mpe wrote on 2021-04-20, 12:14:

I was wondering how to identify bad caps that need replacing.

It is a cap? Replace it. A cap is just a couple of cents. If you start replacing them, you might as well go all the way and replace all of them.
It won't break the bank and you eliminate a possible point of failure.

Small ones like that are cheap to replace but the really problematic ones are quite expensive if you have to replace them all.

It's often cheaper and easier to just get a flawless board.
If the board is rare then you are better off replacing all elcos.

RandomStranger wrote on 2021-04-20, 12:20:

mpe wrote on 2021-04-20, 12:14:

I was wondering how to identify bad caps that need replacing.

It is a cap? Replace it. A cap is just a couple of cents. If you start replacing them, you might as well go all the way and replace all of them.
It won't break the bank and you eliminate a possible point of failure.

I doubt money is the factor that makes people wary of doing it, but time

unless one is very skilled and experienced it takes us amateurs some time to replace them and there's always a slim chance we'll make some mistake and cause more work

so if the board looks ok, works ok and isn't whistling or obviously bulging / spilling contents then why use up the time now?

if it doesnt work, then of course, there is a task to be done

I follow the EPRI template I use at work for replacing electrolytic capacitors. Most have a usable life of 10 years until they start to break down and degrade. High quality caps will last longer, but 10 years is a good rule of thumb. If you have capacitor plague era components, those were lucky to get 5 or 6 years before they popped their shell or degraded beyond their factory specifications and tolerances.

To save money, I buy capacitors in bulk from Mouser. Buying large amounts of each type you need will cut costs a lot, and you'll have enough to do several boards worth of replacements.

But yes, it helps to have the right tools to make the job go quick and painless. It's also good to get some perma-proto boards if you're new, so you can practice installing and removing capacitors a few times before working on an actual board.

I'm no professional, but having the right tools makes the job a hell of a lot easier.

A temperature controlled iron with a couple of types of tips (a fine tip will mostly be what you use, but it's worth having a choice) for fitting components, but I wouldn't even bother trying to use an iron to remove parts; for through-hole components a vacuum desoldering gun makes it a breeze, and for surface mount parts (which you may not even need if it's just cap replacement) a hot air gun.

A couple of pairs of fine, long handled tweezers, some good flux, cleaning fluid for removing flux residue, a silicon mat to work on (you will put the iron down or drop it at some point!) are all useful.

Don't, whatever you do, try to soldier (there's a pun there somewhere) on trying to remove parts with just an iron and desolder braid... you'll end up applying far more heat than needed and the chance of ruining other parts is far higher.

Miphee wrote on 2021-04-20, 12:25:

It's often cheaper and easier to just get a flawless board.
If the board is rare then you are better off replacing all elcos.

Some entire classes of board are prone to failure. KZG caps like you've shown will, inevitably, fail, as will Nichicon ultra low ESR lytics, even if they're not part of the defective batches which were sold. These were used heavily during the Pentium IV era for motherboard VRMs. Finding a clean motherboard from that era might be harder than you let on.

I think replacing everything could be one strategy. I personally prefer not to touch vintage stuff with solder unless I have to. But when troubleshooting issues, it helps if you know that there is (or was) a problem.

I was thinking about using oscilloscope to identify weird looking edges, but this is not easy to use and needs some knowledge about the circuit the capacitor is a part of.

I was wondering if anyone successfully used ESR/LCR tester on motherboard caps with some success.

megatron-uk wrote on 2021-04-20, 13:30:

A temperature controlled iron with a couple of types of tips (a fine tip will mostly be what you use, but it's worth having a choice) for fitting components, but I wouldn't even bother trying to use an iron to remove parts

I usually preheat whole board with a preheater or by blowing hot air from a distance, but I found that it is usually possible to replace caps easily by heating both leads at once with a wide tip. High powered temperature controlled station is requred. I found T12 no-name station with 19V 75W laptop power supply to be OK for most of the time (though I had to modify it to ground the tip properly)

gdjacobs wrote on 2021-04-20, 14:03:

Some entire classes of board are prone to failure. KZG caps like you've shown will, inevitably, fail, as will Nichicon ultra low ESR lytics, even if they're not part of the defective batches which were sold. These were used heavily during the Pentium IV era for motherboard VRMs. Finding a clean motherboard from that era might be harder than you let on.

It's not impossible because there are tons of boards available from that era and most of them work fine. Those boards are always cheap ($5-20) so it's pointless to invest too much time & money in fully recapping one. Of course people do whatever they want with their boards but it's just not worth the effort IMO.
A generic P4 (478-775) board is almost free nowadays but other "modern retro" boards are very cheap and common too (754, 939, AM-series).
You could buy 2 pcs of working P4 boards for the price of a proper recapping job.

Miphee wrote on 2021-04-20, 16:48:

It's not impossible because there are tons of boards available from that era and most of them work fine. Those boards are always cheap ($5-20) so it's pointless to invest too much time & money in fully recapping one.

P4 OEM boards are inexpensive. S754, S939, and server boards are worth a little more. Having said that...

My contention is you can't necessarily buy two copies of board X with the expectation that one will work if the other doesn't. I contend that it'll be more profitable to avoid boards known to use problematic cap series including NCC/UCC KZE, KZJ, KZJ, Nichicon HM and HN, and Rubycon MBZ or assume the need to replace them.

That involves a lot of different boards.

gdjacobs wrote on 2021-04-20, 18:49:

My contention is you can't necessarily buy two copies of board X with the expectation that one will work if the other doesn't. I contend that it'll be more profitable to avoid boards known to use problematic cap series including NCC/UCC KZE, KZJ, KZJ, Nichicon HM and HN, and Rubycon MBZ or assume the need to replace them.

Or you just throw them away if they are in really bad shape or buy tested boards. It's not life insurance either but it's better than untested. I have a feeling that OP wasn't talking about these boards but maybe older boards with 20-30 years old elcos on them. The caps on a S3 or S7 board are much cheaper to replace and there aren't that many of them to begin with.

mpe wrote on 2021-04-20, 12:14:

I was wondering how to identify bad caps that need replacing. I think it is somewhat easier for bigger caps which are visibly bulging or leaking. But those tend to came on later boards Pentium II+

Visual inspection is by no means a conclusive test. Caps usually just "dry out" and stop working rather than blow up, so most bad caps won't look bad at all (though you may see the leakage on the board around them).

The easiest and cheapest way to test caps for being "good" (read go/no go) is to use one of these cheapo little testers: Ebay link

If you don't mind spending some money on this as a hobby, and you think you might do it more than once, you can get yourself a desoldering gun that make removing the caps so easy that you can do it safely in 30 seconds. This is the one I have and it works quite well: amazon link

Both items are undoubtedly available in your region from other places.

Caps should generally be within 20% of value and have a reasonably low ESR which will grow with the size and value of the cap from next to zero all the way up to 30-100 ohms and sometimes higher.

If they are visibly leaking or burned they should be replaced no matter what the meter says. Sometimes they are glued on the board and people think they are leaking; glue usually won't come off with alcohol, leakage will.

Ha, and there it is again - the original Fake Cache, connected to... ...itself. You just can´t make this stuff up.

Miphee wrote on 2021-04-20, 19:00:

Or you just throw them away if they are in really bad shape or buy tested boards. It's not life insurance either but it's better than untested. I have a feeling that OP wasn't talking about these boards but maybe older boards with 20-30 years old elcos on them. The caps on a S3 or S7 board are much cheaper to replace and there aren't that many of them to begin with.

Indeed, older boards don't tend to have the same exotic polyphase regulator setups, so you're replacing far less caps. Still, I think some basic soldering skills are invaluable when dealing with older gear. Cap replacement is the simplest job available and is an inexpensive fix, all things considered.

The OP question has actually been addressed previously and there's several schools of thought. Some people straight up replace all caps for a suspect board. Others try to be more selective, either by removing and checking caps from different parts of the board or evaluating which circuits on the board will be subjected to more heat stress and more likely to fail.

The bottom line is that visual inspection is only a reliable way of finding caps which have definitely failed. Caps that look good won't necessarily function properly. You also can't reliably test caps in circuit, except for those that have failed short which happens more often with ceramic caps than wet electrolytics.

majestyk wrote on 2021-04-20, 20:11:

Ha, and there it is again - the original Fake Cache, connected to... ...itself. You just can´t make this stuff up.

Waddayamean ?
I'm not grasping the context... care to explain?

weedeewee wrote on 2021-04-20, 20:36:

Waddayamean ?
I'm not grasping the context... care to explain?

I think he means this: http://redhill.net.au/b/b-bad.html

7F20 wrote on 2021-04-20, 20:38:

I think he means this: http://redhill.net.au/b/b-bad.html

I know about the fake cache crud though fail the see the connection to this bad caps thread...
Hell, I worked in a shop for a two years that only sold pcchips mainboard, so please don't remind me of those years. 😒

weedeewee wrote on 2021-04-20, 20:52:

I know about the fake cache crud though fail the see the connection to this bad caps thread...

Pictures above (still want to hear opinions about the cap) are showing Aptron DX-9700 which features a dummy cache.

However, the DX-9700 has also COAST slot that can host a real cache. It is actually a good late 486 motherboard if you happen to own the cache module for it.

mpe wrote on 2021-04-20, 21:24:

Pictures above (still want to hear opinions about the cap) are showing Aptron DX-9700 which features a dummy cache.

However, the DX-9700 has also COAST slot that can host a real cache. It is actually a good late 486 motherboard if you happen to own the cache module for it.

Thanks , Now I can quietly go to bed and rest peacefully 😀