I haven't done a recap yet, but when I start testing and fixing my large collection of boards (which should start in a month or two), if I see one cap that needs replacing, I'll change them all. If one has failed, the rest won't be far behind. It's not like they are expensive, even when buying high quality parts, especially if you buy enough for several boards in one go, and the boards should then outlive me (I'm 61 and have been too ill to work for half my life, so don't expect to survive another 20 years).

debs3759 wrote on 2023-08-13, 08:42:

I haven't done a recap yet, but when I start testing and fixing my large collection of boards (which should start in a month or two), if I see one cap that needs replacing, I'll change them all. If one has failed, the rest won't be far behind. It's not like they are expensive, even when buying high quality parts, especially if you buy enough for several boards in one go, and the boards should then outlive me (I'm 61 and have been too ill to work for half my life, so don't expect to survive another 20 years).

Full recap is of course always the safe bet in the sense that you sure get 100% working caps. But all caps aren't the same. From this era, the low ESR electrolytics are by far the most unreliable and mostly the cause for the bad reputation of capacitors in computer hardware of the era. When we are talking about general purpose caps, things tend to be very different. For example TEAPO low ESR caps are considered of bad quality, but general purpose is mostly just fine or at least decent. Also and depending of course from circuit, simple filter caps can be quite far away from the spec until you get problems. Story is very different for low ESR VRM caps.

Cost is not the reason why I'm asking, although very low ESR caps required here aren't cheap compared to GP caps. As I said, making repairs where not needed isn't advisable either. For example, in this board's case, recapping only low or low'ish ESR stuff means rougly 15-20 caps. With full recap I estimate that number doubles. That also means double the chance of ripping the pad or destroying via on the board. This is the reason I focus on doing the work that is necessary, especially when we are are talking about relatively complex electronics compared to 80s and early 90s stuff.

debs3759 wrote on 2023-08-13, 08:42:

I haven't done a recap yet, but when I start testing and fixing my large collection of boards (which should start in a month or two), if I see one cap that needs replacing, I'll change them all. If one has failed, the rest won't be far behind. It's not like they are expensive, even when buying high quality parts, especially if you buy enough for several boards in one go, and the boards should then outlive me (I'm 61 and have been too ill to work for half my life, so don't expect to survive another 20 years).

Well, it's not the amount of years what it counts but the value of your life, what you do and how you live. Your impact on your family, community and country. Speaking of that, you have a cool website, I've used it in the past. Kudos to you!

About recapping...
I'm by no means an expert (there are some on this board), but I've recapped many boards. You get better with practice.

I change them by groups, as you have stated. If there's one bad cap near a VRM, I change all in that region (with similar value caps) so that one doesn't get stressed more than the others.

If caps look fine and the board is working fine, I subscribe to your policy of "if it ain't broke, don't fix it".

I don't change lower value caps <22µF if there are no problems or visual damage.

To decap use plenty of flux and solder, a good soldering iron, high temperature for a very short time so you don't damage the PCB, that's the most crucial step, as well as cleaning the hole.

And always measure capacitance and ESR before soldering.

tauro wrote on 2023-08-13, 10:13:

Well, it's not the amount of years what it counts but the value of your life, what you do and how you live. Your impact on your […]
Show full quote

debs3759 wrote on 2023-08-13, 08:42:

I haven't done a recap yet, but when I start testing and fixing my large collection of boards (which should start in a month or two), if I see one cap that needs replacing, I'll change them all. If one has failed, the rest won't be far behind. It's not like they are expensive, even when buying high quality parts, especially if you buy enough for several boards in one go, and the boards should then outlive me (I'm 61 and have been too ill to work for half my life, so don't expect to survive another 20 years).

Well, it's not the amount of years what it counts but the value of your life, what you do and how you live. Your impact on your family, community and country. Speaking of that, you have a cool website, I've used it in the past. Kudos to you!

About recapping...
I'm by no means an expert (there are some on this board), but I've recapped many boards. You get better with practice.

I change them by groups, as you have stated. If there's one bad cap near a VRM, I change all in that region (with similar value caps) so that one doesn't get stressed more than the others.

If caps look fine and the board is working fine, I subscribe to your policy of "if it ain't broke, don't fix it".

I don't change lower value caps <22µF if there are no problems or visual damage.

To decap use plenty of flux and solder, a good soldering iron, high temperature for a very short time so you don't damage the PCB, that's the most crucial step, as well as cleaning the hole.

And always measure capacitance and ESR before soldering.

As I said, I have done quite a lot recap work and other repairs, including SMD stuff so I'm not actually hesitating this from the perspective of the work itself. I have good equipment, including powerful soldering station from KurzErsa. It is just I have no experience from this era boards and the general condition of the capacitors, other than the fact that the I know that the low ESR stuff is mostly crappy, even from many reputable brands. But again, those Sanyos on board, for example, look just fine and they are 100% there for the 5V VRM.

That is the reason why I'm asking opinions, especially how you guys have approached those series of caps that still at least look externally fine and/or are from a good manufacturer. I find it extremely annoying to repair something and building the system and later on finding out that it still works unoptimally and would've required more work. I'd rather do the required work only once and that's it.

CharlieFoxtrot wrote on 2023-08-13, 10:30:

tauro wrote on 2023-08-13, 10:13:

Well, it's not the amount of years what it counts but the value of your life, what you do and how you live. Your impact on your […]
Show full quote

debs3759 wrote on 2023-08-13, 08:42:

I haven't done a recap yet, but when I start testing and fixing my large collection of boards (which should start in a month or two), if I see one cap that needs replacing, I'll change them all. If one has failed, the rest won't be far behind. It's not like they are expensive, even when buying high quality parts, especially if you buy enough for several boards in one go, and the boards should then outlive me (I'm 61 and have been too ill to work for half my life, so don't expect to survive another 20 years).

Well, it's not the amount of years what it counts but the value of your life, what you do and how you live. Your impact on your family, community and country. Speaking of that, you have a cool website, I've used it in the past. Kudos to you!

About recapping...
I'm by no means an expert (there are some on this board), but I've recapped many boards. You get better with practice.

I change them by groups, as you have stated. If there's one bad cap near a VRM, I change all in that region (with similar value caps) so that one doesn't get stressed more than the others.

If caps look fine and the board is working fine, I subscribe to your policy of "if it ain't broke, don't fix it".

I don't change lower value caps <22µF if there are no problems or visual damage.

To decap use plenty of flux and solder, a good soldering iron, high temperature for a very short time so you don't damage the PCB, that's the most crucial step, as well as cleaning the hole.

And always measure capacitance and ESR before soldering.

As I said, I have done quite a lot recap work and other repairs, including SMD stuff so I'm not actually hesitating this from the perspective of the work itself. I have good equipment, including powerful soldering station from KurzErsa. It is just I have no experience from this era boards and the general condition of the capacitors, other than the fact that the I know that the low ESR stuff is mostly crappy, even from many reputable brands. But again, those Sanyos on board, for example, look just fine and they are 100% there for the 5V VRM.

That is the reason why I'm asking opinions, especially how you guys have approached those series of caps that still at least look externally fine and/or are from a good manufacturer. I find it extremely annoying to repair something and building the system and later on finding out that it still works unoptimally and would've required more work. I'd rather do the required work only once and that's it.

I personally have recapped a few boards now from this era and slightly newer such as socket 939 and my general feeling is to initially replace the obviously dead caps and to then test the board on my bench (Slots, IO ETC), if it all works then I leave the rest alone. I never do full recaps without testing as its generally far more work than necessary and unless its a rare/unique board not worth the cost in both time and money. (It also reduces the chance of damage to the board, it shouldn't happen but damaged pads do happen with older boards)

The only board I own that got a full recap was a Dual Tualatin board that was in pretty bad shape cap wise, it was an expensive board so was worth the cost of a full repair.

I also never willingly touch parts with Tanty caps ...last thing I need is electrical components throwing tantrums on my test bench ..kids are bad enough as is 🤣 (I prefer to pass these parts on to people who love the smell of burning parts)

Trashbytes wrote on 2023-08-13, 11:21:

I personally have recapped a few boards now from this era and slightly newer such as socket 939 and my general feeling is to initially replace the obviously dead caps and to then test the board on my bench (Slots, IO ETC), if it all works then I leave the rest alone. I never do full recaps without testing as its generally far more work than necessary and unless its a rare/unique board not worth the cost in both time and money. (It also reduces the chance of damage to the board, it shouldn't happen but damaged pads do happen with older boards)

The only board I own that got a full recap was a Dual Tualatin board that was in pretty bad shape cap wise, it was an expensive board so was worth the cost of a full repair.

I also never willingly touch parts with Tanty caps ...last thing I need is electrical components throwing tantrums on my test bench ..kids are bad enough as is 🤣 (I prefer to pass these parts on to people who love the smell of burning parts)

Thanks mate! I personally don't consider that much the market value of the vintage electronics when I do this kind of stuff. It is a hobby and it just feels so damn good when you get some broken and/or abandoned stuff refurbished and after that, hours of joy out of them. For example, I've fixed and restored couple of completely disgusting and black screening C64s. Although I paid very little of the units, economically repairs and refurbs didn't make any sense, because I can't sell them for profit, especially if I count the hours I put into them. But having fully working and good looking clean systems is still nice, although I have several of them.

I continued the mapping today and in my case, Sanyos need to be replaced with polymers as no equivalent electrolytics are available. I decided to order lower capacitance polymers with smaller ESR and new matching or better electrolytics for everything else at or above 470uf capacitance. At least I have then everything I need for all the beefier caps, but I may do first just like you did, that is I change the already bulged caps and every cap of the same series and rest of the large capacitance GSC caps. And then run some tests and see how the board is behaving. If not, I continue work from there.

And then some 10uF capacitor will dry up near the clocking IC, and the motherboard will start to bug.
And the forum threads will begin to multiply "all capacitors look like new, but it doesn't work !?"
How much money and time will you save by not replacing small capacitors?

shevalier wrote on 2023-08-13, 19:17:

And then some 10uF capacitor will dry up near the clocking IC, and the motherboard will start to bug.
And the forum threads will begin to multiply "all capacitors look like new, but it doesn't work !?"
How much money and time will you save by not replacing small capacitors?

Thats easy, have I replaced any caps on this board ?

if yes then time to replace the others I didnt replace.

If no then Im a dumbass and need to go do that.

As for saving time and money, a large portion of the time a lot, I have found small caps are less prone to bloating and dying than the larger ones and I would rather test them first before spending a silly amount of time replacing perfectly fine caps. (Small ones are a pain in the arse to replace, fiddly little buggers that they are)

If you are using board from time to time, like 2 hours per week, you don't need to recap it, even if there's buldget cap. That's my experience.
My main testing board had all capacitors all right (optically), and after 1 year of testing (my guess is, it ran about 50 hours), two of them are very slightly buldged. (not from running... it would happen anyway in box... happened to 3 MSI boards, they were stored in box, and one or two capacitors buldged, after about 2 years... but this testing board is HP Vectra https://theretroweb.com/motherboards/s/hp-vectra-vl410 (note: as you can see, on photo, they are all right (the upper left next to socket buldged, photo is about two year old)

I will not change them. They are buldged like half year (before I didn't notice), but progress had to be very slow, but maybe, I didn't notice in first stages.
It is going strong like another 20 hours of testing, completly stable.

I would change them only in case, I would use board as only retro board in only one retro computer, and had plenty of time with it (basically daily, at least 2 hours), and expected to use it at least one year. Otherwise, I would not care.

That's my personal experience, but different people can have different one. For me, not single capacitor exploded or leaked out, and I'm doing with retro computers daily for about 3 years, in average at least 1-2 hours. I have about 100-120 motherboards, most of them were run at least once for test, if they are functional.

I find capacitor stuff and exchanging them little bit overplayed. I can understand it when the board is used as primary board in its time (4-6 hours daily or more), or at least half of it as retro computer, but similiar way ("I have one (retro) computer, and I run it almsot daily, and have lots of fun time with it"... it's basically then as primary computer). Yes, then you should care to exchange them. If you have more boards, more retro computers, and you have like 4 hours per week to play with them totally, don't bother with capacitors (unless situation is obviously bad, like board is failing, or they are leaked and open).

Trashbytes wrote on 2023-08-13, 22:43:

replacing perfectly fine caps.

There are no perfect-looking capacitors, there are capacitors with ESR and the capacity corresponding to the datasheet.
Capacitors are not looked at, its are measured.

shevalier wrote on 2023-08-14, 03:59:

There are no perfect-looking capacitors, there are capacitors with ESR and the capacity corresponding to the datasheet. Capacito […]
Show full quote

Trashbytes wrote on 2023-08-13, 22:43:

replacing perfectly fine caps.

There are no perfect-looking capacitors, there are capacitors with ESR and the capacity corresponding to the datasheet.
Capacitors are not looked at, its are measured.

Please everyone, guess the riddle.
What will happen to the gigabyte_ga-k8n51gmf_rev_1.0 motherboard when this small 100 uF * 16V capacitor perfect-looking but dries up ?

The southbridge power regulator starts to generate with an amplitude of 1V.
At a nominal value of 1.6 (IIRC), jumps start from 1 to 2.5 volts.
But the motherboard continues to work and is almost not buggy.
How do I know this?
I have this board and an oscilloscope.

I will agree if its a mission critical cap, if its just some filtering cap then taking the time to replace them all is likely not worth it, if the board works fine and passes all its tests including all IO, slots and on board peripherals including board stability then I see no reason to take the extra risk.

If you have the scope, skills and datasheets then measuring the caps makes sense.

I mean we all have to make do with what we have access to, you look to have a much better repair setup than most here.

I do have a question that perhaps you can answer, is there any special requirements for replacing Normal caps with Poly ones ? can they simply be switched with ones of the same ratings ?

shevalier wrote on 2023-08-13, 19:17:

How much money and time will you save by not replacing small capacitors?

If you read what I wrote, I never mentioned money or time as a reason why I don’t want to do a recap. I said I don’t want to do recap unnecessarily, because it always involves a risk. I’ve done decent amount of soldering and repairs, but I’m not perfect. The more complex the system being repaired is, the more there is risk. For example, screw up just one via on the board and it may be unusable and unrepairable. With double sided boards you can always easily bodge a broken trace or via, but four layer board like this is another ball game.

Sure, if you want to be sure that every cap is 100%, you change them all. However, even if those smaller caps are degraded somewhat, most if not all smaller caps on this board are general purpose type used for filtering where they can be pretty far out of the spec not to work or causing issues on the board. I have much older stuff where GP capacitors have been completely fine when I’ve measured them. What I do not know is that are these early 2000s era GP capacitors generally as unreliable like the low ESR stuff, which would mean that replacing them is essential. I read badcaps forums about these boards and to me it generally seems that high capacitance low ESR stuff is the problem and smaller filtering caps are rarely even mentioned in the context. Then again, most of those forum posts are at least a decade old at this point.

This is also the one reason why I created the post to have opinions how other have approached recapping of similar era boards.

Your opinion is certainly important here. I have GP caps of different values readily available. If recapping the larger caps goes smoothly, I will change every cap on the board at the same time.

Among the capacitors that Epox used in that time period, the 2 that are least likely to fail are the Teapos and the Sanyos. If you'd rather minimize the number of caps you replace, I think it's reasonable to skip those until you see signs of them failing. Once you do see them fail though, the rest on the board are not trustworthy.

Teapo is one of the most famous "bad cap" brands, but I'm not sure why. They don't actually fail that often. The late 2000s web was thick with reports of bad Teapos but whenever there's a picture attached, it usually isn't a Teapo. More often it's a Tayeh or something else.
A long time ago I went through around 50-60 Epox socket-A boards (not this model). Virtually all of those boards had bulged caps, unless they were filled with Teapos (or Sanyos). I know the lack of bulging doesn't mean they were good, but their condition was a striking contrast to all the other cheap brands Epox used, and whenever I did remove a Teapo it never tested bad.
What this tells me, if nothing else, is that Teapo was not using the same generic (bad) electrolyte formula that the notorious "bad caps" were using. Their behavior in the field was completely divergent from the other cheap brands.
That was a long time ago though. Recently I looked at a few Epox boards I have left, and did start finding some bulged Teapos. It took them way longer to reach that point than the other brands, though.
Teapo is not one of the reputable Japanese brands, and so I wouldn't use them as replacements, but I don't think they're that bad. If the capacitor lottery gave you Teapos, consider yourself lucky - most caps on that era of Epox boards are a lot worse.

I also never saw a bad Sanyo on the Epox boards. I always wondered if those were replacement caps that Epox installed for warranty claims, because they're so out of character with the cheap brands Epox otherwise used.
A couple years ago I did find some bulged Sanyo WG caps on my Tyan server board. So that's the age thing again, these "ultra Low ESR" caps don't have the most stable chemistry.

The very small general purpose caps scattered around a motherboard rarely fail. Of course they could, but it's rare. It's the "ultra Low ESR" grade caps that are the most problematic.

So there's 2 schools of thought - replace all the caps so you know they're all good and fresh, or just replace the failure prone brands and leave the rest.
If you're taking the latter approach, then I suggest the Sanyos, the Teapos, and the small caps are the ones you can leave alone and hope for the best. But given the age, the idea of wholesale replacement has gained some validity.
You have to decide based on how you feel about doing the soldering, how risky you think it is, and your needs/expectations for the board to not have another failure.
Worst case is that a failed cap leads to other motherboard damage, such as blowing a MOSFET, which does happen sometimes. Lack of noise filtering can also stress the chipset or other ICs.

shamino wrote on 2023-08-14, 06:22:

So there's 2 schools of thought - replace all the caps so you know they're all good and fresh, or just replace the failure prone […]
Show full quote

So there's 2 schools of thought - replace all the caps so you know they're all good and fresh, or just replace the failure prone brands and leave the rest.
If you're taking the latter approach, then I suggest the Sanyos, the Teapos, and the small caps are the ones you can leave alone and hope for the best. But given the age, the idea of wholesale replacement has gained some validity.
You have to decide based on how you feel about doing the soldering, how risky you think it is, and your needs/expectations for the board to not have another failure.
Worst case is that a failed cap leads to other motherboard damage, such as blowing a MOSFET, which does happen sometimes. Lack of noise filtering can also stress the chipset or other ICs.

Many wonderful points to think about in your post, thanks!

Yeah, not a single Teapo or Sanyo on the board show any signs of being bad, but of course that doesn’t mean that they aren’t degraded or going out. I think I make sure that I have all the caps required for full recap and I start with all GSC ultra low ESR caps (LE and RE) which few have bulged (which all are RE series). If everything goes smoothly, then I’ll do Teapo RE and finally SH series, which are 470uF, but general purpose caps.

I checked the Sanyos and I need to install polys there if I change them. I just couldn’t find electrolytics with that capacitance, ESR and ripple current unlike with all GSC replacements. I might also change those straight away as those are 100% for CPU 5V VRM and pretty essential for stability.

I wonder how much cooling has affected to these caps as there definitely is variance how badly or well they have lasted? I mean for example, if you had water block on the CPU, you may get really little air flow on the caps especially around the CPU even if you have proper case ventilation. When manufacturers give lifetime estimates to their caps, they are usually like 5000 hours@105C. If temperature is significantly less, you can expect a lot more hours from the cap. And vice versa.

Just a small update what I have done so far and what my plans are for the recapping now.

I decided to start the process by removing the caps that are either bulged or of the same series as well as 2200uf GSC caps on the column just right to the IO ports, although they did look externally just fine. All these caps are of low ESR type. Desoldering went just fine, although working with this kind of board is certainly different compared to older double sided boards. I had only one via where for some reason I struggled a bit, but I managed to get the hole clear without any significant damage in the end. Rest went really easy. My main tactic was to add LOTS of fresh solder to the feet, then flux, little bit of heat to the board with hot air and finally sucking solder off with desoldering gun. Overall pretty fast job, except for some reason that one via put up a fight.

After decap, I decided to measure the 10 or so caps I removed from the board. Every bulged GSC cap was pretty much dead as I didnt get any ESR values for them and my meter showed probably the smallest capacitance value it can, that is 32pf (yeah, that is pico). Every bulged cap had this. Meter I use is this almost kit type of tool which identifies components and measures their values. It probably isn’t the most sensitive tool for measuring actually failed parts or those that are almost failed, but it does show correct values for fresh and for those that are still somewhat functional. According to it, these were as good as dead which is hardly surprising.

What was surprising instead are the four 2200uf GSC caps on the column right of the IO. Yes they did look okay externally, but I was still expecting bad results because of GSC reputation making one of the worst caps for motherboards. But they were all nicely in spec! Capacitance was around 2000uf for every one of these caps (which is well within the normal 20% tolerance) and ESR was still low. As well as being well within the specs, there was very little variance between the measured values of all four caps, which to to me means that none of these caps were actually even in the process of dying soon.

This finding also made my plan for the recap clear. I will only change those bad caps and those fully working GSC caps I removed and leave everything else as is, that is those Sanyos and every other cap, which are pretty much regular general purpose type. If I find some unstability during testing, then I’ll continue recapping, but as I wrote, I am now pretty confident that other caps on the board still have good hours left in them and there is no actual benefit of doing a full recap at this point.

My general rule is this:

- recap VRM regardless whether the caps are checking fine or not unless the caps are Sanyo/Rubycon/Panasonic/Nichicon(except 2001-2004 HM and HN)/Chemicon (except KZG)
- leave small caps (1000uF and lower) alone, as long as they're OST or any of the above mentioned brands, except Chemicon KZG as those bloat without usage, as well as the bad Nichicon HM/HN batches can fail silently short.
- leave alone any of the 100uF or lower minuscule caps, unless the series is so bad that they are known to fail - didn't have to do this on many boarda but my ABIT BE6-II and Soyo 6BA+IV do have those replaced just because Wendell and Jackcon aren't to be trusted in any forms. GSC/Evercon also applies here, and to a degree some G-Luxons also qualify here though not as much (at least not the post-Teapo merger ones, with black sleeving instead of metallic green.).

Areas you can play with capacitance are usually VRM low, as long as you know target capacitance for the VRM low.