VOGONS


Capacitor replacement

Topic actions

First post, by kanecvr

User metadata
Rank Oldbie
Rank
Oldbie

I usually hand pick the caps depending on use. I go with 10v caps for the CPU (16v for socket A and socket 370 machines) and 10v for the caps around the AGP slot as well. For the rest I use identical spec ones (usually 6.3v). My favorites by far are Panasonic low-esr caps. I have boards that went from unstable to rock solid after recapping with these, even tough the ones installed were perfectly fine looking (not so much when removed and tested).

For example my GIGABYTE GA-7VRXP KT333 voodoo 4 rig had 4 bloated 3300uf 6.3v caps - I replaced all of those with 3300uf 16v panasonic low ESR long life caps, even added two more since there were spots on the board where an extra 4 capacitors could be installed and it can now run CPUs at 166+MHz FSB w/o issues.

On socket 370 boards you usually find 1000 or 1500uf caps, or a combination of both. I used 10v 1000 and 1500uf caps for my Abit ST6 and it can now run my tualeron at over 1333Mhz stable. The only downside is panasonic low-esr caps are pretty expensive...

Reply 1 of 54, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie

Here's a very interesting and venerable 11-year old (@ 2015) post from Topcat (the badcaps.net boss) on how playing with capacitance influenced overclocking:

Abit VP6 Capacitor Mods and Overclocking Hints.

Let the air flow!

Reply 2 of 54, by kanecvr

User metadata
Rank Oldbie
Rank
Oldbie

I like to stick to stock capacitance myself. I just up voltage so the caps will be less "stressed".

Reply 3 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
TELVM wrote:

Here's a very interesting and venerable 11-year old (@ 2015) post from Topcat (the badcaps.net boss) on how playing with capacitance influenced overclocking:

Abit VP6 Capacitor Mods and Overclocking Hints.

Perhaps you should reread his conclusion: "Was it worth the aggrivation?! SHIT NO! Not for a measly 3MHz"
.
I'm surprised TC would leave anything that ignorant posted. It's an excellent example of an old myth.
What actually improved the board's performance was that the 2200uF had lower ESR than the 1500uF.
The capacitance didn't have anything to do with it.

In other words when he raised the capacitance he also lowered the ESR. It was the lower ESR that helped.

I saw that mistake made many times in many forums in the mid 2000's.

That was posted in 2004.
I know for a fact TC has learned better since but he never cleans the old rubbish out of his forum.
I've suggested he do that multiple times but he just won't.
For example BCN still lists KZG as good caps in several places even though it became well known later that they aren't.
As a result you need to take anything you read at BCN with a grain of salt. Particularly if it's old.
.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 4 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
kanecvr wrote:

I usually hand pick the caps depending on use. I go with 10v caps for the CPU (16v for socket A and socket 370 machines) and 10v for the caps around the AGP slot as well. For the rest I use identical spec ones (usually 6.3v).

While I applaud hand selecting the caps, you seem to be going about it wrong.
The voltage rating has very little to do with anything regarding a capacitor's performance or operating characteristics or anything else.
The voltage rating is just the maximum voltage limit for the circuit they are used in.

Your approach can lead to trouble.
For example it is very common to have a 12 circuit (a 12v powered VR) associated with AGP slots so using 10v caps just because it's near an AGP can be a problem.

If the original cap is 16v it is never a good idea to replace it with a 10v or less cap UNLESS you have actually measured the voltage on that cap under all operating conditions.

It's not the same for original 10v caps ON A MOTHERBOARD. On a motherboard if it's not 12v (which need a 16v or higher voltage cap) then it is 5v or less so 6.3v and 10v caps can be used interchangeably.
That does not work in other equipments because they may use 9v somewhere - but 9v is not used on motherboards.
.

Last edited by PCBONEZ on 2015-12-13, 23:03. Edited 2 times in total.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 5 of 54, by kanecvr

User metadata
Rank Oldbie
Rank
Oldbie

For a proper quality fix I use Panasonic (first choice) Nichicon or Rubicon caps. Panasonic caps are the most expensive by far - the 16v 3300uf low-esr ones I use on socket a / 478 boards are 1 to 1.4 euro / cap - so if you have 10 cap to replace it might cost you more then you payed for the board you're trying to fix... Nichicons are quite a bit cheaper but I usually can't find the size / voltage I need, and Rubicons are getting harder to find.

Samxon cap are a decent cheap alternative. These do well up to 4700uf and don't mind high temperatures as much as the other cheap brands. They are hit an miss tough. The green / gold or green / silver ones usually suck bad, but the black / gold and black / silver ones are pretty solid. You might also be able to get the black / gold ones really cheap - like 10-20 euro cents, but make sure you get the black ones. More often then not I order the black ones and get sent the shitty greens.

Here is an example of a good Samxon cap: http://www.adelaida.ro/c-2200uf-16gf-low-esr- … -12.5x20mm.html - I've seen these things still going strong on socket 7 and socket 370 boards after years of abuse. In contrast the green ones are almost always bloated.

These are my favorites to use for mainboards -> Panasonic: http://www.adelaida.ro/c-2200uf-16v-condensat … -12.5x25mm.html and for power supplies -> Nichicon: http://www.adelaida.ro/2200uf-35v-condensator … c-nichicon.html

PCBONEZ wrote:
While I applaud hand selecting the caps, you seem to be going about it wrong. The voltage rating has very little to do with anyt […]
Show full quote

While I applaud hand selecting the caps, you seem to be going about it wrong.
The voltage rating has very little to do with anything regarding a capacitor's performance, lifespan, operating characteristics or anything else.
The voltage rating is just the maximum limit for the circuit they are used in.

Your approach can lead to trouble.
For example it is very common to have a 12 circuit (a 12v powered VR) associated with AGP slots so using 10v caps just because it's near an AGP can be a problem.

If the original cap is 16v it is never a good idea to replace it with a 10v or less cap UNLESS you have actually measured the voltage on that cap under all operating conditions.

It's not the same for original 10v caps ON A MOTHERBOARD. On a motherboard if it's not 12v (which need a 16v or higher voltage cap) then it is 5v or less so 6.3v and 10v caps can be used interchangeably.
That does not work in other equipments because they may use 9v somewhere - but 9v is not used on motherboards.
.

You misunderstand. I use HIGHER VOLTAGE RATED caps. For the AGP slot, I found these range from 6.3 for signaling to 10v for power (I've seen quite a few boards use 10v caps on a 12v line). The 10v ones are bloated 90% of the time. I replace the 6.3v ones with 10v ones and the 10v ones with 16v. 50% of the time the 6.3v ones are also bloated. I guess the caps were so bad the 6v rating was a huge overstatement since they could not handle 3.3v over a long period of time. Not trusting voltage ratings myself, I use more expensive higher rated caps for most jobs.

Here's another pertinent example - > last week my sister was bugging me that her laptop (a dell precision M6700) suddenly started BSODing in Fallout 4. I went trough the usual routine - reinstall windows, drivers - problem persisted. I then removed the MXM R9 m290x and put the fireGL card it originally came with - same problem. Just before I was ready to give up, I noticed some black gunk near two 2.5v "polymer" aluminum smd capacitors just to the left of the CPU socket. "that can't be electrolyte, those are polymer caps" I tough - but to be thought I took one off the board and measured it. It had 40% of rated capacitance... turns out hey were SMD aluminum electrolytic capacitor, not polymer caps - so I replaced them both and the problem went away. This time i used 6.3v parts with the exact same capacitance. I was afraid to use real polymer caps so I used panasonic SMD aluminum electrolytic caps (220uf or 470uf, can't remember exact capacitance).

Last edited by kanecvr on 2015-12-13, 23:19. Edited 1 time in total.

Reply 6 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie

Rubicon are counterfeit Rubycon.

You need to go by the series not the color.
Within a brand the same colors get used for many series and their ESRs vary widely.

Yes, Samxon does have several series that are horrible, but again, you need to go by the series not the color.

Never been particularly impressed with Samxon.
They are a bit better than run of the mill Ch/Tw caps but I wouldn't call them a premium brand.
I only use Samxon if there is no alternative and that's only happened when I've needed 3300uF 16v in 10mm for PSUs.

Last edited by PCBONEZ on 2015-12-13, 23:27. Edited 1 time in total.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 7 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
kanecvr wrote:

(I've seen quite a few boards use 10v caps on a 12v line).

The only time you would ever see that is if some idget before you has recapped the board.
.
If you truly understood capacitors you would know why that's a VERY bad idea.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 8 of 54, by kanecvr

User metadata
Rank Oldbie
Rank
Oldbie

Where exactly did I say samxon are premium caps? I just said they are better then most cheap capacitors, and that's only if you can find a certain type of samxon caps. Only to be used for short periods of time and in emergencies.

Oh and I also test them for capacitance even when bought new. Sometimes they're 20-30% off spec (happens for 1 in 10, but still).

Consider I have a few friends bitten by the retro bug who despite my warnings refuse to spend 10-15 euro on quality capacitors but still want to get a part redone - so I'm forced to use samxon witch I can get for less then half that.

The rubicon thing was just a misspell on my part. Just took one out of my new caps box and it reads "Rubycon", but for some reason I seem to insist on spelling it with an "i" 😜

PCBONEZ wrote:
The only time you would ever see that is if some idget before you has recapped the board. . If you truly understood capacitors y […]
Show full quote
kanecvr wrote:

(I've seen quite a few boards use 10v caps on a 12v line).

The only time you would ever see that is if some idget before you has recapped the board.
.
If you truly understood capacitors you would know why that's a VERY bad idea.
.

We're having a serious communication breakdown here. Let me spell it out as clear as I can:
- even tough I'm a doctor, I know how capacitors work. My hobby, time and monetary concerns have forced me to do extensive research on stuff like this. I might not know exactly how a more complicated active electronic component like a n channel mosfet gate works, but I'm fine with the basic stuff.
- some manufacturers deliberately use lower voltage capacitors in places even tough "it's a VERY bad idea" to pinch pennies. I've seen this on PCChips, Lucky Tech, ECS and Matsonic boards. Even Asus is guilty of doing this, just look at the A7V600.
- there is no harm in using capacitors with a higher voltage rating in place of lower rated ones AS LONG AS THEY HAVE THE SAME CAPACITANCE VALUE. In some cases it might even save you from the situation presented above. That voltage rating tells you the maximum voltage that can pass trough said capacitor, so of course using a 10v rated cap on a 12v line is a stupid idea. Implying I don't understand that bit of common sense is basically calling me an idiot, and I don't remember insulting you at any point.

Last edited by kanecvr on 2015-12-13, 23:44. Edited 2 times in total.

Reply 9 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
kanecvr wrote:

Oh and I also test them for capacitance even when bought new. Sometimes they're 20-30% off spec (happens for 1 in 10, but still).

This happens when the oxide layer erodes away in storage.
There is a process that restores the layer call "reforming". It will bring them back into spec. It is not difficult to do.
That they can do that is why in the old days Aluminum Electrolytic caps were called "Self Healing" capacitors.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 10 of 54, by kanecvr

User metadata
Rank Oldbie
Rank
Oldbie
PCBONEZ wrote:
This happens when the oxide layer erodes away in storage. There is a process that restores the layer call "reforming". It will b […]
Show full quote
kanecvr wrote:

Oh and I also test them for capacitance even when bought new. Sometimes they're 20-30% off spec (happens for 1 in 10, but still).

This happens when the oxide layer erodes away in storage.
There is a process that restores the layer call "reforming". It will bring them back into spec. It is not difficult to do.
That they can do that is why in the old days Aluminum Electrolytic caps were called "Self Healing" capacitors.

Did not know that, that's actually very useful, thanks.

Reply 11 of 54, by TELVM

User metadata
Rank Oldbie
Rank
Oldbie
PCBONEZ wrote:

... What actually improved the board's performance was that the 2200uF had lower ESR than the 1500uF.
The capacitance didn't have anything to do with it.

In other words when he raised the capacitance he also lowered the ESR. It was the lower ESR that helped. ...

Agreed! 😉

kanecvr wrote:

... (I've seen quite a few boards use 10v caps on a 12v line). The 10v ones are bloated 90% of the time ...

No wonder, that's recipe for disaster. 😵

Let the air flow!

Reply 12 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
kanecvr wrote:

- there is no harm in using capacitors with a higher voltage rating in place of lower rated ones.

However what you said (and what I was commenting on) was that you are using lower voltage caps in place of higher voltage caps.
Very bad idea.
.

kanecvr wrote:

- AS LONG AS THEY HAVE THE SAME CAPACITANCE VALUE.

While the best practice is to keep the uF as original it is not the most critical spec to match for this type of capacitor. (Meaning the 6.3mm and larger wet lytics on mobos.)
- The ESR is far FAR more important.
The caps themselves have capacitance rating of +/-20%.
The design of the board would typically have a tolerance of +/- 50% for a given cap. (For this kind of cap.)
Contrary to what you said earlier this kind of cap is not used for signaling so the uF isn't critical for that. They are filters and have to do with clean power to various things.
It (uF) can be important for the power-up timing sequence because if x-voltage doesn't get to y-% of it's value within z-milliseconds the PSU will kick back off.
You would have to go pretty far from the original value to hose up the timing though.
- The reason to stick to the original uF is that there is no way to know if the cap that got installed was near the high end or low end of the motherboard's design tolerance.
That said it's generally okay to go up one standard uF value because it's close enough to the +/-20% and uF goes down as caps age. (Moves towards the original value.)
.

kanecvr wrote:

I've seen this on PCChips, Lucky Tech, ECS and Matsonic boards. Even Asus is guilty of doing this, just look at the A7V600.

I seriously doubt this.
I think if you took actual voltage readings on those caps you would find you are mistaken.
Even if it were true. - Why on earth would you follow suit with a bad practice you know is wrong.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 13 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
BSA Starfire wrote:

Nice, sadly all my abit borads have died from bad caps, seems they really went cheap on that side despite the great features and designs.
Hope you have fun with that monster, 2 PIII 1.4s is a force to be reckoned with 😀

During that time virtually everyone was using cheap caps.
Abit was one of the first to transition to all Japanese caps.
They just didn't make a lot of noise about like Gigabyte and Asus who did so years later.

Last edited by PCBONEZ on 2015-12-14, 00:59. Edited 1 time in total.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 14 of 54, by ODwilly

User metadata
Rank l33t
Rank
l33t

PCBONEZ: Reading through the posts it looks like he has tried clarifying that he has been replacing them with higher voltage caps multiple times, easy to get lost in big walls of text 😀 which is where the communication breakdown and argument is originating from. Just thought Id chime in as a third party observer here.

Main pc: Asus ROG 17. R9 5900HX, RTX 3070m, 16gb ddr4 3200, 1tb NVME.
Retro PC: Soyo P4S Dragon, 3gb ddr 266, 120gb Maxtor, Geforce Fx 5950 Ultra, SB Live! 5.1

Reply 15 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
ODwilly wrote:

PCBONEZ: Reading through the posts it looks like he has tried clarifying that he has been replacing them with higher voltage caps multiple times, easy to get lost in big walls of text 😀 which is where the communication breakdown and argument is originating from. Just thought Id chime in as a third party observer here.

I'm not certain that's what he's saying - but thanks.
.
There are a few other things he said that showed a need for a knowledge upgrade.
I think I've addressed them all now except I'm still not sure he knows what ESR is.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 16 of 54, by ODwilly

User metadata
Rank l33t
Rank
l33t

Im just enjoying the knowledge, big reason why I browse the forum is that the conversations that start up usually turn out to be pretty interesting. Plus you guys have remained fairly civil so far, so the mods havent had to come in and erase a bunch of stuff which is nice 😊

Main pc: Asus ROG 17. R9 5900HX, RTX 3070m, 16gb ddr4 3200, 1tb NVME.
Retro PC: Soyo P4S Dragon, 3gb ddr 266, 120gb Maxtor, Geforce Fx 5950 Ultra, SB Live! 5.1

Reply 17 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
ODwilly wrote:

Im just enjoying the knowledge, big reason why I browse the forum is that the conversations that start up usually turn out to be pretty interesting. Plus you guys have remained fairly civil so far, so the mods havent had to come in and erase a bunch of stuff which is nice 😊

I'm not trying to cut him down. I'm trying to teach him.
My hope is that other people will benefit too.
.
I learned all this stuff the hard way. Much of the info is hard to find or to understand.
I'm quite happy to share what I've learned - but some people take it wrong.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 18 of 54, by PCBONEZ

User metadata
Rank Oldbie
Rank
Oldbie
kanecvr wrote:

I like to stick to stock capacitance myself. I just up voltage so the caps will be less "stressed".

This is another misconception.
Using a higher voltage cap won't reduce stress at all.
The voltage present in the circuit is what it is and using a higher voltage cap isn't going to change it.

If you are concerned with stress on the cap choose one with a higher ripple current rating.
That's were most of the stress is.

Often you will see a mobo manufacturer using 10v caps where 6.3v caps would be fine - but it's not about the voltage.
If you are staying within the same series of capacitor and at the same uF the 10v caps tend to have larger cans. - It's about the size of the can.
(Within the same series or grade of capacitor) a larger can means lower ESR and a higher ripple rating.
Using a 10v in place of a 6.3v was just a side effect of trying to improve the ESR and ripple ratings.

In doing repairs you can get better ESR and ripple ratings choosing a higher grade cap to start with.
.

GRUMPY OLD FART - On Hiatus, sort'a
Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.
You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Reply 19 of 54, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

As Bonez said, using a cap with higher breakdown potential will not improve capacitor life. All else being equal it's not a problem to use a cap with a higher voltage rating, but there is no substantive benefit once you're safely over the circuit maximum.

Heat is the critical factor for capacitor life. There are three ways to combat capacitor failure.
1) Ensure adequate cooling reaches your caps, especially those in proximity to heat emitting chips and chokes. This is why small form factor computers are known as capacitor killers -- they just don't have enough airflow.
2) Use 105deg rated caps. The electrolyte is (or should be) formulated to operate at the specified temperature and achieve the MTBF numbers for the caps in question. Because temperature exists as a statistical distribution (this is why evaporation happens), a capacitor with electrolyte rated for a higher temperature will tend to dry out less quickly than a lesser rated capacitor given equal heating within the capacitor casing.
3) If possible, use low ESR caps. Power lost in the capacitor will be proportional to the square of the RMS ripple current and proportional to the ESR at the frequency of the ripple current. If you reduce the ESR, you reduce the amount of power which must be removed from the capacitor casing. There may be cases where higher series resistance is required to dampen out potential LC resonances and higher ESR has provided that. Low ESR caps may not be usable here.

One reason the fallacy may persist is because of casing size. To explain:
Given equivalent dielectric formulation, a higher voltage rating will require more layers of dielectric material. Equivalent ESR indicates the surface area of each layer will be similar. Thus, within the same capacitor series, a higher voltage rating will most likely result in a larger capacitor casing. Heat will exit the capacitor through the surface of the cap, so a capacitor with similar ESR (same heat generated in the casing) and larger casing (larger surface area to radiate heat) will be able to maintain a lower internal temperature and last longer.

All hail the Great Capacitor Brand Finder