PDA

View Full Version : I met the capacitor plague



Stephen Tashiro
01-01-2013, 11:42 PM
I recall reading in computer magazines years ago that motherboards were having a problem with bad capcitors. I figured the big companies would soon fix it and didn't notice any problems with capacitors on the motherboards that I bought. But now, I've discovered that some motherboards that I bought in about 2006 do have the problem.

I bought three Gigabyte GA-K8N Ultra-SLI motherboards when they were a relatively new product. One system would hang intermittently and I tried various remedies without success. The other two motherboards worked OK. Eventually they were replaced with newer systems. Last year, when I gave away the bad motherboard to a local surplus electronics place, a technician immediately told me that some of the capacitors were swollen.

I expected that a bad capacitor would be burst open or leaking. He showed me that the tops of some of the capacitors were not flat. Instead they looked like shallow cones. Later I examined the other two motherboards and they also had swollen capacitors. Even though those boards gave me years of good service, this episode has made me cautious about buying any used electronics from the capacitor plague years. There's an interesting article on the plague on the Wikipedia.

Jerome Stanek
01-02-2013, 7:31 AM
My wife's Dell desktop quite one morning and I checked the capacitors it had 10 that were swollen. I ended up buying some off ebay resoldered them in and it is running as good as new.

Rich Engelhardt
01-02-2013, 8:18 AM
LOL!
For real???

Caps on boards have actually gone bad?

We used to joke about bad caps all the time becuause of IBM's warranty policy. If we turned in a warranty claim w/IBM, the bad part had to be 100% proven bad, otherwise IBM would back charge us for the part.
A whole lot of times, with intermittent problems, you went with your "gut" feeling on what part was bad. 999 out of 1000 times, it fixed the problem.
We'd smear some soldering flux on the board around an electrolytic cap and list the problem on the warranty claim as a "bad capacitor" or "leaking capacitor".

This is too funny - to find out there really is/was such a thing.

Phil Thien
01-02-2013, 8:58 AM
LOL!
For real???

Caps on boards have actually gone bad?

We used to joke about bad caps all the time becuause of IBM's warranty policy. If we turned in a warranty claim w/IBM, the bad part had to be 100% proven bad, otherwise IBM would back charge us for the part.
A whole lot of times, with intermittent problems, you went with your "gut" feeling on what part was bad. 999 out of 1000 times, it fixed the problem.
We'd smear some soldering flux on the board around an electrolytic cap and list the problem on the warranty claim as a "bad capacitor" or "leaking capacitor".

This is too funny - to find out there really is/was such a thing.

Actually specifically with IBM I've worked with some clients with tons and tons of machines w/ bad caps. Still under warranty.

I've seen bad caps elsewhere, too. I've replaced caps on everything from parts from expensive automobiles to older computer still in service (that needed to remain in service).

The biggest lot of bad caps I've seen have been on Intel-made motherboards.

The Asus boards have had fewer bad caps, but we're beginning to see more there, too. Asus has switch to solid caps on many of their boards. I've had one or two of those go on a video card once, it is like someone is firing a little cap gun.

There were all sorts of stories a few years ago about counterfeit caps and bad compound. But I think a lot of the problem stems from board design.

Jerome Stanek
01-02-2013, 11:42 AM
I had a cap explode in a computer power supply it sounded like a shotgun going off also one exploded in a power brick for an external hard drive that one was small and sounded like a cap gun.

Greg Portland
01-02-2013, 2:19 PM
There were all sorts of stories a few years ago about counterfeit caps and bad compound. But I think a lot of the problem stems from board design.
There have always been bad designs out there that could stress components. However, this was not the case for major MB suppliers... they unknowingly used counterfeit capacitors or used caps with equal specs but poor quality electrolytes.

It should be mentioned that the capacitor plague was most prominent in the early 2000s. However, the problem (3rd party vendors sourcing inferior product) still exists. This is why you'll see some manufacturers advertise the use of more expensive or higher quality capacitors in their premium lines.

Personally, I got bitten with two different Asus motherboards.

Rich Engelhardt
01-02-2013, 4:44 PM
It should be mentioned that the capacitor plague was most prominent in the early 2000sThat probably explains why I'm just now hearing about it.
I got away from doing actual computer repair in 1998 and moved into strictly network/networking systems support.
From 1998 until I retired last year, I probably only ran across 5 or 6 IBM servers in all that time.

It (popping caps) just stuck me as funny that there was actually a real problem.

Phil Thien
01-02-2013, 6:21 PM
There have always been bad designs out there that could stress components. However, this was not the case for major MB suppliers... they unknowingly used counterfeit capacitors or used caps with equal specs but poor quality electrolytes.

It should be mentioned that the capacitor plague was most prominent in the early 2000s. However, the problem (3rd party vendors sourcing inferior product) still exists. This is why you'll see some manufacturers advertise the use of more expensive or higher quality capacitors in their premium lines.

Personally, I got bitten with two different Asus motherboards.

Not to be argumentative, but I disagree with most of that. There is a great article on Wikipedia about all of this. I'm quoting some of it below, and adding emphasis.

In terms of the bad caps we've seen, there have been far fewer on Asus boards. I fix PC's/servers for a living, working on literally hundreds of machines every year, for 25 years.

But I assure you, the problem is ongoing and has not slowed at all. We see plenty of machines that were made in the last 1-2 years with bad caps.

Wikipedia link: http://en.wikipedia.org/wiki/Capacitor_plague


*****
Continuing failures
The first publicized press releases about the widespread problem with premature failures of Taiwanese electrolytic capacitors appeared in September 2002. It might be assumed that by mid-2003 the affected capacitor manufacturers would have changed their production process and used a "correct" electrolyte mixture. With a typical shortened life span of about 1.5 to 3 years for the failing capacitors from mid-2003 up to mid-2006, the last of the bad capacitors should have failed by 2007. Commentators on the Internet often predicted the year 2007 would be the end point for "bad capacitors".

In theory, the defective capacitors should have "failed out" of active use, and very few new incidents should occur. But even after the year 2007, the year in which the failure of Taiwan bad capacitors with the wrong electrolyte should really be over, new complaints about failed capacitors are reported on the Internet.[32]

Burst capacitor with the date code V9A, (production date September 2007), failed after about 2.5 years of operation
The problem of bursting electrolytic capacitors still exists, and recent images of failed capacitors show identical effects with open vents and expelled rubber plugs. Affected by these failures are capacitor series for rated voltages from 6.3 V to 100 V, which have one thing in common, they have a water-based electrolyte with a high water content of up to 75%. In the catalogs of the manufacturers, they are characterized by the catchword "Low-ESR" capacitors or "Low-Impedance", "Ultra-Low-Impedance" or "High-Ripple-Current" capacitors.

These electrolytic capacitors must not be confused with aluminum-polymer solid-electrolyte capacitors, which are also often called "Low-ESR" electrolytic capacitors. The failures discussed here affect only aluminum electrolytic capacitors with a non-solid electrolyte. If failures occur now in these type of capacitors with a newer production date than 2007, it cannot be blamed on the incomplete electrolyte formula of earlier Taiwanese production. If, in an SEM and EDX analysis of the failed capacitors, aluminum hydroxide is still detected, the electrolyte is not necessarily faulty because it also can be caused by improper circuit design. Therefore, two questions must be answered first:
1.Was the ripple current and temperature stress on the capacitors within specifications?
2.Did the failure occur prematurely, before or after the end of rated lifetime?

Steve Meliza
01-02-2013, 7:17 PM
Electronics designed to last 10 to 20 years or more use no electrolytic capacitors because of their relatively short lifespan even if properly made. No, I have no article at hand to back that up, just industry experience is designing electronic systems with capacitors.

Phil Thien
01-02-2013, 8:01 PM
Electronics designed to last 10 to 20 years or more use no electrolytic capacitors because of their relatively short lifespan even if properly made. No, I have no article at hand to back that up, just industry experience is designing electronic systems with capacitors.

Related to that is the fact that, until the advent of high-speed internet in homes (DSL, cable, etc.), people typically turned their computers off when they weren't using them. Maybe they ran them 1-2 hours/day, if even that.

With those high-speed internet offerings, and youtube, and ebay and online games, etc., people use their computers a lot more and often leave them powered up around the clock.

And when did the real exponential growth of high-speed Internet start to happen? Somewhere around 2000.

It was really a perfect storm of circumstances (new higher power CPU's and video cards, etc., more runtime, more computers period).

It is my feeling that much of this is a design issue. Designing motherboards for PC's that are no longer run intermittently is a challenge. Some do it better than others.

Ken Fitzgerald
01-02-2013, 9:09 PM
Component manufacturers having bad manufacturing runs is not just a new event.

My employer got burned by a bad batch of high power RF diodes almost 30 years ago.

Engineers mistakenly specifying parts that don't meet circuit requirements or operating conditions isn't a new thing either.

Brian Elfert
01-02-2013, 11:13 PM
My employer has HP/Compaq and Sun servers that have been running 24x7 since 2003/2004. The only failures have been a couple of power supplies and it is unknown if it was caps that failed in those. We have a Sun server that has been running since 1997 or 1998! Maybe these servers use better parts.

Ken Fitzgerald
01-02-2013, 11:34 PM
Brian,

The environment in which any electronics operates has a lot to do with reliability too!

If operated where the temperature is too warm....and the humity too low or too high.......electronic failures will increase.

Dan Hintz
01-03-2013, 6:35 AM
Electronics designed to last 10 to 20 years or more use no electrolytic capacitors because of their relatively short lifespan even if properly made.
My experience shows nothing of the sort... quite likely based upon differing design fields / projects. Electrolytics are alive and well (at least until they dry out).

Bill Huber
01-03-2013, 8:27 AM
I would like to have a $1 for every monitor that people have pitched just because of a few caps that were bad.

I have repaired about 30 HP PCs and another 40 Dell and HP monitors and all was the same thing bad caps.

It was so bad that there are people on the net where you can buy cap kits for your monitor. For me it has slowed down and the last bunch of monitors have been fine now for about a year.

Steve Meliza
01-03-2013, 8:44 AM
Yes, they are alive and well because they're cheap and effective while still within their lifespan. I was simply trying to point out that there are companies out there purposely avoiding electrolytic caps due to their relatively short lifetime as compared to the expected lifetime of the electronics being designed and manufactured.

Those not familiar with electronics probably won't be surprised to learn that capacitors come in various lifespans, rated in thousand(s) of hours of operation at a certain temperature and more money buys you caps that will last longer and cheap buys you something that needs replaced in a year or two.

Brian Elfert
01-03-2013, 8:46 AM
From what I've seen with these sorts of failures is it is often isolated to specific models. My co-worker had to replace some caps that blew in his TV, but that model was known for the problem. I had a Dell laptop adapter that was built in 2011 pop and quit working. I assume it was a cap. It was under warranty so Dell replaced it.

I've been lucky I guess. I haven't had any electronics fail in years other than the Dell laptop adapter.

Dan Hintz
01-03-2013, 10:31 AM
Our bedroom TV (a Sharp) likely has caps that need replacing. About 1 in 10 times the picture does not come on with the TV, and when it happens it usually takes a few on/off cycles before the picture comes back. Prior to this, the 1 in 10 shot had the pic come on, but it was either extremely low on contrast (but very whiteish) or had a major hue to it (as if one of the color signals wasn't getting through).

Greg Portland
01-03-2013, 3:03 PM
Not to be argumentative, but I disagree with most of that. There is a great article on Wikipedia about all of this. I'm quoting some of it below, and adding emphasis.

In terms of the bad caps we've seen, there have been far fewer on Asus boards. I fix PC's/servers for a living, working on literally hundreds of machines every year, for 25 years.

But I assure you, the problem is ongoing and has not slowed at all. We see plenty of machines that were made in the last 1-2 years with bad caps.

Wikipedia link: http://en.wikipedia.org/wiki/Capacitor_plague

It seems that we completely agree. I'm not sure what you're disagreeing with?

There have always been bad designs out there that could stress components. <-- true

However, this was not the case for major MB suppliers... they unknowingly used counterfeit capacitors or used caps with equal specs but poor quality electrolytes. <-- true (from your wiki article)

It should be mentioned that the capacitor plague was most prominent in the early 2000s. However, the problem (3rd party vendors sourcing inferior product) still exists. <-- true (from your wiki article). Perhaps you thought "inferior product" was referring to the electrolyte formula? I meant that companies are using cheap (poorly spec'd or failed prior to EOL) parts.

This is why you'll see some manufacturers advertise the use of more expensive or higher quality capacitors in their premium lines. <-- You can see an example of these "enhanced durability" products on Gigabyte's website

Personally, I got bitten with two different Asus motherboards. <-- This is my personal experience and not up for debate

Greg Portland
01-03-2013, 3:07 PM
Maybe these servers use better parts.
You are correct, they do. It's not difficult (for a good engineer) to design a motherboard with a longer MTBF. However, the parts cost is prohibitive for most of the consumer market. Dell and HP aren't going to drop $300 motherboards into their $500 entry level computers.

Phil Thien
01-03-2013, 6:38 PM
It seems that we completely agree. I'm not sure what you're disagreeing with?

There have always been bad designs out there that could stress components. <-- true

However, this was not the case for major MB suppliers... they unknowingly used counterfeit capacitors or used caps with equal specs but poor quality electrolytes. <-- true (from your wiki article)

It should be mentioned that the capacitor plague was most prominent in the early 2000s. However, the problem (3rd party vendors sourcing inferior product) still exists. <-- true (from your wiki article). Perhaps you thought "inferior product" was referring to the electrolyte formula? I meant that companies are using cheap (poorly spec'd or failed prior to EOL) parts.

This is why you'll see some manufacturers advertise the use of more expensive or higher quality capacitors in their premium lines. <-- You can see an example of these "enhanced durability" products on Gigabyte's website

Personally, I got bitten with two different Asus motherboards. <-- This is my personal experience and not up for debate

I thought your point was that the problem occurred mostly in the early 2000's, and was related to counterfeit components that were the result of industrial espionage?

You have to read the entire Wikipedia article to make sense of it. It is written almost chronologically, so context is important.

Here, I'll quote the section I feel is relevant again:

If failures occur now in these type of capacitors with a newer production date than 2007, it cannot be blamed on the incomplete electrolyte formula of earlier Taiwanese production. If, in an SEM and EDX analysis of the failed capacitors, aluminum hydroxide is still detected, the electrolyte is not necessarily faulty because it also can be caused by improper circuit design.

What they are alluding to is the possibility that the industrial espionage explanation may have been overstated.

The stories of the industrial espionage all read like a cheap paperback novel. That is, no names of the scientist that took the formula, no names or acknowledgement of the companies he sold it to, or the outfits that distributed the product. Just a bunch of whispers by a few people that don't want to name their sources. No stories in any of the magazines that cater to these industries about reimbursements paid to Intel or any others by an embarrassed manufacturer or distributor. I'd think that with companies like Apple, Intel, and Dell getting burned to the tune of millions of dollars, we'd have a lot more details.

So I disagree this had anything to do with industrial espionage, or that the problem was most prominent in the early 2000's.

The problem continues to today, and probably has a lot to do with the extended time computers are left powered-up due to the utility people enjoy with high-speed Internet and all their favorite sites.

Phil Thien
01-03-2013, 6:44 PM
You are correct, they do. It's not difficult (for a good engineer) to design a motherboard with a longer MTBF. However, the parts cost is prohibitive for most of the consumer market. Dell and HP aren't going to drop $300 motherboards into their $500 entry level computers.

You can find plenty of complaints about Panasonic and Nichicon and Sanyo and other brands of Japanese caps. And the price differences between those Japanese products, and anything made in China/Taiwan, is not that significant on a per-board basis.

Many of the caps I've replaced were Nichicon. A lot of Nichicon. There were not counterfeit. So the board already had the cost of the Japanese caps built into it, and those caps still failed.

Greg Portland
01-04-2013, 6:54 PM
The problem continues to today, and probably has a lot to do with the extended time computers are left powered-up due to the utility people enjoy with high-speed Internet and all their favorite sites.

"Problem" to me means that the systems are not meeting the manufacturer's claimed FIT rate and MTBF (this was happening in 2000 when large numbers of boards were dying in 6 months). Are you claiming this?


You can find plenty of complaints about Panasonic and Nichicon and Sanyo and other brands of Japanese caps. And the price differences between those Japanese products, and anything made in China/Taiwan, is not that significant on a per-board basis.

Many of the caps I've replaced were Nichicon. A lot of Nichicon. There were not counterfeit. So the board already had the cost of the Japanese caps built into it, and those caps still failed.
I'm not talking about brand name, I'm talking about 85C versus 105C+ caps plus their MTBF and FIT rates when I say "higher quality parts"...

Phil Thien
01-04-2013, 10:46 PM
[/COLOR]
"Problem" to me means that the systems are not meeting the manufacturer's claimed FIT rate and MTBF (this was happening in 2000 when large numbers of boards were dying in 6 months). Are you claiming this?

We never saw large #'s at six months. By the time we see them, they're typically 10-14 months old. And yes, it still happens.


I'm not talking about brand name, I'm talking about 85C versus 105C+ caps plus their MTBF and FIT rates when I say "higher quality parts"...

Okay, got it. Yes, that is clearly a large part of the problem, IMHO.