Gigabyte GA-8ID533 went pop!

[stj]

Re: Gigabyte GA-8ID533 went pop!

YXJ are not ultra low-esr.
they have half the ripple handling and rubbish esr figures.

if you wanted to use rubycon then the ZLx range would be a starting point.

read datasheets.
even the FR is not really as good as the HM - but it's a lot closer.

[chrunchyhobo]

Re: Gigabyte GA-8ID533 went pop!

YXJ are not ultra low-esr.
they have half the ripple handling and rubbish esr figures.

if you wanted to use rubycon then the ZLx range would be a starting point.

read datasheets.
even the FR is not really as good as the HM - but it's a lot closer.

that's incredibly helpful, thank you.

so the basics for choosing capacitors, if I have got this right, is getting one with the same or higher voltage rating, identical uF, as close as possible dimensions, -40c to 105c, and an identical or preferably higher ripple rating?

obviously choosing from reputable brands only.

now I know what I am looking for on those data sheets I think I have pretty much got this nailed down thanks to you guys!

[ChaosLegionnaire]

Re: Gigabyte GA-8ID533 went pop!

I am NOT questioning your advice here, but what makes the Panasonic ones better?

I would like to know what to look out for in the future.

FR has longer life too from 6000-10000 hours.

when picking caps for mobo use, the specs like esr, ripple and capacitance must be within ±20% of the original. yxj doesnt state its esr value in the datasheet. if it doesnt, its assumed to be general purpose which are unsuitable for mobo use unless its for the onboard audio/lan decoupling or fan header filtering. other than that, caps for mobo use must be ultra-low esr esp. around the cpu vrm area otherwise the board will be unstable or glitchy.

so the basics for choosing capacitors, if I have got this right, is getting one with the same or higher voltage rating, identical uF, as close as possible dimensions, -40c to 105c, and an identical or preferably higher ripple rating?

your above statement is right also.

obviously choosing from reputable brands only.

actually what u need to be aware of are the problematic series. not all japcap brands had a bed of roses with their ultra-low esr caps. nichicon has problems with their hm/hn line with the 2001-2005 datecodes. chemicon has problems with their kzg/kzj line; and sanyo has issues with their wf/wx line tho that line isnt considered ultra-low esr.

[Wester547]

Re: Gigabyte GA-8ID533 went pop!

actually what u need to be aware of are the problematic series. not all japcap brands had a bed of roses with their ultra-low esr caps. nichicon has problems with their hm/hn line with the 2001-2005 datecodes. chemicon has problems with their kzg/kzj line; and sanyo has issues with their wf/wx line tho that line isnt considered ultra-low esr.

I'm not aware of WX being "bad", maybe a bit heat sensitive but not terrible by any means (I think WG is more likely to fail, being lower ESR...). I'd like to add that although Toshin Kogyo is also a Japanese brand and that OST is one of their distributors (although not manufacturers), their ATWY/ATWB series should also be avoided at all costs. YXJ is miniaturized YXH, which is the long life version of YXG, and all those caps certainly have their ESR values reported in the datasheet at 100KHz. They are not GP by any means.

[chrunchyhobo]

Re: Gigabyte GA-8ID533 went pop!

FR has longer life too from 6000-10000 hours.

when picking caps for mobo use, the specs like esr, ripple and capacitance must be within ±20% of the original. yxj doesnt state its esr value in the datasheet. if it doesnt, its assumed to be general purpose which are unsuitable for mobo use unless its for the onboard audio/lan decoupling or fan header filtering. other than that, caps for mobo use must be ultra-low esr esp. around the cpu vrm area otherwise the board will be unstable or glitchy.

your above statement is right also.

actually what u need to be aware of are the problematic series. not all japcap brands had a bed of roses with their ultra-low esr caps. nichicon has problems with their hm/hn line with the 2001-2005 datecodes. chemicon has problems with their kzg/kzj line; and sanyo has issues with their wf/wx line tho that line isnt considered ultra-low esr.

I have read a few things on various websites that say that the higher the ripple rating, the better.

Can going "too high" (like over 20% as you stated) cause issues?

And I was under (the possibly incorrect) impression that you cannot change the capacitance AT ALL, as both higher or lower values can mess stuff up.

Was what I read a load of rubbish?

[stj]

Re: Gigabyte GA-8ID533 went pop!

it's complex, the capacitance depends on what it's used for.

the ripple handling, higher is better but the cap gets bigger - so doubling it is not something you could usually do anyway.
it wont do any harm though.

[chrunchyhobo]

Re: Gigabyte GA-8ID533 went pop!

it's complex, the capacitance depends on what it's used for.

the ripple handling, higher is better but the cap gets bigger - so doubling it is not something you could usually do anyway.
it wont do any harm though.

so the safest thing to do would be to find a capacitor that isn't from a poor manufacturer, isn't part of a known bad batch, has the best ripple rating that you can get but at the same time keeping the size the same (because some caps are the same size, but are "general purpose" with low esr/ripple) , and is otherwise identical (voltage, uF, etc.)?

I do apologise for all the (possibly stupid) questions, I just want to make sure I have it all right!

[stj]

Re: Gigabyte GA-8ID533 went pop!

yes, but also the replacement should have a similar or slightly lower esr.

i know it's tricky, when you have 4 different parameters to deal with aswell as the size, but that's what you need to do.

[ChaosLegionnaire]

Re: Gigabyte GA-8ID533 went pop!

man, i made such a mess of my posts in this topic... i guess this is what happens when u multitask tv and forumming at the same time.

I have read a few things on various websites that say that the higher the ripple rating, the better.

Can going "too high" (like over 20% as you stated) cause issues?

i forgot to explain that for ripple, its not lower than 20% of the rated value. i say this because in some situations, due to availability issues for example, u cannot find a cap with such high ripple handling specs.

i was faced with this when i needed a 2800mA ripple cap for my mobo but the highest i could get was 2550mA. i was told this was okay because of the "20% tolerance" and 2550mA was within the acceptable tolerance limit so i could get away with using a lower rated ripple cap.

one of the cons with using just a tiny bit lower rated ripple cap is that the cap may not last as long as cap lifespan is dependant on how close the ripple of the circuit is to the cap's max rated ripple.

next, the pros of using higher rated ripple caps is that the cap can last longer. it *may* also improve the overclocking of the cpu tho that is not guaranteed. overclockability is more dependant on the asic quality of the chip. it can also allow u to get away with using a junk psu for the short to medium term. however, i dont recommend u use a junk psu on an irreplaceable antique mobo as a junk psu can kill things other than caps. its also a major chore to recap all the 20-30 caps on a mobo.

And I was under (the possibly incorrect) impression that you cannot change the capacitance AT ALL, as both higher or lower values can mess stuff up.

for what u read, that probably only applies if the cap is used for high frequency signal decoupling (like for the bus clock or cpu clock), noise filtering or some other application. however, electrolytic caps arent used for that purpose. high frequency signal decoupling/noise filtering is typically handled by ceramic caps instead and changing the capacitance for that application can mess up the clock signal or noise filtering.

however, the electrolytic caps are used for power rail filtering on the motherboard. that application has more leeway and is not so "strict" with capacitance. esr matters more instead. however, u cant go too low on the esr, as in some cases, the voltage regulator design may not like it and it'll malfunction. keep to the ±20% rule for esr. similar to me, in case u cant find a cap with such low esr, u can get away with using a higher esr cap as long as it is within the 20% limit specified.

hope that explains things about what u read and it wasnt entirely rubbish. caps are like medicine. the right cap and medicine for the right application/illness keeps things running smoothly.

[chrunchyhobo]

Re: Gigabyte GA-8ID533 went pop!

yes, but also the replacement should have a similar or slightly lower esr.

i know it's tricky, when you have 4 different parameters to deal with aswell as the size, but that's what you need to do.

What part of the datasheet for a capacitor should I be looking at to see its esr?

I think I'm being blind here.

[stj]

Re: Gigabyte GA-8ID533 went pop!

ESR / Impedance / OHMS symbol (like a horseshoe)

[chrunchyhobo]

Re: Gigabyte GA-8ID533 went pop!

ESR / Impedance / OHMS symbol (like a horseshoe)

thought it was that, just had to make sure!

thanks!

[momaka]

Re: Gigabyte GA-8ID533 went pop!

so the basics for choosing capacitors, if I have got this right, is getting one with the same or higher voltage rating, identical uF, as close as possible dimensions, -40c to 105c, and an identical or preferably higher ripple rating?

That actually depends on your skill level. If you know what the circuit does, you can tweak parameters as needed. I often use 6.3V caps in place of 10V caps in computer equipment. Seems like a no-no, but in computers, the main voltage rails are 3.3V, 5V, and 12V. A 6.3V cap will be suitable for both the 3.3V and 5V rails, as would be a 10V cap. Obviously neither a 6.3V nor 10V cap would be found on a 12V rail, so that's why 6.3V and 10V caps are interchangeable. For the 12V rail, always use either 16V or 25V caps.

But to keep it more simple for novice and/or inexperienced techs, I would say do these:
* same or higher voltage rating
* same or lower ESR (if you can't find suitable replacements, up to 20% higher ESR rating will usually be okay too)
* capacitance within 20% of original cap or higher capacitance (but no more than 2x)
* ripple current: the higher, the better the replacement cap (but as stj said, the size of the cap will increase, and that can be a limitation on what you can use)
* size: doesn't matter, as long as the cap will fit and not interfere or get in the way of anything else. That said, it helps to get caps with the same diameter as the originals, as the lead spacing will be the same and make it much easier to solder the new replacement caps.

I presume that the chances of this being caused by my CWT featherweight PSU are slim?

Be careful with cheap CWT units, especially the ISO line as they use a 2-transistor 5VSB circuit. To keep it short, let's just say that a circuit like that can malfunction and easily sent much higher than 5V on the 5VSB line, thus possibly killing the motherboard. If you are recapping the motherboard and really have no other better PSU to use, then get some caps for the PSU as well. In particular, make a thread so we can see what is inside and point if there is anything that needs attention or what caps need replacing.

All the ebay listings for them say "for audio" or some such nonsense, I presume this is just a marketing term to get more hits?

Avoid eBay and Amazon for caps. Too many counterfeit caps there, especially anything that has shipping from Asia.

And I suppose I should ask about the other caps on the mobo, which are 4x Rubycon MBZ 1500uF 6.3v T0245, 2x Rubycon MBZ 3300uF 6.3v T0242, 3x Rubycon MBZ 1500uF 16v T0241, 4x CHOYO MR 22uF 16v 11YJ and 6x TEAPO SS A3 100uF 10v.

Any of those out of a bad batch?

CHOYO is a terrible brand, but the ones on your board are probably too small to matter. However, since you are already ordering caps for your board, you might as well replace these too. After all, there are only 4 of them. Same goes for the Teapo SS. 10 new small caps won't break the bank / empty your wallet for sure.

Rubycon MBZ is a good brand. Leave them. If you are still worried about them, you can do the following for the ones around the CPU: replace half of them with polymer capacitors. Mixing poly caps with regular electrolytic is actually a good way to keep the capacitance close to the original and get very low total ESR in the circuit. The polymers can have half of the original capacitance.
*OR*
You can actually use Nichicon HM or HN for those Rubycon MBZ caps. That may sound silly, since your board just had a few Nichicon HM caps pop - but those are from the years when Nichicon had problem with their HM, HN, and HZ series.
Nichicon HM and HN are okay to use as long as they don't have the following date codes:
x01##, x02##, x03##, x04##, and x05##

These date codes signify production year 2001, 2002, 2003, 2004, and 2005, where "x" can be either "A", "H", or some other letter (it signifies where the cap was made, but it doesn't really matter for us.) And "##" is just a 2-digit number that varies with what week the cap was made (but again, we don't care about that).
Also, not all HM and HN caps with a x05## date code are affected by the problem. 2005 was actually a transition year. But we recommend replacing them anyways. I've also had problems with Nichicon HZ caps with 2005 date codes, so I think the above is also applicable to Nichicon HZ.