goes bad inside a device, chances are the rest of the caps from the same brand and series in that device are likely going to fail too, because they are likely from the same production batch.
And with CS-logo caps, they are just bound to fail at some point. Case in point, I have a 250W Deer PSU that I still run in a classic P3 PC. I never recapped the 3.3V rail, because the caps on it (CS-logo brand) looked fine and even measured fine when I checked them a few years later. But over the years (approx. 10 years after my recap of the PSU), they just failed. The worst part is that the ripple on the 3.3V rail took out a memory module. It was only a 128 MB stick, so no huge loss. But annoying to deal with, nevertheless. The Japanese caps that I put in the same PSU are still going strong and likely will for the useful life of the PSU / computer.

Now I don't mean to judge your build or anything like that, but if you're going to (or already have) drop some cash on VooDoo cards, getting a bunch of new caps (even a few extra ones) shouldn't make the price that much higher on the build... though I suppose that also depends on what kind of prices RS is asking for the caps and what is the minimum order really.

On that note, you can also try looking on eBay. Normally this is not an option I recommend because of all of the fake / counterfeit caps that are sold on there (well, same with Amazon and Ali too.) But if you're careful, you can get genuine Japanese caps for cheaper than RS... or at least here in the USA, I sometimes find listings for caps that are cheaper than Digikey. The few that I bought were genuine. One way to know is to look what else the seller has in their store. If it's a China-mall store-front, forget it - the caps will more than likely be counterfeit. But if it's a seller in your country or somewhere in Europe, the caps should be genuine. Actually, you can try contacting member Behemot here. He's based in CZ and usually has a good selection of caps.

Glad to hear the PSU is not a complete POS, and glad I saved it from being thrown into a skip 😊

Nice!
Yeah, those old Macrons are pretty good for retro builds.

Depending on how much you want to repair/upgrade, replacing those 330uF input caps with 470uF or 560uF parts and the MJE13007 clones with MJE13009 clones might bump up the power capability and/or improve the output noise.

I wouldn't worry too much about the input caps. They can be pricey and the upgrade won't be necessary if the power consumption is low, which should be the case with the Pentium II CPUs.
The output caps might be a good candidate for an upgrade, though, depending on what Macron used in this PSU. I would say each rail should get 2x 2200 uF caps minimum (with the exception of the 12V rail, which should be OK with a single 2200 or 3300 uF cap or whatever is in there already that has equivalent capacity to this.) ... and the -5/-12V rails, of course - those are usually happy with just a 470 or 1000 uF cap.

Is the fan decent or a POS?

My guess is it would be a PowerLogic or Superred. Likely the former, though. Macron used these a lot in their PSUs. They are pretty good fans. Sleeve bearing, of course, so they might need a drop of oil if they've seen many hours of use. But apart from that, they are build well and will last a long time with a proper oil job. The Superred I did in one of my MPT-301 units still spins freely and is whisper quiet after 15 years now.

[ChaosLegionnaire]

The Pentium II CPUs would use around 30-40W max each... so you're looking at 80W there for the pair. And the VooDoo cards... I don't think much of the cards from that era really went past 10-15W. If they have passive coolers, put them at 10W each or maybe 15W just to be on the safe side. Then add 10W for the chipsets and another 10 for inefficiencies and other gremlins inside the machine. With an HDD or two, add another 20W... So the grand total would likely peak at no more than 140 Watts. Maybe 150W if you really want to be on the safe side (though I personally doubt you would see that high of a number, ever on such an old rig.)

So yeah, even a 200W PSU would do the job just fine.

yea thats what i like about systems based on the p6 architecture. its typically 3x less power consumed for the whole system compared to netburst. my p3 system with 1 ghz coppermine cpu undervolted by 0.1v to 1.65v, 1 stick of 512mb pc133 ram, 128mb radeon 9600 non-pro video card, 1 hard drive and 1 optical drive idles at 32.5w and loads at 55w according to my wattmeter. a p4 system with a similar setup would consume 150-160w at load. lets just say i wont be eating electricity as food during a famine! as i am not a cyborg! hahaha!

the only thing is i really regret not making a p3 system earlier when it was going for cheap!!

That's an interesting analogy.

In the case of women, though, pregnant usually does not mean failed

that analogy was in reference to caps as the subject, not women tho. so its a binary or boolean condition: bloated or not; failed or not; pregnant or not.

[chunk69]

A little update for you all, since I found a little time this evening to remove the caps on the output side.

Firstly, I can confirm it does indeed have a 'PowerLogic' fan. It was pretty grim, i've given it a good clean and it feels smooth as silk. It was very quiet when I briefly tested the PSU as well, so i'll oil her up as suggested before that goes back in.

Whilst I've got the PCB to hand, I thought it would be a good idea to share the model number for reference.

MPT-98068-P
Rev. A3

Now, on to the caps. I've removed them and looked at the PCB, and have figured out the following are currently present:

+3.3v	2200uF 16v
+12v	2200uF 16v
+5v	2200uF 10v, 1000uF 16v (x2), 4400uF total
-5v	220uF 16v
-12v	220uF 16v

The -5 and -12v rails use 'fuhjyyu' capacitors from what I can determine. No vent in the top, and they are quite small. Not sure if these are worth replacing, but the rest of the others I will replace.

Now, to find some suitable replacements - i'm seeing some with higher capacitance values and/or higher ripple current values, and i'm not really too sure how far I can deviate from original specifications if I can't find an exact match.

momaka - regarding your comment on upgrading these, it looks like the +5v and +12v rails are fine, but perhaps the +3.3v rail not so good with only 2200uF. However, there isn't really much space to add additional caps or up the diameter of the cap.

Thanks for your tips and advice, this has been a great learning experience so far!

[stj]

replace the 'fuhjyyu' - we call them fuckyu because they always fail
just fill it with panasonic FR

[momaka]

Whilst I've got the PCB to hand, I thought it would be a good idea to share the model number for reference.

MPT-9806b-P
Rev. A3

Fixed that for you.

So essentially it is the same exact PCB model and revision as the Casing Power PSU I liked to above. Thus you can use it to see what I did there when I recapped and modded mine.

Now, on to the caps. I've removed them and looked at the PCB, and have figured out the following are currently present:

+3.3v	2200uF 16v
+12v	2200uF 16v
+5v	2200uF 10v, 1000uF 16v (x2), 4400uF total
-5v	220uF 16v
-12v	220uF 16v

The -5 and -12v rails use 'fuhjyyu' capacitors from what I can determine. No vent in the top, and they are quite small. Not sure if these are worth replacing, but the rest of the others I will replace.

Now, to find some suitable replacements - i'm seeing some with higher capacitance values and/or higher ripple current values, and i'm not really too sure how far I can deviate from original specifications if I can't find an exact match.

momaka - regarding your comment on upgrading these, it looks like the +5v and +12v rails are fine, but perhaps the +3.3v rail not so good with only 2200uF. However, there isn't really much space to add additional caps or up the diameter of the cap.

The filter cap spot for the 3.3V rail is 12.5 mm, so that gives you plenty of options to install a bigger capacitor. I put a United Chemicon LXV 6.3V 3900 uF cap in mine.
For the 5V rail, you can do 2x or 3x 2200 uF caps in 10 mm diameter. Yes, that's quite an increase in the capacitance, but the PSU will be fine. Just DON'T put very low ESR caps like Panasonic FR/FM/FS or similar. I used 2x Rubycon YXJ 10V 2200 uF in mine and left one of the CS-logo 1000 uF caps alone (was running low on my cap inventory.) Figured even if that CS cap failed, the 2x Rubycon YXJ's would still do the job and keep the PSU running.
For the 12V rail, you can do 2200 uF or even 3300 uF (@ 16V), whatever you can get easier. I used 1x Nichicon PW 16V 2200 uF simply because I already had that on hand already (pulled from a motherboard that I upgraded the caps on.)
For the 5VSB, I replaced only one of the two 2200 uF caps with another Rubycon YXJ 10V 2200 uF. The other 2200 uF cap (CS-logo 10V 2200 uF) I replaced with the 12.5 mm one from the 12V rail, since that one still read OK on capacitance and ESR and didn't seem like it would be as prone to failure... but time will tell, I suppose.
Finally, for the -12V rail, I replaced the 16V 220 uF cap on there with a Nichicon PM 16V 220 uF. The Nichi PM was slightly thicker (8 mm diameter) compared to the 6.3 mm diameter of the original cap, but it worked. So yes, you can squeeze in a slightly bigger cap in there. If I had larger capacity at 8 mm, I would have put 470 uF. It's usually not a big deal, since the -12 and -5V rails aren't really loaded much, but increasing the capacity to 330 uF or 470 uF would be preferable. 220 uF is really a bit minimalistic.

With all of that said, try going with more entry-level low-ESR caps, like the following brands and series:
Panasonic FC
Rubycon YXJ and YXF
Nichicon PW, PS, PM, and PJ
United Chemicon LXV, LXZ, LXY

The following would also be acceptable:
Panasonic FK
Rubycon ZL
Nichicon HE
United Chemicon KYB and KY

Oh, and there is also a 25V 470 uF cap to replace for the secondary side aux. rail, if I remember correctly. This is the rail that filter power to the PWM controller and 339 comp., so it's better if they get cleaner power.

Another thing: remove the 22-Ohm (or whatever 20-something values is on there) "dummy load" resistor on the 5V rail and replace it with a 100-Ohm 1/2W or 1W. There's no need to dissipate almost 1W of power and heat up the nearby caps when 0.5W will do just fine... especially since this PSU is going to be used on a 5V-heavy system. Again, see thread I posted above for details.

replace the 'fuhjyyu' - we call them fuckyu because they always fail
just fill it with panasonic FR

Not always.
In my experience, Fuhjyyu is rather inconsistent (with typically more on the side that they would fail than they wouldn't... but again, not always the case.) If they have made it after all of these years, then they should be OK. I still have the original pulled Fuhjyyu caps from my other MPT-301 and two from a PowMax PSU, and they are all still in spec. But most Fuhjyyu caps that I've seen in CWT-built PSUs were always failed. So not sure what makes them fail there more often.
Either way, I do agree that the caps on the -12V and -5V rail should be replaced while in there.

What I don't recommend is using Panasonic FR, FM, FS, and equivalently Nichicon HV, HW, and UCC KZE, KZH, and KZM for this particular PSU. I did my first MPT-301 with KZE on the 12V rail and FM on the 3.3V rail and 5VSB rail. The PSU works OK, but whines very loudly when it's not loaded. It's not an unsafe operation, but it is a result of using caps with ESR much lower than the originals... and can be a bit annoying to listen to. Once loaded up, the PSU does get more quiet. IIRC, someone else did a similar recap on another Macron PSU awhile back and came up with the same conclusion. So I think these older Macron PSUs are better suited with not-so-low ESR caps.

yea thats what i like about systems based on the p6 architecture. its typically 3x less power consumed for the whole system compared to netburst. my p3 system with 1 ghz coppermine cpu undervolted by 0.1v to 1.65v, 1 stick of 512mb pc133 ram, 128mb radeon 9600 non-pro video card, 1 hard drive and 1 optical drive idles at 32.5w and loads at 55w according to my wattmeter. a p4 system with a similar setup would consume 150-160w at load. lets just say i wont be eating electricity as food during a famine! as i am not a cyborg! hahaha!

the only thing is i really regret not making a p3 system earlier when it was going for cheap!!

Nice system there!
Looks very similar, spec-wise, to my P3 PC that I have running in the summer house (post here and here)
Yeah, P3's are great efficiency-wise... though I do find the P4's do play nicer with newer games. Of course, if going that route, might as well go Athlon 64. Even the most basic 3000+ will crush a P4 at gaming. At video encoding and decoding, though, the P4 wins easily over both the P3 and AMDs. So it's down to what you want to use it for.

Also, you live in a hot place, so the P3 is probably better for you. For me, right now when it's winter, I'll take an inefficient P4 and leave it running all day long with the door to my room closed, so it heats it up nicely. I've left the house thermostat temperature at a chilly 17.5C... but the P4 in my room keeps it nice and cozy at close to 19C. Coupled with a CRT monitor and incandescent lights, and I can get it up to 20C after a few hours of use, which feels a lot nicer and not as cold as 17-17.5C. Brrr..

[CapLeaker]

replace the 'fuhjyyu' - we call them fuckyu because they always fail
just fill it with panasonic FR

LMAO!

[ChaosLegionnaire]

The -5 and -12v rails use 'fuhjyyu' capacitors from what I can determine. No vent in the top, and they are quite small. Not sure if these are worth replacing, but the rest of the others I will replace.

the -5v rail is required for isa cards, so if u intend to use a demanding isa card like a creative sound blaster isa card in the system, recapping the caps on the -5v rail may be necessary for a proper running sound card.

the -12v rail is used for the parallel port. if u intend to run something off the parallel port like a printer, recapping the -12v rail may also be necessary for a properly functioning printer. dont want the printer printing gibberish and wasting paper!

hope this information proves useful to the OP whether which rails on the psu are worth recapping and which are not.

[chunk69]

Incredible information as always, thanks so much!

I'll see if I can make an order (and collect) some caps from RS at some point this week before they close for Christmas. I've got a few things sat on the desk i'd like to recap now (and some other soldering work to do on a few other bits and bobs) so that would be a nice little project to keep me busy next week.

Just need to find some time to sit down and research all the caps I need and get them ordered..! Will report back once i've got some caps 😀

[stj]

dont overlook really small ones - like 100uf or smaller,
they have the shortest lifespan

[chunk69]

Thanks for the tips stj and ChaosLegionnaire - I may possibly run an ISA sound card in the rig so that's good to know.

momaka thoroughly enjoyed reading your post where you recapped and modified a similar PSU, very informative and I'm learning a lot (which is why i'm here) 😀

I've compiled a list of what i've found so far in terms of replacements - i'm struggling with the 220uF and 470uF caps as there's only two options I can find, although i'm not sure if a voltage of 350v is too high in this application (I do understand it's a max working voltage, but thought I'd ask).

	Voltage	Capacitance	Dimensions	Replacement	Replacement Dimensions	Link	Min. Order
PSU
+3.3v	16v	2200uF	12.5mm x 20mm	Panasonic FC 3900uF 6.3v	12.5mm x 25mm	https://uk.rs-online.com/web/p/alumi...citors/3150271	5
+12v	16v	2200uF	12.5mm x 20mm	Panasonic FC 3300uF 16v	12.5mm x 35mm	https://uk.rs-online.com/web/p/alumi...citors/3150502	5
+5v	10v	2200uF	10mm x 20mm	Rubycon YXJ 2200uF 16v	10mm x 20mm	https://uk.rs-online.com/web/p/alumi...citors/7497142	5
+5v	16v	1000uF	10mm x 17mm	Rubycon YXJ 2200uF 16v	10mm x 20mm	https://uk.rs-online.com/web/p/alumi...citors/7497142	5
+5v	16v	1000uF	10mm x 17mm	Rubycon YXJ 2200uF 16v	10mm x 20mm	https://uk.rs-online.com/web/p/alumi...citors/7497142	5
+5VSB	10v	2200uF	10mm x 20mm	Rubycon YXJ 2200uF 16v	10mm x 20mm	https://uk.rs-online.com/web/p/alumi...citors/7497142	5
+5VSB	10v	2200uF	10mm x 20mm	Rubycon YXJ 2200uF 16v	10mm x 20mm	https://uk.rs-online.com/web/p/alumi...citors/7497142	5
-12v	16v	220uF	6.3mm x 11mm	Nichicon VR 470uF 350v	8mm x 11.5mm	https://uk.rs-online.com/web/p/alumi...citors/7153205	10
-5v	16v	220uF	6.3mm x 11mm	Nichicon VR 470uF 350v	8mm x 11.5mm	https://uk.rs-online.com/web/p/alumi...citors/7153205	10
Aux	25v	470uF	8mm x ?	Nichicon VR 470uF 350v	8mm x 11.5mm	https://uk.rs-online.com/web/p/alumi...citors/7153205	10

The minimum order qty makes some of these replacements a little pricey, but I'd rather make sure to get genuine caps. I would appreciate your thoughts on suitability or whether there are any better options than what I've chosen above, particularly for the small caps as RS don't seem to stock much for those. I haven't really looked too hard at the likes of Farnell/DigiKey etc yet as I found their search a little more painful to use whilst i'm still getting to grips of what i'm searching for 🤠