Random Shutdowns

[muppet]

I've started observing random shutdowns recently on my modified Acer Aspire 5520g with ASUS 8600m GT 512MB.

I modded my laptop a while back now, and I've developed an amount of sensor software to monitor GPU, CPU and core temps, as well as fan speed and GPU clock speeds into perfmon on my 5520g. I baselined the thermal profile (much as it was before - thanks to the AS5 I guess and more fan activity) I collected and played a few games on the laptop then it became a glorified web browser due to the real world intruding. I run it on a Zalman cooler anyhow, just to be safe.

Having left it turned off for a few weeks, I started it back up and got a shutdown (total power down, no warning) - requiring the power cord to be removed before I could start it up again...

Long story short - it happens on power, battery, power and battery in, - I striped the machine down, used compressed air, re-applied AS5 etc, updated to 1.33 BIOS, updated all drivers, removed Vista SP1, used different power supply. Still had the problem, thermals normal.

In profiling the thermals, I noticed that it almost always occurred after a period of low usage when the system was idling between no fan, letting the GPU/CPU heat up a bit then applying the fans. In fact, I can't remember a shutdown under stress.

I've found a growing number of people with random shutdown issues - and then hit on The Inquirer's articles regarding NVidia's G84 and G86 chips:

http://www.theinquirer.net/gb/inquirer/new...chips-defective

Also, NVidia themselves have identified "field failures" (in addition to GPUs) with their MCPs (of which the 5520 also uses):

http://www.nvidia.com/object/io_1215037160521.html

I've seen that some vendors are changing their fan profiles to try and reduce the thermal stress by running the fan more (some argue to keep the machines issue free long enough to exit warranty) - however nothing from Acer that I've been able to find. With this in mind, I've made a quick and dirty tool to set the CPU temp to 67C all the time via the embedded controller so the fan ramps up - and I haven't experienced the fault (though it is so damn random I've gone 5 days straight without issue then had it hit). Hopefully this will resolve it for long enough that I can use the laptop for a reasonable functional life.

Anyhow - a heads up on the issue and potential stop gap for the 5520 that I'll clean up a bit if it proves to be of value.

Edited September 10, 2008 by muppet