Bad Capacitors.

Status
Not open for further replies.
Joined
Aug 15, 2006
Messages
10,060
Location
Central Washington
I recently replaced my Moms machine. I was going to use the board/cpu/ram to replace my server. The Nforce 2/3000+ XP/1.5GB DDR is not much of an upgrade over my Duron 1800+ machine, but I dont exactly need that much power to pull router duties. The main reason I need to replace that one is it has a couple caps leaking. Although not too bad yet.
The DFI NFII I pulled out of her machine is a different story, however.
DFIICAPS.JPG

We have four leakers and one that tried to play rocket.
No wonder it kept crashing! Im amazed it ran at all. Out of 6 caps handling the low voltage side of the VRM, 5 were obviously toasted and the other probably is even though it looks fine. I was going to throw it away, but I got the idea to recap it. I could probably get another Socket A board off ebay for the amount spent on new caps, but what would be the fun in that? Then itll probably have garbage for caps as well.
DFIINOCAPS.JPG

Here it is, decapped. Im just waiting on the new ones to arrive.
 
One of my monitors is shutting off. Looks like it is bad transistors in the inverter. Ordering them off ebay. On the bright side, if it wasn't for the internet, I'd be buying a new monitor.
 
I heard that too, supposedly it was Rubycon that was ripped off, but they had an incomplete recipe, so to speak.
These are not those, They are United Chemi-con KZG model. The problem with these ones is they cant take heat even though they have "105C" all over them. This is a hot area; they get heat off the MOSFETs and the CPU fan blows hot air on them.
 
Ahh, poor engineering!

Fortunentally, capacitors are easy to replace, i have saved many less well off clients this way. Replaced millions of capacitors on Gigabyte video cards.
 
The secret part of the recipe was how to keep the electrolyte from gassing. Now when you see swollen and leaking caps, it's probably Chinese.

Dell got bitten bad on this years ago.
 
I have an ECS here, but few of them meet the ratings, and knowing ECS it has garbage caps as well. My server has an ECS.
There is also my Gigabyte board with all solid caps. That would be cool, and I hear you can cut capacitance in half going from electrolytics to solid. But the highest rating is 820uF 2.5v and these are 3300uF 6.3v with 1500uF 16v behind them.
 
I had a Soltek SL-75FRN2-L that I had to remove and replace about 6 of the caps around the CPU socket. I think its still working to this day for a friend as a linux box or whatever he used it for. That was a great board. If the board is good. It's worth re-capping :D
 
Over the past decade or so, I've experienced many electronic failures due to faulty electrolytics. This includes two satellite receivers, one model that was limited production. Obviously some poor quality components have been floating about globally. During my college years I interned in the R/D facility for Mallory capacitor, and caps from that era (early 80's) rarely failed like ones do these days. I have experience with pcb design and fab using most capacitor types (electrolytic, tantalum, mica, polyester, etc.) and correctly spec'ed components were never a problem (except when assembly reversed polarity!).

In my latest build, I used a Gigabyte motherboard with all solid capacitors simply due to the problems I've seen over the past several years
 
Originally Posted By: Touring5
During my college years I interned in the R/D facility for Mallory capacitor, and caps from that era (early 80's) rarely failed like ones do these days.


Not necessarily true, not where electrolytics are concerned. It depended on the circuit designer, and the cap vendor selected. All electrolytics from every era are rated for hours at temperature. Even so, smaller electrolytics from any era, particularly axials, are prone to develop seal leaks over time and dry out. Most smaller 'lytics don't short and blow out from overheating; the vast majority die a slow quiet death from dryout.

In my other passion, vintage audio restoration, old electrolytics are nearly always the cause of most problems. While I occasionally encounter large, old HV multisection FP cans (typically in tubed instruments, made by Sprauge or Mallory) that still work well after 45-50 years, recapping old smaller electrolytics (and anything paper) is virtually a given with anything over 25-30 years of service if you want the equipment to operate to specifications. Everything I see from the early '50s into the mid '70s falls into this category. The stuff from the '80s is now starting to reach that age. It's even worse to service the newer instruments, as the number of electrolytics has trended upward with more modern circuit designs.

Good first-tier makers of 'lytics today include Nichicon, UCC, and Vishay. Anyone doing decent field service will usually use these brands. But unfortunately, these brands often fall above the OEM price point for most consumer electronics nowadays. And once you start diving into the second tier manufacturer and rebrander markets, you can start winding up with Sino product. Back in the '80s, you tried to avoid the Taiwanese caps.
 
Its running now. Im doing a couple passes of Memtest86+ because I installed 3 sticks of ram (two 512s and a 256) since it seemed to hate that config before. Then Ill do something that stresses the processor hard.
 
I decided Im going to do the rest of the important ones. 8x 1000uF 6.3V and 2x 470uF 6.3V is the rest of the caps with vents. Im letting it crunch on Mersenne Primes until I get these.
 
Did 200 hours of Prime95 then I replaced the other ones around the RAM and AGP slots. Then I let it crunch on Memtest86+ for a few days. No errors or crashes even with the Athlon XP-M (Barton) 2400+ actually clocked at 200x12=2400MHz. I just got finished copying a Tarball of my server over and recompiling the whole thing.
 
Status
Not open for further replies.
Back
Top