T23 on the fritz, bios battery prehaps?

[Marzzle]

My T23 randomly crashes now for some unknown reason. It all happened one day as it was merrily chugging along and then all of a sudden it became horribly unstable and even crashed in the bios once. This was kind of a gradual decline in which the system would las for like 2 hours then 30 min then 5 minutes and then not even to the bios.

Flash forward to the day after and I got it to stay up all night long getting my stuff off of it onto another hard drive (about 20GB so it did take a while). This may or may not have something to do with pulling the bios battery (I had had similar seemingly random problems caused by dying batteries in the past). However the next day it started the whole crash thing again (though this time it was limited to the OS) so I just set it down for a few weeks since I didn't have the time to actually trouble shoot it.

The otherday I turned it on and left it shut. I set up a ping -t and then remote desktoped onto it when I went to sleep. I wake up in the morning and awesomely it's still going. I tried to use it and BAM crash. To clarify when I say crash I don't mean BSOD. It just sort of locks up and nothing moves on the screen. The pings stop getting returned and the RD gets lost. This was after about 14 hours of continuous uptime. I had even been web browsing and some other stuff via RD so it wasn't because I started using it.

Is there some inherent problem with me opening the thing up and using the track point that crashes it? I have used it both on and off battery power and it seems to of made no difference. The only thing thus far that has actually improved it seems to of been pulling the bios battery. Currently there is no bios battery still since I'd like to make sure it works before I get a new one.

Just for some spec stuff:
D-link external 802.11g card (tried removing it first)
Windows XP (dual booted with...)
Fedora core 4 (havent tried to see if it crashes in that yet)
Stock cd drive
no internal wireless
1024x768 screen
40GB stock HD
2x256mb pc133 (or whatever it uses I can't remember)
1.2 ghz PIII mobile
Oh and I changed the nub out for a cup on the track point

So what can you folks make of the situation?

[Orevin]

Do you get any error messages?
You should check the RAM first. Run the machine with one at a time. If it freezes with one and runs smoothly with the other, you may have found something. Check with both RAM slots too.

-Orevin-

[Marzzle]

No errors, like I said it just freezes. I started it up this morning and then it crashed when I tried to log in on Gmail (took like 5 min). Right now I'm running memtest on it and it's been going for like 30min. It doesn't really make any sense I suppose. Every once in a while it seems like it's just going to work but then it crushes my hope with a freeze of the screen.

[danage]

sounds 99% like a ram issue. can you "borrow" some ram anywhere to test it out? might be the mainboard or processor, but unlikely. ram would be good, since out of the three, thats the cheapest to replace.

[danage]

sounds 99% like a ram issue. can you "borrow" some ram anywhere to test it out? might be the mainboard or processor, but unlikely. ram would be good, since out of the three, thats the cheapest to replace.

[madmaven711]

I would suggest trying different memory. if you have 2 chips try using one only and if it still happens change slots. I have seen the blue screen/ memory error flash by so fast you can miss it. or not at all even.
If its not ram. I would also try a cooler under it . If you dont want to wait or invest about $15 for the cooling padm you can prop the back up a couple of inches and run a small fan to flow under it. I have seen many freeze up if the processor is running too hot. and a cooling fan solves the problem I would also take out any extras like the pcmcia network card, the Cd rom bay the large battery. I would not run it without the Cmos battery. Put it back and see if your clock looses time. if not it is probably not the battery. PLus if you didnt get the error code for battery or bad check sum, on boot it is even less of a chance it it the CMOS battery.

[madmaven711]

Sorry but sometimes I just assume you have good antivirus and ANTI SPYWARE. you should run them both especially the anti spyware. try 2 or 3 or 4 different products. like Microsoft antispyware beta, spybot, adaware, xsoftspy, xoftspy isnt free but it will give you a list of any spys you have .then ask for $40 to fix it. If you can eliminate this as a problem. then you can concentrate on the hardware.

[Nolonemo]

If its not ram. I would also try a cooler under it . If you dont want to wait or invest about $15 for the cooling padm you can prop the back up a couple of inches and run a small fan to flow under it. I have seen many freeze up if the processor is running too hot. and a cooling fan solves the problem

Re cooling, I have heard that the heat sink compound between the CPU and heat sinkl assembly can degrade over time. You could try opening the TP, removing the heatsink/fan, and replacing the old heat sink compound. Arctic Silver is a highly recommended brand.

[Marzzle]

Wow thats alot of responses! I've been barking up the RAM tree by running memtest for about 6 hours now. It was stable and it crashed sometime I wasn't looking. This was with both sticks in. I think I may try removing the after market chip (even though it's IBM approved).

Also I don't actually run any resident AVs or anti spyware stuff. I'll occasionally do a scan with clamAV (latest definitions etc...) but it's never pulled anything up. I haven't had a virus for 3 years now and the one time I did get one it was kind of on purpose anyway.

[Marzzle]

Re cooling, I have heard that the heat sink compound between the CPU and heat sinkl assembly can degrade over time.

This might be a very plausible explination. Why else would it happen after years of consistent functionality? Thats why I thought it was the bios battery since it seems like the only thing that could of degraded over time. The problems do seem to resemble an overheating of the processor. I think I may install a software temperature monitor and then stress the cpu for a bit and see if that wears it out. Running the ram for six hours probably wasn't that CPU intensive but it was sitting on my bed the whole time.

[Nolonemo]

Use the power settings or battery maximizer or whatever to force the CPU to run at the lowest rate, set the TP on a couple of pencils to get some air under the base and see if that extends the time before crashing. If so, would point to a temp problem.

[madmaven711]

I read on another forum that not using antivirus and anti spyware is like sleeping around without using a condom. It is not a matter of if you will get something but a question of when...
NOT A PRETTY METAPHOR
If you are on the internet you have to be REALLY careful where you browse if you dont use protection.

[Marzzle]

I think that with a little intelligence it's pretty easy to stay away from malware stuff. I'm using firefox so there's no IE exploits to worry about and if I'm downloading an executable then I just scan it manually if it's from a developer or place I don't trust already. I'm not to sure how much of a safety factor it is but I'm running BBlean for windows (a shell replacement thats fast and light, kind of what I'm going for with this old T23). The computer has auto update notifications enabled so it's upgraded pretty much as soon as MS releases a patch. With those precautions I think I can get by without using my scarce resources on a AV, but it's not for everyone.

I'm starting to definately think it's overheating now. I got speedfan onto it and it's reading 63 C. This seems incredibly hot to my A64 running preconceptions (I get upset when it hits anything over 45), but I know it's a different architecture. What kind of temps you you folks see on your PIII tualitons?

[Nolonemo]

My T22 1GB runs between 53 and 58 when web surfing, you should be idling in the 40s....

[Marzzle]

I have reason to beleive that the 63 C reading is wrong. I rebooted and it went to 48 and during a sandra burn in it just stuck there. The SMART reading went up but as far as I can tell there's no way to read the thermal resistor in this TP. Am I missing something? Speedfan only comes up with this incorrect reading and the smart temperature.

[Nolonemo]

Try using MobileMeter (that's what I have on my TP), shows CPU speed, temp, & HD temp, download link is on this page:

http://www.geocities.co.jp/SiliconValley-Oakland/8259/

[Marzzle]

I'm apparently hitting about 82 C max then the fan starts in and is bringing it down to around 73-75 C. Is that normal? I'm runing sisoft sandra's media bench on loop. It was idling around 48 C on start up.

[Nolonemo]

OK, I did a little testing. I have a 1GHz T22 with stock heat sink compound replaced with Arctic Silver. I used MobileMeter and Task Manager. The laptop was sitting on a wood desktop.

First, I rendered some video, which pegs the CPU (running at 1GHz) at 100%. It took about 5 minutes for the temp to rise to and level out at 82 degrees C, where it stayed for the 10 minutes or so I let the render go on after that.

Then I closed everything other than MobileMeter and Task Manager. Task Manager showed the CPU idling at between 1% and 10% (there's some background service that kicks in every few seconds and momentarily spikes the CPU to 10%, must be doing some kind of polling) at 700MHz. The temp eventually dropped to 53 degrees C. and seemed to level off there.

I didn't try to see what the idle temp was from a cold start. I seem to recall that when I'm web surfing with the laptop on my lap the temp fluctuates between the high 40s and the low 50s.

I guess the question for you is how much sisoft's media bench is taxing the CPU, if the CPU's pegging, you're running a lot cooler than me.