Epox 8KHA+ leaky capacitors

[time]

When opening a PC to replace the CD writer - and purge it of the loathsome zip drive that lurked in its shadows - we noticed the tops of most of the electrolytic capacitors were 'corroded' and bulging ...

The motherboard was an Epox 8KHA+ purchased in October 2001. Before the zip :eekers: drive was removed, the PC paused for five minutes after Windows shutdown with the disk controller light on solid.

Afterwards, it seemed to shut down okay once, then started pausing for one minute in apparently the same place. After a while, it also lost the ability to drag and drop, cut and paste, or move objects on the desktop. In the couple of days since, this now seems to happen every time the PC is used, although it can be fixed by rebooting.

Concerned, we opened a file server with the same motherboard and vintage - owned by the same customer - and found the same capacitor problem, only worse. And lo, when it was powered up again and then rebooted, it exhibited the same inexplicable one minute delay at shutdown.

The file server runs Linux.

From then on, it seemed to deteriorate every time it was rebooted. When we returned the next day to finish installing a new version of Linux, it required repeated reboots to get anywhere, culminating in a gray screen.

It's worth noting that both machines have been running 24x7 for the last 2+ years; reboots were highly unusual, and in the case of the Linux box, didn't happen unless there was a power outage.

I think I see a pattern here, and I don't like it. Both PCs are just over two years old and out of warranty - and I supplied them. I also supplied the same motherboard to a couple of other people around the same time. At this stage, I'm expecting them to be stuffed as well.

The local branch of the supplier is closed until Monday, but I'm thinking of shaking up their head office tomorrow to see what their story is. Google throws up a couple of references to leaky caps on this board, but not (yet) the avalanche I would have expected.

This was a really popular board. What are other members' experiences?

 

[Mercutio]

I don't have a solution but I've never received any of the in-warranty boards that I've RMA'd to Northgate solutions, Epox's authorized repair service in the USA. They've cashed the checks for the return shipping, too. To date I'm out $60.

They were good boards but over time I had my share of problems with them, mostly losing either the USB or PS2 ports on the board.

 

[time]

Well, I just confirmed my third board today. All just outside the two-year warranty. :evil:

Fortunately, a different Epox model I supplied to this customer at the same time looks just fine.

From what I've read, nearly *all* brands were affected by this malaise, but it's impossible to be certain about a particular model. It generally takes two years for the problem to become obvious.

I've been wondering if this is a time bomb for dealers like Tannin and Buck in particular ...

 

[time]

This is an epidemic. I wonder how many people have these caps?

 

[i]

Wow, that's crazy. :eekers:

 

[CougTek]

I have never seen a motherboard with so many leaking capacitors. 2, 3, 4, ok, but almost all the caps, no, never.

Call Guiness.

 

[blakerwry]

i concur. Even my abit that eventually died of bad caps only had at most 6 leaking ones(they leaked from the underside)... all the big ones were bulging though.

 

[Handruin]

Damn, my dead MSI wasn't even that bad:

http://www.handruin.com/link_img/pc-death/cap1.jpg
http://www.handruin.com/link_img/pc-death/cap2.jpg
http://www.handruin.com/link_img/pc-death/cap3.jpg
http://www.handruin.com/link_img/pc-death/cap4.jpg
http://www.handruin.com/link_img/pc-death/cap5.jpg

 

[i]

Submit a story to Slashdot and ask if others are running into the same problem.

Just don't point any links here. We wouldn't survive (load-wise or troll-wise).

 

[Howell]

I work with IBM desktops exclusively. I myself have replaced four MBs in four weeks. This is an ongoing problem of about one a week. The capacitor appearance ranged from slightly bulged to spewing powder.

IBM has acknowledged the problem and has sent us spare boards in advance.

I have noticed that if you let the boards cool off a bit. They will boot OK but eventually they start to lock up and reboot themselves.

 

[Tannin]

Ouch! Nasty!

I only have one picture to post in return.

I guess that's one way to get my attention.

Now, Epox boards and corroded caps: I don't recall seeing any, Time. Two possibilities occur to me: (a) that you are the victim of a bad batch (perhaps your wholesaler bought several hundred and you wound up with three or four shipments all from that same lot), or (b) that it's the steamy Brisbane climate that is bringing on these failures faster, and that I will meet the same problem in, say, another year or two.

I'll keep my eyes open.

 

[Mercutio]

Submit a story to Slashdot and ask if others are running into the same problem.

Just don't point any links here. We wouldn't survive (load-wise or troll-wise).

There have been at least two stories on /. on that very topic, and one on an individual selling capacitor replacement kits as well as offering resoldering service.

This *is* an epedemic, and it'll probably take some time to resolve itself.

A couple years ago I actually had a cap explode off an Abit board, banging the side of a case. I thought for a second that a gun had gone off.

 

[blakerwry]

i heard a similar loud pop with my abit board... i didnt find the cap... dont know if it was missing or just plain popped....

 

[LiamC]

ASUS suffering as well?

http://www.theinquirer.net/?article=13674

 

[jtr1962]

Looks to me like they're using capacitors with too high an internal series resistance. At the high currents observed when filtering the microprocessor power supply (this could be tens of amps), the capacitors get too hot, and eventually fail. I also wonder if the fact that many boards just happen to fail right after their warranty is up isn't somehow planned for by using these cheaper capacitors. When selecting parts, a capacitor has a maximum ripply current rating, and you should get a part designed for at least twice the current it will actually encounter in the circuit for longevity purposes. You never design a consumer product right on the bleeding edge. I wouldn't be at all surprised if some egghead in accounting with no knowledge of electronics had the final say over which parts were used.

 

[time]

Looks to me like they're using capacitors with too high an internal series resistance. At the high currents observed when filtering the microprocessor power supply (this could be tens of amps), the capacitors get too hot, and eventually fail.

Nice theory, shame about the facts. A higher internal resistance would actually reduce the power the caps have to handle. In any case, this is just one low-level aspect of how the capacitor achieves its claimed rating.

I also wonder if the fact that many boards just happen to fail right after their warranty is up isn't somehow planned for by using these cheaper capacitors. When selecting parts, a capacitor has a maximum ripply current rating, and you should get a part designed for at least twice the current it will actually encounter in the circuit for longevity purposes. You never design a consumer product right on the bleeding edge.

Well, I don't subscribe to the conspiracy theory. In fact, I believe the designers went over the top trying to ensure stability in their product - witness the large number of large capacitors (21 by my count). The proof of the pudding lies in the fact that the photographed board was still happily serving only a month ago. It's sibling is still in use, albeit increasingly erratically. I for one would give a big round of applause for the engineers.

I wouldn't be at all surprised if some egghead in accounting with no knowledge of electronics had the final say over which parts were used.

Yes, well, I think this sums it up exactly. And of course, some egghead in accounting with no knowledge of electronics always does have the final say ... :evil:

BTW, from something I read on LostCircuits, I gather that the brand of capacitor (GSC) is (or was) highly regarded. I haven't yet checked, but I suspect that they're still featured on Epox (and other) motherboards. :eekers: Mind you, as LiamC pointed out with his link, some other brands (Asus & RLZ) are far worse.

 

[Howell]

I have noticed that all failed caps have a plus or X indetation in the top of the cap and the replacements have a K marking. I assume this is a manufacturer specific marking.

It is truly amazing how certain parts with certain specific specifications can still cause problems. Back at my old job site they had only two sources for a specific kind of weld wire. They wanted to use the less expensive one but it would jam the machine. Same specifications, different result.

Rumor was, the cap problem was caused by a reformulation of the fluid in the capacitors for cost. FWIW, I don't know who makes IBMs MBs.

 

[jtr1962]

Nice theory, shame about the facts. A higher internal resistance would actually reduce the power the caps have to handle. In any case, this is just one low-level aspect of how the capacitor achieves its claimed rating.

A higher internal resistance at a given ripple current means more heat in this case. Remember V=IR and P=IV? At least that's what I learned. You're thinking of a resistor across a fixed voltage, and here you are correct. I'm talking about filter caps whose purpose is to smooth the ripple current coming from the switching power supply inductor. Think of the capacitors in this case as an AC resistance which should ideally be zero. Or put another way, if you short a charged capacitor, the one with lower internal resistance will discharge faster and put out a higher peak current (or will put out a given current with less voltage drop). At least that's my theory on this. I welcome alternate theories.

I think what happened was they paralleled many capacitors to attempt to share the current (nice in theory), but thanks to the board layout some (or one) got more than their share of current and failed prematurely. After that, it would only be a matter of time before the rest of the caps failed, one at a time. This is why these boards get progressively less stable as time goes by and the number of capacitors doing the filtering decreases. More ripple current = more variations in voltage = less stability.

 

[jtr1962]

I have noticed that all failed caps have a plus or X indetation in the top of the cap and the replacements have a K marking. I assume this is a manufacturer specific marking.

That's not a marking. That's a pressure relief "valve". They purposely weaken the top of the cap so that if it overheats the gasses will force the cap open at the weak point before pressure gets so dangerously high as to make a violent explosion. At least that's the way it works in thoery. They also put some sort of pressure relief on batteries for the same reason, although I once saw a taximeter whose 1/2AA lithium battery exploded and took out the car's windshield. Some technician had mistakenly soldered one side to the 12V line.

 

[Dr Bombcrater]

Now, Epox boards and corroded caps: I don't recall seeing any, Time.

Epox have this problem too, alas. I've personally lost several 8KTA2s to leaking caps, and I've seen a few 8KHAs with signs of it.

The Soltek 75KAV is also a known offender. I've two of them sitting right here, killed by bad caps.

 

[time]

A higher internal resistance at a given ripple current means more heat in this case. Remember V=IR and P=IV? At least that's what I learned.

Actually, I *do* remember. In fact, I 'learned' it way before any school physics class.

Or put another way, if you short a charged capacitor, the one with lower internal resistance will discharge faster and put out a higher peak current ...

Yep. As you said, "P=IV". My interpretation of this is that P (power) will increase with I (current) - strangely enough ...

I think what happened was they paralleled many capacitors to attempt to share the current (nice in theory), but thanks to the board layout some (or one) got more than their share of current and failed prematurely. After that, it would only be a matter of time before the rest of the caps failed, one at a time. This is why these boards get progressively less stable as time goes by and the number of capacitors doing the filtering decreases. More ripple current = more variations in voltage = less stability.

You've been reading too much Anandtech. Because Intel was using fewer but larger capacitors at one stage, Anand championed the idea that this was the 'ideal' solution, and that Taiwanese board manufacturers sprinkled caps around their boards to save money.

Well, yes, but it's kinda useful to sprinkle them around, too. :-?

You could just as easily describe placing capacitors in parallel as 'over-engineering'. And by God, it seems to have worked with the 8KHA+! The pictured board was still working as a server only a few weeks ago.

In contrast, even though capacitor ratings are similar, modern boards seem to use only a bare minimum of caps. Presumably the manufacturers now know exactly what they can get away with.

 

[time]

The Soltek 75KAV is also a known offender. I've two of them sitting right here, killed by bad caps.

But it's not universal. We checked an SL-75KAV at the same time as inspecting the 8KHA+ boards, and its was fine. And it's had a *hard* life.

I don't suppose you identified the brand of caps on those failed boards, Dr Bombcrater?

 

[jtr1962]

In contrast, even though capacitor ratings are similar, modern boards seem to use only a bare minimum of caps. Presumably the manufacturers now know exactly what they can get away with.

Exactly what I said about designing on the bleeding edge, only in this case I would say the accountants were more responsible than the engineers. After all, if you get rid of one $0.40 cap and make 100,000 boards you save $40,000. I'm sure the engineers weren't thrilled to go along, but they reasoned that most boards would at least last the warranty period, and that seems to be the case.

I rarely visit Anandtech. I came up with that theory based on my knowledge of electronic engineering. Remember, I had to design my own switching power supplies when I made my thermoelectric chamber, so I'm thoroughly familiar with the issues involved.

 

[Dr Bombcrater]

But it's not universal. We checked an SL-75KAV at the same time as inspecting the 8KHA+ boards, and its was fine. And it's had a *hard* life.

I don't suppose you identified the brand of caps on those failed boards, Dr Bombcrater?

I think it's just taking differing amounts of time for the boards to show signs of cap death. I have a couple of 75KAVs right here now - both F5 revision, both with the same brand of capacitors, both have had very hard lives - but only one has leaking caps. The non-leaky board is still in use so it'll be interesting to see how long it lasts.

The caps on both the dead and working 75KAVs are green/gold coloured ones marked 'GSC' just like the ones on your Epox, Time. Interestingly, I've just checked the caps I pulled off an old Epox 8KTA2 when repairing it and they are exactly the same design except the name on them is 'Yona'.

 

[Buck]

It would seem to me that the problem is not a fault of the capacitors, as they seem to be used in an out of spec situation.

 

[time]

Dear Sir,

Thanks for your picture.Base on your serial no.It is already out of
warranty.So sorry,our dealer or us can't replace another one or repair
without charge for you. But we can repair this board with charge,if you
need.

Thank you for your interest in EPoX products and contacting our support
department. If you reply ensure to include all previous E-mail text. It will
prevent unnecessary delays and guarantee the fastest possible response.

Best Regards,

Technical Support (AFA)
EPoX Computer Co., LTD. (Head office)
E-Mail : support@epox.com.tw
Web : www.epox.com


----- Original Message -----
To: <support@epox.com.tw>
Sent: Monday, January 19, 2004 11:59 AM
Subject: Re: Leaky capacitors

Here are some photos of one of the three faulty boards, serial number
011017000129.

I note that although having a later serial number, the replacement board
offered is identical to the faulty boards, including the same brand and type
of capacitor: GSC.


> Dear Sir,

> I'll contact our agent (Wasten) to help you.Before that,can you take
> picture
> about leaky capacitors of your boards.

> Thank you for your interest in EPoX products and contacting our support
> department. If you reply ensure to include all previous E-mail text. It
> will
> prevent unnecessary delays and guarantee the fastest possible response.

> ----- Original Message -----
> To: <support@epox.com.tw>
> Sent: Thursday, January 15, 2004 6:59 AM
> Subject: Re: Leaky capacitors

>> > Please provide the serial no. of your boards for us checking.I'll
>> > inform
>> > your
>> > request to related department.
>>
>> Please check 02052000059.
>>
>> It is from one of the *replacement* boards we have been offered. Our
> supplier does not know if they are part of a bad batch or not, so we would
> appreciate your confirmation.
>>

>>
>> > ----- Original Message -----
>> > To: <support@epox.com.tw>
>> > Sent: Monday, January 12, 2004 6:02 PM
>> > Subject: Leaky capacitors
>>
>>
>> >> Hi
>> >>
>> >> We have discovered that three 8KHA+ motherboards that we sold have leaky
>> > capacitors. The worst affected board has become unusable and the next worse
>> > appears to be heading the same way.
>> >>

>> >> As we did not become aware of the problem until outside the two year
>> > warranty, the local distributor in Australia (Westan) has indicated that
>> > they are unable to replace the boards without charge, although this is not
>> > in line with Australian law.
>> >>
>> >> I decided to contact you in case you have a policy for this problem that
>> > Westan is not aware of. But more importantly, is there any way you can
>> > guarantee that replacment boards will *not* have the same problem? I
>> > appreciate that the problem is industry-wide.
>> >>

 

[time]

I've just had an Epox 7KXA (slot A) fail. Same caps, same problem - just much older board.

At this stage, I suspect everything we've sold in the last three and a half years is more likely than not to fail - it's just a question of time. :evil:

Anyone who hasn't already read the posts here probably should.

 

[blakerwry]

thanks Time, I've checked the status of my soyo board and see that it is within 2 years of the purchase date. I'm going to contact them and see if I can get a replacement.

 

[Buck]

Any experience with Aopen boards and this capacitor issue? Any positive or negative experience with Aopen at all?

 

[Mercutio]

Good boards, but they cost too much for what they are. They need to be in line with Epox, and they aren't.

 

[blakerwry]

Dear customer,



All SOYO products have one year limited warranty from the date of purchase,
however, we can still repair or replace your out warranty products with a
flat fee $35 (except for AK, HI and international customers.) For details,
please refer to SOYO RMA policy online http://www.soyousa.com/rma/policy.php





If you have any out of warranty SOYO products need to be replaced or
repaired, please apply for an RMA number at www.soyousa.com/rma before send
in the RMA item. Your reference number is 40121C02


We appreciate your business and thank you for understanding.

----- Original Message -----
From: "Blake" <blake@you can pm me.com>
To: <rma@soyousa.com>
Sent: Wednesday, January 21, 2004 7:20 AM
Subject: Soyo Dragon+ with bad capacitors



>> Hello, I have a Soyo Dragon+ board that has developed bulging
>> capacitors. As I understand it this was a widespread problem 1-2 years
>> ago and on many boards is just now becoming a problem. My board was
>> purchased less than 2 years ago. I saw that your warranty normally
>> covers only 1 year. Since this is a widespread problem and it is
>> impossible to tell beforehand if the capacitors are bad will you replace
>> or repair the board before the capacitors leak and eventually explode?
>>
>> Sincerely,
>> -Blake Hudson
>>

 

[blakerwry]

Looks like they want the board to explode. This is worse than abit, however. Even though my Abit board was under warranty, they refused to fix the problem (even after admitting it was a problem) without me paying them a ~$15 fee + S&H.

No point in paying $35 to soyo when I can get a new non-soyo board for $50

 

[Mercutio]

Came into work this morning. My server was down. Bluescreened with different errors on every reboot. Opened it up to pull its main drive, noticed some bulging caps.

So I pulled it off its shelf, and took a look in the light.

Two caps had leaked brownish capacitor gunk over themselves, and 11 (I count 14 on the board of any size) other caps were bulging.

Board was an Asus 7V8X-MX

 

[curtnuggets]

I can confirm that Gigabyte motherboards have the same problem. My blue screens of death have been ever increasing until it crashed in the midddle of disk-check (after previous crash) and fubared my C drive to the point where i had to format it.

i have 4 caps that have already burst and 2 that are bulging. I put in an email to Gigabyte...we'll see what they say, but i doubt they'll help. I did find this:

http://www.spectrum.ieee.org/WEBONLY/resource/feb03/ncap.html

 

[CityK]

Yesterday, while venturing into my case for unrelated reasons, a quick visual inspection led to the discovery that the same two capacitors that took out Doug's recently departed MSI K7T-Pro2A are also now bulging and leaking in the case of my K7T-Turbo. I certainly wasn't shocked when I saw this, but I have to admit that I was a bit disappointed, as having had no previous indications of problems (was last inside the case during Christmas time), I was hoping that my board might remain immune to the problem.

Incidentally Doug, a quick search through MSI's forum gave plenty of testimony of those two blue guys just above the AGP slot being common culprits. A couple of people said that they didn't experience any stability problems, and so far none in my case either. But I don't hold out much faith as we all know what happen to yours.

Even if the board still fell within the 3 year limited warranty (which it just falls out of), RMA would not be worthwhile since the limited warranty is 2 years Parts & Labor, and the third Year Parts ONLY.....Applying the Blake analysis, it is apparent that shipping and labour costs alone would likely be as much as what it would cost to buy something in range of a new KT600.

I don't think I'll bother with a homemade remedy --> going out and finding replacement caps, then removing the bad ones and then soldering on good ones sounds distinctly like a lot of personal aggrevation.

Of course, there are alternatives (example), but again, you would really have to be quite fond of your board to pay that much to get it fixed, as opposed to the money much better spent option of buying a brand new board.

 

[LiamC]

It may pay to keep an eye (i?) on this.

http://www.theinquirer.net/?article=13962

Could lead to some positive action

 

[Buck]

I'm seeing the same for my VIA chipset based boards, such as the Soltek SL-75DRV4. Boards that I have sold based on the original nForce chipset or those based on Intel chipsets have not exhibited this problem yet.

 

[Buck]

Still seeing this problem. I just replaced another Soltek SL-75DRV4 and saw a customer's system running an i845 based board with 6 oozing capacitors.

 

[blakerwry]

Friend of mine just got a new PC because his old Athlon 850 Gateway2000 PC had leaky capacitors... they actually hissed as well.

 

[time]

Well, I now have a 100% failure rate on Epox boards sold between two and four years ago (at least, on the 8 or so boards I have been able to keep track of). The last two, bought two years ago (8RDA+ and 8RDA3+) failed within a week of each other - the former without warning but the latter had been deteoriating for months.

There are none more recent because we stopped selling Epox.

All except one Soltek 75KAV that I've managed to check had leaks, but only one has become unusable. I'll put that down to over-engineering. My beloved overclocked Soltek 75FRN-2 also soldiered on for some time with nearly every capacitor in ruins.

Stopped selling Soltek ages ago.

Now I just need to replace our Asus A7N266-VM - the caps are still only bulging on it.

Unfortunately, I've found myself still buying Asus.