shevalier wrote on 2023-10-20, 15:41:

Khm, its ESR looking like .... "brand new japonise LowESR capacitor"
And what`s the problem?

Maybe... but if it's out of spec it means the materials inside are degrading... So 20mOhm now perhaps, but maybe 500mOhms later if put into use. Polymer caps don't reform (read self-heal) like electrolytics do. Once they're out of spec, they're finished.

Hoping wrote on 2023-10-20, 15:45:

Hopefully the current KZGs are better than those of the first decade.

There's no way to know. Yes, you can find KZGs that test within spec from 2014 (they're no longer made, you won't find any "current" ones), but unlike later Nichicon HM/HN/HZ which are all good, we don't know *which* KZG are good... Maybe one batch is, maybe one isn't. The KZGs in my ASRock I partially re-capped yesterday tested fine. I replaced them anyways. KZG should be replaced on-sight. The rest of the board had Samxon GK and a few Teapo SEK. Teapo SEK are generally ok, Samxon GK are batch dependant. I left them alone because they looked ok and were not located at the CPU VRM.

The electrolytic capacitors that I stopped distrusting years ago are the blue Rubycon ones from the 80s and maybe earlier, Amstrad uses them a lot in their computers and the two CPCs I have are one of the first 464 from 1984 and a 6128 from 1986, and they never failed.

Yes, the turquoise Rubycons and the old general purpose Panasonics seem indestructable.

mockingbird wrote on 2023-10-20, 16:40:

Maybe... but if it's out of spec it means the materials inside are degrading... So 20mOhm now perhaps, but maybe 500mOhms later if put into use. Polymer caps don't reform (read self-heal) like electrolytics do. Once they're out of spec, they're finished.

Matsuhita acquired Siemens=Epcos, Kemet acquired Epcos, Yageo bought Kemet.
Siemens has been making capacitors longer than I've been alive.
They think so

https://www.arrow.com/en/research-and-events/ … lytic-capacitor

mockingbird wrote on 2023-10-20, 14:37:

I don't know about other Taiwanese brands (Apaq, Lelon, Capxon, etc...).

Apple takes this easier.

Hoping wrote on 2023-10-20, 15:45:

Amstrad uses them a lot in their computers and the two CPCs I have are one of the first 464 from 1984 and a 6128 from 1986, and they never failed.

I have also one quite early 6128 and no problems with caps either. I can't remember the brand as I haven't had the case open for years. I also haven't had cap problems with my four C64 longboards, although I swapped the caps to those while doing other repairs. My Atari 130XE has had zero problems, and as far as I know Amiga 500s seem to have cap issues quite rarely (I have one with 6A MB and it works just fine). Then again, especially later MSX2/2+ systems sold for Japanese market have issues with leaky caps and which can even destroy the board. Amiga 600/1200 computers are famous for leaky SMD electrolytic caps, I have recapped both and especially my 600 was a complete mess when I got it.

To me it seems, that cap issues started to become more common in the late 80s, although there clearly is a large variance. And my 90s PCs haven't had any issues with electrolytic caps, up till ultra-low esr VRM cap era. Only exception to this was IBM PS/2 model 35SX floppy drive, again with SMD caps and which required recapping.

CharlieFoxtrot wrote on 2023-10-21, 07:41:

To me it seems, that cap issues started to become more common in the late 80s, although there clearly is a large variance.

Repair of Tektronix 24xx series oscilloscopes begins with the point “wash the board from leaking SMD capacitors.”

shevalier wrote on 2023-10-21, 05:10:

Matsuhita acquired Siemens=Epcos, Kemet acquired Epcos, Yageo bought Kemet. Siemens has been making capacitors longer than I've […]
Show full quote

Matsuhita acquired Siemens=Epcos, Kemet acquired Epcos, Yageo bought Kemet.
Siemens has been making capacitors longer than I've been alive.
They think so

https://www.arrow.com/en/research-and-events/ … lytic-capacitor

Apple takes this easier.

The graphics in the Kemet article that you link are self-explanatory. I liked the article.
I have seen articles and videos that indicate that the use of MLCC capacitors in parallel to polymers or electrolytics, if the capacitance of the MLCC is calculated correctly, does not have to be very high, and the MLCCs have greatly increased the capacitance with new technologies, you can still reduce the ESR further at the typical switching frequencies and improve the ripple suppression, I misunderstood everything, or it is more or less like that.
I think that Apple uses or used capacitors of dubious quality is not new, and I have seen videos that indicate that the Apples of the 1980s already used poor quality capacitors.
Better not to talk too much about Apple and its quality and decisions because otherwise we would get lost in infinity.

shevalier wrote on 2023-10-21, 05:10:

Matsuhita acquired Siemens=Epcos, Kemet acquired Epcos, Yageo bought Kemet.
Siemens has been making capacitors longer than I've been alive.
They think so

Ok, so compensate for the drift with electrolytics when you can. I'll take a compensated circuit designed around electrolytic caps when feasible over one that only works with polymer caps any day. Again, I'm all for polymers in motherboard VRMs, but there is no empirical data on the longevity of the best quality polymers that's older than perhaps 20 years... There is however a lot of evidence that suggests that polymers degrade over the decades where an electrolytic equivalent would not.

Aha! Interesting... I suspected Kemet was Taiwanese junk... Now if you can find out who makes those red Würth Elektronik junk caps all the Youtubers are crazy about I'd be really impressed.

Also, stay away from CDE (Cornell Dubilier) caps. Junk.

mockingbird wrote on 2023-10-22, 00:51:

Now if you can find out who makes those red Würth Elektronik junk caps all the Youtubers are crazy about I'd be really impressed.

🤣, I also noticed that, they really love those red things! Are those so much cheaper/easier to get than Japanese ones? Apparently it's a German company known for other types of components (connectors, inductors) and they have their own plant in China for capacitor manufacturing:

https://www.eevblog.com/forum/projects/_wurth … -caps-good-bad/

Obviously there doesn't seem to be a lot of feedback about their long-term reliability yet, but in any case I'm not sure why anyone buying new caps would get these instead of known quality brands unless they purchase them in such quantities that price difference has a noticeable effect. Or maybe they're sponsoring YouTube videos...

Hoping wrote on 2023-10-21, 14:59:

I have seen articles and videos that indicate that the use of MLCC capacitors in parallel to polymers or electrolytics, if the capacitance of the MLCC is calculated correctly, does not have to be very high, and the MLCCs have greatly increased the capacitance with new technologies, you can still reduce the ESR further at the typical switching frequencies and improve the ripple suppression, I misunderstood everything, or it is more or less like that.

https://ksim3.kemet.com/capacitor-simulation
It depends on the conversion frequency.
On older motherboards with powdered iron cores, it is not high (up to 100kHz). There, MLCC do not really work due to high impedance (not ESR) due to low capacitance.
Modern regulators are capable of frequencies of several megahertz, and polymer capacitors are no longer really needed.

So, from what I think I'm understanding, the typical switching frequency of voltage regulators from the era of the capacitor plague should be around 100khz or 150khz, usually less. I don't have an oscilloscope, so I try to look for a general rule of thumb to take into account, even if it is not exact, but it falls within a margin of tolerance.
From the graphs of the articles shown, I think I understand that polymer capacitors are better than electrolytic ones at those frequencies.
Which would justify the use of polymer capacitors in old hardware, unless it is a linear voltage regulator.

An intresting video that I think is aplicable to al the A7N8X motherboards. https://m.youtube.com/watch?v=Y671jiXgYjc

mockingbird wrote on 2023-10-17, 04:07:

Here you go: […]
Show full quote

Here you go:

https://www.digikey.com/en/products/detail/ke … AAE013/13420067

This is 10mm, has double the ripple capacity (7785mA) and the same ESR (13mOhm), but it's 2200uF. You do not need to match the capacitance. I would even recommend going lower than 2200uF.

I hesitate to recommend Kemet, so feel free to choose a better brand if you want.

question for you

the "around HALF the uf rule using poly" goes for caps around the VRM.

but how about the caps near the Memory Slot, and near the AGP slot. would those caps fall into the same "half the UF rule using Poly" or you would follow the exact "uf" replacement for those?

TheMobRules wrote on 2023-10-17, 05:16:

Polymers are pretty much the only choice for VRM caps on motherboards that used ultra-low ESR electrolytics (starting around the […]
Show full quote

SETBLASTER wrote on 2023-10-17, 04:47:

So reading the posts, now it is fine to use Polymer capacitors near the VRMs? , if i remember correctly on polymer you need to use the same voltage from the original cap. but i did not know that you could change a 3300uf electrolytic cap with a 2200uf polymer cap! because the UF value is a rule to follow when changing electrolytic capacitors!

Polymers are pretty much the only choice for VRM caps on motherboards that used ultra-low ESR electrolytics (starting around the XP/S478 era or so), simply because electrolytics of that type like Rubycon MBZ are no longer made (and they were not very reliable to begin with either, especially in hot areas like the VRM -- in the best scenario they had a more limited lifetime and this applies even to good brands).

The "capacitance rule" is not necessarily per-capacitor, you have to look at the whole circuit. In the case of a motherboard, the VRM input and output have certain requirements such as a minimum total capacitance and a maximum total ESR. Manufacturers select capacitors to comply with those requirements. When using electrolytics, they had to go with much more total capacitance than required in order to meet the ESR requirement. This is because within the same series, larger caps have lower ESR (in fact, looking at datasheets you can easily tell there is a direct relationship between case size and ESR).

So, if you go with polymers, which provide much lower ESR than electrolytics at the same capacitance, you can use less total capacitance as long as you are above the minimum. Now, how to calculate the minimum is the tricky part and requires certain knowledge about the VRM circuit design which is way beyond my level. However a member of this board once provided a "guesstimate" that around half the original capacitance would be enough if using polymers. In that sense, YMMV.

Regarding the voltage rating, it's just an upper limit of what the capacitor can handle. So, on a 12V VRM input you will probably want to use 16V caps, and on the output more than whatever the max. CPU voltage supported by the motherboard is (after adding a certain safety margin of course).

question for you

the "around HALF the uf rule using poly" goes for caps around the VRM.

but how about the caps near the Memory Slot, and near the AGP slot. would those caps fall into the same "half the UF rule using Poly" or you would follow the exact "uf" replacement for those?

SETBLASTER wrote on 2023-11-05, 19:47:

the "around HALF the uf rule using poly" goes for caps around the VRM.

but how about the caps near the Memory Slot, and near the AGP slot. would those caps fall into the same "half the UF rule using Poly" or you would follow the exact "uf" replacement for those?

No. just use some modern low-ESR part in those areas. Panasonic FM/FR/FS, or Rubycon ZLx or YXx, or UCC KZx, KYx, or Nichion Hx, etc...