Boohyaka wrote on 2023-10-01, 18:20:

I've looked into recapping many motherboards over the years but always delayed it in the end, just set them aside to fix them "l […]
Show full quote

I've looked into recapping many motherboards over the years but always delayed it in the end, just set them aside to fix them "later". Now I'm ready to get into it.

I've searched vogons and the internet on information about how to choose the right capacitors and I'm a bit overwhelmed with the information. It's one of these subjects where it seems everyone has a different opinion, often insubstiantiated any more than "trust me bro", and going to mouser.com I'm also overwhelmed by the offer...And I'm pretty sure that as usual, I'm overthinking it because it's new to me and don't want to make mistakes.

So I guess I'd just like some confirmation about a few factoids I'll write down, and that my current choice makes sense. For context I'm really looking into motherboards only at the moment (understand soundcards are more sensitive and require audio grade capacitors, that's not the case here). Also, I want to fix motherboards with caps that already leaked (or obviously bulging), I'm not trying to fix things that aren't broken, and I understand I should change all caps of a particular group even if some have not (yet) leaked. Let's use my first real-life case here: an ASUS A7A266 where I need to change 6x 1500µF/6.3V and 3x 680µF/6.3V caps.

- I've read about Japanese brands being the most recommended. Nichicon, Rubycon. Still accurate? Other notable mentions?
- Physical size is obviously important (close to the original ones)
- Capacitance should ideally be the very same rating, but voltage can be higher. Makes sense to me as it is a max V rating, so getting a highest rating for a similar capacitance is no issue.
- Current caps are rated for 105°, so I'll stick to the same.
- Some people recommend grabbing Aluminum Organic Polymer capacitors instead of electrolytics. Apparently they don't leak, last longer and have the advantage of very low ESR. Also read that if going that route I should take lower capacitance rating? (that's all from memory so maybe I'm confusing different things, sorry). From my quick checks it seems the prices are pretty similar. Is that good advice, and any good reason to make that choice? Feels to me that if I can find similarly rated eletrolytics capacitors I'm taking no risk and I'll be good for the next 20+ years anyways...
- A lot of Nichicon caps I'm looking for on mouser are classified as "low impedance". From more searches it seems it's particularly important for PSU's? Should I be looking for general purpose or any other type of capacitors to have them cheaper, or it doesn't really matter?
- Any other specs I need to look out for? Lower ESR = better? What about the "ripple current" rating, is it important?

I think that's it to start...so here's what I'm looking at right now (focusing on Nichicon to reduce the scope of search results 😁):

- 1500µF/6.3V: https://www.mouser.ch/ProductDetail/Nichicon/ … 2F01acxtg%3D%3D
- 680µF/6.3V: https://www.mouser.ch/ProductDetail/Nichicon/ … PzCrlBYOg%3D%3D

Are these good choices in this situation? I'd be happy to get other recommendations, if possible with the explanation that goes with it so I try and learn something useful 😀

Thanks!

Hello,
I never bought capacitors, so my posting is not very important.
But in the first link: There is life: 4000 hours, in the second: life: 6000 hours.
Is that normal?

I can recommend reading the sticky topics here: https://www.badcaps.net/forum/forumdisplay.php?f=24

Anyway, probably your precious vintage motherboard won't be used 24/7 in 365 days/year, so even if your choice is not optimal, it doesn't matter 😉
BTW. Your assumptions are correct. In short: low ESR, 105', same dimensions...

You're on the right track.

The only thing I'd add is being aware of polarization. Some caps are polar (e.g. positive and negative sides), some are non-polar.

Motherboard caps will typically be polarized. Just make sure when recapping, that you line up the respective positive and negative sides correctly.

Thank you! Well I understand that my precious vintage motherboard won't see heavy load, but when electronics are still kinda like voodoo magic to you, you don't know where the cursor precisely is between "not optimal but it's not that important" and "it just won't work". But thinking about it, as I've seen motherboards still run with leaky capacitors, it makes sense that in the end it's not *that* important. Still the occasion to learn something and get a little help to make a good choice 😉

About low ESR: I've seen that mentioned many times, but what's an acceptable range for a motherboard? Or is it again a matter of "get the lowest you can reasonably get, it doesn't matter that much?"

@coffeeone: I don't think this expected lifetime matters that much. Let's say I'm using the computer every single day for 2 hours, we're looking at 10 years expectancy. And that's probably at max load, I doubt these capacitors will see that much of intense action during these 2 hours, so that would be a safe and conservative assumption that we're looking at at least 1.5x that. 15 years sounds like a good deal to me 😀

Shponglefan wrote on 2023-10-01, 19:16:

You're on the right track.

The only thing I'd add is being aware of polarization. Some caps are polar (e.g. positive and negative sides), some are non-polar.

Motherboard caps will typically be polarized. Just make sure when recapping, that you line up the respective positive and negative sides correctly.

Thank you. Yes I was aware of that but as I didn't mention it in my post it's a good thing you mentioned it again, also maybe for other readers.

I'm still curious about the organic polymer alternative 😁

Boohyaka wrote on 2023-10-01, 19:20:

About low ESR: I've seen that mentioned many times, but what's an acceptable range for a motherboard? Or is it again a matter of "get the lowest you can reasonably get, it doesn't matter that much?"

It depends on the motherboard design. VRM circuits on newer boards usually require much lower ESR, which is one of the reasons why pretty much all motherboard makers ended up switching to polymers since ultra-low ESR electrolytics had many issues (short lifespan, overly sensitive to heat, etc.). On the other hand, older boards, such as 486 or early Pentiums can do just fine with general purpose caps since they mostly use linear regulators.

Best approach is to find the datasheets of the original capacitors (if possible) and find similar replacements from good brands. If you have a list of the motherboards you want to recap I can suggest some series that have worked for me in the past.

There are many information on caps, but - in practice - I follow just a few simple rules:
- do not buy cheapest ones,
- use low ESRs,
- always check capacitance (and polarity) before soldering a cap to the board, for new caps I consider 5% discrepancy OK (some say 10%),
- use same capacitance as the cap replaced had,
- if a cap did leak/bulge, replace it with one-step-up voltagewise one (eg. if the leaky cap was rated 10v, replace it with 16v, if it was 6,3v, replace with 10v)
- good flux doesn't hurt

Low ESR is generally good, but be aware that some analog power circuits rely on having a certain amount of ESR and having it too low can actually cause problems. The original capacitors they probably just used whatever was cheapest and didn’t really pay attention to esr at the time. Many circuits benefit from low ESR but others are actually designed for a certain level of ESR particularly analog power circuits.

You can use a higher value capacitor in a pinch if you have to (eg using 35 instead of 25) but be aware that this should actually be avoided if possible because the electrical properties are not actually identical. These will work but if you can it’s actually best to use the exact same capacitance and get as close to the original specs as possible.

Look at the number of hours the cap is rated for and its temperature tolerance, higher is generally better for both.

The best thing is always to try to match the original spec exactly but most of the time you can get away with almost anything.

I'm not hinting at anything, but the formula
Ripple Voltage = ΔI x ESR
is directly stated in all datasheets.
The same Pentium 3 consumes under 30 Amps, and this does not equal to light up the LED.
https://www.analog.com/media/en/technical-doc … heets/3720f.pdf
https://kentie.net/article/deskproupgrade/vrm8.5.pdf

TheMobRules wrote on 2023-10-01, 20:11:

all motherboard makers ended up switching to polymers

Some never left.
Yeah, it's a Pentium Pro with Sanyo OS-con.

PS. The photo is not mine, but I would like to have one.

midicollector wrote on 2023-10-01, 22:57:

Low ESR is generally good, but be aware that some analog power circuits rely on having a certain amount of ESR and having it too […]
Show full quote

Low ESR is generally good, but be aware that some analog power circuits rely on having a certain amount of ESR and having it too low can actually cause problems. The original capacitors they probably just used whatever was cheapest and didn’t really pay attention to esr at the time. Many circuits benefit from low ESR but others are actually designed for a certain level of ESR particularly analog power circuits.

You can use a higher value capacitor in a pinch if you have to (eg using 35 instead of 25) but be aware that this should actually be avoided if possible because the electrical properties are not actually identical. These will work but if you can it’s actually best to use the exact same capacitance and get as close to the original specs as possible.

Look at the number of hours the cap is rated for and its temperature tolerance, higher is generally better for both.

The best thing is always to try to match the original spec exactly but most of the time you can get away with almost anything.

Low ESR and high ripple current capability were definitely a consideration for the socket A era motherboard VRMs. The problem was that manufacturers decided to buy those ultra-low ESR caps from bunch of mainly Taiwanese companies, which offered these for a very low price undercutting big japanese manufacturers. These companies used unstable electrolytics formula in them and it caused those capacitors failing within couple of years, hence the name and era of "capacitor plague".

But yeah, the correct approach is to check the datasheets of the original capacitors and find equivalent caps. However, finding modern replacements for some of these electrolytics may be impossible and you need to install polymers instead. Polymer capacitors usually max out at or around 2000uf, but you can replace original VRM caps with roughly half the capacitance the original had (for example, 3300uf electrolytic replaced with 1500-2000uf polymer).

CharlieFoxtrot wrote on 2023-10-02, 09:28:

Polymer capacitors usually max out at or around 2000uf, but you can replace original VRM caps with roughly half the capacitance the original had (for example, 3300uf electrolytic replaced with 1500-2000uf polymer).

^^^^ this. I did few polymods (for VRM region caps) by sticking to the half capacity rule and all of them were completely stable.
https://badcaps.net/forum/showthread.php?t=82050

zyga64 wrote on 2023-10-02, 10:54:

^^^^ this. I did few polymods (for VRM region caps) by sticking to the half capacity rule and all of them were completely stable.
https://badcaps.net/forum/showthread.php?t=82050

but why?
https://eu.mouser.com/ProductDetail/Wurth-Ele … ZdD7pWHiA%3D%3D
1500uF*6.3V 10mm dia
WCAP-PTG5 series by Wurth Elektronik
https://www.we-online.com/en/components/products/WCAP-PTG5
First of all, red. Everyone knows that the DA RED WUNZ GO FASTA.
Secondly, it does not look like polymer
And, in fact, this is a direct replacement, for example, of the Nippon KZG series.

shevalier wrote on 2023-10-02, 12:29:

but why? https://eu.mouser.com/ProductDetail/Wurth-Ele … ZdD7pWHiA%3D%3D 1500uF*6.3V 10mm dia WCAP-PTG5 series by Wurth Elektron […]
Show full quote

zyga64 wrote on 2023-10-02, 10:54:

^^^^ this. I did few polymods (for VRM region caps) by sticking to the half capacity rule and all of them were completely stable.
https://badcaps.net/forum/showthread.php?t=82050

but why?
https://eu.mouser.com/ProductDetail/Wurth-Ele … ZdD7pWHiA%3D%3D
1500uF*6.3V 10mm dia
WCAP-PTG5 series by Wurth Elektronik
https://www.we-online.com/en/components/products/WCAP-PTG5
First of all, red. Everyone knows that the DA RED WUNZ GO FASTA.
Secondly, it does not look like polymer
And, in fact, this is a direct replacement, for example, of the Nippon KZG series.

Because you can't always find a direct replacement, that's why. For example, try to find a 3300uF polymer capacitor which would directly replace a 3300uF ultra-low ESR cap. There is none, at least not widely available as 2000uF is pretty much as far as polymers go. Gladly you can use lower capacitance polymers as replacements. This has happened to me. 1500uF isn't a problem, but going to 2200uF and 3300uF isn't just possible as direct replacements and you need to settle with lower capacitance.

Thank you for the interesting info and discussion so far. So yeah...looks like caps will always generate debate 😁
@zyga64 thanks in particular for the badcaps sticky posts, lots of interesting stuff that answered some of my questions.

OK so let's stay on the use case of the ASUS A7A266.

Main culprits are those Sanyo leaky bastards:

So trying to do my due diligence and not wait on you guys to do the job for me, I found out those were manufactured by SunElec (under the SunCon name) already but rebranded Sanyo, and since then Suncon have purchased the electrolytic range from Sanyo altogether and then started branding them SunCon instead. Not really relevant to the discussion but helped me find the specs.

So they are Sanyo WX, S.E.14, 1500uF and 6.3V.
I found this: https://www.sunelec.co.jp/common/pdf/eng/ME-WX.pdf

Looking at the 6.3V 1500uF specs, and comparing to what I initially picked (Link)

- Size: 10x20 - check
- Temp: -40 to 105°C - check
- ESR: 23 mOhms - check
- Tolerance: 20% - check
- Ripple: 1.82 A - check

...is it me or did I pick a perfect replacement on my first try? Beginner's luck? 😁

CharlieFoxtrot wrote on 2023-10-02, 14:07:

For example, try to find a 3300uF polymer capacitor which would directly replace a 3300uF ultra-low ESR cap.

Since the era of P3 Coppermine/Athlon XP, the supply voltage of processors has dropped below 2 Volts.
Therefore, 2.5/4V polymer capacitors can be used.
Namely, there are a number of such capacitors with a capacity of 3,000 μF, starting from Samwha FB / Lelon OCRZ and up to Kemet/Nichicon/TDK.

It should be taken into account that for electrolytes, ESR is measured at the end of its life, i.e. the new capacitor has ESR 20-50% less datasheet.
Thus, a modern polymer capacitor is generally a direct replacement.

shevalier wrote on 2023-10-02, 15:21:

Since the era of P3 Coppermine/Athlon XP, the supply voltage of processors has dropped below 2 Volts. Therefore, 2.5/4V polymer […]
Show full quote

CharlieFoxtrot wrote on 2023-10-02, 14:07:

For example, try to find a 3300uF polymer capacitor which would directly replace a 3300uF ultra-low ESR cap.

Since the era of P3 Coppermine/Athlon XP, the supply voltage of processors has dropped below 2 Volts.
Therefore, 2.5/4V polymer capacitors can be used.
Namely, there are a number of such capacitors with a capacity of 3,000 μF, starting from Samwha FB / Lelon OCRZ and up to Kemet/Nichicon/TDK.

It should be taken into account that for electrolytes, ESR is measured at the end of its life, i.e. the new capacitor has ESR 20-50% less datasheet.
Thus, a modern polymer capacitor is generally a direct replacement.

This figure is really frightening. So electrolytic caps get close to end of life (or are already EOL) after 4000 hours.
That is really nothing, it is about half a year.
I don't understand the argument: We speak about retro sytems, so it won't run 24/7. What about retro servers or something.

Boohyaka wrote on 2023-10-02, 14:47:

Thank you for the interesting info and discussion so far. So yeah...looks like caps will always generate debate :D @zyga64 thank […]
Show full quote

Thank you for the interesting info and discussion so far. So yeah...looks like caps will always generate debate 😁
@zyga64 thanks in particular for the badcaps sticky posts, lots of interesting stuff that answered some of my questions.

OK so let's stay on the use case of the ASUS A7A266.

Main culprits are those Sanyo leaky bastards:

sanyo1.jpeg
sanyo2.jpeg
sanyo3.jpeg

So trying to do my due diligence and not wait on you guys to do the job for me, I found out those were manufactured by SunElec (under the SunCon name) already but rebranded Sanyo, and since then Suncon have purchased the electrolytic range from Sanyo altogether and then started branding them SunCon instead. Not really relevant to the discussion but helped me find the specs.

So they are Sanyo WX, S.E.14, 1500uF and 6.3V.
I found this: https://www.sunelec.co.jp/common/pdf/eng/ME-WX.pdf

Looking at the 6.3V 1500uF specs, and comparing to what I initially picked (Link)

- Size: 10x20 - check
- Temp: -40 to 105°C - check
- ESR: 23 mOhms - check
- Tolerance: 20% - check
- Ripple: 1.82 A - check

...is it me or did I pick a perfect replacement on my first try? Beginner's luck? 😁

Lol, life span 2000 to 5000 hours.
So don't run it 24/7, otherwise you have to re-cap 4 times per year. 2000 hours is 83 days.

CoffeeOne wrote:

This figure is really frightening. So electrolytic caps get close to end of life (or are already EOL) after 4000 hours.
That is really nothing, it is about half a year.
I don't understand the argument: We speak about retro sytems, so it won't run 24/7. What about retro servers or something.

Keep in mind that hourly rating is typically given for running at a peak temperature. Running at lower temps will extend their lifespan.

Under normal use, quality caps will last decades.

Shponglefan wrote on 2023-10-02, 17:16:

Keep in mind that hourly rating is typically given for running at a peak temperature.

Not just max temp (usually 105C) but also max ripple current, basically worst case scenario.

Obviously under normal conditions they last much longer, otherwise there wouldn't be 30-40 years old hardware with caps still doing their job.

But since capacitor lifespan depends greatly on the environmental conditions the manufacturers have to provide estimates under the most hostile conditions the cap can be used.

Even polymers have similar lifespan ratings.

CoffeeOne wrote on 2023-10-02, 17:07:

This figure is really frightening. So electrolytic caps get close to end of life (or are already EOL) after 4000 hours.
That is really nothing, it is about half a year.
I don't understand the argument: We speak about retro sytems, so it won't run 24/7. What about retro servers or something.

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