Hardware slowly dying? Random lockups, freezing in BIOS and

[ngfrazier]

I'm completely baffled right now. I'm hoping that someone can give me some ideas on how to fix this.

I have a T43p which is now exhibiting weird lockups/freezing. Here are my symptoms:
1) Random freezing in my OS (Ubuntu) and in the BIOS. Sometimes the system recovers (1-2 minutes of COMPLETE freeze--no ctrl-alt-del, nada, nothing works); sometimes it stays stuck.
2) Extremely slow boot (VERY long to initialize). Sometimes it freezes when I enter the BIOS to make any changes.

Troubleshooting: First, I reset the BIOS. Next, I tested my RAM (2gbs total) with Memtest+ for 8 hrs. My crucial memory tested OK. In a last-ditch effort to solve this problem, I reloaded Ubuntu. No help. Random freezing continues.

The problem is hard to reproduce. Sometimes I can work for several hours without a hiccup, sometimes it starts the minute I boot.

Please help I don't wanna ditch my thinkpad!

[Johan]

Weird, I agree, and I also never recall having heard of anything like this before. Now that you've tested the RAM, I suggest that you also test the rest of the system (you don't mention if you've already done this?), to see if e.g. the CPU exhibit any intermittent errors - use either IBM's bootable PC Doctor or use the newer Lenovo Systems Toolbox (PC Doctor 5+). What type (manufacturer) of HDD do you have? If being a Hitachi, try the Hitachi Drive Fitness Test - if another brand, use the appropriate tool (all HDD vendors have these tools, check their support pages).

Johan

[rkawakami]

Your comment about "sometimes the system recovers..." reminds me of a previous troubleshooting experience.

Assuming that you can successfully load up the DOS version of PC Doctor, select the RTC test from the Systemboard menu, press the F2 key (Test Options), select "Set Pass Count", change the value of "1" to "10", press the Enter key, then press Esc to exit the Test Options menu, then with the highlighted bar selecting the RTC test, press the Enter key which will begin running the Real Time Clock test for 10 loops. While the test is running, flex the chassis around a little bit and see if causes the RTC test to fail. By "flexing" I mean to take the front left corner of the system and slightly lift it up, in relation to the rest of the base. Do the same for the front right corner. Repeat the process by also pushing down on the corners; hanging the system off the edge of a table or putting your fingers under the system while pushing up, while at the same time using your thumb on the top of the corner pushing down. What you are trying to do is flex the motherboard around to see if it affects the running of the RTC diagnostic.

What I'm suspecting, and it may be totally off-base, is something I've seen with a T23 motherboard awhile back. A loose crystal on the motherboard was making the system act like it was put into a suspended state at random times. The system would sometimes boot normally but freeze during POST. Putting some pressure on the corner of the system would "un-stick" the system and allow it to proceed but once the external force was removed, the system would again freeze. Your T43p may have a similar problem with one of the system clock generators.

ref: http://forum.thinkpads.com/viewtopic.php?f=28&t=40648

[ajkula66]

Hopefully, Ray's advice will resolve your issues...

I hate to be a downer, but I just had a planar swapped on one of my T43s with the symptoms that sound extremely similar to yours...

Good luck.

[ngfrazier]

Wow, thanks for all the suggestions! I'm going to try what you suggest and report back.

I'm kinda sentimental about this laptop--I've had it since May '05 and it has been a joy to use and has been FAR more rugged and reliable than other laptop brands I've seen. Silly how we can grow to love a tool that "just works."

Anyway, for those interested in my hardware configuration, here are the details:

T43p motherboard w/ ATI FireGL v3200, with fingerprint sensor, type 2668 machine
14" 1400x1050 LCD
2gb DDR2 (2x 1gb Crucial)
250gb Western Digital BEVE
Ubuntu 8.10

Cheers!

Nathan

[killer]

Nathan,
It is a good idea to put those details in your signature as it saves a lot of questions if or when you make a posting with a problem.

[ngfrazier]

Your comment about "sometimes the system recovers..." reminds me of a previous troubleshooting experience.

Assuming that you can successfully load up the DOS version of PC Doctor, select the RTC test from the Systemboard menu, press the F2 key (Test Options), select "Set Pass Count", change the value of "1" to "10", press the Enter key, then press Esc to exit the Test Options menu, then with the highlighted bar selecting the RTC test, press the Enter key which will begin running the Real Time Clock test for 10 loops. While the test is running, flex the chassis around a little bit and see if causes the RTC test to fail. By "flexing" I mean to take the front left corner of the system and slightly lift it up, in relation to the rest of the base. Do the same for the front right corner. Repeat the process by also pushing down on the corners; hanging the system off the edge of a table or putting your fingers under the system while pushing up, while at the same time using your thumb on the top of the corner pushing down. What you are trying to do is flex the motherboard around to see if it affects the running of the RTC diagnostic.

What I'm suspecting, and it may be totally off-base, is something I've seen with a T23 motherboard awhile back. A loose crystal on the motherboard was making the system act like it was put into a suspended state at random times. The system would sometimes boot normally but freeze during POST. Putting some pressure on the corner of the system would "un-stick" the system and allow it to proceed but once the external force was removed, the system would again freeze. Your T43p may have a similar problem with one of the system clock generators.

ref: http://forum.thinkpads.com/viewtopic.php?f=28&t=40648

Wow. You are exactly right. Totally bizarre. I can confirm the problem is related to "flexing" in the motherboard (perhaps a solder joint?). When the motherboard sits *just* right (usually on my desk, not atop paper or other objects), it sometimes boots normally and passes DOS PC Doctor perfectly fine. Other times, however, I have to flex the lower right side to get the system to "unstick."

That said, I have two questions:
1) Can I disable something in the BIOS which would circumvent this problem?
2) If not, how would I go about determining which solder point (assuming that is the problem) needs repair?

Thanks again for all the suggestions and support

[rkawakami]

1) Don't think that there's anything you can do in the BIOS which would "bypass" the problem.
2) That's the million dollar question .

For that T23 that I described, it was a matter of determining that the RTC diagnostic failed in PC Doctor. Since I'm somewhat familiar on what that test is probably looking at, I went searching on the motherboard for any IC or part associated with timing or clock generation. Crystals are the major component for generating clock signals in computer systems. I was able to identify a couple of them so that's were I concentrated my inspections. It may have been totally lucky that I saw the tiny cracks in the solder joint on two of the legs of that problem crystal.

Surface mounted crystals are typically in a couple of different types of packaging. You might find them as square parts in metal containers, long, thin black plastic packages, or square plastic parts. I'll try to get some pictures off my T23 motherboard and post them in a few minutes.

edit: Digging around my web site, I see that I already did this some time ago....

http://www.kawakami-ca.com/images/t23_crystals.jpg

It was posted last year in this thread: http://forum.thinkpads.com/viewtopic.php?f=2&t=65247