Is there such a thing as TOO COLD?!?

[engjohn]

Well I have the pelt on the ol p///700 @ 1G and with seti running it is sitting at -23c according to MBM5. When MBM started, just before seti, the CPU was at -31c.

So the question is, Is there such a thing as too cold???
oh yea, ist a 80w pelt on a 15v @ 10a PS, and the pelt is water cooled.

 

[William]

i am not completely sure but here is my thoughts. Your cpu cannot be too cold. The only problem I forsee is if your temps vary a lot that you could stress out the cpu from the expansion and contraction. Other than that no.

And yes there is too cold, that would be absolute zero, but I am sure you are not going to the get that far....

 

[Snuffleuppigus]

engjohn (Apr 03, 2001 07:35 p.m.):
Well I have the pelt on the ol p///700 @ 1G and with seti running it is sitting at -23c according to MBM5. When MBM started, just before seti, the CPU was at -31c.

So the question is, Is there such a thing as too cold???
oh yea, ist a 80w pelt on a 15v @ 10a PS, and the pelt is water cooled.

Well.... my PIII 650 @ 1G will run about -20c under load and -50c idle (not sure below -50 as the PIII doesn't seem to read below that). as for temp stress effecting it, i have constant fluctuations of 15c or so that doesn't seem to be the issue.... only issue i have is that i have to re-tighten the 4 screws that mount my waterblock to my processor about once a month.

 

[engjohn]

Thanks for the info guys...

 

[BigAir]

If condensations forms, that is when you get the problems...

 

[BigAir]

* I meant to say condensation

 

[engjohn]

I have it insulated pretty good. I have dielectric grease filling the socket, a square of neoprene in the center of the socket, and I have applied two layers of PlastiDip to the slocket itself. Oh yea, I have hacked up a mouse pad and made a neoprene sandwitch around the slocket. Everything has dielectric grease on it to keep the air out. So far so good...

 

[stool]

I don't think cold is ever going to be a problem for the CPU.. It will only be when you get into real supercooling that you'll have problems with peripherals. Check out some of the supercooling ****(liquid nitrogen, etc) and you'll find cards, boards, and drives taht don't really like the cold

 

[Thelemac]

Actually at around 50 below or so the silicon starts contracting too much and starts pulling apart. I would assume that to be bad.

 

[dimmreaper]

Thelemac (Apr 03, 2001 11:47 p.m.):
Actually at around 50 below or so the silicon starts contracting too much and starts pulling apart. I would assume that to be bad.

BINGO . . . . Thelemac has the ticket, cold can kill as easily as heat. A CPU at -50C or colder may seem fine for days before it dies, but it ain't fine.

 

[The Faceless Rebel]

Generally, if you're going to Peltier something I hear the recommendation is to get rid of stuff like CPUIdle/Rain/Waterfall/etc. It's even more beneficial to run something like SETI@Home/Prime95/Folding@Home/etc. that keeps your CPU going at 100% all the time. Hey, it can't hurt, and by running these programs you're helping a good cause of some kind or another.

 

[engjohn]

I run seti@home and my cpu stays st -24c. I have not let it just sit idle, as I am too scared to see how cold it will get. I have seen -35c. But that seems kinda cold.
Funny thing is, it will run -25c with seti running at 1107Mhz, but at
1130 it idles at -5??

Not so stable at 1120+Mhz yet...

 

[Fink]

Jeff Evans (Apr 04, 2001 07:23 p.m.):

Thelemac (Apr 03, 2001 11:47 p.m.):
Actually at around 50 below or so the silicon starts contracting too much and starts pulling apart. I would assume that to be bad.

BINGO . . . . Thelemac has the ticket, cold can kill as easily as heat. A CPU at -50C or colder may seem fine for days before it dies, but it ain't fine.

I cycled a 50 x 90 mm PCB (a high sensitivity IR detector) down to -78C about 60-70 times over the course of a year then it failed. The traces delaminated (came off the board) and there were some stress fractures the the 4-layer PCB itself. There are techniques for warming and cooling detectors slowly down to -78C or liquid N2 temps (-178C) to prevent destruction of immersed PCB boards... again their biggest problem is stress fractures caused by too rapid cooling or heating.