A crazy fix for a motherboard problem or not?

gneier

Folks,
I've isolated a power problem to the motherboard, I think, and now am thinking of a Rube Goldberg fix. Am curious to know what others think.

The problem: computer doesn't power on. I have an Athlon XP 2600+ Is not the psu (have tested, and swapped). Is not the switch (have tested). When the switch is hit (or shorted), the computer powers on for a half a second, then turns off. The switch's function seems to be simply shorting the connection briefly when you press and release.

My theory as to whats wrong: perhaps the part of the mobo that receives the power switch "on" signal is broken. It isn't able to signal to the psu to turn on.

The potential fix: before buying a new motherboard, why not try permanently having the mobo "on" by shorting the green wire on the 20-pin connector to a black ground on the same 20-pin connector? This is what would tell the psu to turn on. Seems like something to try.. since the only downside, I think, is a more broken mobo.

Crazy, or a good idea? Any ideas, or things I've got completely wrong? I haven't messed with motherboards much, so this will be a first.

-gary

JDMurray

Did the computer just suddenly stop booting, or did it stop after you made a change to it?

Make sure that you try to start the mobo without any cards installed or peripherals plugged in. A bad card or USB device can keep a mobo from starting. A bad CPU may keep a mobo from starting too. Also try starting without any RAM installed to check if it will sound any POST error code beeps. No error beeps under those conditions means that the CPU or mobo (or both) are bad.

mgeorge

What model computer and/or board is this? Several boards have issues with the VRM capacitors that leak out and become unstable. Typically these capacitors are 16v 1500uF. If the processor does not get power during attempted post test, the board will just power on and kick off automatically. This sounds like your issue. I have seen this issue on many dells, compaq, hp and IBMs.

dtlokee

COmponent level repair is a thing of the past, if it's bad, replace it. The time invested in any sort of repair when a new one can be found for $50-100 isn't worth the time. Even if a fix is found, again it may not be reliable and will just lead to frustration in the future.

mgeorge

dtlokee wrote:

COmponent level repair is a thing of the past, if it's bad, replace it. The time invested in any sort of repair when a new one can be found for $50-100 isn't worth the time. Even if a fix is found, again it may not be reliable and will just lead to frustration in the future.

Yeah i agree dt, but some of us like a challange *laughs*

gneier

Thank you Guys,
I will try removing peripherals and retest. I am not sure what board this is, I bought the computer for my folks from ecollegepc 3 years ago. Its probably not too fancy, just has a couple of pci slots, agp, 2 dimm spots, and integrated video. If you really want to know, I can take a picture. I
I like the challenge of going beyond the simple PC repair tasks, even though my knowledge of mobos is pretty pedestrian. Motherboards are mysterious little beasts.
I'll post to this thread what I end up finding and deciding to do.
-gary

JDMurray

dtlokee wrote:

COmponent level repair is a thing of the past, if it's bad, replace it.

This is also true because electronic components are usually very expensive to buy in very small quantities. The multi-layer printed circuit boards are very easy to damage when worked on using conventional soldering irons, and the surface-mount parts do require special tools to replace.

I worked in printer circuit board manufacturing in my youth, and it's amazing how far the technology has come in the past 30 years. Just like modern cars, modern printed circuit assemblies can't be easily repaired under "a shade tree."

oldbarney

JDMurray, I totally agree. Times have changed. Replace the faulty component and be done. It isn't usually worth the time to troubleshoot and repair unless you have the equipment and expertise. Besides, warranties for quality boards and memory are very good these days.

Back around 1997, I ran into a lot of 150 motherboards for old socket 5/7 processors. Some worked although most were bad. Intel repaired 3 under warranty. For other boards, a friend had a scope and some experience with board-level troubleshooting. We knew a gal who worked as a level-2 soldering tech for a nearby tech manufacturer (since closed due to outsourcing).

We were able to repair around 20 more boards. Some required replacing simple stuff. If a controller or bridge was bad, we tried replacing these components from bad boards, but eventually ended up scrapping many. I think after selling the repaired boards on eBay back then, my profit after paying the soldering tech and my friend was only a little over $100.

I've used a couple of tricks to bring dead processors back to life, mainly Pentium Pro or Socket 7. One time, I put 10 CPUs in an oven set as low as possible (325 F, I think), and left the chips in there for 10 minutes. After cooling, one came back to life. I ran the processor for a few weeks and then sold it. The rest were tossed.

Another trick dealt with Pentium 60-66 MHz processors. Intel recalled most over a design flaw. Divide-by-zero error, I think. Back then, if a P60 CPU was dead, you simply called an Intel number even if it was out of warranty. You were patched through to Glasgow, Scotland. There, you told gave the rep the number from the CPU. They sent a new replacement out.

FP and EDO RAM SIMMs were not too difficult to make in the mid-90s. A parts dealer in my area produced SIMMs by taking chips from decommissioned mainframes and soldering on to SIMM boards. The shop made a lot of money selling 30 and 72 PIN SIMMs. Am I bringing back memories?

At my current employer, we have a lab that builds prototype boards for design bids. One tech, she's in her 60s, and really good. It's amazing how steady she is. She's been around the industry since the early 80s. I support facility IT requirements, by the way. She can put together a working board with CPU socket, controllers, IO, SATA controllers, etc. in a couple hours. These boards are not for PCs, but use many of the same components. Just incredible to see the craft still practiced.

gneier

Hi Guys,
Very nice replies to my initial post. From what I've read, and seen, these components really are getting too complicated for the majority of people, like myself.

I tried one of the the suggestions posted. I remove all the peripherals and just tried powering on the motherboard (no agp,pci,ram components connected). Same problem (see video below).
I next tried to force the motherboard to power on my wiring the green and a black wire on the 20 pin connector together (the technique to test psus).. resulting in no change to the problem.
What was interesting is that when I tested the green-black wiring (unplugged) .. but the 4 pin connector was still plugged into the motherboard .. I got the same problem with the power coming on for half a second. BUT, when I unplugged the 4 pin (no with no power to the motherboard), the psu started up fine. Just an interesting bit. I'm done with this motherboard and going to buy another on ebay.

For those still interested, here is a video of the motherboard coming on
http://www.youtube.com/watch?v=98UfyIIf74U
It is a "K7 Motherboard" SIS 741GX Chipset Athlon XP
Here is a picture of the motherboard
http://www.flickr.com/photos/gneier/1660124488/

Any other thoughts or is this board toast?
thank you all for the ideas and giving me some perspective

-Gary

dtlokee

Inthe picture you didn't disconnect the HDD and floppy frives, I would disconnect them too to make sure it's not a short in one of those devices or a bad cable. Also remove any fan connectors to the PSU and only connect to the MB and try again.

gneier

Thank Dt,
I gave that a shot.. removing almost all of the other wires, same problem. Then reconnected them and removed some other ones (speaker), without luck.
-gary