Jep. Bad. They are like rockets that are ready to launch.

If the board was sold as working, and it currently works, I really don't think you qualify for a refund.

Cyberdyne wrote on 2025-04-28, 12:32:

Jep. Bad. They are like rockets that are ready to launch.

Ouch, that's bad news. Do you have any advice in which capacitor replacements to get?
I read from https://www.badcaps.net/forum/general-topics/ … s-are-there-any that KZG 1000 uF 16V can be replaced by Nichicon LG 470uF 16V polymer caps?

Although I replaced caps before on a Pentium 3 machine given to me for free, that was before I learnt about ESR and ripple current, so I think it's a miracle that the machine worked after I was done with it.

I looked up the spec sheet of KZG 1000 uF 16 V and found https://www.digikey.co.uk/en/htmldatash ... zg-series , but there is no ESR figure there.

jmarsh wrote on 2025-04-28, 12:38:

If the board was sold as working, and it currently works, I really don't think you qualify for a refund.

That's one of my worries, although I did make sure to ask the seller to confirm that the caps are not bulging before buying, which he confirmed. Even now with these new pictures, he still doesn't see a problem with them.

These caps are bad, but being realistic you can't expect a board from mid cap plague era not to have cap problems unless specifically recapped. So if the seller literally claimed the caps were good, he's wrong and that could be ground for a return. However that gives you whatever you paid for this to spend on another shot in the dark. Instead I'd suggest keeping it: this board works and the issues you can see are entirely to be expected and are eminently fixable. The only thing better is an actually re-capped board, but you're sure to pay a lot more for that.

dionb wrote on 2025-04-28, 16:35:

These caps are bad, but being realistic you can't expect a board from mid cap plague era not to have cap problems unless specifically recapped. So if the seller literally claimed the caps were good, he's wrong and that could be ground for a return. However that gives you whatever you paid for this to spend on another shot in the dark. Instead I'd suggest keeping it: this board works and the issues you can see are entirely to be expected and are eminently fixable. The only thing better is an actually re-capped board, but you're sure to pay a lot more for that.

It's not as straightforward as that.

I'll address KZG-class series capacitors (that means to say: ultra low ESR)

KZG caps should all be considered unreliable. UCC never fixed that series but kept selling them even after 2010.

I would trust early Rubycon MBZ, but I do not trust very late produced parts, same for Suncon (Sanyo) WG.

Panasonic FJ are trustworthy and reliable.

Nichicon HM was fixed sometime in 2005, so parts with a 2006 datecode and later are acceptable.

For the 8mm KZG parts common on Asus boards from this era - for the 6.3V parts, replace with 2.5V or higher polymer capacitors, 680uF or higher is fine (to replace 1500uF). For the 16V parts, use 16V polymer caps, 270uF or more is fine (to replace 1000uF).

dionb wrote on 2025-04-28, 16:35:

These caps are bad, but being realistic you can't expect a board from mid cap plague era not to have cap problems unless specifically recapped. So if the seller literally claimed the caps were good, he's wrong and that could be ground for a return. However that gives you whatever you paid for this to spend on another shot in the dark. Instead I'd suggest keeping it: this board works and the issues you can see are entirely to be expected and are eminently fixable. The only thing better is an actually re-capped board, but you're sure to pay a lot more for that.

That makes a lot of sense actually. I guess the only alternative is not to have any Pentium 4 era boards at all, or hunt for a specific board that is known to have good caps? I don't know any board like that though

mockingbird wrote on 2025-04-28, 17:47:

It's not as straightforward as that. […]
Show full quote

dionb wrote on 2025-04-28, 16:35:

These caps are bad, but being realistic you can't expect a board from mid cap plague era not to have cap problems unless specifically recapped. So if the seller literally claimed the caps were good, he's wrong and that could be ground for a return. However that gives you whatever you paid for this to spend on another shot in the dark. Instead I'd suggest keeping it: this board works and the issues you can see are entirely to be expected and are eminently fixable. The only thing better is an actually re-capped board, but you're sure to pay a lot more for that.

It's not as straightforward as that.

I'll address KZG-class series capacitors (that means to say: ultra low ESR)

KZG caps should all be considered unreliable. UCC never fixed that series but kept selling them even after 2010.

I would trust early Rubycon MBZ, but I do not trust very late produced parts, same for Suncon (Sanyo) WG.

Panasonic FJ are trustworthy and reliable.

Nichicon HM was fixed sometime in 2005, so parts with a 2006 datecode and later are acceptable.

For the 8mm KZG parts common on Asus boards from this era - for the 6.3V parts, replace with 2.5V or higher polymer capacitors, 680uF or higher is fine (to replace 1500uF). For the 16V parts, use 16V polymer caps, 270uF or more is fine (to replace 1000uF).

Wow, thanks for such an informative comment!

Could you explain why a 6.3V KZG electrolytic cap can be replaced with a 2.5V polymer? I thought at least the voltage has to match?

Also, if by any chance you know a replacement for
Ltec 1500uF 6.3V?

And this board has 10 of the smaller 680uF 4V surrounding the socket that I don't know which brand they belong to. They have an (M) symbol.

The "M" symbol means "Panasonic" ( M like Matsushita).

majestyk wrote on 2025-04-29, 04:26:

The "M" symbol means "Panasonic" ( M like Matsushita).

Fantastic! Thank you!

So I looked into polymer replacements for my caps. What I know so far is that:
- Voltage should probably match. Although in some boards, the original caps are overspecced for the voltage. However I have no way of knowing this, so better match voltage.
- Capacitance doesn't have to match, can be half the value, with no explanation why. I still have no clue about this.
- ESR should be equal or lower.

What I found is that modern polymers have a much lower ESR compared to these old electrolytic caps.

For example, the KZG 1000uF 16V caps have an ESR of 0.021 ohms,
https://www.digikey.co.uk/en/htmldatasheets/p … /0/1/kzg-series
while the Chemi-con 1200uF 16V has 0.008
https://uk.rs-online.com/web/p/polymer-capaci … rs/2514823?gb=s
(The 1200uF ones are strangely the cheapest 16V polymer caps on rs-components, cheaper than 1000 and 470 ones)

That means it's a good replacement right?
Can I simply replace all the caps with polymer ones, as long as the voltage matches and the capacitance is more than half the original's, and ESR can be wayyyy lower than necessary?
I have seen someone saying that too low ESR might not be good, but there was no explanation why.

As you can see, I'm struggling to find a good source of knowledge on this topic. The information seems to be scattered with little explanation, as they are mostly anecdotes.

Is there a definite & comprehensive guide for replacing electrolytic caps with polymers? I looked at capswiki, but it's just a collection of hardware with known capacitor issues, and some replacements that might not be polymer, and my board is not on there.

randomdavid wrote on 2025-04-28, 18:11:

Could you explain why a 6.3V KZG electrolytic cap can be replaced with a 2.5V polymer? I thought at least the voltage has to match?

Motherboard VRMs have VRM high caps and VRM low caps. VRM high caps are run at 12V (from the 4-pin 12V auxiliary connector nearby), so the manufacturer uses 16v rated parts. VRM Low caps are running at the CPU vcore. 775 CPUs ran at significantly less than 2.5V, so you'll never need a cap there with a higher voltage. In your case, you can leave your VRM-low caps alone. They are Panasonic (FJ or FL series).

Also, if by any chance you know a replacement for
Ltec 1500uF 6.3V?

You will have to post the series and diameter of the cap, it's written on it.

And this board has 10 of the smaller 680uF 4V surrounding the socket that I don't know which brand they belong to. They have an (M) symbol.

Yes, these are the VRM low caps. You can leave them be.

- Voltage should probably match. Although in some boards, the original caps are overspecced for the voltage. However I have no way of knowing this, so better match voltage.

No, see above.

- Capacitance doesn't have to match, can be half the value, with no explanation why. I still have no clue about this.

The reason is that the VRM design of these boards are geared towards bulk ripple suppression capability of the capacitor bank, and not the bulk capacitance total. Polymer caps have much better ripple ratings than electrolytic capacitors, so you are losing capacitance but gaining ripple suppression.

Can I simply replace all the caps with polymer ones, as long as the voltage matches and the capacitance is more than half the original's, and ESR can be wayyyy lower than necessary?

For the VRM you can reduce the capacitance... You should also do this for other VRMs, like the chipset or memory VRM. But for the other capacitors on the board, if you're going to use polymers, then you should match the original capacitance. An example of this are the 6.3V 820uF (or 1000uF) capacitors scattered throughout. Particularly, as an example, the USB filtering cap runs at 5V and uses a 6.3V part and should be replaced with a capacitor of an equivalent value and should not be halved, though you could get away with it. But you don't need polymer for that and would gain nothing there.

What electrolytics lack for in ripple suppression capability, they make up for in endurance rating and price. Why use a 2000-hour polymer cap which runs for $1-$2/piece to filter a USB port when you can get a 5000-10,000 hour electrolytic cap for 30 cents.

randomdavid wrote on 2025-04-28, 13:56:

jmarsh wrote on 2025-04-28, 12:38:

If the board was sold as working, and it currently works, I really don't think you qualify for a refund.

That's one of my worries, although I did make sure to ask the seller to confirm that the caps are not bulging before buying, which he confirmed. Even now with these new pictures, he still doesn't see a problem with them.

Well, be glad at least the seller somewhat knows something about bad capacitors. 😀 I've seen far more sellers with boards sold as "100% tested and working" when there are visibly bulged capacitors... and even when I went to point the obvious issue to them (before even attempting to buy, if say, the price was too high), they still completely dismissed my claims.

In the case of your KZG caps seen in the above pictures - it looks like they are just starting to build up pressure and probably still are mostly within spec (at least that's my experience with KZG and KZJ caps starting to bulge.) Now, it is indeed strange that they are going from the bottom (rubber bung) rather than the top. But this is not so uncommon for KZG in 8 mm diameter... and even more for TMV/TMZ (which don't have vents). Whereas KZG in 10 mm dia. almost always bulges from the top.

Anyways... in short, I think your caps are just starting to fail (early stage of electrolyte breaking down). But indeed as the others noted here, the seller probably won't take that as a reason for a return, even if you showed him/her this thread. And to be fair, I think @dionb kinda hit the nail on the head with the fact that at least you have a working board here, so probably better off recapping this one than looking for another board from the same era that likely will also need a recap too (not that many boards from the P4/Athlon XP/64 era used good caps).

mockingbird wrote on 2025-04-28, 17:47:

I'll address KZG-class series capacitors (that means to say: ultra low ESR)

KZG caps should all be considered unreliable. UCC never fixed that series but kept selling them even after 2010.

Indeed.

However, from my observation seeing these in various computer equipment, I'd say KZG made past 2010 are certainly a lot better... though it really seems to depend on the specific voltage+capacitance spec - e.g.: 6.3V 820 uF and 1000 uF should *never* be trusted. Meanwhile, 16V 1500 uF KZG, both old and new, seem to do OK most of the time with very few failures. 16V 1000 uF (the one seen in the above pictures) are a somewhat mixed bag - I've seen few of these fail, but certainly more often than the 16V 1500 uF KZG. Same goes for the 10V 1000 uF parts. With that said, I tend to leave 16V 1500 uF KZG on my own boards and boards I give to family (that I can recap if trouble ever appears... though so far it's been OK). But 6.3V 820 uF and 1000 uF parts I usually replace - if not all of them, at least all of the ones I determine to be important on a board (e.g. any buck-type VRMs, linear regulator filters with only single cap filters, and etc.)

mockingbird wrote on 2025-04-28, 17:47:

I would trust early Rubycon MBZ, but I do not trust very late produced parts, same for Suncon (Sanyo) WG.

Panasonic FJ are trustworthy and reliable.

Nichicon HM was fixed sometime in 2005, so parts with a 2006 datecode and later are acceptable.

Rubycon MBZ indeed tend to do pretty good in my observation.
But with the MCZ series, it depends. These don't like heat and I've seen more than a few wrecked / spewing their guts in small hot SFF/USFF boxes (HP/Compaq DC5000 and D530 SFF/USFF boxes with P4 CPUs.) Otherwise, MCZ is mostly OK with me.

Sanyo/Suncon WG, I check periodically, as random failures of these are not that uncommon at this age.
Sanyo/Suncon WF - replace on sight!

Nichicon HM, HN, and HZ - as you noted, good only with 2006 (H06--) and later date codes. Any ones from 2005 and earlier can be problematic... though worth noting that I have not seen any HM or HN caps with a 16V rating and 2005 date codes go bad yet. So it seems the higher voltage -rated HM and HN tend to be OK with 2005 date codes. No data (of my own) about 2004 and earlier... but I'd say these are likely to be affected and not very reliable.

For Panasonic, I treat the FL, FJ, and FJS series with a little more caution now. Actually, not sure which one of these in particular it was, but I know with one of these was a common failure on the ASUS/Pegatron M2N68- and M2N78 OEM board for Dell and HP. Perhaps it's possible ASUS/Pegatron might have screwed the CPU VRM design (or just used too few caps, as I often see empty spots on these boards). But what I can confirm is that I have seen more than a dozen of these boards on eBay with failed caps on the CPU VRM and also on my own Pegatron M2N78-LA I had one fail. Again, not sure if it was FL or FJ or FJS (I think FL). So I treat these with a bit of caution now.
All in all, though, Panasonic was pretty solid even with their ultra-low ESR series, so no need to change these usually.

And finally United Chemicon.
Besides the KZG series, the KZJ series is very closely-related to the KZG series, and thus has the same problems.
Other ultra-low ESR series include TMV, TMZ, and one or two more TM* series that I forget now. These are usually smaller 4V 680 uF caps on motherboards and on the CPU VRM. They tend to bulge from the bottom and do fail quite a bit too.

randomdavid wrote on 2025-04-28, 17:59:

That makes a lot of sense actually. I guess the only alternative is not to have any Pentium 4 era boards at all, or hunt for a specific board that is known to have good caps? I don't know any board like that though

Well, the bad cap era goes quite far back - as early as socket 7 boards, if not earlier. The only reason why caps started failing a lot more during the P4/Athlon XP era was because these era of CPUs have much higher power draw compared to the previous generations of CPUs, and thus required high-power "beefy" VRMs that themselves created a need for ultra-low ESR caps... which became developed specifically for that era of VRMs. Solid polymers were also known at the time, but just too expensive. So the cap industry started making ultra-low ESR caps as a cheaper alternative... and as with any new product, there were "quirks" and "hurdles" along the way, as noted with some of the specific series above.

That said, most non-Japanese electrolytic caps are bound to become a problem on motherboards of any era (it just happens to be quicker on boards from the P4/AXP era) and even quality Japanese brands can sometimes exhibit failures now at this age (20+ years easily for some equipment.) So the reality is, a recap will sooner or later be needed on most retro gear. As such, might as well just get the equipment and learn how to recap... well, most people should be able to learn.

randomdavid wrote on 2025-04-29, 12:26:

So I looked into polymer replacements for my caps. What I know so far is that: - Voltage should probably match. Although in some […]
Show full quote

So I looked into polymer replacements for my caps. What I know so far is that:
- Voltage should probably match. Although in some boards, the original caps are overspecced for the voltage. However I have no way of knowing this, so better match voltage.
- Capacitance doesn't have to match, can be half the value, with no explanation why. I still have no clue about this.
- ESR should be equal or lower.

Yup, that more or less should work for any board recap.

As far as cap voltage goes, one place where you can certainly reduce the voltage without worrying too much is the CPU VRM low side - i.e. all of the small 4V/6.3V -rated caps that form a group right around the CPU. As mockingbird noted, these won't see even 2V (unless you're dealing with an old Pentium II/3 -era board - in that case, the voltage might be higher) so it's OK to use 2.5V or 4V caps. For the rest of the board, stick with 6.3V with any caps that were rated for 6.3V or 10V and 16V for the ones that were originally rated for 16V. There are some exceptions to this as well - e.g. Gigabyte GA-6oxt, which uses 25V -rated caps for many of the "small" caps scattered around the board... but in reality, these caps filter 5V and lower -rated voltage rails, as I've measured.

Speaking of which, if you have a multimeter, you can check the voltage on each cap and make a list based on that. This is what I do, except... I don't measure the voltage on each cap, but rather try to see which rail each cap is connected to. The ones that are connected to the PSU 3.3V and 5V rails will show near-zero resistance from their positive leads to the 3.3V and 5V rail pins on the ATX connector. And any subsequent caps connected to a MOSFET that is powered from the 3.3V rail or 5V rail will likely see that rail's voltage or less.
But anyways, only sharing this info for the interested reader here or anyone who might want to get more "in-depth" with recapping motherboards.
If you want to keep it simple, your own findings/advice that you posted above is a pretty safe way to proceed about motherboard recapping.

randomdavid wrote on 2025-04-29, 12:26:

Is there a definite & comprehensive guide for replacing electrolytic caps with polymers? I looked at capswiki, but it's just a collection of hardware with known capacitor issues, and some replacements that might not be polymer, and my board is not on there.

Not really, because what might work with many motherboards could be something that doesn't with just a few specific ones.
So recapping with polymers is mostly on a "per-board" basis... and really, there aren't any specific rules or guides, because you can often deviate quite a bit from the original design spec.

mockingbird wrote on 2025-04-29, 13:27:

randomdavid wrote on 2025-04-28, 18:11:

And this board has 10 of the smaller 680uF 4V surrounding the socket that I don't know which brand they belong to. They have an (M) symbol.

Yes, these are the VRM low caps. You can leave them be.

I'd actually suggest to the O/P to replace these as well... well, only if he/she doesn't find the rest of the board too hard to recap first, as the VRM low-side caps are the most difficult to replace, especially for beginners.
If O/P gets a good iron and can replace the 16V 1000 uF KZG's with easy, then I'd say go for the CPU VRM low side caps too. While these Panasonic caps aren't anywhere near as problematic as other ultra-low ESR series in that small size/can, I do recall seeing some sleeveless FL go bad on the CPU VRM low side like this. Pentium 4's are indeed good source of heat, so these Panasonic caps may not have that much life left in them.
That said, O/P should change them only if he/she is confident it's within his/her ability to do so. If not, then they can be left alone for now and replaced only if any issues arise.

*EDIT*

randomdavid wrote on 2025-04-28, 18:11:

Also, if by any chance you know a replacement for
Ltec 1500uF 6.3V?

If it's LZG series, Rubycon ZLH or ZLQ series should be fine.
In fact, I use the above two for most of the non-CPU-VRM caps on many boards I recap.
The most common sizes I use are ZLH @ 6.3V and 820 uF along with ZLQ @ 6.3V and 1200 uF. I use these to replace mostly 820-1000 uF 6.3V "small caps".

Panasonic FR, FM, and FS should also be fine if those are easier to find.
If you want to go with Nichicon: HV and HW series will do. HE also OK for smaller caps near the PCI/AGP/PCI-E slots
And if you want to keep the board fairly original-looking and go with UCC (United Chemicon), KZE, KZH, and KZM series will do for most non-CPU-VRM caps.

momaka wrote on 2025-04-29, 15:22:

Indeed.

However, from my observation seeing these in various computer equipment, I'd say KZG made past 2010 are certainly a lot better... <snip>

Agreed. The really late ones you tend to find in post-2010 Dell or HP machines don't seem to bulge as much as the earlier ones... Yes, I admit, I personally do ignore them with the intention of replacing them if needed, but so far this hasn't been necessary.

For Panasonic, I treat the FL, FJ, and FJS series with a little more caution now. Actually, not sure which one of these in particular it was, but I know with one of these was a common failure on the ASUS/Pegatron M2N68- and M2N78 OEM board for Dell and HP. Perhaps it's possible ASUS/Pegatron might have screwed the CPU VRM design (or just used too few caps, as I often see empty spots on these boards). But what I can confirm is that I have seen more than a dozen of these boards on eBay with failed caps on the CPU VRM and also on my own Pegatron M2N78-LA I had one fail. Again, not sure if it was FL or FJ or FJS (I think FL). So I treat these with a bit of caution now.
All in all, though, Panasonic was pretty solid even with their ultra-low ESR series, so no need to change these usually.

Interesting... This is the first time I am hearing of this.

I'd actually suggest to the O/P to replace these as well... well, only if he/she doesn't find the rest of the board too hard to recap first, as the VRM low-side caps are the most difficult to replace, especially for beginners.

Yes -- exactly... Let him first deal with the VRM high caps first before he starts ripping out the VRM-low through-hole vias in frustration 🤣.

Good to see you here.

Also, if by any chance you know a replacement for
Ltec 1500uF 6.3V?

You will have to post the series and diameter of the cap, it's written on it.

I think the series is Ltec LMB series 1500uF 6.3V.
Turns out:
- There's a few more other Ltecs on the board too, from LMB and LZP series.
- A few OST branded RLX series
- One Chemicon KMG
- A few that has "GSC" and "T44A" markings that I don't know from which brand
- One with "4652U" marking with a logo that has a capital F letter with horizontal lines on top and bottom of the letter.

I have compiled a spreadsheet with all the caps on the board to aid myself and others with the same board in the future:
https://docs.google.com/spreadsheets/d/1wCS0B … iew?usp=sharing
There's brand, series, locations on the board, values, a picture, and possible replacement. I believe all the caps are listed, but the information is nowhere near complete.

By the way, thank you mockingbird and Momaka, you guys are awesome, so much knowledge is being shared here!

randomdavid wrote on 2025-04-29, 18:39:

I think the series is Ltec LMB series 1500uF 6.3V.
- There's a few more other Ltecs on the board too, from LMB and LZP series.

For LMB, you will need to substitute with a polymer type. For LZP, use one of the series momaka suggested. Panasonic FR or FS should be equivalent to it, ESR-wise, off the top of my head. You can also use polymer if you want, but don't halve the capacitance for it.

- A few OST branded RLX series

For OST RLX, use polymer.

- One Chemicon KMG

KMG is a general purpose capacitor. You can use pretty much anything here, but you can also leave it alone if you want. If you do replace it, I highly suggest you do not search for a "4V" part, as this is very uncommon for electrolytics and I don't know why Asus custom ordered it. Perhaps they needed something that was reduced height that would not interfere with peripheral cards. Panasonic FS (reduced height version of Panasonic FR) would work well here, look for 6.3V or higher.

- A few that has "GSC" and "T44A" markings that I don't know from which brand
- One with "4652U" marking with a logo that has a capital F letter with horizontal lines on top and bottom of the letter.

GSC should be replaced, but people have stated elsewhere that the small caps may be left alone. I always replace them. Panasonic FR or FS comes to mind again as an adequate replacement for these.

The yellow cap is a Fujitsu part (RE Series) and is "hybrid polymer". Replace with a polymer cap with the following spec or better if you are so inclined to do (that means to say - you can also choose to leave it, it's probably fine): diameter: 10mmx12.5mm, ESR:13 mOhm Ripple rating:4840 mA.

By the way, Tech Tangents has a capacitor database wiki just for this purpose. Feel free to post the list there if you want.

mockingbird wrote on 2025-04-29, 16:27:

Agreed. The really late ones you tend to find in post-2010 Dell or HP machines don't seem to bulge as much as the earlier ones... Yes, I admit, I personally do ignore them with the intention of replacing them if needed, but so far this hasn't been necessary.

Yes, exactly how I treat them too.
I take pictures of the equipment they are in so I can keep track of what might come back later on.
I suspect I will get those systems back as retirees before their (newer-made) KZG caps go bad. 😀

mockingbird wrote on 2025-04-29, 16:27:

Interesting... This is the first time I am hearing of this.

It's something I tracked down a few years back when I was more active on Ebay. Unfortunately, I didn't take screenshots of all of the boards I ran into with that issue, just one or two... and my own one. And also was busy with other stuff (moving) at the time, so wasn't posting much online.
It doesn't seem to be too widespread of an issue... but IIRC, I saw 5 or 6 of these M2N68/78-LA boards on Ebay with this issue - all of them OEM boards for Dell and HP systems. So it's possible this could be a design error.

mockingbird wrote on 2025-04-29, 16:27:

Good to see you here.

Thanks you, you as well! 😀
Seems a good deal of former BCN members have "flocked" to this place. Not sure if it's because of interest in retro PCs (which doesn't seem to get that much attention there anymore) or the forum update. For me it's a little bit of both, I suppose. Though I only discovered VOGONS a little over a year ago. Still active on BCN too, just not as much. The new forum update is PITA on many levels.

mockingbird wrote on 2025-04-29, 21:24:

For OST RLX, use polymer.

Hehe, RLX is supposed to be one of OST's lowest ESR series OTOH, probably 2nd or 3rd after RLA.
But in reality, the larger RLX caps aren't that great for ESR.
Have several Antec Earthwatts PSUs with these as the original caps. Had no problems replacing them with Panasonic FM and Rubycon ZLH. I think even UCC KZE came relatively close and used a few of these as well.

mockingbird wrote on 2025-04-29, 21:24:

The yellow cap is a Fujitsu part (RE Series) and is "hybrid polymer". Replace with a polymer cap with the following spec or better if you are so inclined to do (that means to say - you can also choose to leave it, it's probably fine): diameter: 10mmx12.5mm, ESR:13 mOhm Ripple rating:4840 mA.

Funny how so many ASUS P5G-- series of boards have just one or two of these, and it's always by the RAM slots / Northbridge on a buck VRM circuit - ASUS knew a crappy cap brand won't last in that one spot, so they used something good.

I usually leave these yellow Fujistu caps. Haven't had any failures with them yet.

The Ltecs should go, though.
GSC is about the most terrible brand... though those small "T" series are usually really small caps and thus rarely used on anything important.

randomdavid wrote on 2025-04-29, 18:39:

By the way, thank you mockingbird and Momaka, you guys are awesome, so much knowledge is being shared here!

You're welcome!
It's nice "oldschool" forums like this (vogons, BCN, etc.) where I learned this info, so I'm simply trying to pay it forward.

Tech Tangents has a capacitor database wiki just for this purpose

You mean capswiki? I didn't know that site was created by him.

So, with your bountiful suggestions, I will finish up the replacement list, gradually acquire all the necessary gear (a decent desoldering gun and maybe an ESR meter), and if not too much time has passed since then, I will post an update here!

You guys are the most awesome people I've met in a long time.

Before this thread disappears into oblivion, can I get your help one more time with advice for these?
These caps that I found on the board either have no data sheet, or doesn't appear in their series data sheet, possibly custom ordered, so I have no idea about their ESR and Ripple rating.

What do you advice me to do here?
In my line of thinking, knowing the exact original specs possibly only matters for the VRM low ones, but maybe not for the rest, especially if the replacements are polymer?

• Panasonic FL 680uF 4V
  Position: VRM Low 10
  Note: Can use 2.5V and half capacitance or higher
• OST RLX 820uF 6.3V
  Position: PCI 2, Chipset/PCIe 2, Chipset North 2, RAM 5
• United Chemi-Con KMG 470uF 4V
  Position: PCIe 1
  Note: Custom ordered by Asus. Look for 6.3V or higher
• Ltec LMB 330uF 6.3V
  Position: Between Floppy & RAM 1
• GSC 100uF 16V
  Position: Audio 6, Chipset South 1, Chipset North 4, VRM 1

Also:

The yellow cap is a Fujitsu part (RE Series) and is "hybrid polymer". Replace with a polymer cap with the following spec or better if you are so inclined to do (that means to say - you can also choose to leave it, it's probably fine): diameter: 10mmx12.5mm, ESR:13 mOhm Ripple rating:4840 mA.

I found this data sheet: https://www.datasheet.directory/pdfviewer?url … 3RE681M-R5C.pdf
Which says the 680 uF 6.3V cap has the ESR of 5 mOhm and a Ripple rating of 7220 mA. I'm not sure if I have made a mistake, because these numbers differ from what you mentioned before.

randomdavid wrote on 2025-05-04, 20:11:

<snip>or doesn't appear in their series data sheet

The small GSC capacitors are general purpose. Use something like Panasonic FR/FS here. I recommend these series for 5mm capacitors because they have a high endurance rating for smaller diameter capacitors whereas equivalent series from other manufacturers tend to drop the endurance rating with the smaller parts.

Panasonic FL 680uF 4V
Position: VRM Low 10
Note: Can use 2.5V and half capacitance or higher

To date, no one has been able to obtain a datasheet for Panasonic FL.

That said, I have many specimens of both the black-sleeved and sliver lacquered 8mmx8mm variants, and they're almost all reading between 40-50mOhm ESR at 1Khz:

I found this data sheet: https://www.datasheet.directory/pdfviewer?url … 3RE681M-R5C.pdf
Which says the 680 uF 6.3V cap has the ESR of 5 mOhm and a Ripple rating of 7220 mA. I'm not sure if I have made a mistake, because these numbers differ from what you mentioned before.

Don't pay attention to that datasheet, it was a preliminary release meant to overstate the product's quality to appeal to board manufacturers. In other words, it's more an advertising brochure than it is a datasheet. Two later datasheets (last one from 2004) correctly give the ESR as 13mOhm and specify the ripple rating at 4840 mA. My own test confirms this: