Lock ups when whole case is cooled

[MsNath]

VP6
Single piii 700e OC 933 1.70v

Before I start committing wads of cash to this case-cooling project. I decided to do a few test runs by cooling the whole case and checking on the performance.

Living in Canada in the middle of the winter makes this very easy. Its –21 outside today so I just opened the door.

Before I started temps were.
Room 27c
CPU1 39c running seti full time
Chipset temp 23c

I then dropped the room temp down to 10c

Cpu1 dropped to 15c
And the chip set dropped to 18c

I started getting lock ups at these temps. I remember seeing on Intel’s web site that the rated temps for this chip were 18c to 80c <unfortunately Intel’s site is down today so I can’t double check these numbers>

I always thought that colder was better…

What could be causing these lock ups?
Is the chip not happy or is it something else on the board that doesn’t like the cold temps?
Should I be backing off my voltage at these cooler temps?

 

[MsNath]

Update. I managed to get onto Intels web site.

I must have been halucinating the last time I was there. It seems I was mistaken about the minimum temp. There is no listing fo a minimum temp.

So Im still stuck with the problem of why Im getting lock ups when temps drop below 15c.

 

[Eriksson]

Hd problem I guess, but I threw my comp outsite once with 0c° and it ran fine for few hours.

 

[OpenFriday]

my GUESS (stress that word) would be that the cold just did not want to work with one of your components. i live in Canada aswell and i would just like to know where you buy all you equiptment?

 

[Eriksson]

OpenFriday (Feb 11, 2001 06:20 p.m.):
my GUESS (stress that word) would be that the cold just did not want to work with one of your components. i live in Canada aswell and i would just like to know where you buy all you equiptment?

You could be wrong I guess.....

 

[Tim-]

This kind of problem is most likely due to mechanical connections (electrical plugs, card edge connectors, etc) where the contacts are losing tension or shrinking away from their mates. The thermal properties of different materials cause them to shrink and expand at different rates and this can cause intermittent connections.

Computer hardware is designed for a nice, typically warm environment so little thought is put into "ruggedization" for temp extremes. I worked in aerospace many years ago and dealt with computers designed to go into fighter aircraft, and space craft and they were built for and tested at cryogenic temperatures. Need less to say their hardware didn't resemble a PC very much!

 

[MsNath]

Tim- (Feb 11, 2001 07:37 p.m.):
This kind of problem is most likely due to mechanical connections (electrical plugs, card edge connectors, etc) where the contacts are losing tension or shrinking away from their mates. The thermal properties of different materials cause them to shrink and expand at different rates and this can cause intermittent connections.

Computer hardware is designed for a nice, typically warm environment so little thought is put into "ruggedization" for temp extremes. I worked in aerospace many years ago and dealt with computers designed to go into fighter aircraft, and space craft and they were built for and tested at cryogenic temperatures. Need less to say their hardware didn't resemble a PC very much!

--------------------
The temps dropped almost 20c in 5 minutes so your probably right on the money. The key maybe a slower cool down period. Hopefully then everything will shrink more evenly...

 

[Mr_Goat]

Shrinkage? You guys are talking 0.000018" - 0.000007" for the materials inside the puter case...a single strand of hair is about 0.002"

 

[Colin]

I have seen a strand of hair keep a system from posting while stuck in a ram slot. It's possible.

 

[markedmundb]

If there ain't no contact, there ain't no contact!

 

[GauntMan]

Another possibility I´ve been stuck with once. The sudden shift in air temp made my system lock up once too. It turned out to be condensation on some pins and the small short circuit lower the voltage enogh to lock the system.This might be your problem. Lower the temp inside your room before you turn the system on. This will help prevent condensation. It might help. Try it.

/GM

 

[Ammethyl]

GauntMan (Feb 12, 2001 09:38 a.m.):
(...)
The sudden shift in air temp made my system lock up once too. It turned out to be condensation on some pins and the small short circuit lower the voltage enogh to lock the system.This might be your problem. Lower the temp inside your room before you turn the system on. This will help prevent condensation. It might help. Try it.

/GM

When taking a cold device (a PC, for instance) and bringing it in a humid, hot place: CONDENSATION can occur (if the device's temp is lower then the Dew Point). But puting a hot device in a cold environment, it's almost impossible to see ourself confronted with CONDENSATION.

 

[GauntMan]

In this case it might be condensation after all since the humidity in the room has to go somewhere. In my case it wasn´t much but apparently enough.

/GM

 

[klosters64a]

This is a very interesting post, MsNath. Thanks!
Dang! I've been dreaming of -200C ever since a man from Japan reached this temp with his C300A. He did this very carefully, in a proper laboratory. Got 800 Mhz +, too. Something about seeing one of your fingers shatter and fall to the floor sorta...brought me back to reality.

 

[Ammethyl]

GauntMan (Mar 05, 2001 10:38 a.m.):
In this case it might be condensation after all since the humidity in the room has to go somewhere. In my case it wasn´t much but apparently enough.

/GM

You're right when you point out that humidity has to go somewhere: It goes outdoor in this case, since the room gets cooled by air exchange (convection), the warm-humid-indoor air gets gradually replaced by outdoor-cold-dry air (in this case we are talking about absolute humidity and NOT relative humidity).
Since the room's wall and especially the electricity-heaten parts of the PC are warmer than the Dew Point...:No Condensation!

Though I may be wrong, I'd appreciate constuctive, informed criticism....

 

[Ammethyl]

Here I should add One specific situation where condensation might occur:

Suppose that you closed your computer room's door (the one that connects the room to the rest of your humid-home).
Then, you open your window/outdoor's door/etc. the room's temp drops to -10: No problem.... Do whatever you feel, staying in the room for a quite long time (say a few hours). Then, get hungry, close the outdoor's door/blabla, and open your room's door to go eat and LET the door open.
Come back and take a look at your condensation crashed computer.

In this case, your home's humid air would come in contact with your computer's COLD surfaces ( Temp lower than the Dew Point = Bad

Anybody correct me if you feel it is appropriate...

 

[surlyjoe]

"I don't get why my soundcard still samples at 44,1Khz in spite of my OCing... (and why it's still 16bit)."..........LOL, aw man ,I almost hurt myself laughing at this

 

[batboy]

I've seen CPUs that were cooled below freezing with pelts and had frost covering the whole heatsink and chip. I'm guessing it's not the CPU, but maybe a component, connection, or maybe a harddrive like was suggested. It would be "cool" (hehe) if you could duct some of that cold air from outside right at the CPU cooler.

By the way MsNath, nice pic, lookin' good! Stay warm, spring is on the way.

 

[LimeyGreg]

Dry joint or poor connection, at normal temps contact is being made, but as cooling occurs then contraction occurs until it is just enough to cause a crash - just needs 0.001" gap and it will quit.