Talk:Capacitor plague

From Wikipedia, the free encyclopedia

This is the talk page for discussing improvements to the Capacitor plague article.
Article policies

Contents

[edit] Still Going on

I am a service tech and I have replaced quite a few power supplies in 32" and 40" LCD displays made by a major Korean electronics manufacturer. The mfg. dates on these displays is as new as 08/2006.


[edit] 2000s?

What is "in the early to mid 2000s" supposed to mean? 2000-2005? 2000-2050? 2000-2500? I can figure it out by ellimination, but it seems like an awful way of saying that.

[edit] Initial comments

  • Bushytails 04:20, 30 Dec 2004 (UTC) First revision. Not bad for a couple hours of writing. (IMHO, at least). I have a fair pile of motherboards waiting to be recapped (my last order of caps is taking a bit long to show up... (censored) christian holidays slowing up the post!); will take some photos of them tomorrow to show what failing caps look like. Will probably also fill in some of the missing sections, expand some parts of it, re-check for typos, then sit back and fluff my tail happily.
  • --Wiki Tiki God 12:42, 26 October 2005 (UTC) I first heard about this in Nov 2003 when building a computer and a lot of articles were commenting on capacitor failures on motherboards by companies that were (previously) known for rock solid quality (like Asus, and MSI). It was one of the reasons I just went with a genuine Intel board. But I haven't heard ANYTHING about this in the last year (except in 'economy' motherboards like ECS, which always sufffer from poor quality control anyway). At the time I heard it was because a lot of motherboard manufacturers were in a cutthroat, micro margin, price war (like memory makers sometimes do), and that now, most MB makers have realized that loosing a customer for life (like me) is not worth saving 23 cents per board. Perhaps an update is needed?
  • Novous 17:15, 11 April 2006 (UTC) I think the section on "recapping" a motherboard should be rethought. It's currently written from a scared consumer's point-of-view. Almost anyone who can solder can easily replace blown capacitors on a motherboard, but this section sounds like motherboards are some crazy magical black art that should never be tampered with. People who don't know what they're doing shouldn't touch them, but I see no reason why this article should scare people away from it. But then again, I imagine anyone who knows how to would ignore this section anyway.

--timeheater 03:29, 4 May 2006 (UTC) i agree that it could be written a bit differently, but i do think that stressing the complexity of motherboards as well as other "cutting edge" electronics is necessary - i work in production at an electronics company and don't have problems with SMD work, but some of the boards scare the hell out of me! pc boards with more than two layers is tricky even if you have lots of experience; they probably weren't designed to be repaired in the first place. i'd be more worried about lifting pads than ruining nearby components... and given that the caps are bigger, as well as power sections usually being a bit more sparse, someone with experience shouldn't have a big problem replacing them, but anyone who isn't fully competent should keep in mind that they may fail. if i get a chance in the next couple days i'll try and rewrite it a bit.

[edit] Motherboard companies not to blame?

Motherboard companies not to blame for capacitor plague?  Who do you think chose  the bad components for installation in the first place? The user of components is ultimately to blame when they choose low-ball parts to save 23 cents per board, fail to do due diligence in qualifying the parts and qualifying the vendor, fail to perform ongoing incoming quality inspection of parts and don't perform statistical process control of the finished products.—Preceding unsigned comment added by Hydrargyrum (talk • contribs)

I agree with the above statement. This seems like a possible NPOV infringement. Stovetopcookies 17:49, 30 November 2006 (UTC)

It's hard for me to agree with the assertion that the motherboard companies are responsible. They want to buy the cheapest parts they can in order to maximize their profit margin. Otherwise, their investors, shareholders, and owners will think they're irresponsible, too. I'm not sure where the figure of 23 cents per board comes from, but over millions of units and dozens of parts per unit, the prices add up fast, particularly in an industry with no margin.
Meanwhile, the blame statement is unreferenced though I'm not sure it's POV. -- Mikeblas 22:25, 30 November 2006 (UTC)

The page now says " The motherboard companies assembled and sold boards with faulty caps sourced from other manufacturers (see below)", which doesn't seem right, either. At the time the motherboard manufacturers bought the capacitors, they didn't know they were faulty. -- Mikeblas 12:12, 4 December 2006 (UTC)

Motherboard manufactors ought to do their own quality ensurance check. They decide what shoddy sourcing deals to make. Not the end user. Any uncertainty in regards to reliability is likely to get them ditched due no benefit of dault. Because of how corporations are run these days. Electron9 18:36, 12 August 2007 (UTC)

[edit] Too x86-pc focused?

  • --AaronWL 03:03, 20 June 2006 (UTC) This is a strange article, in that it talks about a common failure mode for capacitors in general, a common electronics component, yet is entirely centric to some specific application. Surely PC system integrators are not the only people who feel they experience electolytic capacitor failure? I'd like for an expert who is familiar with the technical details on this mode of failure to transform this discussion into something more generic, with a specific section for PC motherboards.

[edit] Dangerous Voltages

I see this uncited passage has been added again to the article:

Recapping on a power supply unit should NEVER be attempted by a non-professional as dangerous voltages can remain present across the unit's functioning capacitors for as long as days even after being unplugged.

Can someone explain that, or better yet provide a citation that does? What capacitor in a computer power supply will cary dangerous voltages? For days after power is removed? Unlike a television, the power supply doesn't produce lethal votages. There are no capacitors which are charged to thousands (or even hundreds) of volts on the output side of the transformer. -- Mikeblas 15:31, 4 July 2006 (UTC)

The filter capacitors and tank capacitor in a switching power supply don't hold voltages much larger than the output voltage of the supply. For the power supplies used in PC's, this means they're not much over 12 volts DC. The capacitors aren't large enough to hold a charge for very long as their own leakage current (and the unswitched load shown to the PSU by the motherboard and peripherals) will dissipate their charge in a time much shorter than "days".
Without any references to the contrary, then, I'm removing the "recapping" warning. -- Mikeblas 15:39, 11 July 2006 (UTC)

Sorry but you are wrong. The input filter condensers of a computer power supply can be rated as high as 200 WVDC, as these are used in rectifying an AC signal directly from the source.Stovetopcookies 21:17, 30 November 2006 (UTC)

Why would they be rated so high? The secondary output of the transformer is almost always less than 50 VAC, so the rectified and filtered DC voltage is less than 70 VDC or so. It makes no sense to put electrolytic filter capacitors directly across the source, as you're suggesting; it'll also make it pretty hard to get UL certification. Because of the limited space, capacitors with a working voltage of 200 are never used on motherboards. In a computer power suppy, I think you'll have a hard time finding caps with a working voltage higher than 100 or 150 volts. The working voltage, of course, is a specification about the capacitor and does not indicate that 200 volts is present in the circuit. -- Mikeblas 22:01, 30 November 2006 (UTC)
If you readup on switched power supplies you will see that they will rectify your 110/240V AC into DC. Stabilise this intermediate DC with a capacitor. Chop with a power transistor in the kHz range. Feed an high frequency transformer. The output is then again rectified and stabilised with a capacitor.
The output from the PSU are in some sections of the motherboard subject to the same process again to provide the obscure CPU voltages.
The shock risks are mainly with the high voltage input DC capacitor stage (~339V). In theory maybe the coils can be a potential hazard aswell. But I haven't heard of such occurence. The PSU output stage, and the on-motherboard switched psu should be harmless.
This means any PSU, UPS, CRT, "Inverter", etc.. must be treated with great respect.
The Voltage rating is just what the capacitor can handle in normal operation, not what it actually contains. —Preceding unsigned comment added by Electron9 (talk • contribs) 22:05, August 29, 2007 (UTC)

For the record, in North American mode, computer power supplies use a voltage doubler, while in European mode they just rectify the line voltage. This means that the input capacitor will be charged to ~340V DC worldwide. The mains voltage bulk capacitor is marginally capable of delivering a lethal shock (based on a lethal current range of 6-200mA, and a required duration of 1-3s[1], and using a capacitance charge/discharge calculator [2]). Certainly many people have survived this kind of capacitor discharge, but it is in dangerous territory, and really very unpleasant.--76.10.128.59 (talk) 04:50, 10 December 2007 (UTC)--


And since the metal pcb tracks conduct the danger around the board, merely warning people about charged caps does not stop some of them getting bitten. Tabby (talk) 01:50, 12 January 2008 (UTC)

[edit] References?

This article indicts many vendors by name, but provides no references for that indictment. Is that responsible? -- Mikeblas 02:44, 6 July 2006 (UTC)

I've removed the uncited section. -- Mikeblas 20:29, 22 July 2006 (UTC)

[edit] "Capacitor Plague"?

Why is this called the "Capacitor Plague"? Searching the web, I find very few references to the term. -- Mikeblas 12:45, 20 July 2006 (UTC)

It needed a name and this one is fitting as such. Capacitors (or condensers) are involved, and a plague, or "illness" if you will, strikes them. This problem was widespread, as is a plague. Stovetopcookies 21:17, 30 November 2006 (UTC)

I disagree. I don't recall seeing any technical texts (consumer level computer magazines don't count) refer to the problem in these terms, and the term does not describe the problem. "Capacitor plague" implies a problem with the very presence of capacitors when they are essential components. 79.77.101.45 (talk) 21:08, 8 June 2008 (UTC)

[edit] Capacitor brands.

Is it really necessary to remove all bad and known cap brands. Obvisously Chhsi, Tayeh, Choyo nad Jun Fu are bad as there a pictures dipicting these bad caps. I have quite a few boards with leaking G-LUXON, OST, GSC and Licon on them (The Licon caps are on a Point Of View GeForce FX-5500)

On the contrary, I have ECS motherboards littered with G-LUXON, OST and GSC caps and they are fully stable and they are not leaking so I guess it depends on the model of the Caps. —Preceding unsigned comment added by Mjr4077au (talk • contribs)

Without any verifiable sources that show those brands as affected, yes, they should be removed. The pictures show that single units of the particular brand failed, not that there's something wrong with the brand of capactiors as a whole. Those units that failed may have failed for any of several reasons, not just bad electrolyte. (Even then, I would expect that a particular line from the manufactrer to be affected, not all capacitors made by the brand.) As I indicated above, beyond Wikipedia policy for verifiability, I think it is irresponsible to indict the products of a company without presenting any relevant evidence whatsoever. -- Mikeblas 02:12, 5 August 2006 (UTC)

Thanks to person who added the IEEE references (2002, 2003, 2006), this was covered -- and I believe that the manufacutrer (Taiwan?) was mentioned in this industrial espionage case. As far as branding -- if it is a brand that this mfg. used -- this is exactly what brand marketing profesisnals hate - turning around a tainted product name or brand -- due to management failures (accidental or deliberate). Beatgr 02:10, 11 October 2006 (UTC)

No problem; I have my nose in the journals most of the time, anyway. There is a very long list of manufacturers who have been implicated in the issue, but I've never seen any hard evidence in any journal about which brands really were guilty. As far as I have seen, countries where the espionage took place haven't said anything about court cases or settlements. Becasue the reputations of involved companies are at stake, I think it's very important to make sure we get very good references before listing specific names. -- Mikeblas 22:15, 30 November 2006 (UTC)

I have again removed this list; it was replaced without references. -- Mikeblas 17:00, 22 August 2007 (UTC)

The best references you will find are the badcaps.net fourms, where way back when, I took the original, cited list from. Until someone went ahead and deleted it. If you're not going to consider their forums an acceptable reference, then it is unlikely you will ever find a reference you like, as just about all the manufacturers deny it. As a list of commonly failing brands is highly useful for the article, giving users an idea of the scale of the problem, the tendency of copycat brands to appear, information that will aid in checking their own computers, etc, one should be provided. Bushytails 18:02, 22 August 2007 (UTC)

It's not that I don't accept links to forums as references; it's that Wikipedia doesn't. I this case, I think it's hard to disagree that the people posting in forums don't know for sure if the capacitors failed because of bad electrolyte or because of other reasons. Magazines like EDN, Electronic Times, some of the IEEE magazines have all run articles about the problem, and I think they might provide reliable sources about the issue. Unfortunately, I don't have back issues that old. -- Mikeblas 20:27, 29 August 2007 (UTC)

[edit] AV Equipment

What about including a discussion of aging caps in vintage stereo equipment? -- Slowmover 16:44, 15 November 2006 (UTC)

This article is about a bad batch of capacitors in the early 2000s. The natural aging of capacitors would be best covered in the main capacitor article, I think. -- Mikeblas 22:35, 30 November 2006 (UTC)

[edit] ABIT was not the only one.

Juventas 04:53, 7 January 2007 (UTC) Intel did post a support article for their products that were effected, but have since removed it. I assume this is because their effected products are now outside of their support period. Google has a cached copy here: [1] I'm quite new to wikipedia editing, so I'll let someone else correct it if they feel its justified.

also, Dell totally denied the problem. wondering if I should include that... Dell sucks tbh. ~Crazytales (Talk) 12:57, 7 January 2007 (UTC)

Crazytales: As far as I know Dell never denied the problem, they simply didn't address it in any manner. If you have a url that shows otherwise, please post it. Juventas 03:08, 12 April 2007 (UTC)

[edit] Failure Analysis Rubbish

Found this at the end of the article.

The second theory is that as the capacitance decreases and the ESR increases, the buck controller for the voltage regulator increases the switching frequency to compensate for the load. Since most of the MOSFET's heat output is produced during the switching transitions, the increased frequency causes them to overheat.

This is complete garbage. The frequency doesn't change at all, rather, the duty cycle does. As the capacitor fails and the duty cycle increases, eventually the duty cycle will reach 100% as the regulator tries to make up for the cap. This is the reason for failure if it is a reason at all... not some garbage about increasing frequency. Buck regulators operate independently of frequency.

--Noishe 23:42, 4 February 2007 (UTC)

While I'm not sure the article's version is correct, your explanation is equally rubbish.  :)
The regulator is a feedback loop... In a fixed frequency, variable duty cycle regulator (not all are this type, many controller chips used a fixed on-time with variable off-time), the lower the voltage, the higher the duty cycle. Increasing the duty cycle, especially anywhere near 100%, will cause the voltage to rise unacceptably high, regardless of what the capacitors are doing... which the regulator won't do, as a higher voltage would cause the duty cycle to decrease. If regulators did increase the duty cycle as the capacitors failed, the system would crash due to cpu overvoltage or overtemperature well before the mosfets melted. (not to mention that no regulator chip worth using would keep increasing the duty cycle beyond what's needed to give a reasonable approximation of the desired average output voltage). Looking at it from another point of view, if you assume the current draw of the powered components remains the same, the average current through the mosfets must also remain the same. And the on-time isn't really relevant anyway, as the mosfets specced for VRMs usually have very low Rds(on), on the order of a couple miliohms... any increase in on-time enough to cause substantial additional heating of the fets would have already fried every powered component.
If you have a better theory, please update the article with it... but the increasing-duty-cycle one is not all too sound. Bushytails 04:21, 5 February 2007 (UTC)

[edit] Failure background info

This would be a better article if it contained more info about traditional capacitor failure modes, over the decades, and cross-references and links to extensive info about that more general subject. In particular, I seem to recall that when resurrecting very old electronic equipment (old vacuum tube radios), there is usually a problem with the capacitors (they dry out?) and there is a special technique to bring them up to operating voltage slowly, to try to re-form them. Or you can just replace them all first, on principle...-69.87.199.150 11:47, 14 September 2007 (UTC)

But this article isn't about general capacitor failures, it's about the plague of ones shipped in a certain year range with faulty electrolyte. An article about, say, a specific earthquake, wouldn't need to give information about the causes of other earthquakes... Bushytails 16:15, 14 September 2007 (UTC)
I guess the question is, is this particular incidence of "plague" a unique occurrence? Did capacitors not fail like this in any other recorded instances? If it is the case that every capacitor that fails in this mode is traceable to the plague, then the article lead-in should perhaps be more explicit about this -- but it's also a bold claim that requires some verification. If this isn't the case, then the article should be restructured so that the plague is discussed as a special case of this occurrence rather than the actual cause and source of these bad caps. It isn't obvious to the casual reader what the actual nature of the situation is. 24.6.88.7 02:37, 26 October 2007 (UTC)
I am afraid I have to be a bit sceptical about this whole article. I repair consumer goods - TVs, VCRs etc - and the humble electrolytic is the most unreliable component there is. A capacitor next to a hot heatsink in a power supply is almost guaranteed to fail after a year or two, I am changing them all the time. It takes a very brave person to say that this instance is caused by sabotage rather than the basic unreliability of a cheaply produced item. Dsergeant 20:25, 2 December 2007 (UTC)
If it were only ones next to hot heatsinks in power supplies, that might be a reasonable assumption - but some of these are simple stiffening capacitors, located down on the cold end of a board, with no ripple to speak of... And if you want brave people saying that, please consult IEEE's Spectrum, one of the first places to research the origin of this problem. It's also limited to a certain group of fly-by-night brands, all from the same area, all with near-identical products, many even intentionally misrepresenting themselves (Rulycon vs Rubycon) or producing blatant forgeries (several major brands have had issues with said fly-by-night producers putting their brand name on things!), and did not happen before or after this incident. I'd add a list of some of the brands that do this, but "someone" keeps deleting the list, so you'll just have to live without one. I do a fair bit of electronics repair as well, and aside from other products that got the same brand capacitors as are failing on motherboards, they're far, far, far more reliable... and when they do fail, there's usually an identifiable cause, or they slowly lose capacitance and increase esr over time, not the gas pressure-related bursting typical to ones made with this electrolyte. Bushytails 23:26, 2 December 2007 (UTC)


[edit] Marginal design

Lytics normally decrease in capacity and increase ESR over time. Marginally designed circuits can work fine when new, but falling cap effectiveness causes equipment failure later. Too often bad caps are blamed, when in reality the problem was corner cutting design, and the caps are not faulty, just too small for long term reliability.

Another type of marginal design occurs where high ripple is used in caps unable to handle it long term. The high ripple causes cap degradation, but again the problem is the circuit design rather than caps with faulty chemistry.

This could all do with clearer explanation in the article. Tabby (talk) 01:59, 12 January 2008 (UTC)

That is all correct, however this article is about caps that fail well before the manufacturers claim they should, due to a specific reason. I have seen capacitors of the various brands (the list of which keeps getting deleted from the article) explode after a couple years of zero-ripple, low-temperature use... An electrolytic is expected to gradually decrease capacitance and increase ESR with age... not to explode! Some studies have found the capacitance _increases_ with time on some of these capacitors, peaking shortly before they catastrophically fail. Others just increase esr at very high rates. Some seem perfectly normal, then explode in your face.
It actually surprises me how well some motherboards work as the capacitors fail... the computer is still running, despite most of the capacitors having sprayed electrolyte all over the board, and measuring essentially zero capacitance and/or infinite ESR... I don't think the motherboard design is to blame. Especially since replacing the capacitors with name-brand capacitors of the same rating (and sometimes inferior ratings... the ones I use are slightly higher ESR than some of the exploding ones claim to be, just because they're easier to get) never results in a repeat failure. Bushytails (talk) 05:15, 12 January 2008 (UTC)