Hello. Possible 8KNXP problem

[pol098]

Just joined, see I'm asked to post at least one message ASAP, this is it. Hello everyone.

No real questions. The reason I came here: I have an 8-year-old Gigabyte GA-8KNXP (rev 1) running off an 8-year old PSU that came with a case. Never given any trouble until recently. It's not a terribly slow machine for my purposes despite its age (2400MHz hypertheading Northwood, 4GB ECC RAM); mainly used as RAID file server with WinXP (patched to support software mirroring). A couple of weeks ago it began not to start when switched on from cold; never started first time, but eventually started and booted (I would leave it on for a minute in case it needed to warm up in some way). I decided, in my wisdom, that the PSU caps must be getting tired, and replaced the PSU. On putting the replacement in, I'm not getting any startup (no startup=no BIOS display, no beep, no disc activity). I saw a few electrolytic capacitors in the motherboard that looked possibly a bit bulged and possibly with signs of past leakage (dried-up). I'm going to clean the machine a bit to see if it's really leakage, look for stupid things like forgetting the 12V2 connector (no I didn't) and decide what to do. Repairing an 8-year-old computer seems a bit of a waste of time. I've got the necessary tools (ESR meter, POST card, PSU tester, etc.), but just don't feel like doing this.

Very glad I'm using software RAID; if I used motherboard RAID might not be able to access data without repairing/replacing with similar motherboard (though it's all backed up in at least two places).

Anyway, I'm searching this forum for "8knxp". Not found much yet, but just started; I think it missed the capacitor plague.

Tip from the old days: if your machine is set up to boot from floppy disc, you can see the basic bits (PSU, motherboard, RAM) are working by removing all cards including display card and disconnecting all drives except floppy, putting in a boot floppy, and seeing if the machine seems to access it as expected (light flashing, floppy noises). I'm not totally sure that a modern machine will work happily this way.

Anyway, this message basically says hello.

pol098

[PCBONEZ]

Re: Hello. Possible 8KNXP problem

There are more than one Plague.
-
One that age probably got bit with HM/HN [w/dates 2001-2004] or KZG/KZJ.
If they or non-Jap caps are present it's probably an easy fix.
-
Hello back.
.

[pol098]

Re: Hello. Possible 8KNXP problem

Thanks for the comment. Frankly I wouldn't call 2003 capacitors which may be dying in 2011 victims of whatever plagues there may have been when they were made, but well-enduring veterans. Aluminium electrolytics often have a design lifetime of about 5,000 hours (~ 1 year), and this machine ran full-time until I implemented Wake on LAN a few years ago. Rather than replace the capacitors, which may or may not be the cause, I'm inclined to update motherboard, processor, and RAM. Thought I might just try buying an exact replacement motherboard for a peppercorn price, but was astonished to find that this antique motherboard is still offered on eBay, USED, for up to US$144 (delivered from Hong Kong).

My best badcaps story is about a Samsung monitor in the early 1990s, an early multisync design; I had bought about 50. After about a year (just out of warranty) they started failing in the same way, with dramatic smoke, failed tripler, IC, power transistor, and resistor as far as I remember. Expensive and time-consuming. Was able to find pre-failing machine: all voltages way too high (35,000V where there should have been 25,000 - X-ray hazard?). I analysed the circuit and found a 1 uF 50V electrolytic in the power supply regulation feedback circuit; it was going high-ESR, feeding back a reduced voltage thus telling the regulation circuit that voltage was low. ESR wasn't a hot topic in those days, no cheap meters; I worked out how to measure it using the square-wave calibration signal from an oscilloscope and a resistor to display how the capacitor charged. Sounds like the manufacturer's bean counters overrode the designers to put in a 1-penny capacitor where a 2-penny computer-grade one was required; replaced them all (quick and cheap), and had no more problems.

[PCBONEZ]

Re: Hello. Possible 8KNXP problem

To me, 105C lytics dying in 8 years is dying very young.
I still have 486, socket 7, and P3 boards that work fine with original caps.
I've also worked on 15+ year old 24/7 shipboard industrial equipment [read 'harsh environment'] that still worked fine with original caps and even at that age cap problems were rare.
And when people come in here fixing their ancient TVs or audio gear the problem caps are nearly always Chinese and Tw cap brands.

The 'Lifetime' you think means 'Useful Life' isn't that at all.
The fact that 5000 Hr = 208 days should make that obvious.
The fact that higher grades of caps [MBZ, MCZ for example] generally have a 2000 Hr [83 days] 'Lifetime' should also make that obvious.
Search "Endurance" in the forum, there are already a few threads around that go into all that in detail.

HM/HN were defective due to manufacturing error [overfilled].
KZG/KZJ have a poor electrolyte formula.
Chinese & Tw caps have either impurities in their aluminum, poor electrolytes, or both.
Time to failure is largely dependent on time and temp exposure.
Were they not defective they would probably last 15-30 years.

The popularized 'Capacitor Plague' term is used to describe issues with a defective electrolyte formulation that was on the market for 3 or 4 years which often caused caps to fail in a few months of use.
It was only used in Chinese and Tw cap brands.
Nothing to do with HM/HN or KZG/KZJ. [Which are Japanese.]
Nothing to do with Chinese & Tw produced after ~about~ 2004.
So, it has nothing to do with the previous stated problems, yet the previous stated problems are clearly also plagues on the electronics industry.
-
Most news reports on the 'capacitor plague' topic were wrong in that they blame the 'famous' Dell problems on the defective electrolyte formulation which only affected Chinese & Tw caps.
- But Dell never used Chinese & Tw caps.

Those news reports lead to a lot of confusion because they imply there was only one issue with caps.
- There are LOTS of issues with caps - and they aren't all related.
.