Power problem with T21 (split from T20 not booting)

Message

db1 · #1 Post by **db1** » Tue Aug 05, 2008 6:51 am

No help I am afraid but I have a very similar problem with my T21. I am sure it is power related but don't know where to start. The PSU is putting out 16v off load but for some time now sometimes it thinks it is charging the battery and sometimes not. Checking the battery gives two 11V and two 12V. It misbehaves differently with the combinations of no bat. bat only a/c & bat. In that sometimes it won't even shut off - it just blinks all the leds on the RHS. Other than that it is the same as you with it randonly stopping. In my case I still have the panel lit but not even the mouse moves and I have to power off by holder the power key for ages. Sometimes this does not work and I have to pull the power & the battery. Tried with and without various bits such as the DVD plugged in and it makes no difference. Fan is running.

Moderator note: This thread was split out from T20 not booting... just after vlyne's post of Wed Aug 6 PDT.

db1 · #2 Post by **db1** » Wed Aug 06, 2008 6:52 am

Thanks for the pointers.

It boots most of the time but dies around half an hour. This is with the top bezel offf and nothing in the ultra bay. So I assume that if it is a power problem then it must be heat related else it would not boot in the first place??

I have cleaned the memory and tried booting without various bits and get beep codes so that looks good to me. I have tested the incoming voltage with the box powered up but no keyboard - so it was stuck on the password screen - and white to red on the onboard connector reads 16.4v. This is with or without the battery in.

Cleaned it and put fresh heat paste on the CPU but have not tried reseating that yet. (Cleaned it last night before I swa your post.)

When it dies on me it can take one of two forms. One there is still output to screen (LCD or external) but nothing moves or it just goes black. Either way I have to completely remove power to turn it off. The power buttons do nowt - I have it on a port replicator.

Thanks again. Will try some more of your hints later.

vlyne · #3 Post by **vlyne** » Wed Aug 06, 2008 7:21 am

db1: Your problem seems a bit different, and yes definitely heat-related but I'm not sure it a power issue as well. I'd give the memory a thorough work out - memtest loop, if you're already done the eraser cleaning trick. The other possible problem is a dodgy timing chip. The main one is I think under the harddisk cage near one of those sticky pads. Check the resistor which should be in the megaohms range (but test it on the pads on the motherboard) and check that the it is connected to the motherboard. It could still also be a CPU shutdown issue - I'm assuming you're running the right fan (i.e., either a T20/21 fan with the sponge, or a T22/23 fan with thermal paste/arctic silver). If not, that's the first thing to fix.
Cheers

db1 · #4 Post by **db1** » Wed Aug 06, 2008 8:34 am

Yeah, I was not convinced it was a BOD. PC Doctor runs clean on all tests. Will try things later. SWMBO says we have to go shopping

db1 · #5 Post by **db1** » Wed Aug 06, 2008 1:06 pm

Hmm. I am afraid checking resisters etc. is beyond me - and I can't find that fuse either

With it all back together, only two screws left over

- It boots fine and fails fairly quickly - but it will reboot at once, apart from having to remove poewer most times to get it to reboot. If it were a thermal trip would it reset that fast?

The failure is now consistent in that it always black screens. Whereas when it started to fail last Friday it was keeping the display up.

madkat · #6 Post by **madkat** » Wed Aug 06, 2008 2:39 pm

from my point of view - is a "cold solder point" somewhere - maybe even on the power supply circuit...
and your taking apart an putting back together "disturbed" it a little ...

i know my remark doesn't help much... but check with a magnifying glass the solders in the supply "zone"... that's what i'd do...

vlyne · #7 Post by **vlyne** » Wed Aug 06, 2008 6:45 pm

We're hijacking this thread from the OP but yes, a cold joint is a distinct possibility, and so is an overheating CPU, or some other heat sensitive power regulator. As for the missing screws, just make sure it's not from the fan or from the mainboard to backend panel connection. Are you absolutely sure the fan is doing its job and that you've got the right heat transfer medium between the fan and CPU? If the fan doesn't have the "springy feet" you must use the sponge, otherwise use the arctic silver paste. And, you've tried it with and without devices (including battery) attached? Memory is the other possible issue. If all that checks out and you're not sure how to test resistors etc, it's time for another motherboard I think.
Cheers

#8 Post by **rkawakami** » Wed Aug 06, 2008 7:11 pm

Try downloading and installing MobileMeter and run a couple of PC Doctor tests (if they have not been run before). Details are in this post:

http://forum.thinkpads.com/viewtopic.ph ... 568#207568

db1 · #9 Post by **db1** » Thu Aug 07, 2008 2:26 am

Yes, with and without devices. CPU has the blue patch and the "springy feet" .

I can't run the suggested s/w as I don't run windows. on it. PC Doctor runs clean but I will double check the points in the referenced post. Thank you.

Now this is getting stranger. This box is running OS/2 ECS - Now T21's, OS/2 and 3Com 3CN3AC1556B mini PCI cards don't get along. They run at about one byte a week. That was the card it came with when I bought it new and I swapped it it for an Intel Pro. About a year ago something on the modem side of that card failed in such a way as to stop the boot process in its tracks. A friend kindly supplied another Intel Pro and off we went again until now.

With the Intel card in place and a ten second ping loop running, when it hangs the link light goes off on the switch and it stays up maybe 20 minutes. With the 3Com card, but no ping test, it still hangs but the link light stays on and it *appears* to stay up longer. Last night it was over three hours. I am just trying again with no card at all. Then I will rerun the PC Doctor tests as suggested.

Sorry about hijacking the thread. It looked so similar to begin with and I think to start a new one now would lose a lot of information/suggestions.

madkat · #10 Post by **madkat** » Thu Aug 07, 2008 2:42 am

how does it work without any of the two cards?

db1 · #11 Post by **db1** » Thu Aug 07, 2008 9:35 am

So far 7.5 hours and still going.

Of course the questions now are?
Is it the supply to the mini PCI bay?
Are those two cards bad in some way and what would happen if they were put in another Tpad?
Are the PCMCIA slots on the same power bus? ie I might get away with using a WIFI card in there. Will try later. Hate to stop it right now

madkat · #12 Post by **madkat** » Thu Aug 07, 2008 9:40 am

so - without the minipci is working fine?
try to borrow a pcmcia card - and see what happens - my guess is that it will work ok

db1 · #13 Post by **db1** » Thu Aug 07, 2008 10:44 am

Yes without a mini PCI card it has stayed up for over eight hours. Got bored and stopped it. I found that I had an old version of PC Doctor so got the latest, thanks Ray, and the thermal and fan tests run clean. Currently running a full memory test .

I have an Artem WIFI card and that is the next test.

db1 · #14 Post by **db1** » Fri Aug 08, 2008 3:23 am

Thanks for splitting the thread mods - Did not know you could do that.

Working fine with no mini PCI cards and a WIFI card in the PCMCIA slot. - Apart from a couple of s/w problems that don't concern us here.

So, do I have dodgy mini PCI bay connections?
Do I have three duff NICs?
Is there an ethernet PCMCIA card around?

madkat · #15 Post by **madkat** » Fri Aug 08, 2008 4:15 am

yes, there are ethernet pcmcia cards

anyways - i think it's ok it's only a minipci problem...

db1 · #16 Post by **db1** » Wed Aug 13, 2008 12:55 am

Yes there are PCMCIA ethernet cards - But ones that work with OS/2 are extremely thin on the ground - Even on eBay.

However it seems my conclusions were mistaken. Yesterday I needed to open an Excel file. The only way for me to do that is with a bootable Linux CD which has OpenOffice on it.

Now Linux is *very* chatty when it boots which makes a bootable CD a very useful diagnostic. Anyway, one of the things it tells you is the processor speed - which was very low 167 MHz. I plugged in the AC charger and rebooted. This time it said a higher figure as expected, 670 MHz but not the full speed of the CPU 850MHz. Then I recalled that in my messing about one of the things I tried was disabling the Intel Speedstep. So I turned it from disabled back to automatic. Within minutes the box had crashed. I tried again with a setting of always max but that too crashed.

I then tried again with disabled but with the offending network card, as I thought, installed. No crash. Put back the card drivers and it is still going. So it looks like the speedstep.

#17 Post by **Harryc** » Wed Aug 13, 2008 6:26 am

If it's speedstep, it's rare but you might have a bad CPU there. Could even be bad L1 or L2 cache on the CPU itself.

db1 · #18 Post by **db1** » Wed Aug 13, 2008 7:19 am

Did not realise it was internal to the CPU. Assuming it *is* a failing CPU - what is the fastest CPU I can put on a T21 mobo?

sarbin · #19 Post by **sarbin** » Wed Aug 13, 2008 7:27 am

just a random thought...
since your t21 crashes when you tweak the speedstep control in bios, what bios version are you running?

http://www-307.ibm.com/pc/support/site. ... IGR-4JSQHB

db1 · #20 Post by **db1** » Wed Aug 13, 2008 9:21 am

Good point, but I have the latest available.

Edit. I have found where the three screws should have gone

sarbin · #21 Post by **sarbin** » Wed Aug 13, 2008 9:51 am

oh, well... check that one off.

Power problem with T21 (split from T20 not booting)

Power problem with T21 (split from T20 not booting)

Who is online