retardware wrote on 2021-10-11, 17:09:

I always thought, in particular where the DC-DC converters are, low ESR would be helpful.

Helpful - yes, in general low ESR is better. But typically not required. In fact a standard electrolytic or tantalum plus some low inductance foil (or better yet, ceramic) capacitor in parallel will not only provide good power rail decoupling, that extra non-aging capacitor will make sure the actual ESR value (for high frequencies) of the aging electrolytic is of little importance. This is not true for the output filters of high current switching mode PSU though, here you have some 20+ kHz spikes when the cap is recharged, and then discharged by load. Even though these PSUs use Pi type filters with pretty beefy inductor between capacitors as primary energy storage, these caps have a hard life at higher loads.

In other words these caps make most sense in the PSU itself, not in the target system the PSU is powering - in theory the system should not care where the power comes from, it should be clean and stable on both linear and switching mode PSUs. If the system in question becomes more stable with switching mode PSU when caps are replaced to low ESR ones, I would always bet the problem was not the caps but the noisy PSU in the first place. So can it help? Yes, but it just masks the original problem, not solves it. This of course assumes the design of both PSU and whatever it powers is not so terrible that it needs all the help it can get to work properly - but that too can be fixed by adding some extra ceramic caps in places where power delivery is marginal.

For the motherboard, Nihicon PA is the only good choice on the list... The rest are general purpose or audio caps (which are just marked up in price general purpose caps).

For the PSU, there are two issues:

1) That particular cellulose brown glue will become conductive over time. You will need to remove all of it.
2) This PSU uses a two transistor forward 5VSB design. That means the PSU always draws 1 or 2 amps on that rail as long as the switch is powered (or if there's no switch, as long as it's plugged in)

For the caps you selected, discard Elna audio capacitors. There are much better choices. I wouldn't use anything on that list.

That said, I checked my PSU repair notes for my Z350-ATA, but the capacitor locations don't match up so that's that. Good luck though.

Does anyone have a definitive list of caps to get for the abit bf6 motherboard im sure some people must have recapped it?

I'm getting the caps from mouser so if anyone has bought the caps from there could you give me a list?

I also managed to remove 14 of the 21 caps from the power supply.

I measured the esr apparently all the small caps have a esr of about 2.4 (2 of them had 1.5 though) while the larger caps have a esr of 0.24

I'll probably finish decapping the psu tommorow and I'll check what values the output caps have.

Deunan wrote on 2021-10-11, 18:52:

retardware wrote on 2021-10-11, 17:09:

I always thought, in particular where the DC-DC converters are, low ESR would be helpful.

Helpful - yes, in general low ESR is better. But typically not required. In fact a standard electrolytic or tantalum plus some low inductance foil (or better yet, ceramic) capacitor in parallel will not only provide good power rail decoupling, that extra non-aging capacitor will make sure the actual ESR value (for high frequencies) of the aging electrolytic is of little importance. This is not true for the output filters of high current switching mode PSU though, here you have some 20+ kHz spikes when the cap is recharged, and then discharged by load. Even though these PSUs use Pi type filters with pretty beefy inductor between capacitors as primary energy storage, these caps have a hard life at higher loads.

In other words these caps make most sense in the PSU itself, not in the target system the PSU is powering - in theory the system should not care where the power comes from, it should be clean and stable on both linear and switching mode PSUs. If the system in question becomes more stable with switching mode PSU when caps are replaced to low ESR ones, I would always bet the problem was not the caps but the noisy PSU in the first place. So can it help? Yes, but it just masks the original problem, not solves it. This of course assumes the design of both PSU and whatever it powers is not so terrible that it needs all the help it can get to work properly - but that too can be fixed by adding some extra ceramic caps in places where power delivery is marginal.

But aren't 6th gen motherboards and up usually equipped with switching mode VRMs too so their output profits from low ESR capacitors?

I'm not againt replacing capacitors, the bad ones have to go, that much is obvious. All I'm saying is recapping whole PCBs has become almost a religion that cures any and all problems you might have. Not only is that not true, and you learn nothing from it (other than soldering, maybe), it can be expensive and risky for amateurs. So what to recap, and with what, is a valid question and it shouldn't be answered with "everything with low ESR caps".

Without any tools (and a good idea of what a is a good / bad result) to test ESR and maybe also capacitance in-circuit I would opt to recap the PSU (the more tired it looks the more important it is, although with said tools a lof caps can be left as-is if keeping it original is a priority). Then anything obviously bad would also be replaced on logic PCBs but the rest stays. A good, clean power supply helps a lot, even with aged capacitors in the other parts of the system. Less work, less chances of unintentional damage.

But to answer that question diretly, yes, VRM circuits would benefit (at least it won't hurt or be pointless) from low ESR caps.

Deunan wrote on 2021-10-12, 14:03:

All I'm saying is recapping whole PCBs has become almost a religion that cures any and all problems you might have.

Indeed. And this is even crazier when people do not consider at all the quality of the power supply output.
I mean, what is the point of recapping the mobo when the PSU has excessive ripple? I have seen >3V ripple at 12V and >2V ripple at 5V on "working" PSUs that showed perfectly fine voltages on the multimeter.

Deunan wrote on 2021-10-12, 14:03:

A good, clean power supply helps a lot, even with aged capacitors in the other parts of the system.

Completely agree.
I don't recap my mobos (at least not those whose caps look fine and run stable).
Instead of doing that, for my most valued retro computers I use some selected good PSUs that have DC-DC converters (eg no crossloading headaches) and have ripple <25mV at 12V and <10mV at 5V.
For the others I consider <40mV at 12V and <20mV at 5V as acceptable (even most old Seasonics are in this range, even without recapping), provided that they pass the crossloading tests (eg not more than+-5% variation no matter the load ratios). The crossloading issue is a bad one, as I've seen many PSUs that manage to reach almost 15V on 12V when nearly exclusively loaded on 5V, and I seriously doubt that this is good for HDDs or floppies.

Matth79 wrote on 2021-10-11, 14:16:

Also, trying to figure out what would need 10V caps on a motherboard
The ATX PSU delivers 12V (16V caps), 5V (6.3V caps, 10V could be there), 3.3V (maybe 4V caps, otherwise 6.3V), and the board generates core and VRAM voltages
SDRAM being 3.3V and CPU Vcore being max 3.5V

Place I was at years ago had a house design rule of 3x nominal voltage and 105C for electrolytics. Reasoning was that the specified lifespan of them is at their given rating and is often only in the low thousands of hours, which wasn't much good for stuff that was on 24/7 (1000 hours ~= 7 weeks, 9000 hours ~= 1 year). But the lifespan had some sort of exponential increase the further away from the ratings you stay (I can't remember the details). 3x is probably overkill, but that was for industrial stuff. I normally try to stay 2x / 105C if I have to replace anything, PCB space allowing, as a 6.3V cap on a 5V line will go out of spec a lot quicker than a 10V and I don't like changing caps more than once.

There's also a problem running ceramics close to their voltage rating. Their capacitance decreases with DC bias, so running too close will end up with less capacitance than you designed for. Still fine for decoupling, less good for bulk.

I think the capacitor plague of the late 90s/early 00s scarred a lot of people, so they made it a requirement to always change all electrolytics everywhere, no matter what. I had a socket 939 (2005ish?) that was fine, then in a cupboard not powered up for ~10 years. On digging it out again, most of the caps >1000uF had bulged and some had leaked (fortunately/by design from the tops). Didn't try turning it on, just swapped out all the big caps and it's now fine. I did test one of the smaller caps just to be sure, and it was in spec.

Look I wouldn't be replacing the caps unless it was necessary, I've got a nice pentium 1 pc with a compaq delta psu with pristine caps that haven't been replaced and the motherboard also has nice caps that haven't bulged and I wouldn't replace.

The problem is that this motherboard here is an abit p2/p3 motherboard which was a victim of the capacitor plague of the late 90s and it has bulged caps , as does the power supply one which was vented and lost like 2 thirds of its capacitance. Trust me id rather not have to bother with it , but its not my fault some engineer got greedy in the late 90s early 00s and ruined the entire capacitor industry by stealing plans.

AppleSauce wrote on 2021-10-12, 17:15:

Look I wouldn't be replacing the caps unless it was necessary, I've got a nice pentium 1 pc with a compaq delta psu with pristine caps that haven't been replaced and the motherboard also has nice caps that haven't bulged and I wouldn't replace.

The problem is that this motherboard here is an abit p2/p3 motherboard which was a victim of the capacitor plague of the late 90s and it has bulged caps , as does the power supply one which was vented and lost like 2 thirds of its capacitance. Trust me id rather not have to bother with it , but its not my fault some engineer got greedy in the late 90s early 00s and ruined the entire capacitor industry by stealing plans.

The capacitor "plague" is a mish-mash of many different things that have been converged into one idea in an attempt to focus hardware failures into one narrow definition.

There are several reasons for capacitor failures:

1) Prohibition of the use of certain toxic materials in capacitors, which lead to the use of experimental electrolytic formulas, often with negative results (this is why many types of caps from the 60s to 80s still test good today)
2) The begining of production of capacitors by Chinese manufacturers who used really primitive electrolyte formulas
3) The increasing stress placed upon capacitors by ever more hungry VRM designs
4) The additional stress placed on already lousy PSUs by newer hardware

The idea that the entire problem rested upon the Chinese 'stealing' the electrolyte fomulas but not getting them right is a myth... In the 90s, no one really had good electrolyte formulas, save for perhaps the non-aqueous series of the time. Nichicon and UCC aqueous series of that era, especially those with quaternary ammonium formulas failed en masse. The aqueous electrolytic series of today are superior in orders of magnitude in their longevity and in their ratings.

All this said, there are two basic principles you need to follow when re-capping PSUs and motherboards:

1) Low ESR, high ripple capacitors for PSUs... Don't use polymer capacitors in PSUs.
2) Ultra-low ESR, high ripple capacitors for motherboard VRMs on P4 or newer platforms, Ultra low to low ESR caps for P3 and older VRM designs.

In other words, UCC LXZ, Nichicon PW, Panasonic FC type capacitors for PSUs (non-aqueous series -- though this is just to play it safe, you can use aqueous series in PSUs too), UCC KZE, UCC KY or KYx, Panasonic FM/FR/FS, or Nichicon HE for P3 and older, and POLYMER capacitors only for P4 and newer (since you can't purchase good ultra-low ESR aqueous series anymore).

Stay away from Panasonic FC for the time being, they are having production issues with them.

See, this is why I don't like just saying "low ESR caps for PSU". Surely the output filters, possibly also the big primary side ones (but that's not critical), but what is the point of putting low ESR caps in PWM controller supply rail? Or POWER_GOOD circuit where there's often an electrolytic used as a timing cap, charged once at power on?

And frankly putting polymer capacitors on anything that hasn't had polymers is kinda waste of money, unless it's a pretty recent mobo (say Athlon 64 or newer) and you really care about it. You are not going to use as a daily driver, what is the point, modern aluminum electrolytics are going to last you a lifetime.

mockingbird wrote on 2021-10-12, 18:41:

Stay away from Panasonic FC for the time being, they are having production issues with them.

What's wrong with those, that is my favourite series since I can actually get them unlike Nichicons that I often find out of stock. Never had any issues with them either.

Deunan wrote on 2021-10-12, 19:07:

See, this is why I don't like just saying "low ESR caps for PSU". Surely the output filters, possibly also the big primary side ones (but that's not critical), but what is the point of putting low ESR caps in PWM controller supply rail? Or POWER_GOOD circuit where there's often an electrolytic used as a timing cap, charged once at power on?

The point of writing 'low ESR' is to prevent confusing people, who don't need to be made aware of the minutia. The fact is that low-ESR caps are also more often than not long-life as well. Now, you may ask why should they have to be long-life, well the answer is that they don't necessarily, but when being asked to recommend a cap, one tends to be cautious in their recommendation and opts for long-life capacitors in general, because we don't know where the cap is going... Sure, we might be told which circuit it's going to be used in on the PSU PCB, but will the cap be situated near the diode of another circuit? If it's going to be exposed to a higher level of heat than normal, then its lifetime decreases.

So, to be more specific, and to your point, you can use something like Panasonic EB series outside the pi filters, which is a long life series, but quite ordinary in other respects (though I wouldn't use it... the "not to be sold in Japan" throws up all sorts of red flagd for me).

And frankly putting polymer capacitors on anything that hasn't had polymers is kinda waste of money, unless it's a pretty recent mobo (say Athlon 64 or newer) and you really care about it. You are not going to use as a daily driver, what is the point, modern aluminum electrolytics are going to last you a lifetime.

Without getting into the detail of modern VRM designs, you absolutely do need polymers on P4 and newer VRMs (or to be more specific, you need to replace ultra-low ESR electrolytics with parts that have an equivalent rating, which can only be found in polymer parts today). We are not looking at longevity with VRMs (though longevity is nice too and they do manufacture 5000 hour polymers), we are looking at ripple suppression. Polymer capacitors are more often than not rated as having the capability to withstand several amperes of AC ripple. I think ESR also plays an important role here, with said VRM caps often having a rating of only several milliamps. So yes, you absolutely do need polymer capacitors in a modern VRM.

What's wrong with those, that is my favourite series since I can actually get them unlike Nichicons that I often find out of stock. Never had any issues with them either.

They're either having electrolyte problems or bunge/lead problems because they all leak from the bunge and corrode the lead and leave green residue.

mockingbird wrote on 2021-10-12, 20:06:

The point of writing 'low ESR' is to prevent confusing people, who don't need to be made aware of the minutia.

I don't want to be nitpicking but that is exactly what I am against. We say "go low ESR everywhere, can't hurt" and it becomes a gospel, we already are at a point where "recap everything" is one. And then, when somebody who actually knows what he/she is doing, posts a repair video on YT is ridiculed for not doing just that, because it's the norm, so surely that one must be wrong. This is just being lazy, not helpful. Soon people will be asking what are the other capacitors even for if you always have to put the low ESR ones in.

I try to explain why capacitors need to be replaced (or even IF they need to be replaced) and with what. In the end OP might decide to recap everything anyway, with low ESR variants, but at least he will know that was done because it was easier (though surely not cheaper), and not because it's something you do.

mockingbird wrote on 2021-10-12, 20:06:

Without getting into the detail of modern VRM designs, you absolutely do need polymers on P4 and newer VRMs (or to be more specific, you need to replace ultra-low ESR electrolytics with parts that have an equivalent rating, which can only be found in polymer parts today).

I'm just going to disagree here. OK, maybe the high clocked P4 in particular would benefit from such treatment, since these CPUs were so power hungry and stressed the VRMs more than many moderns CPUs do. I've replaced failed capacitors on quite a few mobos, never used polymers. Some of these mobos worked for years 24/7 since and there were no issues whatsoever.

The original caps these mobos had were not magic, just low ESR aluminum electrolytics. Sure, you can put better caps in there, but that should be an informed choice, not a default action. And polymer caps have their issues, they don't vent but even brief overvoltage events can blast them off to space - something that would only mildly stress a standard electrolytic. And I can imagine how a failure like that could lead to domino effect and more damage in turn.

mockingbird wrote on 2021-10-12, 20:06:

So yes, you absolutely do need polymer capacitors in a modern VRM.

Modern ones, sure. Replace polymer with polymer, if that's required. Never downgrade a capacitor, but also don't upgrade for no reason, and most certainly don't just go telling people to do that because it's easier than explaining why.

OK, I don't want to start any turf wars over capacitors and I think I explained my position, so I'm done here.

Deunan wrote on 2021-10-12, 20:45:

I'm just going to disagree here. OK, maybe the high clocked P4 in particular would benefit from such treatment, since these CPUs were so power hungry and stressed the VRMs more than many moderns CPUs do. I've replaced failed capacitors on quite a few mobos, never used polymers. Some of these mobos worked for years 24/7 since and there were no issues whatsoever.
<snip>
The original caps these mobos had were not magic, just low ESR aluminum electrolytics.

Except for the very niche Willamette P4 motherboards, Northwood and up (that means 845 chipset) almost always used ultra-low ESR caps for the VRM... We're talking Fujitsu Hybrid Polymers (no longer produced), Nichicon HM or better, Panasonic FJ or better, or Rubycon MBZ or better. Intel branded boards of that era often used a combination of an ultra-low ESR electrolytic as well as either Fujitsu FPCAP or some really early Nichicon polymer parts in the VRM low. I'd take pictures but I scrapped all my old P4 motherboards earlier this year.

Asus branded boards of the 462 variety with nForce2-type of later chipsets often used Nichicon HM (which had manufacturing problems for much of its production life).

mockingbird wrote on 2021-10-12, 20:06:

So yes, you absolutely do need polymer capacitors in a modern VRM.

Modern ones, sure. Replace polymer with polymer, if that's required. Never downgrade a capacitor, but also don't upgrade for no reason, and most certainly don't just go telling people to do that because it's easier than explaining why.

OK, I don't want to start any turf wars over capacitors and I think I explained my position, so I'm done here.

I am not going to say "gotcha" or anything like that, but you say to replace one part with its equivalent, but earlier you state that it's ok to downgrade P4 motherboards with non ultra-low ESR parts. So you're not being consistent.

This is not unimportant information - to the contrary. You weaken the pads on motherboards when you re-cap. Better do it right the first time.

I had an Abit BF6 back in the day, that's probably why I posted here in the first place... I don't know how Abit did it, but it was noticeably faster than ordinary BX boards. They must have tightened the default timings somewhere in the BIOS. You could feel it.

Unfortunately I did not know what I was doing at that time and I destroyed that board trying to re-cap it.

Okay i got nearly all the caps off except for three , I lifted a pad one of the last ones by accident , should I just like solder the pad back down?

Yeah, just flatten it and make sure it didn't move too close to any other traces nearby, this looks to be the low voltage section so little cause for concern. If you feel there's enough damage done and too little copper left then just slap a piece of wire on top and solder it to nearby pins, or unmask some copper and solder to that. Just like that piece of wire that's already there to improve GND routing, but obviously it doesn't need to be that thick.

EDIT: Also, problems like this is why I recommend not to just recap everything on more complex PCBs like mobos. I've said all that needs to be said already, just pointing that out.

Deunan wrote on 2021-10-13, 10:07:

Yeah, just flatten it and make sure it didn't move too close to any other traces nearby, this looks to be the low voltage section so little cause for concern. If you feel there's enough damage done and too little copper left then just slap a piece of wire on top and solder it to nearby pins, or unmask some copper and solder to that. Just like that piece of wire that's already there to improve GND routing, but obviously it doesn't need to be that thick.

EDIT: Also, problems like this is why I recommend not to just recap everything on more complex PCBs like mobos. I've said all that needs to be said already, just pointing that out.

Okay then , well I've got one vented cap on the ABIT , so you would suggest like just swapping the one vented cap then?

Also I've made a better diagram of the PSUs layout.