The big caps around the cpu socket, VRM's and memory sockets are the most important as they fail easier. I would replace these first:
Caps of Asus TUV3X (Socket 370): 1500uF/6.3v and 1000uF/6.3v
Caps of Gigabyte GA-7VT600 1394 (Socket 462): 3300Uf/6.3V , 1500Uf/6.3V and 1200uF/6.3v

And what about the smaller ones?

Since it's in good condition, can I leave it there?

I was really thinking in replace the caps that you mentioned, but I'll be happy if I can leave the smaller ones......

Thank you!!!!!

I thinks these 10uF, 100uF, 470uF mostly for power rails filter and they rarely failed, unless your mainboard suffer catastrophe power surge or the capacitor brand is known to failed then you need to replace them.

The standard practice (of professional techs) for a "full" mobo recap (which is not really full but that's what it's called) is to...
.. replace all that are:
1: Physically 6mm diameter or larger
2: Capacitance 470uf or larger.
As long as you use the correct caps and assuming the only problem is caps that will fix 99+% of motherboards.
Most of those caps are devoted to reducing ripple directly from the PSU, VRM or larger POL regulators.
They have the "high stress jobs" and that standard range of caps should get all of them.
.
The small 5mm or less and 220uF or less wet-lytics caps do fail too but they have different jobs and see a lot less stress so they don't fail very often.
(Notable exceptions are boards using OST (brand) and early Intel made P4 boards that used 85°C Nichicons for the small caps. Both of those fail often and can fail en masse.)
470uF are usually on the output of a good sized POL regulator. (An LDO or a MOSFET.)
330uF caps and 470uF @ 5mm are grey areas. They may be used for high or low stress jobs.
The 330uF and larger 220uF were common in onboard sound but they also show up as surge suppressors for I/O ports such as USB, LAN and LPT.
.
.
On motherboards and in PSUs 6.3v and 10v can be used interchangeably. (That gives you more choices for replacement caps.)
On those boards 9v is not used and the next voltage is 12v so you know both 6.3v and 10v are on 5v or less.

You must check by the CPU Vcc for the socket but in most cases the CPU side of the VRM is a low voltage so in many cases 4v and even 2.5v caps are fine there.

For wet-lytics use low ESR Japanese caps. (There are many discussions of why in other threads already.)
If you don't know caps you have a better than 50% chance of getting counterfeits if you buy at rip-off bay.
Some of the higher end ranges are no longer in production by anyone at all. Polymer is the only option left for those.
(Range meaning the Ripple/ESR ratings relative to the Capacitance.)
Suncon (formerly Sanyo) still makes their WG series which is equivalent to Rubycon MBZ and the only still in production wet-lytic cap I know of in that range.
Unfortunately Sanyo/Suncon are very popular with the counterfeiters and some of the fakes are very good visually.
Fake Nichicon and Chemicon are also popular.
Chemicon KZG and KZJ have issues (high failure rates) even though they are Japanese.

Focusing on uF and Volts is a newbie mistake that everyone makes at first.
You need to start your replacements search after looking up the specs for the original caps.

The ESR and Ripple specs matter the more than the uF (Capacitance) for this capacitor application.
ESR has to do with their primary job and should not be 'fudged' to a higher value.
The Ripple rating has to do with their primary job and should not be 'fudged' to a lower value.
Capacitance has to do with their secondary job and can be 'fudged' [up] by one standard value without issue.
Fudging Capacitance down is a bad idea because Capacitance goes down as caps age.
-- VRM wet-lytic caps did not have those large uF values because they need all that uF in the VRM.
-- In order to get the very high Ripple and very low ESR characteristics they wanted in VRM they had to use physically large caps (and/or more caps) and all that uF came as part of the deal. The Ripple/ESR ratings were the driver, the uF was incidental.
(You can see this by evaluating 'transitional' boards meaning boards where the early revs had wet-lytics and later revs had polymers.)
~ NOT Rules, just general observations.~
3300uF/2200uF were commonly replaced by 820 or 680 uF Polys.
1800uF were commonly replaced by 680-560 uF Polys.
1500uF and less were commonly replaced by 560-470 uF Polys.
The "high side" (PSU side/16v caps side) of VRMs sometimes got polys with as little as 4x330uF or 3x470uF.
- The point is that (pertaining to caps that filter PSU/MOSFET/LDO ripple) the Ripple and ESR are the more critical design considerations. Capacitance matters but it is secondary.

If you go the polymer route.
Pay attention to the Ripple and ESR ratings. It's still not all about Capacitance.
They make a wide range of poly caps now. Not all are rated better than the better low ESR wet-lytics.
The brand doesn't matter much. I've heard of no brand that has issues with their polymers.**
[** Not counting early Fujitsu Functional Polymers (now Nichicon "FP Caps"). Those are not solid, but rather a hybrid between wet-lytic and solid polymer. The problems were only with some of the very early black-yellow versions made before about 2001. The later ones (even the later Black/Yellow) are fine.]
- Out of habit I still prefer to buy Japanese brands of polys when I can find what I need.
As long as the ESR and Ripple ratings are as good or better than the original cap you can go to a smaller uF.
How much smaller is a judgement call. Look at what is used on newer boards as a guide.
.

Good info PCBONEZ but I do have an issue with this part:

PCBONEZ wrote:

~ NOT Rules, just general observations.~ 3300uF/2200uF were commonly replaced by 820 or 680 uF Polys. 1800uF were commonly repla […]
Show full quote

~ NOT Rules, just general observations.~
3300uF/2200uF were commonly replaced by 820 or 680 uF Polys.
1800uF were commonly replaced by 680-560 uF Polys.
1500uF and less were commonly replaced by 560-470 uF Polys.

An older motherboard or PSU was designed with specific capacitors for a reason, mostly due to how the original silicon chips/MOSFETs/etc worked at that time. Replacing a 2200uF in a 2003 PSU with 820uF is a very bad idea unless you also replace the silicon with a newer replacement part.
Just my opinion...

Horun wrote:

Good info PCBONEZ but I do have an issue with this part: […]
Show full quote

Good info PCBONEZ but I do have an issue with this part:

PCBONEZ wrote:

~ NOT Rules, just general observations.~ 3300uF/2200uF were commonly replaced by 820 or 680 uF Polys. 1800uF were commonly repla […]
Show full quote

~ NOT Rules, just general observations.~
3300uF/2200uF were commonly replaced by 820 or 680 uF Polys.
1800uF were commonly replaced by 680-560 uF Polys.
1500uF and less were commonly replaced by 560-470 uF Polys.

An older motherboard or PSU was designed with specific capacitors for a reason, mostly due to how the original silicon chips/MOSFETs/etc worked at that time. Replacing a 2200uF in a 2003 PSU with 820uF is a very bad idea unless you also replace the silicon with a newer replacement part.
Just my opinion...

Okay.
I've been an Electronics tech since 1981. My specialty was control systems for nuclear plants until late in the 90's when I became a QA Inspector. Just over 20 years ago I added PC tech to that (on the side) and after retiring from my first career (and after trying some other things) I spent 10+ years repairing motherboards at component level as my primary occupation.
I tend to use wet-lytics to replace wet-lytics (because I have about 20,000 left over) but I have done maybe 75-100 poly upgrades for people and it works fine.
What I said wasn't an opinion.
It's okay.
What you are thinking is what everybody (I did too) thinks before they sit down and actually study what is going on.
That is partly because electronics courses give ESR like 5 minutes or they don't address it at all.
The books for the school I went through are 24 volumes and they gave ESR one short paragraph in the entire set.

The simplified formula for ripple on a capacitor:

Vrip = I/F*C

Where:
Vrip is peak to peak ripple in Volts.
I is the current in Amps.
F the frequency in Hz
C the capacitance in Farads.

ATX PSUs are 'allowed' to output up to 120mV of P-P ripple on 12v.
[Vrip = 0.120 V]
SMPS & VRM Ripple is ~100kHz (This is why this type of capacitor is rated at 100kHz.)
[F = 100000 Hz]
A socket 478 board I have handy has 3x1500uF Sanyo WG on 12v VRM-In
http://www.paullinebarger.net/DS/Sanyo/Sanyo% … WG%20Series.pdf
They are rated for 2550 mA each. 2550 x 3 = 7650 mA = 7.65 A
[I = 7.65 A]

Rearranging the equation to find C. (The C that is ideal for 100kHz.)

Vrip/I/F = C (in Farads)

0.120/7.65/100000 = 0.000000156862745 Farads = 0.156863 uF
The ideal Capacitance to protect the ICs from 100kHz Ripple is a fraction of 1 uF.
(And that's for the sum of all the capacitors on that rail.)

So I will say it again.
The primary goal (to protect the IC's and prevent data corruption) when selecting caps is high Ripple and low ESR.
The large capacitance's you see are incidental. If choosing wet-lytics to meet the Ripple/ESR needs you have to use physically large caps or many.
Although solid poly were already around by socket 8 they were still too expensive to use outside of servers and workstations and most of those used wet-lytics too.

Capacitance does help with the secondary job. Smoothing DC transients. But wet-lytic boards have excess uF for that purpose as well.
That's easier to see. Look at VRM-in (16v caps) on skt 775 boards.
Poly and wet-lytic boards face the same degree of DC transients on the same socket.
Assuming 3 caps, wet-lytic boards usually have 1500-1800 uF there. Poly boards tend to have 470uF.
Clearly 1410uF is plenty to handle DC transients on the main 12v rail. The surplus capacitance is about getting lower ESR for the primary job.
The 3x 470uF Poly there is very common all the way from socket 478 into socket 1150.
.

I really need to say thanks for all of you!

PCBONEZ really explained it all!

All my motherboard are working.

For prevention, I'll recap, like PCBONEZ explained.

Maybe, I'll consider a more rigorous "full" recap.

Replacing everything above 220uF sounds very good to me, instead of the mentioned 470uF.

I'm considering this, cause some of my motherboards have OST caps.

My favourite board is a Soyo Dragon Plus (462), and works with around 55 OST caps!!!

Finally, I'm considering to leave everything below 100uF there!

This is my idea, as a newbie, right now.

Thanks you again, mates!!!!!

If that's one of those boards were Soyo got weird and ordered all the caps with silver sleeves I'd just replace all of them.
Those were horrible caps.
One of those Soyo-OST-Silver boards was one of the few boards I've ever seen where bad caps caused hard drive corruption.
The primary IDE drive would corrupt whatever drive you put on it in about a day. Change drives it would just do it to the next one too.
The secondary port worked fine.
I ended up going cap by cap because it made no sense and I found a single bad 5 or 6mm cap between the chipset and the IDE ports.
I should have traced the circuit to see exactly what it did but I had to get the board back to the owner.

If you don't want to do all that then do as I said earlier, replace:
1: Physically 6mm diameter or larger.
2: Capacitance 470uf or larger.
Then keep an eye out for signs of the little ones getting flaky. (Any kind of inexplicable weird behavior or instability.)

There is also a 'thing' called a "Partial Poly Mod".
To do one of those you upgrade all the VRM and optionally the RAM caps (the most stressed caps) to Polymer but leave the rest of the board wet-lytics.
When polys were still expensive that was a popular way to go.
The way the industry is going (all the appropriate wet-lytics going out of production) that may be the only way to get it done in a few years.
.
Didn't the Asus TUV4x come with mostly Rubycons?
Many of their boards around that time did and I don't think that one was considered low end.
.

Caps of Gigabyte GA-7VT600 1394 (Socket 462):
3300Uf/6.3V 4
[...]

You'd have to check on your particular board, but I'm guessing you have several unpopulated locations near those 3300uF caps around the CPU. If so, use your meter and see if those locations are in parallel with the 3300uF caps.
If they are, then you have the option of using a larger number of smaller caps, instead of needing to find 3300uF caps. This also means the ESR on the individual caps doesn't need to be as low, since it gets divided by a larger number of total caps (see parallel resistance formula).

Gigabyte did that on several of their boards around that time. The board designer laid them out to support more caps but they got manufactured with a few 3300uF ones instead.

shamino wrote:

You'd have to check on your particular board, but I'm guessing you have several unpopulated locations near those 3300uF caps aro […]
Show full quote

Caps of Gigabyte GA-7VT600 1394 (Socket 462):
3300Uf/6.3V 4
[...]

You'd have to check on your particular board, but I'm guessing you have several unpopulated locations near those 3300uF caps around the CPU. If so, use your meter and see if those locations are in parallel with the 3300uF caps.
If they are, then you have the option of using a larger number of smaller caps, instead of needing to find 3300uF caps. This also means the ESR on the individual caps doesn't need to be as low, since it gets divided by a larger number of total caps (see parallel resistance formula).

Gigabyte did that on several of their boards around that time. The board designer laid them out to support more caps but they got manufactured with a few 3300uF ones instead.

Very true.

There is a gotcha when you use multiple caps in parallel.
(Most rails do that though the board so the caps are in parallel but aren't always physically located near each other.)
Most people decide what caps to use in parallel based on their uF and as per the previous discussion that is a mistake.
The caps should be chosen by their ESR and Ripple ratings, not uF.

In parallel the ESR and Ripple work exactly like Resistance and Current in a bank of resistors in parallel.
You can even use the same equations (and online calculators) by using ESR in place of Resistance and Ripple in place of Current.
----
There are many online calculators, I like these two simple ones.
http://www.1728.org/resistrs.htm
http://www.ohmslawcalculator.com/ohms-law-calculator
----
The gotcha is that in parallel more Ripple (Current) goes through the part(s) with lower ESR (Resistance) values.
If you mix different values of ESR/Ripple in the same bank then one (or a few caps) may be handling more Ripple than they are rated for even though the total for the whole bank is fine.
.
This is why if you have a rail with several caps in parallel replacing only a few (without doing the math for each cap) is a bad idea.
Same-same with using caps with vastly different specs in the same rail. Gotta do that boring math.
.

PCBONEZ wrote:

If that's one of those boards were Soyo got weird and ordered all the caps with silver sleeves I'd just replace all of them. Tho […]
Show full quote

If that's one of those boards were Soyo got weird and ordered all the caps with silver sleeves I'd just replace all of them.
Those were horrible caps.
One of those Soyo-OST-Silver boards was one of the few boards I've ever seen where bad caps caused hard drive corruption.
The primary IDE drive would corrupt whatever drive you put on it in about a day. Change drives it would just do it to the next one too.
The secondary port worked fine.
I ended up going cap by cap because it made no sense and I found a single bad 5 or 6mm cap between the chipset and the IDE ports.
I should have traced the circuit to see exactly what it did but I had to get the board back to the owner.

If you don't want to do all that then do as I said earlier, replace:
1: Physically 6mm diameter or larger.
2: Capacitance 470uf or larger.
Then keep an eye out for signs of the little ones getting flaky. (Any kind of inexplicable weird behavior or instability.)

There is also a 'thing' called a "Partial Poly Mod".
To do one of those you upgrade all the VRM and optionally the RAM caps (the most stressed caps) to Polymer but leave the rest of the board wet-lytics.
When polys were still expensive that was a popular way to go.
The way the industry is going (all the appropriate wet-lytics going out of production) that may be the only way to get it done in a few years.
.
Didn't the Asus TUV4x come with mostly Rubycons?
Many of their boards around that time did and I don't think that one was considered low end.
.

Lower ESR will be hard to find!!

THe majority of branded caps here is made by EPCOS, which is hard to find datasheet.

I'm thinking in buying caps a litle above the specs, cause usually it means lower ESR.

Is it true?

And yes, TUV4X caps come with Nichicon and Rubycon caps, but I have another boards with silver OSTs, for example.

Do you know what's funny?

The Soyo Daron Plus, which is my favourite card, have around 50 silver OST caps.....

And x16 22uF/25v, which is very small and a little hard to replace!!

Do you have part #'s for good replacements from say Mouser? I also have a TUV4X and I was about to recap it with some Panasonic caps I bought from mouser. The board has Nichicon caps on it as of now. I was only going to replace the 1500uf and 1000uf caps.

I don't want to destroy/ruin this motherboard or make it further unstable because that was the reason I wanted to recap it to begin with. The AGP has gotten weird at 2X and also the ram won't run at 2T anymore.

gabimor wrote:

THe majority of branded caps here is made by EPCOS, which is hard to find datasheet.

You can give this a look:
http://www.paullinebarger.net/DS/Epcos/

gabimor wrote:

I'm thinking in buying caps a litle above the specs, cause usually it means lower ESR.

That's not necessarily true. There are some rules of thumb when comparing caps in the same series, but different manufacturers will vary electrolyte formulation and physical construction making inferring specs based on capacitance, case size, and voltage an unreliable proposition. Try to locate a datasheet whenever possible.

gabimor wrote:

And x16 22uF/25v, which is very small and a little hard to replace!!

If it's not RoHS compliant with massive ground planes it's not that hard. Find yourself a nice, small stainless steel pin to clear those through holes.

RoyBatty wrote:

Do you have part #'s for good replacements from say Mouser? I also have a TUV4X and I was about to recap it with some Panasonic caps I bought from mouser. The board has Nichicon caps on it as of now. I was only going to replace the 1500uf and 1000uf caps.

I don't want to destroy/ruin this motherboard or make it further unstable because that was the reason I wanted to recap it to begin with. The AGP has gotten weird at 2X and also the ram won't run at 2T anymore.

It's a bit more complex than that.
Just saying Nichicon and the uF isn't enough info. Nichicon has (or at least had) dozens of different series of low ESR caps.

To look up correct replacements I/you/whoever needs to know exactly what the original ones are. The format is thus:
~ Make ~ Series ~ uF ~ Volts ~ Diameter ~ Length ~

The first step in doing a recap is to make a cap-map that shows all the caps you will replace and their polarity.
It does not need to be pretty, it just has to have all the information on it.
You DO NOT want to have caps out and get interrupted to come back later and not know which caps were where and which side was +.
When you make the cap-map also make a list (a BOM) of all the caps you'll need including those six bits of info for each kind of cap.
.
Another reason for a cap-map is because more often than you would expect some of the polarity markings printed on the board are wrong.
It's better to look at the polarity of physical part before you pull it out.
.

gabimor wrote:

Lower ESR will be hard to find!!

You are looking for the same or lower.
The point is to not go higher.
With all the go-to series going out of production it is getting harder to find wet-lytic replacements.

We over at badcaps.net anticipated this would come years ago and that's why we started experimenting with poly mods and sharing our findings.
That's also how this came about.http://www.paullinebarger.net/DS/
Barger did most of it without telling anyone then afterwords everyone contributed missing datasheets from their own collections.
Never understood why he used his wife's name. Guess he's not interested in fame.

gabimor wrote:

I'm thinking in buying caps a little above the specs, cause usually it means lower ESR.

Is it true?

Nope. Not true. - That's a myth that was proven wrong like 15 years ago.
Two caps can be the same physical size and have the same uF but have ESR/Ripple that are orders of magnitude different.
The chemical makeup of the electrolytes can be different.
They etch the aluminum foils differently to get different grain structures in the oxide layer.
They can use different metals for the leads. (Which affects ESR).
The thickness of the paper and foils can be different.

gabimor wrote:

And x16 22uF/25v, which is very small and a little hard to replace!!

Not sure what you mean by hard to replace.
Finding replacements for the small ones is usually pretty easy but sometimes those little buggers are a PITA to get out.
.

Thank you for all the info you provided.

This will be hardest than I thought.

Oldest motherboards I have (all working normally. I don't see bloated caps):

Socket 370
- x4 Asus TUV4X, with Nichicon and Rubycon caps;
- x1 ECS P6S5AT, with crappy caps;

Socket 462
- x1 Soyo SY-KT600 Dragon Plus (black) with silver OST Caps;
- x2 Gigabyte GA-7N400, which I bought this year, selled as new;
- x2 Gigabyte GA-7VT600 1394

Socket 478
- x1 Intel D875PBZ;
- x 1 Gigabyte GA-8IPE1000 Pro;
- x1 MSI PT880 Neo;
- I'm buying a Gigabyte GA-8TRS350MT and a Intel D865GLC.

Socket 939
- Gigabyte K8N-SLI
- Abit KN8 Ultra
- DFI NF4 Lanparty Ultra-D
- I'll buy another...

My two favorite retro boards are a Soyo SY-KT600 Dragon Plus (black) with silver OST Caps and a ECS P6S5AT. The boards are working, but since the caps are bad, I'll try to recap soon. The Soyo have a ton of very small caps, which is hard to work. Maybe, I'll leave this small caps in place.

Boards like Asus TUV4X and Intel D875PBZ, which work with really good caps, at this moment, since everything is working, I'm considering to leave the caps there.

I'll think about the other boards.

My main objective is to have all my builds working till the end of my life.

That's why I have around 4 boards in each build. I'm using one of the boards in each build and the others, I turn on for about 1 hour every year.

I'll buy another socket 939 board. I have one Asus A8N-SLI but it's not working.

I have many other boards, from newer sockets, but the caps are all solid and the quality of the equipment is really good. For this modern systems, I have 3 boards of the same socket for each build.

Later, I'll open a new topic, to show the list of retro hardware I have. We'll talk about conservation of them.

Thank you!!!!

PCBONEZ wrote:

It's a bit more complex than that. Just saying Nichicon and the uF isn't enough info. Nichicon has (or at least had) dozens of d […]
Show full quote

It's a bit more complex than that.
Just saying Nichicon and the uF isn't enough info. Nichicon has (or at least had) dozens of different series of low ESR caps.

To look up correct replacements I/you/whoever needs to know exactly what the original ones are. The format is thus:
~ Make ~ Series ~ uF ~ Volts ~ Diameter ~ Length ~

The first step in doing a recap is to make a cap-map that shows all the caps you will replace and their polarity.
It does not need to be pretty, it just has to have all the information on it.
You DO NOT want to have caps out and get interrupted to come back later and not know which caps were where and which side was +.
When you make the cap-map also make a list (a BOM) of all the caps you'll need including those six bits of info for each kind of cap.
.
Another reason for a cap-map is because more often than you would expect some of the polarity markings printed on the board are wrong.
It's better to look at the polarity of physical part before you pull it out.
.

I have a lot of electronics experience, I was a repairman for 25 years, but I worked on audio equipment not digital stuff. I am aware of proper procedures, and I do them one at a time and note polarity on the board with a sharpie before removal, however the ground plane is pretty obvious.

There's no series on these caps, like there isn't on most (cheap) caps. Just max temperature operating range, value, voltage and polarity. I already checked lead spacing before I purchased to make sure I got the same thing. The 1500uf caps are shorter dimension wise, even if they are the same size by width. I'm not sure if that's a difference in manufacturing technique or if the taller caps are important. I think I'll shelve the idea until I learn more about it and try to find what series these bloody caps are.

RoyBatty wrote:

PCBONEZ wrote:

It's a bit more complex than that. Just saying Nichicon and the uF isn't enough info. Nichicon has (or at least had) dozens of d […]
Show full quote

It's a bit more complex than that.
Just saying Nichicon and the uF isn't enough info. Nichicon has (or at least had) dozens of different series of low ESR caps.

To look up correct replacements I/you/whoever needs to know exactly what the original ones are. The format is thus:
~ Make ~ Series ~ uF ~ Volts ~ Diameter ~ Length ~

The first step in doing a recap is to make a cap-map that shows all the caps you will replace and their polarity.
It does not need to be pretty, it just has to have all the information on it.
You DO NOT want to have caps out and get interrupted to come back later and not know which caps were where and which side was +.
When you make the cap-map also make a list (a BOM) of all the caps you'll need including those six bits of info for each kind of cap.
.
Another reason for a cap-map is because more often than you would expect some of the polarity markings printed on the board are wrong.
It's better to look at the polarity of physical part before you pull it out.
.

I have a lot of electronics experience, I was a repairman for 25 years, but I worked on audio equipment not digital stuff. I am aware of proper procedures, and I do them one at a time and note polarity on the board with a sharpie before removal, however the ground plane is pretty obvious.

There's no series on these caps, like there isn't on most (cheap) caps. Just max temperature operating range, value, voltage and polarity. I already checked lead spacing before I purchased to make sure I got the same thing. The 1500uf caps are shorter dimension wise, even if they are the same size by width. I'm not sure if that's a difference in manufacturing technique or if the taller caps are important. I think I'll shelve the idea until I learn more about it and try to find what series these bloody caps are.

I don't know other people's skill levels and later readers might not be at your level anyway so I default to writing as if the reader is completely new to it.

Now I'm confused.
You said earlier they are Nichicon and gave no series but now you're saying you don't know what they are.
There are alternate ways to idendify most crap brands.
If you post some photos of them then I (or someone) may have seen them before.
For example Tayeh and GSC both made caps with no series markings. There could be others.
GSC used the can color as the series indicator when they weren't marking the caps.
Tayeh are probably a rebranded OEM version of Teapo but there has never been a published datasheet or even a website.
Tayeh used Teapo colors (and sometimes vents) and checking with an ESR meter the specs match certain Teapo so that's a way to work around those.

If all else fails you go by what is usually found in that "spot" on similar boards that do have caps with the proper markings.
You haven't given where they are on the board so I can't go that route either without more information.
.

PCBONEZ wrote:

The small 5mm or less and 220uF or less wet-lytics caps do fail too but they have different jobs and see a lot less stress so they don't fail very often.
(Notable exceptions are boards using OST (brand) and early Intel made P4 boards that used 85°C Nichicons for the small caps. Both of those fail often and can fail en masse.)

Asus liked to use 4mm G-Luxon caps in their audio circuits on some old boards. These should be replaced. They do fail and audio stops working.

The brand doesn't matter much. I've heard of no brand that has issues with their polymers.**

Not all polymer caps are alike. The quality of the polymer varies widely from brand to brand... I *think* Taiwanese polymers are ok (Apaq, Capxon, Lelon) but I wouldn't completely trust anything other than Japanese polymers. Chinese polymers should be avoided at all cost. A _good_ polymer cap has a moist, intact polymer layer inside, even after many years. A poor quality polymer cap will be crusty after a short time.

[** Not counting early Fujitsu Functional Polymers (now Nichicon "FP Caps"). Those are not solid, but rather a hybrid between wet-lytic and solid polymer. The problems were only with some of the very early black-yellow versions made before about 2001. The later ones (even the later Black/Yellow) are fine.]

I know that this is the consensus, but it is inaccurate. Even the later yellow/black ones do not age well. They should be replaced on sight, regardless of their manufacturing date.