Dead TP 240 - is there a way to rescue it?

[JanR]

Hi all,

after hours of reading postings I registered (thanks for the fast approval!) to ask if there is a way to rescue my old TP 240 (Celeron 400, 192 MB RAM, 12 GB HDD):

I bought it in early 2001 and used it as main notebook until I got a X60s from work last year. It survived falling from a university table to the hard floor in 2004 (landing on a WLAN card that probably safed the machine) and operated it on linux the whole time. Last year I even upgraded it by replacing the modem card with a Broadcom WLAN card which worked perfectly.

In late 2006 I first had problems with turning it on (screen remained blank, nothing happend) but I was able to solve it by power cycling. This spring it became worse so I dissassembled it in part to see if there is a problem. I found nothing but the machine was fine without the upper part of the case. After adding some isolation (I guessed there was a blank part of the speaker cable touching the PCMCIA slot) I assembled everything and it was fine again.It was used then as secondary notebook for another few month. Problem once returned, was solved by removing battery for a while. Then we needed it for vacation (me and my girlfriend... so we needed both Thinkpads )... and I was not able to power it on. After vacation (we used an old and heavy Vaio from work instead) I tried the old trick with dissassembling again with no success this time.

Technically, the problem is as follows:

Turning on the machine (AC or battery does not make any difference) lets the LEDs behave the usual way (short on, then off except power) and turns the fan and harddisk on. The screen remains black (I even see no backlight). Then, fan increases speed slowly to maximum without emitting heat. In this state the machine can only be turned off by pressing the power button for longer then 4 seconds (most the times this works, sometimes it only shuts off the fan so power must be removed).

The problem is the same with external monitor and/or keyboard used, it remains also unchanged if I remove the connection to the LCD completely (from the first successful "repair" I know that a TP 240 is able to run with just the board, power and external keyboard and external monitor).

I inspected the mainboard from both sides and found nothing strange there (obviously, I only powered it on with heatsink attached to the CPU). Even in the area of the PCMCIA connector which was probably affected by the impact 3 years ago everything seems to be okay.

In the area of the power supply I was able to find 5V, 3,3V and 12V... so everything seems to be there.

I used the "erase CMOS" jumper, I tried with AC, with DC, with both, with harddisk, without... and, obviously, I also removed the S0-DIMM running it only with the 64 MB onboard. The WLAN-card was also removed during all these tests.

Is there any more idea what can be done with that machine? I would like to repair it with little cost (I 'm already trying to get a new board cheap but I guess this is not really an option as boards are the same price range as complete machines at ebay or defect themselves).

Is there anything I may have overlooked? Any idea?

All the dissassembling was done using the HMM by me and another guy from work (we both are experienced with hardware).

Greetings from Berlin, Germany,

Jan

[tfflivemb2]

Welcome to the Forum!!

I would check in the area of the damage to see if anything on the systemboard came loose, such as an inductor or something. You might need a magnifying glass to see if a connection is working loose. You might be able to touch up some of those spots with a small soldering iron.

[phool@round]

Sorry to hear your venerable 240 is loosing the battle.

I would be very suspect of the area near the PCMCIA socket. If you think about the force of impact directly put on that socket (3lbs) it should give your minds eye a place to start. A trace, a component, and so on.

By the way the BIOS chip is socketed very near the PCMCIA socket. I've done this in the past.....lift that chip out of it's brown socket completely then push it back in. It may have come loose enough to allow oxidation over time to cut off the signal. If the video chip isn't firing up, that's usually an old school reason. They used to call it chip creep, when the expansion/contraction of chips would work them out of their sockets.

And.......I was able to purchase a 240X system board for $15 off ebay. It's a direct replacement of the 240's. So, don't give up hope on getting one for a decent price.

Good luck, post what you find.

[al7kz]

From the description of your problem, no video, fan running, your CPU may be bad.

My 770Z had a similar problem, which was also intermittent, before it became permanently dead:

1. No video
2. Fan running
3. Speaker symbol flashing
4. No other signs of life

I fixed it by replacing the CPU board with one of the same type that was removed from my upgraded 600E (P2, 366, MMC-2).

I'm not familiar with the 240, but from looking at the HMM it looks like the CPU is soldered to the MB. Try replacing the system board/CPU if you can find one. Good luck...

[JanR]

Hi all and thanks for your answers,

I try to address them all:

I would be very suspect of the area near the PCMCIA socket. If you think about the force of impact directly put on that socket (3lbs) it should give your minds eye a place to start. A trace, a component, and so on.

That was my guess too, but the device operated with not a single problem for more than 3 years after that accident. This could only explain some very minor damage like a connection that sometimes has problems. We tried to find this kind of failure by applying pressure to different parts of the board (isolated, of course) while we started it - no success.

By the way the BIOS chip is socketed very near the PCMCIA socket. I've done this in the past.....lift that chip out of it's brown socket completely then push it back in. It may have come loose enough to allow oxidation over time to cut off the signal. If the video chip isn't firing up, that's usually an old school reason. They used to call it chip creep, when the expansion/contraction of chips would work them out of their sockets.

I will try this on monday (the machine is in my office at the moment), but I'm not that sure if the BIOS chip is really socketed (of course I made fotos of everything except this one).

And.......I was able to purchase a 240X system board for $15 off ebay. It's a direct replacement of the 240's. So, don't give up hope on getting one for a decent price.

I monitor german ebay since 6 weeks. In this time there was one working 240 for 150+ Euro and one 240X board (I know there are compatible) with a failure description very similar to mine. I just checked US ebay... the situation is much, much better but shipping is very expensive or not possible at all (most ship only to US). I guess the reason is that IBM sold these machines very, very expensive in germany. I got mine as a special offer AFTER end of production from a local dealer who had it on display. If I find a cheap board on US ebay this would be a solution... but I will spend less than half the money a used X-series machine would cost. Thats why I hope to fix it somehow or at least figure out whats the problem.

I would check in the area of the damage to see if anything on the systemboard came loose, such as an inductor or something. You might need a magnifying glass to see if a connection is working loose.

Thats exactly what we have done with no success.

I'm not familiar with the 240, but from looking at the HMM it looks like the CPU is soldered to the MB.

CPU is soldered and the whole thing is single board. I don't think it is CPU related: I did a lot of testing and in one case (after the first dissassembly and before the second one) the screen started displaying the normal POST messages, I tried to enter BIOS and while waiting ("Entering SETUP...") the machine turned off. After and before it behaved exactly as described in the first post. Hmm... this would also fit a BIOS chip with connection problems (if it is socketed).

Greetings,

Jan

[phool@round]

I would agree with al7kz that the board does sound terminal and would have said so but I hate to condemn the innocent until proven guilty by the overwhelming preponderance of evidence.

I too had a 240 with the same symptoms and it was proven to be the processor. I merely suggested something that can be overlooked while gathering a complete conclusion before spending money.

[JanR]

Hi,

I merely suggested something that can be overlooked while gathering a complete conclusion before spending money.

Right... and I will test this tomorrow.

I too had a 240 with the same symptoms and it was proven to be the processor.

How did you prove that? This would be very interesting for me and my problem with that notebook.

I'm pretty sure that swapping the mainboard will fix the issue because this is a single board machine that can be operated with just that board and a power supply (which is, according to the voltage, good). So, using another board ist more or less another computer with some of the components of the old one (that were already removed from the old board with no impact to the problem).

Are there any ways of diagnosis of such a TP 240 board? At work I have access to a logic analyzer and similar equipment.

Greetings,

Jan

[tom_k]

Hi Jan

I hope you get_it_running_agin, but if everything fails you can buy a mobo from me or another guy who works on 240`s,
unfortunately i offered him my whole stack of parts for the 240ies
2 days ago.
Got no reply up to now (thats usual;) but he is one of the "good guys" and i think i can offer you directly one of my boards if you are in need .

1st is a 240-cel300, lets say for 1337cent
(without postage, but if your location is correct that wouldn`t be so much)
The 240x 500 mobo should be worth some price negotiation by pm or mail ;)
if you prefer a celeron 400 board i could arrange contact with that other guy.

[JanR]

Hi,

you can buy a mobo

You got PN

Greetings,

Jan

[JanR]

Hi all,

By the way the BIOS chip is socketed very near the PCMCIA socket.

Just disassembled it again (its amazing how fast this is possible if you know what to do) and removed + reinserted the BIOS chip. Behavior is exactly the same.

As this was the last option I also made the crazy test of starting with no BIOS chip - suprisingly nothing happened at all. So something is still working... with no BIOS there is no boot code so this behavior is clear but it may be also an indication that there is still some live (especially CPU... because for a dead CPU it should make no difference if there is a BIOS chip or not).

Whatever... I'm afraid this board is really, really dead... so Tom_k is the next hope for that small machine (mail will follow in few minutes).

Thank you all for your help... and I will let you know if this 240 will get a new live with another board.

Edit: Just to make sure... I guess there is not a single 240 that died because of a WLAN card instead of the modem or because of an 3rd party battery pack, right? Because these are the only "non-regularities" with that machine (except falling from table three years ago) - and both worked well for several month.

Greetings,

Jan

[phool@round]

Just to make sure... I guess there is not a single 240 that died because of a WLAN card instead of the modem or because of an 3rd party battery pack, right?

Right.......

They just fail, usually it's the processor and not any other component.

[JanR]

Hi,

They just fail, usually it's the processor and not any other component.

Is it possible that very high load resulting in very high temperatures for a long time may a cause for this?

The last thing my 240 did before I was never able to return it to normal operation was a system upgrade for Gentoo linux - this means heavy compiling for 36 hours. ACPI gave CPU temperature values of 76 C which is still far below the 100 C max. core temperature but is nevertheless pretty high.

What do you think? I guess a notebook should survive 36 hours full CPU load (especially C++ code puts a very high load on the CPU with nearly no chance to "rest" by waiting for IO)... but on the other hand this is really something very hard.

Once I have the new mainboard I will avoid this kind of operation by system generation on another (and faster) machine.

Greetings,

Jan

[phool@round]

I think it's more to do with a weak cooling solution then thermal overloading. Very low fan speed in combination with a heatsink that mounts the fan to the side isn't an ideal combination, it's indirect cooling at best. The Gentoo install probably hastened the processors exit.

Thermal cycling is probably the other contributor, the expansion/contraction may have an effect on the processors solder connections.

In either case, it's not unusual for 240 system boards to quit, it's the most common fault.

Was your Gentoo install successful? I don't have the patience to wait that long although I know once Gentoo is installed its very efficient and fast. I went with Arch Linux with Fluxbox window manager. It was just as fast, being an exclusive i686 distro.

[JanR]

Hi,

I think it's more to do with a weak cooling solution then thermal overloading. Very low fan speed in combination with a heatsink that mounts the fan to the side isn't an ideal combination, it's indirect cooling at best. The Gentoo install probably hastened the processors exit.

The fan speed was not that low... and fan control was operational. Temperatures of 70C...75C are not that high for such a processor, but you are right, it stresses the soldering and internal connections.

The modern X60s also reaches 72C on long compile runs but due to its much higher performance they take not that long. A firefox is build in less than 30 minutes on the 1.66 GHz L2400, while the Celeron 400 needs 6 to 8 hours if I remember correctly.

Was your Gentoo install successful?

Actually, it was an update that rebuilt half the system, but it was not the first of this kind on that machine. All the other were successful... this one I was not able to test. Once I have the new board we will know...

The initial install I made in a vmware instance running on a much more powerful AMD X2 4400+. Tar'ing this and use it than instead of the stage tarball makes install on slow machines much more fun (and much faster).

I don't have the patience to wait that long although I know once Gentoo is installed its very efficient and fast. I went with Arch Linux with Fluxbox window manager. It was just as fast, being an exclusive i686 distro.

Better than many i586 distributions... that way you are at least in 1996, not in 1992 (or whenever P5 came out)

Especially for slow machines with small caches Gentoo is a good way to go - binaries compiled with -Os instead of -O2 run sometimes significantly faster if L2 cache is very small (like on Mendocino-Celerons).

If the TP240 runs with the new board I will make such installs by exporting the whole disk via NFS and chroot into that from another machine... that should reduce stress significantly (or make it with rsync and an vmware image).

Fluxbox is also great... but I prefer Windowmaker. Very fast, low overhead... and perfectly usable also on a slow Celeron and on small screens like the TP240.

Greetings,

Jan

[JanR]

Hi all,

to report good news:

It's alive again! Actually, I type this message on this Thinkpad 240 using Gentoo Linux and a Broadcom BCM4306 in the mini PCI slot.

Many, many thanks go to tom_k and his fast help. I got his 300 MHz board yesterday, swapped the boards today and now everything is perfect again (after a BIOS upgrade to make ACPI working). So... the next 6 years for this little machine can come!

300 MHz is slightly slower than 400 but the difference is not that much.

Greetings and again many thanks to all helpers!

Jan