Repeating power-on self-test fail...finding the culprit

[turnipboy]

I've assembled the last couple of desktops I've owned, so I've got some measure of what's going on, but this has got me stumped. For about the last two months, on about every fourth or fifth start-up, when I push the power button I get a flash out of the LEDs and about two seconds rotation out of the fans, but then everything stops, and then the process repeats again, the LEDs flash and the fans fire up, and then it stops, and it just keeps doing that, over and over. One day when I was poking around I discovered that if I gave the heatsink on the mobo chip a bit of a wiggle, it'd snap out of the start-up/fail loop and go on to boot fine. It was as though there was a connection somewhere that just needed a bit of a nudge. So doing that got me through to last weekend, when no amount of nudging and wiggling stuff would bring it out of the start-up/fail loop. I figured it was either a mobo problem or a PSU problem, but as the PSU is a fairly decent one, and still under warranty, I was leaning toward the mobo. So I went and bought a new mobo and some RAM to suit (no big drama as I'd been thinking about upgrading anyway)...

So the initial specs were:
Pentium 4 3.0ghz (Zalman CNPS7700 fan)
Gigabyte GA-8I945P
4GB DDR2 533mhz
Seagate SATA 750GB
Radeon X700
Seasonic M12 (500W)

But now the mobo and RAM have been replaced with:
Gigabyte GA-EP45-DS4P
G.Skill 4GB DDR2 800mhz

So I swap the mobo/RAM and fire it up--same thing. Can't get past the first POST hurdle, or it's not even making it to POST. So I figure it's a power issue; first I try a different power lead, which doesn't help, then I swap the PSU for the original Antec one. Still no luck. So the only two variables that I haven't altered are the CPU and the Zalman fan. (I figure it's not a hard drive issue because I can disconnect it entirely and the same thing happens; it doesn't even get the chance to go on and say 'Primary drive failure'...same for the graphics card; remove it and it makes no difference.)

So are there any other checks I can do here, or should I try my hand at a new CPU?


Cheers,
turnipboy.

[Calum]

The case could be at fault here, I've had this issue and it was a short.
Can you try another case?  Ideally, run the computer with the bare minimum (PSU, motherboard, CPU +cooler, 1 stick of RAM, graphics card) on a cardboard box, and see if it powers up after disconnecting, checking, and connecting every cable again.
Let us know how it goes.

[turnipboy]

Thanks for your reply, Calum. I tried what you suggested. I set the mother board out on some cardboard, with only the CPU, cooler, one stick of RAM, and graphics card in place. Only mobo connections were ATX/ATX_12V (plus CPU cooler). Same problem! Fires up for a second or two then switches off, fires up again, switches off. Tried it without the graphics card; same deal. I'll pick up a new CPU tomorrow, probably an E8500, and see if that gets her going.


Cheers,
turnipboy.

[jamesj2001]

I'd say it's quite obvious that it's your processor.  The clue should have been when wiggling it made it work.

[turnipboy]

But the wiggling was of the small heatsink of the motherboard chip, not the heatsink on the CPU.

[jamesj2001]

Oh, my bad.  In that case, maybe you could check the CPU fan and heatsink installation.  When I built my first PC, it kept switching off after just a couple of seconds being on.  I was just about to send the MOBO back when I finally got it to bios just long enought to check the CPU temperature.  It was like 120 degrees celcius!  Turns out the heatsink, although it appeared secure, wasn't screwed down quite tight enough.  Just an extra turn of the screwdriver solved the problem.  I'm not suggesting you necessarily tighten the heatsink, but check and make sure it's making full contact with the CPU.