Pretty much all Socket 3 and early Pentium boards use linear/LDO regulators, general purpose electrolytics work just fine on those and will probably outlive other components in the board (consider that these boards are about 30+ years and they rarely suffer capacitor related problems). If you really want to recap these kinds of boards, I would suggest using general purpose caps from a good brand, low ESR is not going to have any significant advantage, and LDO regulators usually have a minimum ESR requirement in the circuit, so going too low (i.e. polymers) might mess things up.

Late Socket 7 and Slot 1 do use low ESR caps in general, but it's more like "entry-level" low ESR meaning stuff like Nichicon PW, Rubycon YXF or Chemi-con LXZ series. But that kind of stuff is widely available today, and again, using polymers here is kind of pointless when you can get pretty much exact replacements for what the original design asked, you would be paying more for something that is not going to yield any noticeable improvement.

Use of ultra-low ESR electrolytic caps started being a common thing around the Athlon XP era, that is where polymers become really useful, not only because there are no replacements produced anymore but also because those ultra-low ESR capacitors kind of sucked, even quality brands had series with all kinds of problems, such as extreme heat sensitivity and very low lifespan in general. What you need to take into account here is that in most cases you won't be able to match the original capacitance of the VRM circuit when using polymers, usually it's not a big problem since you don't need that much but it's necessary to know how the circuit was designed if you want to be really sure. There was a "rule of thumb" from a user on the badcaps forum that using about half of the original capacitance would be enough, but it's just that, a rule of thumb that was based on a specific design.

Is there a general-use ESR value that may work reasonable well on boards? for ~500, ~1500, and ~2200uF caps, are we talking 20mOhm? 50mOhm? I would think a 470uF at 100mOhm would be a bit high for general motherboard use on something like a late socket 370, or socket A.

While I agree that polymers aren't "needed," they look better imo, and polymers can be smaller (easier use of heatsinks in the 300-1500MHz era of boards - some heatsinks are just as much importance for the "look" of a system as anything else), priced similarly at around $0.50USD each, and polymers don't have much of a "shelf life" to worry about versus electrolytics if I were to buy in bulk today, and use them over the next few decades. If I'm lucky, I'll be around for another 40+ years. I do also have concerns of through hole becoming even more rare in the decades to come, so attempting to switch to polymers today makes sense. Plus I'd never worry about any other potential electrolytic leaking onto a board, causing damage - I've seen this in places, some PC, some other electronics. Some Panasonic and Nichicon low-ESR electrolytic caps I purchased about 10 years ago, still new in bag, are showing elevated ESR values today and I wouldn't want to use them due to now being out-of-spec.

There's little available for good caps on lcsc.com now, so I'm generally looking at the $0.50-1/ea retail storefronts now, and the polymer and traditional electrolytics appear to be priced similarly.

Which board are you working on?

Ultra low ESR is equivalent of large MLCC capacitors plus polymer capacitors in parallel.

Cheers,

smtkr wrote on 2023-04-12, 22:41:

Which board are you working on?

In the end, it will be a lot. Below are my favorites in my possession, although some don't POST, others are unstable or have visibly bad caps but still stress test fine. Many don't currently show leaking caps, but they're all aging. I'm looking more for generalized options for keeping boards preserved/restored into the future, and more permanently than electrolytics.

I just gave a couple examples above of two boards that I'd like to fix first, but I want all of the below to be kept running/stable, and others not listed.

1My top 30 boards:
2Socket 3
3Jamicom km-s4-1 - socket 3 PCI SiS496
4PC Chips M912 v1.7 - socket 3 UM8498
5PC Chips M919 v1.5 - socket 3 PCI UM8881
6Shuttle HOT 419 - socket 3 UM8881 OPTi 82C895
7
8Socket 5/7
9AOpen AX59 Pro - ATX socket 7 MVP3
10ASUS P/I-P55T2P4 v3.2 - AT socket 7 HX
11ASUS P5A-B v1.03 - AT socket 7 ALi
12BIOSTAR MB-8500TVX-A v2.3 - AT socket 7 VX
13EPoX EP-MVP3G5 v1.3 - ATX socket 7 MVP3
14EPox P55-IT r0.2U - AT socket 7 FX 
15Gigabyte GA-586TX3 v1.10 -  AT socket 7 TX
16Soyo SY-5EMA+ v1.1 - ATX socket 7 MVP3
17
18Slot 1
19ASUS P2B-D v1.06 D03 - dual slot 1 BX
20ASUS P2B-F v1.00 - slot 1 BX
21ASUS P3B-F v1.03 - slot 1 coppermine BX
22ASUS P3V4X v1.02 - slot 1 coppermine 133A
23
24Socket 370
25ASUS CUSL2-C v1.02 - socket 370 coppermine 815EP
26ASUS CUV4X-DLS v1.03 - dual socket 370 coppermine 133A
27ASUS TUSL2-C v1.04 - socket 370 tualatin 815EP
28ASUS TUV4X v1.01 - socket 370 tualatin 133T
29Abit BP6 v1.0 - dual socket 370 BX
30DFI CA64-T revA - socket 370 tualatin 133T
31
32Slot A
33Gigabyte GA-7IXE v1.1 - slot A AMD750
34FIC SD11 v1.8 - slot A AMD750
35MSI MS-6191 v1.0 - slot A AMD750
36
37Socket A
38Abit KR7A v1.0 - socket A KT266A
39Abit NF7 v2.0 - socket A nForce2
40MSI KT3 Ultra v1.0 - socket A KT333
41SOYO SY-KT600 DRAGON PLUS v2.0  - socket A KT600
42Shuttle AN35N Ultra v1.1 - socket A nForce2 Ultra

Masejoer wrote on 2023-04-12, 18:46:

Is there a general-use ESR value that may work reasonable well on boards? for ~500, ~1500, and ~2200uF caps, are we talking 20mOhm? 50mOhm? I would think a 470uF at 100mOhm would be a bit high for general motherboard use on something like a late socket 370, or socket A.

I've pretty much been using anything that measures 25mOhm or better on my component tester... dunno how right it is.. but the boards seem to work good after so suits me fine... I am mostly using surplus and reclaimed caps, cheap as possible, because I have no idea about whether some of my boards are worth it until they show signs of life. So if any of them find employment in a "sweet spot" system I may actually buy objectively decent caps for them... or use them until they degrade and then put decent decent stuff on. Though I have lucked out on some surplus buys and have at least middling good parts on hand, and only re-use "top half" brands that are reclaimed from other equipment. Don't know if I'd go tooooo far up the quality/price ladder, not building a defense or safety system or anything, might pay $3 a cap if they're big buggers.

BitWrangler wrote on 2023-04-13, 03:24:

Masejoer wrote on 2023-04-12, 18:46:

Is there a general-use ESR value that may work reasonable well on boards? for ~500, ~1500, and ~2200uF caps, are we talking 20mOhm? 50mOhm? I would think a 470uF at 100mOhm would be a bit high for general motherboard use on something like a late socket 370, or socket A.

I've pretty much been using anything that measures 25mOhm or better on my component tester... dunno how right it is.. but the boards seem to work good after so suits me fine... I am mostly using surplus and reclaimed caps, cheap as possible, because I have no idea about whether some of my boards are worth it until they show signs of life. So if any of them find employment in a "sweet spot" system I may actually buy objectively decent caps for them... or use them until they degrade and then put decent decent stuff on. Though I have lucked out on some surplus buys and have at least middling good parts on hand, and only re-use "top half" brands that are reclaimed from other equipment. Don't know if I'd go tooooo far up the quality/price ladder, not building a defense or safety system or anything, might pay $3 a cap if they're big buggers.

Thanks for the input. Yeah, I'd prefer to recap as fewest times possible. Never know when you may leave heat on a pad for slightly too long on a specific board and lift it. Always repairable, but for the last few years I've gotten more into preservation more than anything else. If something has survived this many decades before I came to own it, I don't want to be the one responsible for its destruction. The same is true for how I clean and store my components - if we ever get a big earthquake over here, I've even done what I can to protect my parts during such a natural disaster. Thankfully there isn't much more that I desire to own, other than a few specific boards, and a second, spare of others such as the MVP3G5, a board I owned for years. I still haven't found a PC Chips M570 (or M571?), and have been looking for years!

I'd love to be able to continue to use my parts into early-retirement years. I just want to do my best for the hardware I have, and not need to do it again, or worse, be unable to find replacement parts later. I think with everything going digital/no physical media into the future, this will be even more important.

Masejoer wrote on 2023-04-13, 00:27:

MSI KT3 Ultra v1.0 - socket A KT333

Is your KT3 working?

Masejoer wrote on 2023-04-13, 04:14:

I still haven't found a PC Chips M570 (or M571?), and have been looking for years!

Not sure why you want one of those, they're lovable in a lame duck, underdog kind of way, but if you are still looking there appears to be several on eBay at the moment if you search "TX Pro motherboard" but prices aren't that great. I sometimes define "looking" as "looking for one at half the inflated eBay price to fall in my lap locally" so yah.

cloppy007 wrote on 2023-10-23, 21:00:

Masejoer wrote on 2023-04-13, 00:27:

MSI KT3 Ultra v1.0 - socket A KT333

Is your KT3 working?

It does work, although the AGP slot had a factory defect in that it was installed crooked. Using an AGP card, in a chassis, puts some stresses on the board and/or videocard, but on a desk where the card isn't forced into an expansion slot, everything runs fine without worries.

It's a board that I'll hold onto until I get a KT3 Ultra2, which I ran back in 2002 with a 1700+ and a Ti4200. They're pretty similar boards.

BitWrangler wrote on 2023-10-23, 21:19:

Masejoer wrote on 2023-04-13, 04:14:

I still haven't found a PC Chips M570 (or M571?), and have been looking for years!

Not sure why you want one of those, they're lovable in a lame duck, underdog kind of way, but if you are still looking there appears to be several on eBay at the moment if you search "TX Pro motherboard" but prices aren't that great. I sometimes define "looking" as "looking for one at half the inflated eBay price to fall in my lap locally" so yah.

Correct - reasonably priced.

I'm mostly interested in one so I can relive my K6-300 system. Now that I'm far wiser than my younger years. This was my first set of components that I "picked out," or rather the store salesman suggested for the price point, but it was my first DIY build that I was involved with. I'm mostly curious about how inaccurate my memories were of the system, and I'd like to experience its issues to see what I had wrong. I went from a DX4 with some unknown videocard, a crystal soundcard, and 12MB of RAM to the PC Chips board, a K6-300, Viper V330 PCI, and 32MB RAM, so that upgrade was a huge eye opener.

I'm still not sure if the board model I'm looking for is accurate, but I still have the original driver CD ( V1.8 ). When I reviewed the content, I assume that one of the model folders listed on the disk were what I had, from which I remember installing a driver from to get UDMA working (an amazing performance improvement when I somehow found out about that). I remember "TX Pro," I believe it was written on the heatsink, and it was not "VX," but I also remember it being an SiS driver that I installed, so I limited my search to SiS chipset models. The board did have onboard audio (which I used, and I think I remember it having a poor noise floor), I don't remember it having an AGP slot, but I knew nothing at the time and the slots would have meant nothing much to me. I would have thought that if I had an AGP board, that I would have been sold the AGP version of the V330, not the PCI. I also do not think it supported a faster CPU - I don't think I could run a K6-2 in the motherboard I had, but who knows...my followup system was a K6-2 400 on an EPoX MVP3G5 and a Riva TNT AGP card (with a PCI Voodoo3 added later for UltraHLE emulation). The EPoX board, and overclocking, is where I started to really get into the hardware side of computers, and where I remember a lot more of the details. My following board was a BP6, so it's obvious where I started to pay much more attention to the hardware side of things.

The PC CHIPS motherboard's IDE driver folder contains the following potentially relevant models:
M559 (VX)
M570 (SiS, AGP, sound)
M571 (SiS, PCI only, sound)
M572 (Intel)
M573 (VIA)
M575 (ALi)

So with the above details and the driver CD I still have, I concluded that it must have been a M570 or M571. I was also trying to find board release dates, since my build was the beginning of "summer break" from school in 1998, so it was likely a late June purchase as a birthday gift from my father.

Yes, I've put a ton of thought into trying to remember/figure out what components I had in that system 😉 A lot happens in 25 years.

I really wish I had gotten into retro computing a few years earlier. I would possibly still have my DX4, K6-300, and K6-2-400 systems. My father did pull my first 3d accelerator cards from the systems though, and I have them today - I'm thankful for that at least.

When searching for new capacitors this is the approach I take:

Pentium III and older- Use Panasonic Electrolytics, not polymers, these are the longest life capacitors (rated). Using polymers on boards older than Pentium iii would probably be fine, but SOME older boards use circuits that rely on the higher ESR of electrolytics, there will be no benefit from going polymer in these cases.

AthlonXP/Pentium4 and anything newer, use polymer Nichicons.

Use a site such as digikey, input the capacitors voltage, lead spacing, diameter and max height clearance you have in the space it is going in. Sort by descending and use the capacitors with the highest capacity, highest ripple currents and they usually have the lowest ESR too.

Following this method every single board I have recapped has been rock solid stable, and shown to be a massive improvement on the oscilloscope - VRM temperature has dropped on some boards. Higher clocks achieved in some cases.