T41P GPU problem after fan assembly exchange

[DrMarco]

My 2373-GEG T41P bought in April 2004 recently stopped booting due to the famous "Fan error". I ordered a new fan assembly from IBM and did the swap myself four days ago, following instructions in the hardware maintenance manual. That was quite easy (remove battery and hard drive, keyboard and palm rest and you have access to the fan assembly).

Right after the swap I made a few stress tests by launching CPU-intensive programs for a while. Everything went smoothly until yesterday when I left the computer on with lid closed while I was doing something else. When I got back to it, the screen was blank and I couldn't get it back to life. There was no apparent reaction to hitting the keyboard or the trackpoint.

After switching it off, and restarting it, it first booted fine but immediately after the screen went blank again. Leaving it switched off for a longer period of time allowed the screen to work a few minutes before going blnak so that I suspected a thermal problem. I went back to the fan assembly and then thought I had make a big mistake: I had not peeled the pink pad protecting the thermal grease on the part going over the GPU. First, the hardware stated explicitely that "Only in the case of Fan assembly, long, you need to peel the plastic sheet b off on the thermal rubber before you place it on the system board. Fan assembly, long M10 DOES NOT have the plastic sheet.". Second the pink square looked very much like something designed for thermal management.

I removed the pink pad, try to fit the fan assembly more tightly to the GPU but alas the problem has not gone. It is now about 10 minutes before the computer crashes and the screen goes blank very reproducibly. Needless to say, I have now a very fine computer that is unusable and I am trying to figure out the best to do. Here is my assessment of the situation.

1) Looking in retrospect, I don't believe I made an error at all. Probably the pink rubber-like pad was the thermal rubber alluded to in the manual and there was nothing wrong in leaving it on the fan assembly ? Too bad I throw it in the trash bin. Any way the problem occurred with and without.

2) I was of course suspecting the GPU because of the blank-screen/no video output symptom. But it could also be due to a system crash. By remotely logging into the thinkpad at boot time I was able to verify that although most of the time the system would crash when the screen would go blank (no response from terminal), but this was not the case in some cases. The screen went blank and I could still type in the terminal. Thus I guess this is the GPU that goes south and crashes the kernel some times. In one instance there was a kernel message about an interrupt being switched off.

3) The problem does not seem to due to heating but to physical stress. When I leave the computer alone and launch intensive tasks remotely the computer can be one hour without going blank. Sometimes it goes blank spontaneously but most of the time it is right after I have touched it. Under linux you can monitor GPU temperature with the ibm-acpi driver and I have seen the machine crash at a GPU temperature of 45 degrees Celsius.

Thus it seems that this is another strike of the famous GPU soldering problem due to which the motherboard was already swapped under warranty in september 2005. Even though no brutality was involved in the fan swap, I guess it must have perturbed the GPU at some point, which surfaced when the machine was left with lid closed (which I have done many times before).

Since I am fond of my 1400x1050 14'' screen I guess that besides the reflow trick I have no other choice than to buy a replacement motherboard. IBM France's quote is 900 euros (~1200 USD). I have however found that thinkpad-parts.com will sell the 93P3313 motherboard for 190 USD which is quite reasonable. Also, Ebay seller better_used_eq seems to sells tested sets base+motherboard + CPUs at incredible prices. For example 140 USD for a 93P3313 motherboard and 2373-GGU base + 1.7 GHz CPU : http://cgi.ebay.com/ws/eBayISAPI.dll?Vi ... 0328773029

Have you heard of these? What would you recommend as a reliable and cheap source of thinkpad motherboards? (a good business if I judge from the numbers of faulty motherboards) If that makes a difference I am not that interested in having a FireGL T2 ATI card as I do not use any acceleration, I only want to be able to run the display at the nominal SXGA+ resolution.

Thank you very much in advance.

[FTC]

Hi,

Yes it could be the GPU problem (and most likely it is) but... you need to make sure.

I had a very similar problem that ended up being .. just a bad memory dimm ! and the problem reproduced more easily when charging the battery at the same time.. go figure.

So I wold first think on other recent changes you made to the machine.. did you by chance swap memory or memory slots recently ? or the battery ?.

Then try to isolate the problem, when it happens.. are you by chance charging the battery or running w/battery when the problem happens ? Can you run without one of the dimms OK ? does the problem 'fix' when pressing the GPU ?