Flakey T20: Heat related problem

[vlyne]

After a few months away I got back to trying to rescue a flakey T20 for my daughter. The original owner got tired of it randomly dying. I tried the usual suspect and changed the CPU fan to a brand new one (with preinstalled thermal compound) with the spring loaded feet and replaced the 700MHz CPU with a 900MHz version. I monitored the temperature with mobmeter but it kept dying randomly. even at a reported CPU temperature of 36C. and no applications running apart from mobmeter The LCD would go blank, the AC light would stay on and the harddisk would die but try to restart every second or so. The longest it would stay up was for one hour (but doing nothing). I changed the CPU back to the 700Mhz and the old fan cleaned of dust and refreshed the sponge with a wipe of alcohol to soften it up. Still the same symptoms. I took off the modem card and hard disk and replaced the 512Mb Crucial ram with a cleaned 128Mb stick. Still the same, even after swapping RAM slots…I’ve taken the motherboard off twice and checked major fuses, diodes, inductors and capacitors. I found one of the small capacitors beneath the CPU was dislodged (probably my fault in trying to clean the board with a brush – the bristles may get caught and dislodge small components). Got it resoldered but a bugger of thing to get back on. – Hint: do not try and blow on it after soldering (it will fly away!) and get a good tweezer with a sharp tip. I touched up solder pads on some of the large power components. Yesterday I loaded up tpfancontrol and managed to get the fan to stay on at full blast. I had it running for a good 3 hours but under the stress of installing XP SP2 and the .Net Framework (so I can run NHC) it died. After that It wouldn’t stay up more than 30 seconds (mainly because I couldn’t get tpfancontrol to start soon enough). I took out the CPU and replaced it with the 900MHz unit but still it wouldn’t stay up for more than a few minutes. I hadn’t replaced the screws for the keyboard so I replaced them and managed to get it to stay up for an hour or so but half way through the .Net installation it died. There it stays…arrrghh! Tightening down the keyboard seems to help. So to summarise, I have a heat-related motherboard problem that is not the CPU or fan but someother heat sensitive component/problem on the motherboard. Manually turning on the fan at full blast and putting back the keyboard screws helps. One thing I did notice was the bios version of 1.22 had a date of 1999 (it should be 2004) and changing it down to version 1.21 didn’t help. I’ve also powered down the CPU to Auto Slow mode and disabled all shutoff timers. Any hints/suggestions? Anyone know the whereabouts of any thermal sensors apart from the one in the CPU?
Cheers

[nikki605]

The symptoms you describe on your T20 sound nearly identical to a problem I had with my T21 dating back to the end of last year. I swapped CPUs and heatsink/fan assemblies. I applied new AS5. I installed mobmeter and saw the TP shut down even when the temp was in the 40C range. At times, it would run for several hours before shutting down. Other times, only a few minutes or maybe several days. With or without the a/c adapter attached. Swapped batteries. Etc, etc, etc.

I finally gave in to the fact that it was not heat related and broke down and bought a used mobo off Ebay for $70 plus shipping. There are MOSFET power circuits on the mobo that can go bad and cause these symptoms. It looks like overheating, but its not. I was nervous about buying a used board, but took the chance. I bid on a board where the seller offered a warranty from Square Trade. For $9 more, I got a 2-year warranty on the "used" mobo. If anything goes wrong with it in that time, I can get my original purchase price back.

I installed the board on April 9, and have not had the problem since.

I can't say for sure that your problem is the same, but as I said, the symptoms you described sound identical to the symptoms I was having. In my case, it was not overheating, but rather, a bad mobo.

[vlyne]

Not the sort of news I was hoping to hear but you’re right I may have to get another mobo. The reason I’m thinking heat related is that I can always get it started once I allow it to cool down. The cooler it is to start with the longer it will last. It only dies once it’s “stressed”.

I took out a T21 board I had lying around and found an LM75 temperature controller located under the PCMCIA area on the underside edge. Next is figuring out which sensor location this corresponds to in tpfancontrol and some way of testing it.
Cheers

[nikki605]

I messed around with my problem for several months. Including, sending the TP to a repair shop here in the states that does component level repairs. After getting it back the first time it worked for about 2 weeks, then it started doing the same thing again. I sent it back for warranty repair and they said they never saw it fail for the almost 2 weeks they had it back.

They returned it to me (I paid for all this shipping back and forth) and within a few days it started failing again. That's when I broke down and bought the used mobo.

I was wasting too much time and not able to use the TP. Right now I am traveling and using the T21 to type this reply. Its working fine. It would have been nice to know exactly what was bad on the mobo, but I wasn't making any progress and needed the machine back working. Replacing the mobo was the quickest way to accomplish that.

Good Luck

[vlyne]

I'm glad you've got a functional machine again. The difference in my case is that I can make my machine die by stressing it out - I can almost guarantee when this will happen. So, in some sense it is flakey but it is also repeatable. It is just that by getting the fan to turn on at full blast and starting from a cold start I can run it for longer but it will die once it's warmed up. Unless there is some repeatability or response to some kind of stress (heat, flexing etc) you have no idea which part is failing. In my case, heat can affect a whole range of components but there are components that are specifically designed to shut down the system when overheated. I'm guessing that the one or more of the thermal sensors has gone astray hence the reason for my interest in them. And, I'm reasonably sure that given the symptoms of my machine a competent component level repairer should be able to identify the culprit fairly quickly. Not being all that competent, nor able to afford one who is, I have to do things the hard way
Cheers