Toshiba M45 S359 shut downs, freezes, heat & cpu issues

Attempting to recommission an M45 that was given up on by owner. (owner removed HD).

INITIAL SYMPTOMS:

M45 would shut down (go black) after maybe 30 seconds of on time.
The fan was working on boot.

CLEANED HEAT EXCHANGER:

Removed massive dust bunnys and lint accumulated on inlet to heat exchanger. Flow must have been only a fraction of what it should have been.

Thermal grease was all over the CPU and gray in color. There was so much extra, I simply smoothed it around and reassembled the cooling system.

PARTIAL SUCCESS:

Now M45 will now run for minutes before it now freezes. I tried a number of diagnostics and eventually, it freezes after so many minutes.

WHAT TO DO NEXT and SOME THEORIES:

1. I'm confident freezing is due to cpu heat.
2. Cpu heat failure could be due to:
   • Failing heat pipe heat exchanger,
   • Poor grey thermal grease (this is a stretch)
   • Fan output is low (I doubt this)
   • CPU was damaged by repeated over heating.

EXPERIMENT 1: Lower CPu temp buy lowering ambient temp or increasing flow through heat exchanger.

If this produces stability improvements, then the over heating theory is sound. However, it doesn't determine if it is the heat exchanger or the cpu.

QUESTION 1: Do heat pipe heat exchangers ever go bad?

REPLACE CPU? I read about a poor fellow who had a lower model M45 that replaced everything including mobo, keyboard, display, etc. etc. and finally, after replacing CPU, his problems were solved.

QUESTION 2: Do CPU's which have been marginally damaged by OCing (if there is such a thing) require a lower temp for stability? Obviously, my cleaning of the heat exchanger resulted in a lower CPU temp which dramatically improved stability but did not result in complete stability.

Any suggestions will be greatly appreciated and of course,
Thanks in advance!
Tom

PS. The conduction surface of the heat sink is terrible. It has a poor as cast surface finish but more importantly, it is nowhere near flat. In fact, right in the center, there is a concave area so that the resulting contact area is an irregular ring around the perimeter of the chip. Compared to a flat, polished since surface (2000 grit sandpaper) I'll bet this sink runs 3-5 deg C higher. The concave center defect is so deep, I would be concerned that the amount of material removal to make that sink surface flat.

_________________________________________________

STATUS UPDATE: 7-10-09

Ordered cheap CPU from eBay.

This is how I want to proceed, quantifying stability at each step:

1. Measure temp of current CPU or heat sink (software or pyrometer)
2. Flatten and polish heat sink contact surface
3. Measure temp
4. Install new CPU
5. Measure temp

 

Answer to question 1: As long as the pipe heat exchangers are not bent, or corroded at the surface where the heat exchange takes place (copper is unreactive so this is highly unlikely) they wont go bad. They must also not touch the motherboard to prevent damage from the heat they tranfer (although this is obviously not the case). I would try and fix the conduction problems you mentioned, since this is the most probable cause of the overheating. Someone must have been really "playing" with his laptop

Answer to question 2: The only thing I can think of is the CPU getting overvolted. This can cause temperatures trouble big time, and it will be a really tough to get this if it was done by modifying the BIOS, since you will have to manage to undervolt it within the time limit you are provided before the laptop shuts of due to overheating. If no manual overvolting was done, I don't know if damage on the CPU causes overvoltage. Did the owner mess with the CPU?

How much time will the laptop now run before shutting down?

 

The laptop belonged to a woman. So it was never messed with. The person who pulled the hard drive left the screws in a mini zip log bag where the HD was(!) That was most courteous!

With the air condenser plugged up with lint and dust bunnies, the laptop would shut down in about 10 seconds.

After clean the condenser, no more shutdowns, however, anything that loads the cpu causes a freeze in 5 to 20 minutes.

As far as damaging the cpu, if run w/o cooling, it would burn up in seconds. So if the heat exchanger was serious compromised by dust and then also a factory defective conduction surface, perhaps the cpu could have been marginally damaged. (I have seen abused cpu's work OK under light load, but load them up and they freeze - and this would be in a stock setup where a healthy cpu works without incident.)

Anythoughts on measuring temps while stressing cpu? Something Linux based maybe? I have no HD so it has to run off of memory. Could I run an OS off of a USB memory stick?

Thanks
Tom

 

Hey, some of us women build our own PCs and even laptops on barebones... stop with the sexist assumptions!

Geez.

And, if you can figure out how... probably being a typical male who doesn't read instructions (ha!) clean off your heatsinks and apply arctic silver.

 

Well, don't have Windows any more (why don't you just use a disc to boot windows?)? I was going to say to try Everest (Stresses CPU and provides info like temps)...

 

Sorry Krista.
___________

Also, there is no hard drive so I can only use what gets loaded into memory. The Hiren Boost Disc has Mini Windows XP which has a directory of windows programs, one of which is a siw.exe (Topala System Information) which has a Sensor page that seems to capture a temp that is probably the cpu. (TZCR)

I tried to install SpeedFan into MiniXP but it hangs when it tries to populate its window.

Prime95 will install and run.

siw.exe shows idle temp of 50 deg C and 68 deg C running Prime95

Intel Spec Finder for SL7SM says 100 deg C which seems really high.

I wonder what temp was causing shutdown - I checked the BIOS and there is no such parameters unfortunately.

QUESTION: At 68 deg max (assuming that is accurate) should I not bother trying to resurface the heat sink cpu contact surface?

I think for now, I'm going to concentrate on some memory testing and further quantification of system freezing/hanging.

RESULTS:

• Memtest86+ V2.11: Passed (one pass)
• GoldMemory 6.92: Fails 122/369 for all combinations of memory/slots

Thanks
Tom

 

STATUS UPDATE 7/15:

Installed replacement used CPU and still fails GoldMemory at Test 122

(Currently running GM and have over 400k failures for Test 122 - it looks like every iteration is a failure.)

I will see if I can get past test 122, however, I hope it doesn't do 1MM iterations...

So, I'm basically done.

I don't know what to do other than give up and put on eBay for parts.
______________

actually it did over 1.5MM iterations, so I shut GM down.

Too bad, right now only problem is M45 fails GM.

 

If it's failing the memory test, then most likely the RAM is bad. Since the cooling system was compromised, I bet that the high temperatures inside the laptop helped dramatically shorten the lifespan of the RAM. Do you have any other RAM to try in it?

-J.B.

 

J.B.

Thanks for the reply!

Its failing "not just a memory test" since it is Gold Memory. GM wrings out the entire memory system including cpu, chipset, bus, memory, etc. so you are left with big questions if you failed GM. memtest86+ I did a single pass w/o errors. GM on test 122 of 369 I think I fail every pass of that test since I left it run until it counted over 1.5 million errors. So something was completely not working for that particular test.

I did try other memory and another CPU and had the same failure.

I've read numerous references to the MAXIM chip in the M35 Toshiba coming loose probably because of brittle no-lead solder. I don't know what this chip looks like, however, it may be a blind one (contacts on back face of chip).

I have heard from an expert laptop repair guy that there are designs where they have a big blind chip that comes loose when the motherboard is flexed when you pick up a laptop in one hand from say a corner or side. This particular failure in combination with the new brittle solder sounds like it is going to be popular like Bad Caps.

What is interesting, is that as I have been fiddling around with this laptop, it seems like it has quit freezing after many minutes of use although it still 100% fails GM test 122.

A real test would be to apply pressure on suspect chips when running GM and seeing if anything changes.

I still have to believe there is a plethora of M45 failures since mine sounds exactly like the M35's which were notorious and had the class action lawsuit.

Thanks again,
Tom

PS. I wonder if there is a website specializing in actual motherboard repairing?

 

UPDATE:

Only problem now seems this GM 6.92 test # 122.

Has run 1.5 days GM 5.07 without any errors.

Its been doing the SIW CPU and MEMORY stress test w/o incident.

I'm going to load up Prime 95 and see what happens.

I wonder if the latter GM has a problem with no Hard Drive present???

_______________________

Further late nite update:

SIW ran cpu 63-65 deg C (ran about 3 hours)
Prime95 Torture Test a mere 61 deg (ran about 4 hours and running over night)