A Real "Burn-In" Test....

[ArmoredFist1994]

I've seen and run a bunch of "burn-in" tests. Prime95, Memtest, etc. Acutally with my faulty motherboard with leaking exploded caps (just sent it in for recap a month ago - not back yet), all those tests ran fine when it booted!

The REAL TEST is getting it to boot, isn't it? Will it boot repeatedly? That's what it's all about.

Wouldn't a real reliability test be to just keep turning the thing off and on, off and on, over and over. I know this sounds like abuse of equipment, but if it's going to fail, that's when it'll happen, correct?

Somebody who knows more about electronics than I, please educate me...what's the deal with turning electronics off and on? I've always waited 30 seconds before turning something on after turning it off so (I believe this is correct) the capacitors can discharge!

Any input is welcome, thanks.

[370forlife]

Re: A Real "Burn-In" Test....

Well, turning a computer on and off repeatidly will hurt even the highest quality components.

Sure a PCCHIPS mobo may fail quicker that like a Intel or Asus mobo. And a Bestec psu would fail quicker than a Corsair. But you could still cause damage to the Asus or the Corsair. With a software test like memtest or prime95, it works the cpu and capacitors and ram, but won't really damage them. (except maybe the caps may explode, but you can replace them.)

[ArmoredFist1994]

Re: A Real "Burn-In" Test....

I did run memtest and prime95 on a mobo that had leaking exploded caps, booted only intermittently and it ran both forever.

When I turned it off, though, the ability of the machine to boot to the OS or even to make it past the post tests was 50-50 at best.

Running prime95 test and memtest didn't seem to mean anything. It kept going and going through the torture test...never locked up or anything once it did boot....weird.

Yes, it sounds really stupid to keep turning a device on and off.

What does that do exactly to damage components?

[acstech]

Re: A Real "Burn-In" Test....

Try this:

http://www.xtremesystems.org/forums/...d.php?t=197835

[PCBONEZ]

Re: A Real "Burn-In" Test....

On starting a marginal power supply will slow things down.

During a Burn-in it won't usually show any problems.

[willawake]

Re: A Real "Burn-In" Test....

the best thing would be something that loads the cpu up and then idles it repeatedly

[i4004]

Re: A Real "Burn-In" Test....

i don't think that would do either...
esr goes down as temp goes up, and as soon as you turn the machine on temp goes up...and won't really go that much down with cpu idle..

also this
"to just keep turning the thing off and on, off and on, over and over."
is not godo either, as it takes time to cool down...rapid on/off would just mean it boots ok after first few hickups...

what i'm interested in is why do you need such tests at all.
it's not like the problem is not apparent(you said one of the symptoms, won't boot) and that you don't have time to react after first symptoms....

[ArmoredFist1994]

Re: A Real "Burn-In" Test....

what i'm interested in is why do you need such tests at all.

That is a very good point! I guess I expected that there was some software burn-in or diagnostic that would have detected the problem. To my surprise, the machine passed all software tests I subjected it to.

Yes, I know there's a hardware problem, but no software that I ran detected it or caused a failure. The only thing that "caught" it was the cold boot.

However, running the previously mentioned IntelBurninTest Linpack may have caused a failure....I didn't know about that one until just now.

I guess the purpose is to put the machine through a safe software test and if there's anything wrong, I expect the program to tell me or cause a failure. If a machine passes a test like that, it can be considered reliable.

The need, I guess, is for a good reliability test so I can feel confident the thing isn't going to die on me at work.

[PCBONEZ]

Re: A Real "Burn-In" Test....

There is no software you can run that will say:
Hey: You caps are bad.
Hey: You MOSFETS are failing.
.... Or specific things like that.

Reason is there are no devices to measure the needed parameters from which some software could run a logic routine to determine the problem. (Say for a cap: temp/amps/leakage/maybe internal pressure even.) You would need 100's of different sensors and motherboards might cost as much as a car to do that.
-
Fact is most people had that many sensors to deal with they'd be totally lost anyway..

[ArmoredFist1994]

Re: A Real "Burn-In" Test....

There is no software you can run that will say:
Hey: You caps are bad.
..

I agree....bummer!