Capacitors, Ugh

So I’m looking at an ASUS M3A32-MVP Deluxe board that stopped working some time ago and I had no time to figure out why. The board is from early 2008, not exactly vintage hardware, or at least not just yet. It gave up the ghost approximately in 2013.

I was about to fire up the board when I noticed that about half of the capacitors look like they’re trying to poop out the rubber seal at the bottom. It’s not terribly obvious at first glance and I also can’t find any traces of leaked electrolyte, but those capacitors look very sick.

The interesting thing is that the caps near the CPU, which I would think (perhaps naively) get stressed the most, look just fine. And I’m reminded of another board, an Intel DG965RY, which still works but has 3 or 4 rather unhealthy looking capacitors far away from the CPU.

I strongly suspect that the problem is insufficient case cooling. The capacitors near the CPU get good airflow from the CPU fan which is blasting air at them. That’s part of the design—Intel started requiring omnidirectional heatsink/fans in the early 2000s, to help with board cooling. While the M3A32’s CPU fan only blows out air in two directions, it’s very obvious that that’s where all the high-temp components are, and the other two areas adjoining the CPU are much emptier.

But the caps on the other side of the board far away from the CPU might get very little airflow in a standard case, and were probably exposed to considerably higher temperatures than what they were rated for. After a few years of relatively heavy use, the M3A32 board simply stopped working. The capacitors probably weren’t bad per se but were abused.

It seems like re-capping the M3A32 shouldn’t be a horribly difficult soldering job but I’m not at all sure it’s worth it. I really hate leaky capacitors.

If anyone has tips for reliably identifying bad capacitors without de-soldering, please let me know (if I’m going to de-solder a capacitor I might as well replace it with a known good one).

This entry was posted in PC hardware. Bookmark the permalink.

10 Responses to Capacitors, Ugh

  1. Andreas Kohl says:

    With an ESR (Equivalent Series Resistance) measuring instrument a capacitor can be measured in circuit. But ultra cheap ESR meters are not working reliable in ciruit.

  2. Mihai says:

    It’s not always insuficient cooling, can be bad electrolyte. We’ve had some new monitor power supplies (as service spare parts) on which the capacitors popped their seals while standing in storage – it was quite a surprise taking the new board and seeing the caps like that. I also remember seeing some leakage on another batch of new capacitors while still on their tape.
    Regarding ESR, Andreas Kohl is right, you can use a (good) ESR meter, however they tend to be somewhat expensive (somewhat over $100 for the DE-5000 that we use here).
    Oh and by the way, on older equipment (around ’80s vintage) you can have the unpleasant surprise of electrolyte leaking and corroding traces and/or causing strange defects.

  3. Michal Necasek says:

    The reason I suspect poor cooling is that in both cases the same type of caps look good near the CPU and damaged near the edge of the board.

    I’m unfortunately familiar with leaky late 1980s/early 1990s electrolytic capacitors. With those, the capacitor itself usually looks fine but the nearby area is corroded. And yes, they do (some of them) leak simply with time even when not used/abused.

    I was actually a bit surprised that with the newer capacitors there’s no corrosion. It seems every 5-10 years the electrolyte formula changes and it takes the industry a while to get it fully stable. And unfortunately the problems tend to pop up only after a few years.

    I’ll have to look into the ESR meters. Something in the $100 range isn’t nothing but could be worth it for me in the long run.

    Out of curiosity, when was the board which popped the caps in storage manufactured?

  4. Chris M. says:

    That Asus board uses polymer capacitors which should last a very long time. Its possible they are all solid (no electrolyte, should leak), but hybrid designs that do use liquid electrolyte do exist. Gigabyte boards from the period were specifically marketed as “all solid polymer capacitors”.
    http://www.gigabyte.com/webpage/8/article_02_all_solid.htm

    The wording on Asus’ page for that board is a bit more vague, but the symptoms point to polymer cap failure.

    http://neosmart.net/blog/2008/gigabytes-solid-core-capacitor-gimmick/

  5. Mihai says:

    Re. caps popping in storage: The boards were manufactured around 2006 and the problem took some years to manifest itself (IIRC we found out when doing some out of warranty repairs). I agree that usually it happens because of temperature (or overvoltage, in some cases).

    By the way, I had an Intel Desktop board (D865GBFL) that popped the caps while still under warranty. I am quite certain I did not abuse it (thermally or otherwise) and it was replaced under warranty. I remember that the original board had thru-hole caps, the replacement board had SMD caps, but I didn’t notice more about their type/brand. The replacement one worked fine for another ~ 5 years.

  6. Michal Necasek says:

    I wonder if the D865GBFL is old enough that it might have suffered from the “capacitor plague” of the early 2000s? Because those caps definitely died fast enough that users really noticed. I used an Athlon board which died of bad capacitors sometime around 2003 and was only two years or so old at the time. From what I recall, the plague-bearing capacitors were killed by standard operating temperatures.

  7. Michal Necasek says:

    I can’t tell by looking what they are, or who made the caps (for example on a Gigabyte board of the same era, one claiming to have solid caps, I see they were made by Fujitsu). I’m tempted to invest in an ESR meter in order to stop guessing. Concrete suggestions (like DE-5000) welcome.

    On a semi-related note, at work we’ve been recently affected by a different kind of plague which looks like maybe flash ROMs going bad. The symptom is that a device (server, router) works 24/7 for years but is completely dead after a power off/on cycle (power outage, equipment move, etc.). Although I suppose capacitors could be involved too.

  8. Chris M. says:

    Dell had a big problem with bad Nichicon HN(M) and HM(M) series capacitors on the 865 motherboards found in the Optiplex GX270. I believe the boards were OEMed from Intel and very similar to the D865 series. The boards were produced in 2003-04.

  9. Michal Necasek says:

    Thanks for the suggestions!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.