The mystery of my desktop that locks up when it gets too cold

March 25, 2019

This winter I have been having a sporadic but surprisingly consistent problem with my home desktop computer, which is that, well, here's my Tweet:

So I definitely appear to have the most ironic of things, a computer that doesn't like it when it's too cold. I don't think it's the CPU fan or the CPU, either, which leaves me strongly suspecting the motherboard. I guess it's time to update to the latest BIOS.

(Sadly, updating to the latest BIOS didn't fix the problem.)

What has been happening sporadically all winter is that when the ambient temperature around my home desktop drops too low, the machine will lock up. When the ambient temperature rises again, the machine boots up again. I have not nailed down exactly what 'too low' is, but based on motherboard sensor readings and an external electronic thermometer, it is around 60 F ambient outside my case and no more than 68 F inside it. Since this is a far lower temperature than I'm comfortable with, it only happens at times when I'm nowhere near the computer. The CPU temperature appears to be irrelevant; I have run CPU soakers that kept the CPU temperature fairly toasty warm and the CPU fan actively working away, and still had the machine lock up.

One time I deliberately created a low temperature situation where I could see the machine in this state after it had locked up, before the temperature rose again and it came back to life. When I observed it in its locked up state, the power lights were on and all of the fans were spinning but power was not being provided to my USB keyboard, and power-cycling the machine didn't bring it back to life (only letting the heat come on did).

(One might wonder why I didn't see this last winter, but as far as I can tell I only assembled and put together this desktop in late March of 2018, which is after the really cold weather that induces such low interior ambient temperatures.)

This is a genuine and somewhat frustrating mystery. I have no idea what the cause is, apart from that it seems most likely to be something related to the motherboard, although I suppose it could also be the power supply sagging the voltage on one or more rails when it gets too cold. One of the possibilities that worries me now is that sufficient cold is making various metal parts shrink and move enough that they're creating some sort of short that shuts things down; this could be a fault in the motherboard, or something about how I put the machine together.

(In theory I could completely disassemble and reassemble the machine, to re-seat and re-do all connectors and so on. In practice I have very little enthusiasm for that, especially taking apart the CPU and the CPU cooler.)

Comments on this page:

My gut thoughts are an intermittant open rather than an intermittant short, but this is still conjecture. Expansion and contraction on a bad solder joint could be causing this.

Suggestion: use a can of air duster upside down to freeze certain parts of your motherboard one by one. Start with the southbridge and PMICs.

PMIC = power management IC. These are little ICs with at least 8 pins sitting near the power supplies on your motherboard. Power supplies are areas with lots of large mosfets (big chips with few or no legs) and inductors (coils, often in big grey epoxy boxes).

Source: I repair electronics.

By Me2 at 2019-03-31 21:14:36:

Take apart the computer and put it back together. i.e. unseat the memory and reseat it, unplug and replug everything.

If there is a connection that is not being made and it will be easier to just re-connect everything rather than debug a faulty hardware connection.

By Madhu Gupta at 2019-04-01 02:49:32:

You can try to reapply thermal paste of your CPU.

By Shadi at 2019-04-01 03:16:11:

I ran into something similar before, changing the power supply fixed it, I also noticed at the same time that some of the capacitors on the motherboard were defective, but I'm not sure if that was related or not

By Stew Dellow at 2019-04-02 10:51:05:

I once had a similar problem with a Thomson Wysius Plasma TV. This post from 2006 ( explained why the thing never worked in the cold and how to fix it.

Sure enough the hairdryer on the offending transistor worked like a charm as temporary fix until I soldered a new one in.

Might be of some related interest.

Written on 25 March 2019.
« Prometheus's delta() function can be inferior to subtraction with offset
Drifting away from OmniOS (CE) »

Page tools: View Source, View Normal, Add Comment.
Login: Password:
Atom Syndication: Recent Comments.

Last modified: Mon Mar 25 21:11:23 2019
This dinky wiki is brought to you by the Insane Hackers Guild, Python sub-branch.