"capacitor plague"

[qrz]

from a link posted on the www.wa6ati.com forum 

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

of course , 'lytics can and do fail for a number of reasons .
but, this one is why i stick with certain mfgrs   

[BobA]

Had a couple of motherboards fail due to this problem.   Now I try to get MBs with premium caps even though the caps with this problem were from a while back.   Better quality caps are always a plus.

As an aside are you a ham?

Bob
VE6SN

[grantspain]

I think the biostar geforce 6100-m2 motherboard sums this up

[qrz]

qsl.....16 years and counting .
electronic servicing >30
seen many changes - and not all good 

73 de  N9YOD

btw, check ur log ....
call sounds vaguely familiar . might have worked you - K9WJU ??

[Cothien]

i remember seeing a lot of that at one of my jobs. had a model of the ibm desktop that did it.

I always find it kind of interesting how when one goes there is normally a cluster that goes.

[newmanfamilyvlogs]

My K7 board I built fresh out of high school.
after reinstalling windows three times because of weird crashes I opened the case. Smelled it before I saw it.

[smartbomb2084]

The above picture reminds me of MERIT MEGATOUCH FORCE and MEGATOUCH MAXX UNICORN motherboards. It seems that it is always the 1000MF/6.3 volt caps that go on motherboards regardless of manufacturer. The green Chinese caps of the LUXAN brand name  come to mind, they could be the very caps pictured... are they?

[lilshawn]

The above picture reminds me of MERIT MEGATOUCH FORCE and MEGATOUCH MAXX UNICORN motherboards. It seems that it is always the 1000MF/6.3 volt caps that go on motherboards regardless of manufacturer. The green Chinese caps of the LUXAN brand name  come to mind, they could be the very caps pictured... are they?

yeah, i'm going to go kill myself now. those "G-luxon" caps are junk. the radion boards are all starting to fail now. ive even had some of the ion boards crap out. (OST branded) the first wave of BBH Pro computer boards (before dell) are all capped out now too. (also OST brand)

i up cap the 6.3's to 10 volt (narrow gauge) they are nearly identical in diameter to the 6.3's... though about 4mm taller they still fit alright the the old space. Nichicon are the best ive found so far. i havent had to re-replace one yet.

generally any brand that has CON at the end raycon, taicon, Jackcon, tocon... (Nichicon exempt) are crap.

FYI: if your mobo has bad caps, chances are the power supply caps are toast. I always up the power supply wattage if possible. ive been putting 300w power supplys in the Merrit Ion machines, it's a little bit of work, but they run cooler than the original 125 watt and later 200 watt power supplys... and we know how bad those elite edge machines where/are.

[NoRefunds]

I have at least one of my ION's or Aurora come through each week.  Some of those boards are not even two years old.    Total crap power supplies and MB's.  A 300 watt power supply does seem to solve that issue and run cooler but I'm so tired of recapping those boards.  Like lilshawn said, the Big Buck Pro boards have capacitor problems on the motherboards and power supplies.  I've seen a few JVL's with cap problems as well.  I've seen them on video cards on games from Global VR. 

Cheap power supplies with bad caps are all in the new crane machines like Toy Soldier, Toy Taxi, etc.

[MonMotha]

My experience has been that it's not usually the cheap capacitors themselves.  The "bad cap plague" where things really were just failing for no good reason seems to have subsided a few years ago.  The issue seems to be that, in an effort to save a couple tenths of a cent per unit, incompetent product sourcing bean counters substitute cheap caps for what the engineering people actually spec'd.  They usually do this based purely on case size, capacitance, and voltage rating.  This would be fine except that 1) there are other specs that matter (ESR, ESL, rated ripple, etc.), and 2) The specifications on those cheapo caps are often only good for 2000-3000 hours whereas the quality, name brand caps are often rated 6000+.  The bean counters will also often drop down to 85C instead of 105C ratings.  All this combines to drastically shorten the life of even a properly manufactured capacitor.

When replacing, I always upspec by a voltage step, if I can, go for 105C if it can possibly be made to fit, and get the highest ripple rated (and reasonably low ESR, for SMPS applications) caps I can find.  I figure, at that point, I'm probably close to what engineering originally specified.   The replacements never seem to die, even if they're offbrand, though I go for Nichicon or Panasonic or Nippon Chemi Con for big stuff if at all possible.

[lilshawn]

The issue seems to be that, in an effort to save a couple tenths of a cent per unit, incompetent product sourcing bean counters substitute cheap caps for what the engineering people actually spec'd.  They usually do this based purely on case size, capacitance, and voltage rating.  This would be fine except that 1) there are other specs that matter (ESR, ESL, rated ripple, etc.), and 2) The specifications on those cheapo caps are often only good for 2000-3000 hours whereas the quality, name brand caps are often rated 6000+.  The bean counters will also often drop down to 85C instead of 105C ratings.  All this combines to drastically shorten the life of even a properly manufactured capacitor.

i'd have to agree partially with that... i'll bet dollars to doughnuts that the cap factory is allowing substandard caps to be passed during QA in order to meet quotas, they allow caps with higher than normal ESR to be passed along with the good ones and they later fail. i indeed have a handful of questionable quality caps, brand new, from the same batch even... that read higher than the rest of the caps in the batch.

The specifications on those cheapo caps are often only good for 2000-3000 hours whereas the quality, name brand caps are often rated 6000+.

i know, and its a frikin' joke too... a cap should last more than that... 3000 hours operated 24 hours a day is only about 4 months, and that is gross. Even 6000 is sick. i want them to last forever, but given that is a virtual impossibility, i would be happy with 3 years.

it also doesn't help that when they design the system and choose caps... some will run a 6.3 volt cap on the 5 volt line... isn't the norm to double the voltage?? talk about stress. or using a cap JUST big enough to supply startup current. that's bad too. Ive subbed 1200uf and even 1500uf caps for 1000's in some units, we will see how the strain on the regulators is, but i'm betting the caps will last the remaining lifetime of the unit.

[MonMotha]

The issue seems to be that, in an effort to save a couple tenths of a cent per unit, incompetent product sourcing bean counters substitute cheap caps for what the engineering people actually spec'd.  They usually do this based purely on case size, capacitance, and voltage rating.  This would be fine except that 1) there are other specs that matter (ESR, ESL, rated ripple, etc.), and 2) The specifications on those cheapo caps are often only good for 2000-3000 hours whereas the quality, name brand caps are often rated 6000+.  The bean counters will also often drop down to 85C instead of 105C ratings.  All this combines to drastically shorten the life of even a properly manufactured capacitor.

i'd have to agree partially with that... i'll bet dollars to doughnuts that the cap factory is allowing substandard caps to be passed during QA in order to meet quotas, they allow caps with higher than normal ESR to be passed along with the good ones and they later fail. i indeed have a handful of questionable quality caps, brand new, from the same batch even... that read higher than the rest of the caps in the batch.

I'd guess that the "lesser" brands probably have substandard QA.  I hate to be xenophobic, but at this point, if the CoO is China, I tend to have my doubts.

The specifications on those cheapo caps are often only good for 2000-3000 hours whereas the quality, name brand caps are often rated 6000+.

i know, and its a frikin' joke too... a cap should last more than that... 3000 hours operated 24 hours a day is only about 4 months, and that is gross. Even 6000 is sick. i want them to last forever, but given that is a virtual impossibility, i would be happy with 3 years.

it also doesn't help that when they design the system and choose caps... some will run a 6.3 volt cap on the 5 volt line... isn't the norm to double the voltage?? talk about stress. or using a cap JUST big enough to supply startup current. that's bad too. Ive subbed 1200uf and even 1500uf caps for 1000's in some units, we will see how the strain on the regulators is, but i'm betting the caps will last the remaining lifetime of the unit.

You can get 10000 hour rated ones from the major brands.  At 12 hours a day (reasonable retail day), that's about 2.25 years.  Still something of a joke, though those are at least supposed to be the minimums with almost all the limit conditions valid simultaneously.  Overspec'ing other things tends to increase it beyond that, at least with reputable brands.  Keeping things cool also helps a lot, and therefore the 105C rated ones can be a lot better.

Yeah, it at least used to be standard practice to double the voltage rating on anything where reasonable.  I often triple it or more when spec'ing if I've got the room.  Again, bean counters, though there's also sometimes some size constraints decreed upon the engineering department that forces the issue.

[newmanfamilyvlogs]

So what about these solid state capacitors I hear advertised on some motherboards recently? What's the story on them?

[lilshawn]

So what about these solid state capacitors I hear advertised on some motherboards recently? What's the story on them?

ya we bought a couple of those...5000 hours...still a joke. i'm not sure they are true solid state caps...think they are just running a gimmick. they're not solid state at all. they're just in a different packaging - but they are supposed to have some benefits. They are still an electrolytic cap, just in a different aluminum case. Solid state caps are a completely different beast and the only "high" capacity ones are made of tantalum and are really expensive.

[MonMotha]

Most PC motherboards advertising "solid state capacitors" (whatever the hell that means!) just have surface mount aluminum electrolytics rather than the traditional through-hole.  There's absolutely no electrical difference.  The leads just don't go through the PCB, and the jacket color is usually a silvery gray rather than blue, green, or black.

SOME of those motherboards actually mean that they have tantalum electrolytic rather than aluminum electrolytic capacitors.  Tantalums cost more and tend to be bigger for the same capacitance and voltage than aluminum electrolytics.  They do tend to stand up to "wear and tear" better than the aluminum electrolytics.  They tend to suffer sudden catastrophic failure rather than drifting over time and oozing goo all over your motherboard (and boy do they stink if they do go!).  Tantalum electrolytic capacitors are especially susceptible to surges and voltage transients.  SMPS applications often need special "surge robust" rated caps.  While even properly specified aluminum electrolytic caps will go bad with time, a properly specified tantalum should last about as long as you need it to (i.e. the device will be long obsolete before it goes bad).

The "good" caps are the multi-layer ceramic ones (MLCC).  Properly spec'd, they'll last basically forever without appreciable drift, but the high capacity ones are rather expensive and fairly large.  They are starting to show up in high-end PC motherboard power supply (for the CPU Vcore) applications, though.  They're especially good on the high-ripple inputs to buck converters found on PC motherboards for the CPU.  Sanyo OSCON are also a good choice if higher capacitance per volume is required.

[smartbomb2084]

If there is a silver lining to this 'capacitor plague' it is that the repairs are usually easy to diagnose and perform. 

Also, repair facilities and technicians are making a killing just changing caps and cleaning out some dust.

That will be 85 dollars today, sir, Thank You, and have a nice day.

[ahofle]

Interesting.  I thought it was just a side effect of electrolytics in general.  I'll bet this had a huge role in the demise of the Wells Garder D9200.

[qrz]

If there is a silver lining to this 'capacitor plague' it is that the repairs are usually easy to diagnose and perform.  

Also, repair facilities and technicians are making a killing just changing caps and cleaning out some dust.

That will be 85 dollars today, sir, Thank You, and have a nice day.

SHHHHHHHHHHH !!!!  thats a trade secret.   
as is knowing WHICH one(s) to replace

[smartbomb2084]

Sorry, I lost my head. Hey, aren't you the one that started this thread?

[lilshawn]

you have to watch motherboards with square cased choke coils on them, they are being described as solid state as well.... the whole term solid state is bastardized everywhere... solid state... it's like, not liquid or gas, it solid state!!

durr!

[qrz]

Sorry, I lost my head. Hey, aren't you the one that started this thread?

its those voices again ... they told me to do it ... make it stop 


note to self : no more late night posts 

[dataman19]

OK caps are caps... crap is crap...
..
The whole cap thing was caused by a QC error (magnified by corporate greed and a language issue).
..
In Malaysia a couple of the cap factories were putting the wrong amount of electrolyte in their caps (to much to be exact) .. something about ml being different over there (or they applied a conversion where one was not needed).
..
To complicate things some Indonesian companies were diluting their electrolytes to stretch their supply out and save a few bucks...
..
Add to this the corporate scramble to resource and (like has already been mentioned mentioned) the bean counters got involved and decided the problems were that the engineers didn't know what they thought they knew and the went to a "revised specification".
...
Now you have to understand that this fiasco went on for like 6 months to a year.  A cap factory does like 10-20M caps a months (at least).  Dell was particularly hurt because they rely on three or four different motherboard manufacturers.  And they are all centered in Malyasia.  HP was hit, so were FIC and Gigabyte.  Just about all consumer electronic gizmos were shipped with out of spec caps...
I have even found them in Linksys WRT Routers and one Cisco Switch (although I suspect the Cisco was a repair at one of their domestic shops that used - you guessed it - bad caps).
...
One of my associates was stuck with several million bad caps and almost went under.  He was one of the first ones to scream foul early on.  His shipments are routinely checked for proper tolerances and the specs are double checked.  When he started getting out of spec caps he called Malaysia who promised to verify and re shipped supposedly "checked caps" - which also failed incoming QC.
...
He refused all further shipments and took a bath on the $1.2M he had invested.  Cost him another $2.3M to get back up with a "verified" brand.  Not including his trip to Malaysia and Japan.
...
Indo China factories are notorios for some of the shortcuts they will take to make an extra $0.01 per thousand pieces.
...
In my world almost all caps are suspect.  And testing one or two out of every thousand is the norm.  In my world 10uf with tolerance of + or - 20% means it should be at least 8uf and not more than 12uf.  If ESR ratings are supposed to be 1.2 then they better be 1.2.   At least you can normally return bad caps to a domestic supplier (I know that Digi-Key and Mouser will accept Bad Cap lots that have a failed sampling - I have sent bad lots back).
...
I recapped ten Dell GX-270 Mother Boards with polymer caps and they ran great.  Polymer caps are rated well above 105 degrees and their sizes are somewhat smaller.  On those motherboards the 1,800uf 6.3v caps were replaced with Polymer rated at 1,800uf and 10v.  ESR ratings are more stable with polycon caps as well - so heat is not an issue with ESR stability.  Since they don't have a liquid or gel electrolyte - they aren't subject to electrolyte swell due to long term heat as well. The motherboards ran cooler and seemed to have more pep.  Dell Had a fit.  I kept two of them, they took the other eight back to Texas - don't really know what they are doing with them (one of them is in their corporate office and runs 24/7 - so I have been told ).  But they made me promise never to pull that stunt again.   Never bothered to tell them about the GX-280 motherboards in our tech bench computers - they have all been recapped with polymer (OsCon actually) caps as well.  They are under the bench and do not get a great air circulation - not to mention the dust bunny attacks.  We took one out two months ago and opened the case.  I swear we blew a ton of dust out of that case - which looked like a cement factory inside.
..
So if you want a virtually bullet proof cap for your CPU core voltage regulator - OsCon Polymer Caps are the cats meow - the meow coming from the "OWWW" when you go to pay for them.  This would be the nine 1,800uf 6.3v caps, the four 1,800uf 16v Caps,(both located right next to the long Core Voltage Regulator Heat fence) and the three 1,500uf 6.3v caps as well (For the GX-270 anyways).
....
Yea I know the P4 Socket 478 and 775 CPUs are old.  But they still work fine for what we use them for.  Besides, I still have a 486cpu based router that has been in service since 1986.  It seems to push packets just as well as that new Cisco Router sitting next to it - even has less software issues.
..
Oh well, call me out dated....
But I know what a good cap and a bad cap is... And I know why MBA's should stay away from EE's as well.  Its like sand and water - it will seem to mix well as long as you keep stirring.  But when you stop stirring - the MBAs settle to the bottom of the pitcher, just like the sand.
....
Dave
...

[smartbomb2084]