Thinkpads Forum

benottomex wrote:I guess the responsible of the GPU temp should be the thermal compound used. The only way to know is to open the machine, not for the heart-fainted people, replacing the GPU compound is worth it.

I think the T6x design compared to the T4x series is similar, but Lenovo (I think) made a change in the heatpipes: The T4x GPU's heat was transferred with one heatpipe that was pass very close to the 2 heatpipes that transfer the CPU Heat.
T6x GPU heatpipe did not intersect very close to the 2 heatpipes of the CPU.

In other words: The heat generated on a T6x GPU can't be handled thru one single heatpipe, this is because the CPU heat is lower than in T4x series. While in T4x series, the GPU heat is even invading the 2 CPU heatpipes, resulting in having a hotter CPU (i.e. the two components generated heat is shared, not the same in T6x)

My theory

Thanks for your explaination, your theory is really helpful. So, the fan's problem may be concerned with the position of heatpipes. But, why the temperatures are so different from using X1400 and X1300?

khaverblad wrote:I got an email from Lenovo Sweden that they have open and forwarded my issue to the lab, so let's see what they figure out. But, I highly recomend that more people would do the say and really complaint loud and clear.

I've sent a dozen e-mails to Lenovo Switzerland about complaining the noise issue so far... They also suggested me that they could escalate the case and I would have to send my machine to the Lab ind the UK for testing... In fact, I am not willing to do that because I dont trust them in deed... I posted an e-mail about the GPU's temperature issue to them recently, hope they will give me a comfortable answer next Monday... And, looking forward to hearing from your report!

Well, Benottomex you where right on to it; I had IBM servce people out today to switch out the "fan" or I would rather say the entire heatpipe and fan component for the CPU och GPU. What we both discovered was that the thermal compound used for the GPU first of all looked like a thick layer of tack that you could easy remove with your fingers. The thicknes was around 2mm and in the area where the GPU was pressed against it around 0.5 - 1mm in thicknes - so way too much !!!

For the moment we just replaced with a new fan and heatpipe component without altering anything regarding the thermal compounds. But, I will send an e-mail up to Lenovo Stockholm about this and as well, the technician will report his suggestions (same as mine) to his office.

I did take some pictures as well, but will post them here as well, later on tonight.

The important thing here is if the thermal goop on the heatsink that contact the GPU did seat properly (no gaps, etc..). Did it have a pink film?

Would be great to see some pics. What are the new temp readings for your machine?

Well, I've posted the images at OS2 World (due to max 50kb here), so please check this thread.

The first image here shows that the two pieces of thermal compounds (squared) used for the ATI X1400. You might notice that the layer is really thick - around 2mm!

Second image is a closer view where the left piece of thermal compound was more like tack that you could easy remove with the fingers?!

Well and about the current values, they actually got worse

CPU 42°C (0x78)
APS 33°C (0x79)
PCM 37°C (0x7a)
GPU 71°C (0x7b)
BAT 50°C (0x7c)
BAT 27°C (0x7e)
BUS 37°C (0xc0)
PCI 43°C (0xc1)
PWR 44°C (0xc2)

Well, actually CPU went down a bit (from 46C to 42C) and the GPU went up to 70-71C.

The pink goop is an utter menice. I remember having an ES of one of the mid end ATI cards here, and it had this horrid pink stuff that would not come off on the GPU (and nothing on the ramdacs).

I'm used to lovely substances such as acetone, isoprphyl and benzine for removal, but this had to literally be pryed off. The cooling capabilities weren't all that terrific either, and that card currently holds 3dmark01, 06 WRs.

Could you ask the technicians if reapplication of the thermal paste must be done with a technician present, or whether it could be done as a user repair?

Thanks;)

[Edit] Well, I find some people here complaining about a gpu temp of around 60 degrees - my t60 never leaves the low 70s, and hits mid 80s while running half life 2. However, to me, despite having an obnoxious interest in cooling on a desktop, this is of little importance. My T60, even with a fan running all the time, is quiet, and runs very cool. I don't notice any heat disassipation through the chasis, which is the critical point.

I would rather have the fan on ~3000rpm all the time (which is quiet) than a hot laptop without the fan running. It could be that lenovo uses various HSFs with diffrent fans, some of which are noisier. Mine seems to be fine?

[edit2] the reason why your temps may be higher is, that the HSF was designed for a thermal pad rather than thermal paste, and hence there is bad contact between the two. If you can, try tightening the grip the HSF has on the motherboard, just dont flex it too much

Well, the high temperature as I see it can partly be explained by the thermal goop used and when feeling on it it's not sticky or so and feels more like tack (the Swedish word is "häftmassa" and Danish I think it's "trolsnott" or something like that. Easy to remove and easy to put on again (as you might see on the pictures) and for me what I've learned when building my own system the thermal goop layer is way to thick.

Well, let's add at least one picutre here so we can see the layer:

img]http://www.os2world.com/pictures/2006/t ... 0_pic3.jpg[/img]

For the rest of the pictures, check this thread at OS2 World.

Interesting images. I'm sure that thermal goop is the responsible for the high temperature. Nothing beats having a cooper heatsink making 100% contact with the generating surface (CPU/GPU)

I noticed in this image:
img]http://www.os2world.com/pictures/2006/t ... 0_pic2.jpg[/img]
There's 2 thermal pads, my guess is that the left pad contacts the northbridge chip ??

Why both of the thermal pads have the shape of a square?? Isn't the GPU chip suppose to be even and plain.

Maybe you could post a pic showing where each heatsink thermal pads make contact with the chips on the machine.

Interesting images, again... I would like to make some experiments to improve heat transfer in the T60 series

khaverblad wrote:
For the moment we just replaced with a new fan and heatpipe component without altering anything regarding the thermal compounds. But, I will send an e-mail up to Lenovo Stockholm about this and as well, the technician will report his suggestions (same as mine) to his office.

My understanding is that Thermal Pads, if that's what you indeed have on, are "one shot" only. If you have removed it, you need to slap on a new pad else it will be even worse than before (if that's possible!).

With that sort of thickness, it's no wonder that your chips are getting so hot.

FYI, Arctic Silver and other thermal greases have thermal conductivity around 4W/mK. Recommended thickness is around 0.001", which is like 0.02mm. VERY thin. The conductivities for Thermal pads are less (at least for the ones I looked at, including recommendations from AMD).

If you go through the math, for a 25W CPU and size like the Core Duo, with the proper application of thermal grease, you should get around 40C with ambient of 20C (that's actually close to what I get on my laptop. Whew!). Do the calculation for 2mm thick pad and the numbers go through the roof, so it's no surprise your CPU/GPU is so hot (or rather it's a surprise it's still so cool...)

Those of you not concerned about the fan noise, you should still do your best to fix the issue. Many IC's out there go through reliability testing, usually active at 85C ambient for 1000hrs as this simulates an much longer lifetime in the field, since you usually don't run this hot. If your CPU is already running 70-85C, you will be degrading the life of that thing.

To determine whether the problem is with the thermal pad or with the machine design, is simply detaching the keyboard (lossening) having access to the heatsink and then power-on the machine.

then with ATitool force the GPU to Detect artifacts, when the thing starts to get hot touch the heatpipe that is suppose to be transferring the heat. Is it too [censored] hot, or just midly hot.

I say this, because it should be very hot, meaning the thermal pad is doing its job.

Again, I would like to see the view from above with the keyboard and heatsink detached, a picture like this is what I mean:
http://img81.imageshack.us/img81/3989/d ... gu8.th.jpg
(My T41p exposed)

Well, I won't ripp out the thermal component (not now at least) until Lenovo has a proper solution for this. But, when I have the next opportunity of taking some photos I will.

About the northbridge chip, I didn't pay that much attention to what goes where, but guess you're right since the northbridge usually are placed close to the CPU.

And as well, I will try to stress the system when I have the moment for it.

I don't think Lenovo would remove those termal pads in the next T61 releases. Not entirely sure if those thermal pads are inadequate in the heat transfer, but for sure that is not common to have on a high caliber machine.

If you could take some pictures without the fan (where the CPU, GPU are). Maybe there could be a way to make the copper heatsink contact the GPU by bending it a little (not too much)

Regarding the copper heatsink that covers the GPU and northbridge chip there is actually a locking mechanism to make extra preasure to the two mentioned chips.

To sum up, the change of the fan/heatsink has actually managed to increase the average temperature of the GPU:

CPU 44°C (0x78)
APS 43°C (0x79)
PCM 35°C (0x7a)
GPU 73°C (0x7b)
BAT 50°C (0x7c)
BAT 29°C (0x7e)
BUS 38°C (0xc0)
PCI 43°C (0xc1)
PWR 42°C (0xc2)

70-73C seems to be the more natural temperature now...

khaverblad wrote:Regarding the copper heatsink that covers the GPU and northbridge chip there is actually a locking mechanism to make extra preasure to the two mentioned chips.

Oh, didn't know Lenovo implemented a system to avoid extra pressure. For sure there should be a way to mod and make the copper make 100% contact with the GPU.

My T41p readings with the fan on controlled by the BIOS (2900rpm) are:
CPU 44°C (0x78) <-- Max Performance, 1.196V
APS 44°C (0x79)
PCM 32°C (0x7a)
GPU 46°C (0x7b) <-- Undervolted to 70MHz, Zalman STG cooper to GPU contact
BAT 32°C (0x7c)
BAT 29°C (0x7e)

GPU temp of 70°C is not normal to me. Keep us informed what you come up with.

Just a quick follow-up on this thread and to say that I havenät come around to do anything further about this issue. But, I will check in with the Leonovo in Stockholm next week to hear what they have to say.

khaverblad wrote:Just a quick follow-up on this thread and to say that I havenät come around to do anything further about this issue. But, I will check in with the Leonovo in Stockholm next week to hear what they have to say.

Any updates?

where do I get the program to check the temperatures of my T60P?

dickeywang wrote:

khaverblad wrote:Just a quick follow-up on this thread and to say that I havenät come around to do anything further about this issue. But, I will check in with the Leonovo in Stockholm next week to hear what they have to say.

Any updates?

Nope nothing haven't heard anything, but this reminder from you would give me a kick in the butt to follow up on this. I'll check it and get back.

Just turned it on a few minutes ago. Got wireless running and two internet explorer open, attached to the power supply and screen brightness on max: Fans wasnt running at the time.

CPU 58°C (0x78)
APS 35°C (0x79)
PCM 33°C (0x7a)
GPU 65°C (0x7b)
BAT 50°C (0x7c)
BAT 23°C (0x7e)
BUS 34°C (0xc0)
PCI 43°C (0xc1)
PWR 45°C (0xc2)

T60P 2613HP C2D T7400 2.16 GHZ - 14.1" TFT , 2GB ram, 60GB 5400, ATI M FIREGL V5250

One thing I have noticed is that,
once the fan in bios setting hits rpm>3000
it wont back down to say 2800rpm where it initialy was.

does anyone have any explanation or idea about this, because
logically, once the temperature have come down the fan should
return to its low rpm.

CPU 46°C (0x78)
APS 35°C (0x79)
PCM 34°C (0x7a)
GPU 60°C (0x7b)
BAT 28°C (0x7c)
BAT 26°C (0x7e)
BUS 37°C (0xc0)
PCI 43°C (0xc1)
PWR 41°C (0xc2)

the above is my temperature with screen at full brightness.

ennma

anyone tried taking the battery out whilst working on desktop
really helps in my opinion and case, just dont forget to put it back in
when youare making a move.

CPU 43°C (0x78)
APS 29°C (0x79)
PCM 33°C (0x7a)
GPU 57°C (0x7b)
BUS 34°C (0xc0)
PCI 38°C (0xc1)
PWR 40°C (0xc2)

with no battery just surfing the net and posting.
the way temps are supposed to be IMO.

ennma.