BSOD and Gigabyte Capacitors

[momaka]

Re: BSOD and Gigabyte Capacitors

I just had a ECS AM2 motherboard with input kzg's that showed no signs of going bad, but all the output caps (Sanyo WF's) were bloated. The computer would boot, and win7 started up, and in the middle, it would power off, with the power button still lit up but fans and system down. The KGZ's probably dried out and killed most of the sanyos

Nah, I doubt the KZG's killed the Sanyo WF caps. 16V 1500 uF and 1800 uF KZGs don't fail that much, even the older ones. It's the 6.3V ones to be aware of.

Those Sanyo WF failed because they are not so great, as stated earlier.

That said, was the original mobo an ECS MCP61PM-AM or MCP61SM-GM by any chance?? If yes, those also have lots of problems with the nVidia 6100/6150 chipset going bad over time (and really, any mobo with that chipset will have problems, unless it has a really nice big heatsink on it). The only way to stop or slow down that is to install a fan on the existing chipset heatsink or upgrade to a much bigger one (which can be a problem if you want to use the PCI and PCI-E slots, as that can block them).

In any case, don't chuck those boards, even if you see a dead one. A reflow of the chipset can get them going again. It was a widely-used board back in the day, so there is still a bit of demand for it from folks who are trying to get older PCs going again.

I replaced the bad WF's with used MCZ's and it booted fine. Just enough to change some windows devices so I could boot it off another motherboard that was a much higher quality (but different chipset, AMD AM3 Phenom II X4 955, where the ECS was nvidia, Athlon X2 5600+). Worked Great, no bluescreens. Reactivation went well (did have the sticker on the case.)

If it's an AMD 770, 780, or 790 chipset, those tend to fail too same way as the nVidia's, but nowhere near as often. Really, it's only older Intel chipsets that I trust will work fine, even at elevated temperatures. ATI... maybe. And nVidia - I'm glad to see they don't make chipsets anymore. While they did offer good features, they were always terrible in terms of reliability (save for the ancient nForce 2 and possibly nForce 3. nForce 4 appears to be so-so, and anything afterwards is garbage.)

[Uranium-235]

Re: BSOD and Gigabyte Capacitors

I just had a ECS AM2 motherboard with input kzg's that showed no signs of going bad, but all the output caps (Sanyo WF's) were bloated. The computer would boot, and win7 started up, and in the middle, it would power off, with the power button still lit up but fans and system down. The KGZ's probably dried out and killed most of the sanyos

I replaced the bad WF's with used MCZ's and it booted fine. Just enough to change some windows devices so I could boot it off another motherboard that was a much higher quality (but different chipset, AMD AM3 Phenom II X4 955, where the ECS was nvidia, Athlon X2 5600+). Worked Great, no bluescreens. Reactivation went well (did have the sticker on the case.)

the old one also had a 160GB PATA, I cloned it to a 250SATA and enabled AHCI. Worked even better after that.

[Wester547]

Re: BSOD and Gigabyte Capacitors

Most of the 6.3V, 3300 uF Chemicon KZG caps I pulled that were showing okay at some point now are starting to show very high capacitance (4000+ uF), and a few even bulged in storage - same way my 6.3V, 2200 uF HZ did.

That's a slight cause for concern, but nowhere near as drastic a capacitance increase as the bad datecode Nichicon HM and HN (and TK ATWY and ATWB, in my experience - Samyoung NXC too). Those KZG may see a 30-40% increase in capacitance over time, but I've seen the other aforementioned series skyrocket in terms of capacitance (anywhere from 2x-6x the nominal capacitance, and that's with no bulging or leaking).

The MOSFETs aren't always the hottest parts. Sometimes, it's the coils/inductor that run much hotter than the FETs. I find that on newer boards, it's these square "enclosed" inductors that run very hot, whereas the FETs may or may not run hot. On older boards, especially with large toroidal inductors, it's the other way around: FETs are usually hot (and on that note, usually the high-side FETs too, it seems, as they like to make the high-side filter caps toasty) and the inductors just warm.

Yeah, if you ask me, the “square box”, “enclosed” inductor design is fairly dumb and results in the coils overheating. Newer boards have polymers exclusively in the VRM input/output area, so it may not be as big of a deal as it would be with liquid electrolytics.

I was referencing older boards when I meant sinkless, hot high side VRM FETs. Those will bake any adjacent electrolytics dry. What's also concerning is that many boards tend to use freestanding FETs, which is even worse for heat dissipation.

[momaka]

Re: BSOD and Gigabyte Capacitors

So KZG are prone to failing high capacitance / high leakage current in your experience, like bad Nichicon HM, HN, HZ, and Suncon WF? In my experience, most KZG (and KZJ) just go open circuit when they fail, and the capacitance plummets dramatically, although I’ve seen a few exceptions to that rule of thumb.

Most of the 6.3V, 3300 uF Chemicon KZG caps I pulled that were showing okay at some point now are starting to show very high capacitance (4000+ uF), and a few even bulged in storage - same way my 6.3V, 2200 uF HZ did.

But I can't say that applies for the smaller KZG caps, like the notoriously bad 6.3V, 820 uF. I think those loose capacitance and go O/C or high ESR more often, like you said.

What I have also noticed about VRM input caps on many boards with poorly designed VRM sections is that they are pigeonholed right next to hot MOSFETs (the VRM output caps are located away from the FETs, closer to the CPU), which doesn’t help at all.

The MOSFETs aren't always the hottest parts. Sometimes, it's the coils/inductor that run much hotter than the FETs. I find that on newer boards, it's these square "enclosed" inductors that run very hot, whereas the FETs may or may not run hot. On older boards, especially with large toroidal inductors, it's the other way around: FETs are usually hot (and on that note, usually the high-side FETs too, it seems, as they like to make the high-side filter caps toasty) and the inductors just warm.

[Sparkey55]

Re: BSOD and Gigabyte Capacitors

So KZG are prone to failing high capacitance / high leakage current in your experience, like bad Nichicon HM, HN, HZ, and Suncon WF? In my experience, most KZG (and KZJ) just go open circuit when they fail, and the capacitance plummets dramatically, although I've seen a few exceptions to that rule of thumb.

Yes, me too, but I don't doubt that when the PSU's output filter caps go bad, it puts significantly more stress on the general filter capacitors (and VRM input filtering capacitors) on the motherboard, somewhat shortening their lifespan. What I have also noticed about VRM input caps on many boards with poorly designed VRM sections is that they are pigeonholed right next to hot MOSFETs (the VRM output caps are located away from the FETs, closer to the CPU), which doesn't help at all.

Interesting. Some time ago, I happened across this image on the internet, which was posted ten years ago now, and it shows two bad 3300uF 6.3V Nichicon HM with the exact datecodes that yours have (H0453? Week 53, 2004, made in Ohno, Japan), along with a bulged and leaking 1000uF 16V KZJ (with a datecode of February 2nd, 2005). Alongside them was obviously a Rubycon MBZ which appeared fine.

The only conclusion I can come to is that all electrolytics (according to the documents provided by the manufacturers), insofar as their useful life, are rated with 60% confidence (or -/+ 40% confidence). So even if you calculate the formula for useful life (based on the endurance life spec), it may not provide an accurate idea of how long the cap will actually last in the real world (although good caps will last longer than their rated lifespan).

Well then maybe the issue was not fixed with the H0453 HM's. The boards I had were hardly used though. The quality of the PSU and plenty of cooling goes a long way to extend their life it seems.

[Wester547]

Re: BSOD and Gigabyte Capacitors

Looks like Sanyo WF caps are the new Chemicon KZG - just took them a bit longer to go bad. Like the KZGs, all of the WF ones that went bad on my stuff have gone high capacitance (high internal leakage), especially the ones that sat more in storage. So I can confirm what Wester547 said above.

So KZG are prone to failing high capacitance / high leakage current in your experience, like bad Nichicon HM, HN, HZ, and Suncon WF? In my experience, most KZG (and KZJ) just go open circuit when they fail, and the capacitance plummets dramatically, although I’ve seen a few exceptions to that rule of thumb.

So I too doubt that those bulged WF caps were a result of a bad PSU.
Moreover, I've seen systems with completely gutless PSUs filled with bulged crap caps, and the motherboards crap caps survived and were within spec - multiple times.

Yes, me too, but I don’t doubt that when the PSU’s output filter caps go bad, it puts significantly more stress on the general filter capacitors (and VRM input filtering capacitors) on the motherboard, somewhat shortening their lifespan. What I have also noticed about VRM input caps on many boards with poorly designed VRM sections is that they are pigeonholed right next to hot MOSFETs (the VRM output caps are located away from the FETs, closer to the CPU), which doesn’t help at all.

I have found two socket 939 motherboards with Nichicon HM caps with a date code of year 2004 week 53. Yes. I said week 53, not a typo. Still working just fine on a Gigabyte motherboard. Must be a transitional production batch before the standard run of 2005 came out. Very strange indeed.

Interesting. Some time ago, I happened across this image on the internet, which was posted ten years ago now, and it shows two bad 3300uF 6.3V Nichicon HM with the exact datecodes that yours have (H0453? Week 53, 2004, made in Ohno, Japan), along with a bulged and leaking 1000uF 16V KZJ (with a datecode of February 2nd, 2005). Alongside them was obviously a Rubycon MBZ which appeared fine.

The only conclusion I can come to is that all electrolytics (according to the documents provided by the manufacturers), insofar as their useful life, are rated with 60% confidence (or -/+ 40% confidence). So even if you calculate the formula for useful life (based on the endurance life spec), it may not provide an accurate idea of how long the cap will actually last in the real world (although good caps will last longer than their rated lifespan).

[Sparkey55]

Re: BSOD and Gigabyte Capacitors

Understandable, it isn't exactly straight forward, I only know about it because we ran into issues at work back in 2015 (last time a 53 week ISO year occurred) with some reports that pulled data from multiple systems (some using the ISO 8601 standard and some using other standards that always have 52 weeks) where the weeks on the reports didn't match some of the data.

Thanks, interesting though. I think Nichicon must have found a fix for the formula with them but never implemented them into production until after 2005.

[dmill89]

Re: BSOD and Gigabyte Capacitors

TL;DR

Understandable, it isn't exactly straight forward, I only know about it because we ran into issues at work back in 2015 (last time a 53 week ISO year occurred) with some reports that pulled data from multiple systems (some using the ISO 8601 standard and some using other standards that always have 52 weeks) where the weeks on the reports didn't match some of the data.

[Sparkey55]

Re: BSOD and Gigabyte Capacitors

Assuming they are using the ISO 8601 date standard there can potentially be 53 weeks in an ISO year depending on what day the year starts on (2004 is one of these years).

TL;DR
Somehow I do not think that was it.

[dmill89]

Re: BSOD and Gigabyte Capacitors

I have found two socket 939 motherboards with Nichicon HM caps with a date code of year 2004 week 53. Yes. I said week 53, not a typo. Still working just fine on a Gigabyte motherboard. Must be a transitional production batch before the standard run of 2005 came out. Very strange indeed.

Assuming they are using the ISO 8601 date standard there can potentially be 53 weeks in an ISO year depending on what day the year starts on (2004 is one of these years).

https://en.wikipedia.org/wiki/ISO_week_date
First week

The ISO 8601 definition for week 01 is the week with the Gregorian year's first Thursday in it. The following definitions based on properties of this week are mutually equivalent, since the ISO week starts with Monday:
It is the first week with a majority (4 or more) of its days in January.
Its first day is the Monday nearest to 1 January.
It has 4 January in it. Hence the earliest possible first week extends from Monday 29 December (previous Gregorian year) to Sunday 4 January, the latest possible first week extends from Monday 4 January to Sunday 10 January.
It has the year's first working day in it, if Saturdays, Sundays and 1 January are not working days.

If 1 January is on a Monday, Tuesday, Wednesday or Thursday, it is in week 01. If 1 January is on a Friday, it is part of week 53 of the previous year. If it is on a Saturday, it is part of the last week of the previous year which is numbered 52 in a common year and 53 in a leap year. If it is on a Sunday, it is part of week 52 of the previous year.

Last week

The last week of the ISO week-numbering year, i.e. the 52nd or 53rd one, is the week before week 01. This week's properties are:
It has the year's last Thursday in it.
It is the last week with a majority (4 or more) of its days in December.
Its middle day, Thursday, falls in the ending year.
Its last day is the Sunday nearest to 31 December.
It has 28 December in it. Hence the earliest possible last week extends from Monday 22 December to Sunday 28 December, the latest possible last week extends from Monday 28 December to Sunday 3 January.

If 31 December is on a Monday, Tuesday or Wednesday, it is in week 01 of the next year. If it is on a Thursday, it is in week 53 of the year just ending; if on a Friday it is in week 52 (or 53 if the year just ending is a leap year); if on a Saturday or Sunday, it is in week 52 of the year just ending.

Weeks per year

The long years, with 53 weeks in them, can be described by any of the following equivalent definitions:
any year starting on Thursday (dominical letter D or DC) and any leap year starting on Wednesday (ED)
any year ending on Thursday (D, ED) and any leap year ending on Friday (DC)
years in which 1 January and 31 December (in common years) or either (in leap years) are Thursdays

All other week-numbering years are short years and have 52 weeks.

The number of weeks in a given year is equal to the corresponding week number of 28 December, because it is the only date that is always in the last week of the year since it is a week before 4 January which is always in the first week of the year. Using only the ordinal year number, the number of weeks in that year can be determined:[1]
weeks ( year ) = 52 + { 1 (long) if p ( year ) = 4 or p ( year − 1 ) = 3 0 (short) otherwise p ( year ) = ( year + ⌊ year 4 ⌋ − ⌊ year 100 ⌋ + ⌊ year 400 ⌋ ) mod 7 The following 71 years in a 400-year cycle have 53 weeks (371 days); years not listed have 52 weeks (364 days); add 2000 for current years:
004, 009, 015, 020, 026,032, 037, 043, 048, 054,060, 065, 071, 076, 082,088, 093, 099, 105, 111, 116, 122,128, 133, 139, 144, 150,156, 161, 167, 172, 178,184, 189, 195, 201, 207, 212, 218,224, 229, 235, 240, 246,252, 257, 263, 268, 274,280, 285, 291, 296, 303, 308, 314,320, 325, 331, 336, 342,348, 353, 359, 364, 370,376, 381, 387, 392, 398.
On average, a year has 53 weeks every 400⁄71 = 5.6338… years, and these long ISO years are 43 times 6 years apart, 27 times 5 years apart, and once 7 years apart (between years 296 and 303). The Gregorian years corresponding to these 71 long years can be subdivided as follows:
27 Gregorian leap years, emphasized in the list above: 14 starting on Thursday, hence ending on Friday, and
13 starting on Wednesday, hence ending on Thursday;

44 Gregorian common years starting, hence also ending on Thursday.

The Gregorian years corresponding to the other 329 short ISO years (neither starting nor ending with Thursday) can also be subdivided as follows:
70 are Gregorian leap years.
259 are Gregorian common years.

Thus, within a 400-year cycle:
27 week years are 5 days longer than the month years (371 − 366).
44 week years are 6 days longer than the month years (371 − 365).
70 week years are 2 days shorter than the month years (364 − 366).
259 week years are 1 day shorter than the month years (364 − 365).

[Sparkey55]

Re: BSOD and Gigabyte Capacitors

^ Were those HM caps rated for 16V or 6.3V?

I think the higher voltage parts (16V or more) tend to be a bit more reliable, even from the defective date codes. That's not to say that 16V (and higher) HM, HN, and HZ with date codes before 2004 should be trusted. But for transitional period ones like 2005, I think it is something to consider.

It has been about five years since I had those but I remember them to be in the CPU VRM low side and being 3300uf 6.3V and the high side was Rubycons 16V.

[momaka]

Re: BSOD and Gigabyte Capacitors

^ Were those HM caps rated for 16V or 6.3V?

I think the higher voltage parts (16V or more) tend to be a bit more reliable, even from the defective date codes. That's not to say that 16V (and higher) HM, HN, and HZ with date codes before 2004 should be trusted. But for transitional period ones like 2005, I think it is something to consider.

[Sparkey55]

Re: BSOD and Gigabyte Capacitors

*drumroll* *enters momaka*
Ladies and gentlemen, allow me to introduce myself.......

No I wouldn't try to buy and fix every socket 939 board, but I definitely do like to save the more rare ones (particularly ones with SLI).

Yes, they are slow by today's standard, even with a top-of-the-line FX chip (which are rare and cost an arm and a leg). But for everyday use with a dual-core CPU, they are actually okay. Also extremely reliable if you get a board with an AMD chipset... though those don't support SLI and rarely you'll find a special s939 board with an AMD chipset.


Actually, not so much anymore.

Looks like Sanyo WF caps are the new Chemicon KZG - just took them a bit longer to go bad. Like the KZGs, all of the WF ones that went bad on my stuff have gone high capacitance (high internal leakage), especially the ones that sat more in storage. So I can confirm what Wester547 said above.

So I too doubt that those bulged WF caps were a result of a bad PSU.
Moreover, I've seen systems with completely gutless PSUs filled with bulged crap caps, and the motherboards crap caps survived and were within spec - multiple times.


Exactly.

All HM, HN, and HZ series made after 2006 are pretty safe to use (no less than Rubycon MCZ anyways).
2005 is sort of a hit-and-miss year, as I've found out lately. I have a big jar full of HN and HZ caps from Xbox 360 consoles with date codes ranging from mid 2005 to late 2008. About 1/4 of my 2005 date-code HZ Nichicons (note: they are all 6.3V, 2200 uF) have gone bad - either bulged in storage or in use. The 16V, 1500 uF Nichicon HN caps with the same date codes, on the other, have held up just fine and still reading in spec the same as their newer brothers and sisters.
And, I have enough old motherboards with 2005 date code HM caps (mostly 16V, 1200 or 1500 uF) to confirm that these are okay as well.

But any HM, HN, and HZ series made before 2005 are certainly to be avoided.

That's all from my experimental field here.

I have found two socket 939 motherboards with Nichicon HM caps with a date code of year 2004 week 53. Yes. I said week 53, not a typo. Still working just fine on a Gigabyte motherboard. Must be a transitional production batch before the standard run of 2005 came out. Very strange indeed.

[momaka]

Re: BSOD and Gigabyte Capacitors

god who would want to fix such an old mobo. most the CPU's it can take are beaten by $59 haswell chips

*drumroll* *enters momaka*
Ladies and gentlemen, allow me to introduce myself.......

No I wouldn't try to buy and fix every socket 939 board, but I definitely do like to save the more rare ones (particularly ones with SLI).

Yes, they are slow by today's standard, even with a top-of-the-line FX chip (which are rare and cost an arm and a leg). But for everyday use with a dual-core CPU, they are actually okay. Also extremely reliable if you get a board with an AMD chipset... though those don't support SLI and rarely you'll find a special s939 board with an AMD chipset.

I know, right? Got a MSI K9MM-V in today with a few bulging WF-series Sanyo's. Rare indeed.

Actually, not so much anymore.

Looks like Sanyo WF caps are the new Chemicon KZG - just took them a bit longer to go bad. Like the KZGs, all of the WF ones that went bad on my stuff have gone high capacitance (high internal leakage), especially the ones that sat more in storage. So I can confirm what Wester547 said above.

So I too doubt that those bulged WF caps were a result of a bad PSU.
Moreover, I've seen systems with completely gutless PSUs filled with bulged crap caps, and the motherboards crap caps survived and were within spec - multiple times.

Nichicon HM/HN/HZ made after 2005 should be fine. Those made before 2006 and especially before 2005 are very failure prone.

Exactly.

All HM, HN, and HZ series made after 2006 are pretty safe to use (no less than Rubycon MCZ anyways).
2005 is sort of a hit-and-miss year, as I've found out lately. I have a big jar full of HN and HZ caps from Xbox 360 consoles with date codes ranging from mid 2005 to late 2008. About 1/4 of my 2005 date-code HZ Nichicons (note: they are all 6.3V, 2200 uF) have gone bad - either bulged in storage or in use. The 16V, 1500 uF Nichicon HN caps with the same date codes, on the other, have held up just fine and still reading in spec the same as their newer brothers and sisters.
And, I have enough old motherboards with 2005 date code HM caps (mostly 16V, 1200 or 1500 uF) to confirm that these are okay as well.

But any HM, HN, and HZ series made before 2005 are certainly to be avoided.

That's all from my experimental field here.

[Wester547]

Re: BSOD and Gigabyte Capacitors

Nichicon HM/HN/HZ made after 2005 should be fine. Those made before 2006 and especially before 2005 are very failure prone. Sanyo's (Suncon) WF series is rather failure prone (so I doubt the PSU was at fault), and Teapos are failure prone altogether, especially on motherboards (IE, in high ripple current applications such as the output of a buck regulator). NCC KZG made between 2001 and 2007 have a very high failure rate (KZJ weren't much better if at all). WG aren't the greatest caps in the world, that's for sure. It seems on nsb_'s board that there was one Nichicon HM which didn't bulge or leak (out of the nine), but it's probably out-of-spec anyway.

[stj]

Re: BSOD and Gigabyte Capacitors

i always replace HM/HN on sight.
i never trust those anymore.

[nsb_]

Re: BSOD and Gigabyte Capacitors

all the big nichicon's are bad on the first board.

Yeah, are the Nichicon HM 1000uF and 6,3V. All of those (9 in total) leaking or bulging. I think that were a f***ing bad model/family of capacitors.



Anyway, I think that this system had a life with bad heat conditions..

[stj]

Re: BSOD and Gigabyte Capacitors

all the big nichicon's are bad on the first board.

[nsb_]

Re: BSOD and Gigabyte Capacitors

bulging sanyos? thats rare. A little tight depending on how close they are together (increase from 10mm to 12.5)

the 16v is fine, as long as the impedance is lower and mA s higher. rest looks good

Well, really, one of the Sanyo 1500uF and 6,3V blowed up into a nice explosion. The system (Pentium 4) continued working as if nothing had happened.



In both motherboards (K8nsc-939 and 8PE667) can be seen how all the Chemi-con KZG are bulging..

[Agent24]

Re: BSOD and Gigabyte Capacitors

u could just keep quiet and let the psu kill the mobo again and wait for them to send the board to u again in a few years lol... its good to have "regular customers" eh? hahaha!

That sort of thing will come back to bite you in the ass, not worth it.