T43p Sudden Power Loss

[mark-ibmtp]

A 2687DUU, with both external power and a charged battery, is abruptly powering down every few hours, for no apparent cause. [There has been no sudden movement or mechanical shock in these episodes. Also, no unusual temperatures, or abnormal fan operation.]

To restart the T43, it is necessary to disconnect the external power and to remove the battery for a while.

Is the internal power supply failing? If so, what would be the steps to repair or replace?

I didn't find this symptom in a forum search -- my apologies, if the answer is already here.

PS I know that motherboard failures can cause symptoms like sudden lock-up (freezing), display blanking, etc. Can a motherboard crack also cause power-downs like I am seeing?

[lukee]

Do you use TPFanControl? What are the temperatures in the time near to shutdown? I experienced similar behavior about a year ago - the problem was overheating of the CPU (100 deg. of Celsius was not a problem to reach) which caused whole computer shutdown. New layer of thermal compound (Arctic Silver 5) solved my issue. Before applying it, I precisely cleaned CPU's core, bottom side of heatsink and GPU chip with isopropylalcohol.

[mark-ibmtp]

Thanks for the info, lukee.

You asked about TPFanControl, which I'm not using. Is it better (I mean, less likely to overheat?) Also, I don't have a reading of the temperature -- does TPFanControl show that?

I replaced the entire fan assembly about a year ago, and I think I did a fairly good job of cleaning the CPU, and used the grease that came with the fan (which was an IBM factory assembly).

If I can read the temperature, and see a correlation with the shutdowns, then this would confirm overheating.

Can you tell me what is the maximum temperature the CPU should reach? 100C is obviously very high!

[lukee]

Yes, the TPFanControl shows CPU temperature also. Maximum temperatures which I can reach with a heavy use are in interval 80 - 85 deg. of Celsius and includes almost 100% CPU load and 3D gaming. If the laptop is closed (for example in a dock), the temperatures might be even slightly higher.

Please do following steps:
1. install the TPFanControl: http://www.staff.uni-marburg.de/~schmitzr/donate.html
2. run some application which requires lot of CPU/GPU power
3. in the same time keep the TPFanControl opened and check how the CPU temperature behaves

If the temperature will increase quickly but air which is blowing from the fan remains lukewarm then the problem is bad or stiff thermal compound. And high overheating of the CPU causes automatic safety shutdown of the laptop.

[mark-ibmtp]

A follow-up report, if anyone is interested in this problem...

I installed TPFanControl (nice software!) and have not seen any alarming temperatures. The maximum I noticed (after a couple of minutes of CPU-bound condition) was about 70C.

Because one of the motivations for TPFanControl was apparently to reduce fan noise, its "Smart" mode allows higher temperatures than the BIOS control, which I presume is what governed fan operation before I installed TPFanControl. When I select "BIOS", the temperatures are usually below 60C.

I was hoping to catch the temperature readings not long before a shutdown -- but after about 6 shutdowns in just a few days, I have gone for a comparable length of time with no shutdowns! I don't suppose this has anything to do with the installation of TPFanControl, as (to my understanding) the fan is running on the same "BIOS" algorithm as before, and in any case there doesn't seem to be an excessive temperature issue.

It would be nice to imagine that the gremlin has left, but probably he is only sleeping. My best guess is that I have a motherboard fault (cracked traces, or the like).

[RealBlackStuff]

A motherboard of that age was probably never taken apart yet.
Remove the CPU cooler, clean off the old thermal paste from CPU and cooler, and apply fresh Arctic Silver 5 paste.
That will help keeping temperatures down.

[mark-ibmtp]

@RealBlackStuff (or anyone else, who might know):

I did replace the fan about one year before the shutdowns began (see above). It is not clear to me, that the problem is one of overheating -- but I haven't ruled it out.

Does anyone know what temperature triggers the "emergency shutoff"? As I posted before, the highest I've seen was about 70C.

Thanks, all.

[RealBlackStuff]

The IBM-supplied thermal paste is not the greatest around, and may well have dried out by now.
Replacement won't do any damage.

Another thing to check, albeit a long shot: if you run W7 installed and have screen blanking on after xx minutes.

Recently I was working on a quite new desktop, that spontaneously shut itself off regardless of being used or not.
As it turned out, it was an incompatibility of Windows 7 Home Premium 64-bit (that was installed on it) that for some reason could not work together with a certain type of motherboard (Asus M4A78T-e), if W7 was set to blank the screen after a certain interval of activity.
See e.g. http://social.technet.microsoft.com/For ... 058421a544
This desktop suffered from that problem since day one, and the owners were ignorant enough not to trace down the problem.
It took me nearly a day to find this, and I've been fixing computers since the Commodore 64 days...