Thinkpads Forum

Has anyone tried replacing the heatsink in a T60p with one from a T61? I've seen posts suggesting it might be possible, but haven't been able to find out if anyone actually tried it.

Do you mean the integrated or discrete version?

I have one of each in front of me (discrete versions) and I would say yes.

The mounting holes and outside dimensions are the same.

The little paddles on the GPU/chipset arm look a little lower on the T61 version - a little more pressure: probably not a bad thing!

If you are using a new replacement, don't forget to remove the clear plastic covers off the gray thermal pads...

I did

Yours,
Peter

I meant the discrete version. I'm trying to get better cooling for the GPU in my T60p. What's the FRU number of the T61 heatsink you have? 42X4685? Thanks.

42X4685 is correct.

Yours,
Peter

Does it actually cool better? I can tell the fan on my T60 is on its way out, so I'm interested to hear about better alternatives.

It probably would. Look at this thread for some discussion of the possibility:
http://forum.thinkpads.com/viewtopic.ph ... 31#p323331
However, we won't know if it works until someone tries it. I probably won't attempt it for a while because I don't want to void my warranty yet.

Also, the T61 fan should be quieter: http://lenovoblogs.com/insidethebox/?p=81

If someone wants to donate a T61 heatsink to me I'm happy to try it.

Me too. I will gladly try a T61 heatsink on my noisy T60p.

I am new owner of a T60p and am glad to know I am not the only one concerned about the heat, fan noise and why it runs and can be heard quite well in most environment.

Velani

PhilMichael wrote:Has anyone tried replacing the heatsink in a T60p with one from a T61? I've seen posts suggesting it might be possible, but haven't been able to find out if anyone actually tried it.

I actually have done this and am running a T61p heatsink in my T60p right now.

It's not a "drop in" replacement though - I had to make several modifications to it - most notably to the GPU cooling end of the heatsink - both to accommodate the way a T60p creates pressure against the chips it must cool and also the height differences - the T61p and the T60p did not use the same GPU and the heights are different between them.

In retrospect I don't know if all the effort I made was worth it. The heatsink does a good job cooling the CPU - perhaps better than any T60p heatsink I've tried - but I don't think it's any better than the T60p heatsinks at cooling the GPU and that's the big heat problem - for me anyway.

JR, thanks for taking the time to post a report on the actual job done.

Three questions....

1- Which FRU did you use?....the 42X4685 mentioned above, or if not, exactly which number?

2- You said you had to modify it because the T61 gpu was different height than T60p, but T60p HS wasn't formed down over GPU anyway...they just used a thick ugly pad...

But it's been posted before that T61 HS -is- already formed downwards to make direct contact with GPU.

So please describe better what you meant. It sounds like it did not contact the gpu properly already. Did you just form it downwards more? Or are you saying you had to form it -upwards- ?

3- Does the T61 sink form down further at the very end to make more direct contact with intel Bridge chip than t60 sink did? Or does it use the same thick pad?

PS; If you have the time for precise details, I'd be interested to know what you had to do in regards to the 'hold down' item over the GPU and Bridge chip.

thanks again JR

dozer wrote:1- Which FRU did you use?....the 42X4685 mentioned above, or if not, exactly which number?

I believe that it was the model for the discrete graphics processor - I can't tell now for certain because I didn't write the number down when I had it out and now that's it's in I'd have to take it apart to find out. That number seems familiar though - I bought it by mistake when a fan failed on my T60p and with it sitting around I wanted to try it.

2- You said you had to modify it because the T61 gpu was different height than T60p, but T60p HS wasn't formed down over GPU anyway...they just used a thick ugly pad...

The T60p heatsinks actually have the plate which cools the GPU and the Northbridge at the same level. This plate in the 41V9932 looks to be bent down for the Northbridge but that's deceiving - it's actually the result of an extra plate soldered on the bottom that necessitated the bend but if you use a straightedge across that plate, they are both dead even. The 41W6407 version is dead flat - not bent down at all. Both of these heatsinks rely on a thicker pad to make contact with the Northbridge - the pad for the GPU looks to be about .5mm thick and the one for the Northbridge looks to be about 2mm thick. It's only this difference in pad thickness that allows any contact with the Northbridge. It's a horrible design.

But it's been posted before that T61 HS -is- already formed downwards to make direct contact with GPU.

The plate on the T61p heatsink I modified is formed downwards to contact the GPU but this only makes the situation for cooling the Northbridge far worse. The Northbridge chip is actually .050" lower than the GPU so bending the plate down where the GPU is requires an even thicker pad if you're going to provide any cooling to the Northbridge.

3- Does the T61 sink form down further at the very end to make more direct contact with Intel Bridge chip than t60 sink did? Or does it use the same thick pad?

See above but it uses a pad for both the GPU and the Northbridge. As I don't have to T61p to check I don't know exactly how different the height is between the GPU and the Northbridge on a T61p - the bend may be perfect for it but it's bent the wrong way for a T60p.

PS; If you have the time for precise details, I'd be interested to know what you had to do in regards to the 'hold down' item over the GPU and Bridge chip.

What I actually did was to unsolder the GPU / bridge plate from the T61p heatsink as while it was an interesting design and I liked the notion that it contacts the keyboard to get additional heat transfer, that didn't work with the T60p's spring that applies downforce on the plate. (This was pretty easy to do - they use very low temperature solder to assemble these). I also had a plate from a 41W6407 (the original one with a bad fan bearing) which I removed from the heat pipe (also easy to do as it was a horrible solder job they had done - VERY poor heat transfer). I used Arctic Silver Epoxy to bond a .050" copper plate on the bottom of that - so that it sat level but contacted both the GPU and the Northbridge. I bent the heatpipe on the T61p unit so that it slightly added to the downforce over the GPU / Northbridge end but more to make sure it was slightly raised away from the heatsink over the CPU, and used Arctic Silver Epoxy to bond the heatpipe to the top of the plate.

After the epoxy cured, I was able to use a small bit of AS5 (with no thick gummy pad at all) on both the GPU and the Northbridge because the heights matched, and I assembled the unit this way - AS5 between on the CPU, the GPU, and the Northbridge.

I'd like to say that I think this is the ultimate heatsink for a T60p but I'm not certain that it is. They have really blundered about with heatsinks and there are 3 distinctly different variations - some better than others. In my view the 41W6407 - designed for C2D CPUs and discrete GPU on the T60p is a horrible design. They set about improving the CPU cooling for the C2D which I suppose they did - at the expense of GPU cooling - which is the bigger problem on the notebook. Extending the CPU heatpipe over both "radiators" of the fan housing leaves no proper cold end for the GPU heatpipe and to an extent the T61p heatsink suffers a similar problem. I think the T61p heatsink is a good design for a T61p but the GPU on a T61p doesn't have the heat problems of the FireGL 5250 on the T60p.

The design of the 41V9932 heatsink actually is the best from the standpoint of getting a properly cooled "cold end" for the GPU heatpipe, and the extra copper plates on the 41W6407 (two over the CPU and 1 over the GPU) are good design elements from that FRU. I actually have two 41V9932 heatsinks and I think I'm going to build something up from them - adding the copper plates I took off a failed 41W6407 to beef up the CPU cooling (I have a C2D CPU in my T60p) and adding a copper plate over the GPU - and probably also adding a .050" plate on the bottom to make proper contact with the Northbridge while still making proper contact with the GPU and avoiding the need for thick thermal pads.

I would add that my real objective isn't noise reduction - it's getting adequate cooling for the CPU and the best possible cooling for the GPU which is the real problem with T60p models.

This is an interesting topic. I am looking for a better cooling & fan noise reduction solution for my T60p's V5250 too. And T61/p's heatsink seems to be a good alternative to the original heatsink shipped with T60/p.

So basically the heatsink for T61/p with discrete graphics is the direction and 42X4685 has been confirmed working with T60p.

According to http://www-307.ibm.com/pc/support/site. ... 68229.html, 42X4685 is for 15.4inch T61/p with discrete graphics.

Another two possible T61/p heatsink options are 42W2462 and 44C0557 for 14.1inch T61/p with discrete graphics according to http://www-307.ibm.com/pc/support/site. ... 67911.html. Can anybody confirm if they are working with T60p? or if 42W2462 and 44C0557 are the same as or similar to 42X4685?

I found two pictures of 42W2462 on the Internet: {picture1} and {picture2}.

JR956678,

You bring up some very interesting points regarding heatsink design. In particular, the following:

JR956678 wrote:In my view the 41W6407 - designed for C2D CPUs and discrete GPU on the T60p is a horrible design. They set about improving the CPU cooling for the C2D which I suppose they did - at the expense of GPU cooling - which is the bigger problem on the notebook. Extending the CPU heatpipe over both "radiators" of the fan housing leaves no proper cold end for the GPU heatpipe and to an extent the T61p heatsink suffers a similar problem. I think the T61p heatsink is a good design for a T61p but the GPU on a T61p doesn't have the heat problems of the FireGL 5250 on the T60p.

The design of the 41V9932 heatsink actually is the best from the standpoint of getting a properly cooled "cold end" for the GPU heatpipe, and the extra copper plates on the 41W6407 (two over the CPU and 1 over the GPU) are good design elements from that FRU.

It makes sense from a logical point of view, although I don't know how much difference it would make in real life. It would be interesting to obtain some results comparing the cooling power of 41V9932 (Original T60 Core Duo / ATI heatsink) with 41W6407 (Redesigned T60 Core 2 Duo / ATI heatsink), with all things being equal. Or compare the original heatsinks to the modded version you plan to build.

crazyfrog wrote:This is an interesting topic. I am looking for a better cooling & fan noise reduction solution for my T60p's V5250 too. And T61/p's heatsink seems to be a good alternative to the original heatsink shipped with T60/p.

That was what I thought when I started out with the project. Now - I'm not so certain that it is nor that it isn't.

The big issue I see with cooling on the T60p is cooling the GPU - my CPU stays quite cool even under the severest stress I can generate for it. Now one of the differences with the T61p is they changed out the ATI FireGL 5250 for an Nvidia GPU which generates a lot less heat in the first place. So perhaps they didn't feel as pressured to get good GPU cooling from the T61p heatsink because they reduced the need for better cooling.

An issue I've not seen anyone bring up about these heatsinks is the heat pipe and just how it's "tuned". I'm a mechanical engineer but for the non-mechanical engineers amongst us - the heat pipes are hollow tubes filled with a pressurized liquid - likely some Freon variant. The pressure inside and the liquid choice sets a boiling point somewhere in the temperature range they want to control. Once the hot end of the pipe reaches the boiling point, the liquid boils and turns into a vapor, which fills the heat pipe until it finds a section of the heat pipe cold enough for it to condense - the cold end. The boiling liquid extracts the latent heat of vaporization from the hot end / GPU and releases that heat at the cold end when the vapor condenses and that heat is carried off by the air flow from the cooling fan. This is the same physics that makes you feel cold when you first get out of a swimming pool even though it may be a warm day.

By watching the rate of temperature increase when I stress the GPU (ATI Tools spinning the fuzzy 3-D cube at max frame rate) and then also watching the rate of cooling when I remove the stress, I believe the T61p heatpipe on my notebook begins to boil at an indicated 72-75 degrees C because temperature raises very quickly to that temp and then much slower above that temp. Similarly, it drops very quickly to around 75 and then much more slowly to idle at aroung 64-65. The big problem I see is that the temperature just keeps climbing unless I turn off ATI Tools - which could mean that there's just too much heat for one heat pipe to remove, or perhaps the heat pipe isn't being cooled well enough at the cold end.

It makes sense from a logical point of view, although I don't know how much difference it would make in real life. It would be interesting to obtain some results comparing the cooling power of 41V9932 (Original T60 Core Duo / ATI heatsink) with 41W6407 (Redesigned T60 Core 2 Duo / ATI heatsink), with all things being equal. Or compare the original heatsinks to the modded version you plan to build.

I am probably going to go down this path - at least partly. I don't have a working 41W6407 and at this point I don't think I want to go out and buy one because there's too much I don't like about it even though it's the "correct" one for my notebook. I do have two 41V9932 heatsinks - one of them brand new - and I plan to mod one of them slightly and try it. I actually had one of these on my T60p and while I didn't document anything what I remember is that it cooled very well. I'll keep posting here what I find as there does seem to be some interest.

JR956678 wrote: ........
The plate on the T61p heatsink I modified is formed downwards to contact the GPU but this only makes the situation for cooling the Northbridge far worse. The Northbridge chip is actually .050" lower than the GPU so bending the plate down where the GPU is requires an even thicker pad if you're going to provide any cooling to the Northbridge.

JR, that is exactly the kind of detail I was hoping for....truly an excellent post...thank you.

In regards to the above para, that's backwards.

Reforming the pipes to lower the plate to touch the GPU will of course lower the Northbridge extension as well; making its heat-conduction path -better-, not worse.

I have reformed 41V9932's (the stock T60p HS) for direct contact with GPU; and as logic would suggest, it results in being able to use a -thinner- pad on the Northbridge.

The 9932 properly splits the cold-ends of the heatpipes so that each 'side' of the HS receives only one pipe/heatflow.

It is interesting to note that the GPU pipe goes to the -rear- fins...and -vents-....which are always half-blocked by a trouser-wrinkle when using the T60p on my lap; which is how I use it 100% of the time.

Just another stick piled on back of the poor GPU's thermal camel !

fyi, I recently posted in another thread (but don't have a link for it handy at the moment) some details about both the forming of a 9932 and some other aspects of such a project, which also affect the results (e.g., modifying hold-down plate tabs to apply pressure on -center- of GPU, not edge....using old pads to conduct heat from pipes up to keyboard-plate, etc..).

I have abjured any design which uses a single plate over both GPU and NB, because of the obvious difficulty in ever ensuring perfect flat-seating on both at once with a single plate. (which is of course why lenovo themselves used 2 seperate plates out at the ends of the 9932 pipes).

PS; It can sometimes be counterproductive to solder an excessive length of heatpipe on the hot-end; as the outside of the pipe is the -return- path for the condensate; so a hot 'outside' can lower the volume of fluid returning to the -hottest- spot, the center of the chip.

Notwithstanding that, I too have seen a couple units that were pretty darn skimpy on attach!

Thanks again very much JR for taking the time for such a detailed report of mods and set of related thoughts and observations.

Richard

dozer wrote:Reforming the pipes to lower the plate to touch the GPU will of course lower the Northbridge extension as well; making its heat-conduction path -better-, not worse.

Very true. But the bend I was referring to was the plate itself from the T61p heatsink. It has a kink in the middle such that the portion that contacts the GPU is .010 lower that the portion which contacts the Northbridge. But on a T60p (mine at least) the GPU sits .050" higher than the Northbridge so using the T61p plate just didn't seem to make sense - in addition to making up the .050" height difference between the GPU and the Northbridge I would have had an additional .010" to make up. No amount of bending of the pipe would have made up that difference.

As crude as the design is, I suspect that cooling the Northbridge is perhaps not so critical - it just doesn't create the heat load that the GPU does and that's probably why the extra thick pad was used there.

The single plate design to cool both the GPU and the Northbridge is as you point out very poor - it's very unlikely that the top of both chips would be perfectly parallel even though the plate to cool them is - thus some sort of thermal pad is needed. Sadly.