Jim: PC crashes

Hey, Jim, just read in a German NG where someone had the same symptoms.
Thought is was the RAM modules, had them tested by a shop, RAM was fine,
took a really good look at the mobo and saw that some caps had begun a
slight bulge. Might be worth a look if this happens again on the
affected PC.

 

Thanks, I'll take a closer look.

...Jim Thompson

 

It's been going around for several years now - cheap non-OSCON
high-density caps with inferior substitute organic electrolytes.
One doesn't always see a bulge; often the most reliable symptom
is CPU crashes. It's generally taken that the fix is a new mobo.

 

Most of the time, it's cheaper than trying to replace all the caps.
normally, all the caps are of the same vender and if one starts
going, the others might not be to far behind.
I understand how ever, that a couple of the chinese manufacturers
did have problems with crappie caps and no longer are using them.

Could be hear say, who knows.

 

All mothermakers during around '99-'02 had problems with those caps. They came
from a single manufacturer and showed up *everywhere*. I've personally seen the
problem on motherboards from asus, msi, intel, gigabyte, ecs, and biostar.
I don't think any correlation between the use of the bad caps and the maker's
nationality.

 

Eh? The fix is to change the caps.
http://www.badcaps.net/pages.php?vid=31

 

It can often be diagnosed by holding a scope probe to the rails. If you
see excessive spikes that would be the tip-off. And yes, changing the
caps might be a bear to do. You might occasionally strip the
metallization out and Murphy's law states that this is usually going to
happen on the last one. Obtaining caps that fit the available space can
also be a (pricey) adventure.

 

The caps came from MULTIPLE suppliers actually. The *electrolyte* might conceivably have had a single
source but that's not the name on the cap.

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

Graham

 

I've been fighting that problem for years. For a time, I was
replacing ALL the caps on motheboards. I have a Pace desoldering
station so that was fairly easy. Some motherboards have fairly sloppy
hole clearance for the capacitor leads, so removal and replacement
were easy. Others (i.e. Dell) looked like the caps were pounded into
the motherboard. If I ripped out the plated through hole along with
the capacitor, it usually indicated that I just destroyed the board.

The first indication of trouble is usually the rubber plug on the
bottom of the capacitor trying to wiggle out of the can. If the board
has a number of lopsided caps, it's either the worlds sloppiest board
manufacturer, or the caps are about to go bad. Usually the rubber
plug shows before the top or sides bulge.

I've tried various methods of detecting bad caps. Bulges aren't
always a good indication. I have an old Dick Smith capacitor ESR
tester, which is easily fooled by the large number of parallel
capacitors. It works fine if the capacitor is independent or pulled,
but not on the motherboard power supply bus.

I've also tried far IR imagers. Nothing like $20,000 worth of
borrowed hardware to repair a $100 motherboard. Some digital cameras
can see near IR and don't work for seeing heat. That has found some
overheated capacitors, but more often just shows the reflected
radiation from the CPU. A finger test of the capacitors is sometimes
effective. If the cap is warm, it's bad.

I've tried to use a scope to detect the expected noise on the power
supply lines, with little success. All motherboards have some noise
on these lines. The point of measurement (and grounding) have a huge
effect on this noise level. Making an accurate determination is
difficult.

The degree of failure is also somewhat entertaining. I've seen
motherboards with severely bulged caps, that continue to function
without incident for many months. Eventually, there's a failure, but
the amount of damage these can tolerate are amazing. On the other
foot, I've seen boards with only a barely visible amount of bluging,
generate boot problems and crashes. I suspect that this is an
indication of board quality.

For motherboards that I just wanted to work, and didn't care what they
looked like, I just chomped off the capacitors with a pair of very
large cutters, leaving the leads in the mother board. I then tack
soldered the replacement capacitors leaving about 1/4" of leads
exposed. It looks truely unprofessional, but gets the job done
without destroying the motherboard. I've also substituted tantalum
caps for the low-ESR electrolytics. Although more expensive, the low
frequency ESR is less, so fewer caps can be used. Please don't tell
anyone I suggested doing all this.

Drivel: See comments on tantalum ESR at:
<http://ludens.cl/Electron/esr/esr.html>
in the "using the meter" section. I've seen the same problem with the
same caps. If you think we have problems with low-ESR electrolytics,
get ready for more of the same with tantalums.

In retrospect, it's not worth fixing the motherboard. It's just too
much labour. If it's under warranty, get a replacement. If not, buy
a replacement and install it yourself.

Incidentally, I've seen some random CPU's fail along with the
motherboard when run with volcanic capacitors. I haven't investigated
the connection, but I suspect that non-optimum power supply voltages
and substantial voltage spikes, might not be very good for CPU
survival. In general, the ones that crash while booting or running
will have this problem.

 

Just got done fixing two Sceptre LCD monitors bought in 2004. Both
died close to the same period, about 1.5 years later. Multiple bulging
electrolytics on the outputs of the low voltage switching supplies
were the culprit. I've also seen a oozing capacitor on a View Sonic
monitor cause problems. That was the cap on the output of the mains
rectifier.

Mark

 

That works much better if you build yourself a little toroid transformer
so you can measure differentially. Done that at a client on Tuesday.
Then I needed one that goes down to a few Hertz so I took a large iso
transformer. You should have seen the face of one engineer walking in on
us and seeing a mains plug going to two 1N4148 and then on into the
soundcard input of my laptop. Sure enough that also had to be the day
the fire marshall did his rounds. I quickly tossed a rug over all that ...

With Windows that sort of crashing is kind of normal.

 

Perhaps some MLCC (high capacitance ceramics) in parallel to the leaky
caps may do the trick.

 

Depends on how many coals the switcher in front of it can throw on the
fire because a few 1uF or 10uF MLCCs can only supply juice for a few usec.

 

When those electrolytics leak, they eat the traces off the PC board.


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida