New methods to reduce the fan noise in a T60 or in a T60p

[Johan]

The purpose of this thread is to share a bit of information about how to potentially quiet down the fan in ThinkPad e.g. T60 and T60p models; a couple of tips aimed at reducing the fan noise. I have not tried any of the tips mentioned below; all I want by posting this is to point readers of this forum to potentially useful information published elsewhere. Proceed at your own risk...

As especially many T60p owners have experienced, the quite powerful ATI Mobility FireGL V5200 and/or V5250 graphic controller units (GPU’s) in these laptops are consuming relatively much power, needing the much generated heat to be dissipated, which require effective heat-sinking which in turn leads to that the fans in T60p’s very often run… and run “loud” in many peoples opinion. There are many threads on this and other forums discussing this issue.... "T60p fan noise" – and also discussing possible ways aimed at reducing the fan noise. Common tricks to reduce the T60p fan noise includes improving the thermal contact between the GPU and the fan assembly/heat sink, like described in e.g. the threads Improving T60p heatsink performance and How thick is the gap between GPU & heatsink?

Certain newer ThinkPad fans (e.g. in T400’s and T500’s) have been reported to be more quiet compared to the standard T60p fans; more quiet in the sense making less noise when running at the same speed (e.g. a T60p fan running at 2900 rpm compared to a T500 fan running at the same speed, the noise in these to laptops compared). This observation has triggered a very clever German T60 ThinkPad owner/user to replace the standard T60 fan by a T500 fan! The result is interesting; see the thread (in German): T60 Lüfterumbau. A translation (by the free Google Translate) of this German thread into English can be found --> here; the translation is of course far from as good compared to if a German/English speaking ThinkPad user had translated it, but you’ll get the main points, especially if you consult a dictionary yourself if/when in doubt...

I hereby welcome all T60p owners who are annoyed by the fan noise to review the above-mentioned thread, and if anyone goes ahead and actually carry out the modification as described in the German thread, PLEASE be kind to share your experience here in details – thanks very much in advance!

Another apparently interesting suggestion about how to quiet down e.g. the fan in T60p’s (but the method not restricted to these model ThinkPad's only!) is described at the commercial page Worlds first intelligent ThinkPad fan controller! – there’s a video about this thing at YouTube --> here which shows/explains how it works. This thing has also briefly been mentioned a few times at this forum; see the threads tpfc.mywo.org - cheaper alternative and by forum.thinkpads.com member electron73 in this post, plus in this post and is also briefly mentioned in How to cool down ATI5250 on T60p? Unfortunately, I haven’t been able to dig up any user-feedback or –reports in English about how well it works, but again the (very active!) German ThinkPad’s forum has a couple of thread about this; I have only been able to find Thema: tpfc.mywo.org - wie gut funtioniert das? (again with a Google Translate translation --> here) and GroupBuy - ThinkPad Fan Controller Chip - praktisch lautloser t60 (Google Translate translation --> here). Again, if anyone goes ahead and actually gets and installs one of these devices, PLEASE be kind to share your experience here in details – thanks very much in advance!

I have no association of any kind with any posters mentioned in this post… I only wish to share with you folks so that you can hopefully be even more satisfied ThinkPad owners!

Johan

[tomkrishan]

ok, folks I have posted Dec 05, 2011 this

http://forum.thinkpads.com/viewtopic.php?f=29&t=100932

questionary about a possible replacement of the T60p fan & heatsinck with the one of a T61/T61p.


After the replacement I can clearly state:

Everyone annoyed by the T60/T60p fan noise and/or high temps as shown with the TPFC should instantly change the fan/heatsinck with the succesor fan/heatsinck like the ones (42W2028 42W2822) for the T61/T61p machines!!!

I am totally satisfied & really enjoy the new smooth cooling system. The heat savings are about -+ 5-7 °C (41-44 F) and even with huge workload, such as surfing, watching video and a 3D game all together at one go, the increase of the heat & noise is clearly below what was peaked before which was sometimes up to 92-95 °C for the GPU under heavy load. Now my 30 min. heavy test load has shown just slightly above 80 °C for the gpu.
Idle mode says rarely more than 60 °C which was before at least 64 °C under Win 7.

The replacement goes without any issues. Nearly the same heatsinck/pipe construction. Just some minor differences which are not effecting the change at all.
I`ve added just some fresh thermal compound (not "Arctic Silver" but a no-name brand also with 10% silver) after cleaning the cpu, gpu & heatsinck.


Cheers

[TuuS]

This is intended as a public service, not a self promotion, but if anyone is interested in a late model T61p fan, I have several of them pulled from late model T61p units in excellent condition. They are easily worth $40-50 each, but I'm only asking $20 plus the cost to ship, while they last.

edit: the surplus of these no longer exists, but it was nice to be able to help so many with upgrading their fans

[Lo-q]

Very interesting mod! I´ll really consider getting one of those fan controllers.

Did you read this by the way?

I feel pretty confident about this now, that my T60 with the X1400 GPU and others as well have an issue with the termal sensor in the GPU, causing it to report too high temperatures. This gives the impression that it runs very hot, and causes the fan to uneccesarily run most of the time, at least when controlled using TPFanControl.

[_65536_kb_sec_]

So I've been along the cooling threads, so I've came with quite different solution.
Ever since started dealing with PC's I've been having some of my machines overclocked.
So from my previous experience, I've got some copper plates, not much wider than 0.5mm. So last night I've cut several pieces with quite the same area as the GPU and chipset crystals spacing. For the mod I've used some "silver" paste between each copper piece and between heatsink and the chip surface, and removed the annoying (IMO) stickers.
Have in mind that you can go over with a small size sandpaper (1000-2000 or so) to get the copper surface flatter.
With this mod I've managed to go down with somewhere around 5-10* Celsius.

PS Sorry for the bad English.

[tomkrishan]

Did you read this by the way?

yes indeed, an interesting point. since I have built-in the T61p fan I also wonder sometimes where the fan is as you can`t hear the fan sometimes. before the change I was thinking about why not deinstalling this annoying T60p fan as I know that the cpu and gpu should resist temperatures of about 100 degree. but of course this is not the best solution on the long run.

by the way I ve just posted my experience report with a german thinkpad forum where someone has made the change even with the fan from a T500 machine and in addition the change was for a T60 14.1" not a 15.4"!!!

here: http://thinkpad-forum.de/threads/92788- ... Cfterumbau

it seems the guy has made some minor change as you can see on the second foto and in some posts they discuss just some little difference between copper plate and gpu of about 0,5 mm. the difference is filled with a thin piece of copper or just filled with enough thermal compound. they are all satisfied with the results.

[TuuS]

My personal feeling is that using a copper shim is a bad idea. Ideally you want the heat to transfer across the airgap between the die and the heatsink's cooling platform to be as efficient as possible and by installing a shim, instead of "filling the gap" as it intends to do, it simply creates a second gap that the heat must traverse. My best results have been obtained by doing some careful fitting, making sure the platform fits snug and level across the surface of the die and any existing gap is filled with only as much high quality thermal paste (recommend arctic silver #5) as is necessary.

I initially had a surplus of about 25 of these T61p fans and have installed them in both T61 14.1 (standard) 15.4 (wide), as well as both 14.1 and 15" T60/T60p models. In every case I've found they cool much more efficiently. Most recently I built a T61 14.1standard screen with a T8100 penryn chip and intel graphics and it cooled so well that it would run indefinitely without the fan coming on until you ran something that demanded some work out of the cpu. According to tpfancontrol the cpu temp never reached 60 when the system was idle. With the factory fan it would quickly raise to 60+ where the fan was set to kick in, which is about what to expect. On my T60p with T7600cpu and fireGL5250 it decreased both cpu and gpu temps significantly, on average of 4-7 degrees at idle, but more important was it's ability to quickly bring down the temps by several degrees in mere seconds just by changing the setting to manual. This seemed to be the best confirmation that the cooling system was working efficiently.

One word of caution though... whenever you modify something as important as a computers cooling system, you should do it with care and respect. Don't just throw the parts in and declare success, do it carefully, making sure to properly fit the parts and be prepared to monitor it long term. You're basically altering the way your system is cooled and in doing so, you've become the engineer, so it's now your responsibility that it runs correctly, so it's not a simple case of "set it and forget it". If you're not willing to monitor it long term, then stick with the factory parts, but if you don't mind experimenting and want the added performance, then I'd encourage you to try this mod. I'm very happy with the results from my systems. YMMV

[shalliday]

Thanks to TuuS's recommendation and advise walking me through it, yesterday I completed installing one of his T61/t61p fans in my T60p, which has the T7600cpu and FireGL5250 card. It is not only more quiet than my original T60p fan but so far it is doing a better job keeping the system running cooler. As I was with the original fan installed, I am using TpFanControl to monitor and control fan speed. While one day's use may be to early to declare success, so far I am very impressed and glad I installed the T61 heat/sink fan unit!

After reading so many threads here on the forums regarding all of the various options including those recommending spacers and such, I was really surprised how easy it was to install the T61 heat/sink fan unit into my T60p. It was a perfect fit and required nothing more than:

1. Using Arctic Silver cleaner to remove any old residue on the cpu and gpu, readying it for a new coat of thermal paste.

2. Fitting the T61 heat/sink fan in place to make sure it was sitting level and making contact, as TuuS advised:

usually you can eyeball it to tell if it's fitted properly and if you press down on the cpu platform to make sure it's seated, then if you press on the gpu platform with another finger, there should be no movement... which would indicate you have contact. If you can feel it moving then you have an obvious air gap. One other technique, and you can decide how obsessive you want to get with this, but generally you can use long strips of paper between the gpu and it's platform. when properly seated there should be some friction. You can also judge by moving the paper to different areas if you have even contact across the entire face of the die. If you feel friction on one side of the die but not the other, that will tell you that the platform is not level and you can "gently" adjust it until you get the perfect fit. Trying to determine the perfect fit isn't easy and in the end you have to just do the best you can. The thermal paste and the retention brackets will make up for small gaps, so absolute perfection in fitting is neither necessary or possible.

Luckily mine sat level and there was no movement so no further fitment needed to be done.

I've read where some people use shims to fill gaps, but this is a bad idea because it doubles the amount of gaps the heat has to transfer across. The best cooling is obtained when you have the best possible fitting with the thinest coating of paste. As far as the paste goes, using to much can be just as bad as using not enough. You want to make sure you have enough to cover the full surface of the die and you will get some overspread, but try to minimize that as much as possible. Some people have done a "mock up" assembling with toothpaste so they can determine where it's spreading. If the heatsink spreads it out one side but not the other, this would indicate an uneven seating. I think this is unnecessary, but if it helps you to learn or become more confident in the procedure, then you might want to try it. If you do, make sure to fully clean it up before applying the real paste.[/quote]

3. Next I removed the T61 heat/sink fan unit and proceeded to apply a thin even coat of Arctic Silver paste on both the cpu and gpu. As both ThinkpadLover and TuuS recommended, I applied a small amount of thermal paste about the size of an uncooked piece of rice and then proceeded to level it off using a credit card, making sure there were no air bubbles and that it was spread evenly. Once applied, I used several q-tips and arctic silver cleaner to remove any excess around the cpu and gpu.

4. Placed the metal plate back over the gpu and reinserted the screws into the metal plate and heat/sink fan then tightened them up in stages so that pressure would be evenly applied.

5. Reassembled the laptop, powered it up and started monitoring temperatures using TPfancontrol.

As TuuS stated:

In the end after it's all said and done, the results will tell you how well it's cooling. The CPU part should show immediate improvement, but the GPU results won't be as obvious to tell. The GPU is more or less just "along for the ride" when it comes to cooling. Most of the fan operation is controlled by the CPU temps, but as I said in a prior message, if you manage to get the GPU temp elevated you can switch TPfancontrol to manual so the fan is always running, and watch how well the GPU cools. On an average system your CPU will idle at about 60-65c and the gpu will be 5-10 degrees cooler. If you stress the system and get both CPU and GPU temps upto 80 or higher, the fan will run on high until the cpu gets down below 65ish, by then your gpu could be higher or lower, but the system won't seem to see a need to continue cooling as long as the cpu temp is within limits, so this is where manual mode can answer a lot of questions. You also have your other systems to compare with so you should be able to see a noticeable improvement with this heatsink/fan assembly.

It worked for me.

Hope this is of some help.

Regards,

Scott Halliday

[LASTGUY2GETPS2]

Similar results here. I just installed a T61 42W2028 (42X4685) fan in my T60 (not T60P) and am getting pretty reasonable temperatures (GPU 64C, CPU 50C as I'm typing this post). The machine is also much much quieter than before. The fan speed seldom goes above 1.

For people who want to do this, make sure you also take off the keyboard bezel before installing the fan, otherwise it is going to be a tight fit. A thorough cleaning with a non-lint cloth goes without saying and arctic silver 5 is a must.

EDIT: Forgot to mention that I have discrete graphics by ATI

[iwishiwasinwhistler]

I tried this recently with the T500 fan with great results.

Seeing as I couldn't find much English info on this one I thought I'd post my experiences here in the hope that it'll help someone. BTW the German Thinkpads forum is a mine of technical knowledge if you're willing to take the time and effort with Google Translate and/or a dictionary.

I did this while rebuilding my QXGA T60p. To be clear the parts I'm using are.....

2007-AE7 15 inch 4:3 T60p frame with QXGA mod
42W7872 intel integrated GMA965 planar (pulled from a working 14.1 T61 4:3)
45N5492 HSF fan assembly for the T500 discrete graphics
???????? ATI metal hold down clamp/bracket from T60p - sorry don't have the part no to hand
T9500 processor using AS5

....plus other bits and bobs that I've cherry picked from all over the place and have taken me far to long to collect.

My reason for using the discrete T500 HSF was because I hoped it would cool even better than the integrated HSF. I'd read the T500 fans were supposed to be quieter (owl blade technology?) but would need a few mods to go in. And well, why not, I thought I'd give it a try......what's the worst that could happen?...... (please don't answer that!)

The T500 HSF sits in really nicely with a few considerations. I had to remove 3 of the copper cooling fins in order to avoid the display hinge as detailed in a post on page 4, at the bottom, by seja (with photo) on the german forum here....

http://translate.google.com/translate?l ... Cfterumbau

.....this was really easy with some long nose pliers just gently rocking the fins back and forth. I noticed the OP on that thread had taken all the fins off but this wasn't necessary and I wanted to leave as many on as possible for cooling.

I left the thermal pad on for the intel GPU. I know many folks have had great success with bending the heatpipes and penny mods but I figured the integrated GPU just doesn't run that hot and the thermal pad made great contact as it was. The 45N5492 is also a fair deal more rigid and stronger than the T60/T61 HSF's. It looked like it wouldn't be as easy as the softer 60/61 units are to bend. If anyone does give the bending mod a go on the T500 HSF I'd be interested to know the results. Anyway it all came together and seated down really well using the pretty solid original 15" T60p ATI bracket.

One thing I did notice is that the 45N5492 sat pretty close to the planar (touching??...difficult to see once in place..) in the spot where it would contact the ATI graphics on a T60p board - for want of a better description. I can't think of a better way to explain this - just north of the intel GPU on the 42W7872 planar is an area that doesn't have many components, is relatively unpopulated. Hopefully somebody else can describe this better as you can obviously see I'm struggling. The HSF sat sufficiently close for me to be concerned about the possibility of shorting so I attached some of the high temp tape I had from a spare HSF laying about to alleviate my worry.

It all goes together perfectly, everything lines up well, with no bulges or problems. I've been using a stock W7 64 ultimate install with C300 256 ssd and 8gb. Bios is Middleton's (what a genius - thankyou).

So far it's MUCH quieter than the T60p in it's original state (that was with AS5, clean inside and undervolted). The T500 fan is certainly quieter. More so than when this board was in the 14" T61 with original fan. I'm aware that I've changed a lot in one go and we can expect the T9500 Penryn to be cooler than the original T60p board with Core Duo Yonah 2.16.

Bit of a wordy post that one. In conclusion T500 45N5492 works well in FrankenPad 15" T60p case with 14" 4:3 T61 intel planar. The fan rarely comes on. When it does it really isn't distracting and runs at a similar volume compared to my x200s Sl9400. The fan brings temps down fast and then goes off. The idle temp for the T9500 is around 31-33C. I haven't had time to undervolt or use TPFanControl yet but I will do at some point. Saying that, I could honestly and very happily just carry on using it like it is with the current temp and fan volume when it does occasionally run. All in all well worth the effort.

[Johan]

@ iwishiwasinwhistler:

Thanks very much for taking the time to share your experience - your efforts are certainly greatly appreciated, and I am fully convinced that your contribution will inspire many other T6x/p owners who also wish to quiet down their fans.


Johan

[julukaForever]

Steps to modify the T60 fan rpm levels used in the acpi thermal management
----------------------------------------------------------------------------------------------

I came into a T60 recently (widescreen model with ATI Mobility Radeon X1400). Like many other Thinkpads (including the T22, T41, R51, and G40 with which I have extensive experience), there appears to be a bug in the way the thermal management implements hysteresis. On power-on, the fans run reasonably (at least for the first 0-40 minutes or so), but as time goes on, especially after several hours, the fans are on more or less incessantly (with temps fine, and even with the pc idling). If the fans are reasonably quiet, while somewhat annoying, its ok. With the T60, however, the fan noise is insanely loud. The lowest rpm is around 3000 which is WAY to high, period. The T60, in my judgment, is a bit of a dog of a pc, a far cry from pre-T5X thinkpads; the keyboard is cheap/flimsy, the thinklite shines in your face (not on the keyboard), the pc beep is way to loud (clearly a hw and/or ec sw bug with the hd audio), etc.

I actually started to design a simple circuit (using ff's and a pll) to increase the frequency of the tach signal to spoof the embedded controller (ec) into thinking the fan was running faster than it really was. Such a circuit is pretty easy to prototype, but to make it robust, flexible, and small enough to fit inside the laptop is difficult to do at home; hence, this is really a dead end.. much better to try and solve this issue in software. So, on the hunch that the ec uses a table of 'target' rpms to drive fan speed management, I decided to look at the ec in earnest. It turns out that the offending table exists, and is easily modified.

Of course, this is a DANGEROUS (and possibly unwise) thing to do; if you get a checksum wrong, the pc may not boot, and changing rpm values may have unintended consequences which render the pc unusable. Nevertheless, I decided to do it, and all appears to be well. In addition to the rpm table mod, I also use a kernel module to set up trip points to handle fan control (not optimum, and with obvious pitfalls, eg, a kernel oops => no fan control..), but, for me, acceptable (been doing this since the T22). My T60 temps are now steady at 47-50 C at idle/editing, 49-52 C under moderate activity (eg, web browsing), 55-59 C on Hulu, and 55-69 C under heavy loads (eg, a 5-6 hr gcc build). I consider these temps entirely satisfactory, in-line with (or better than) other thinkpads I've used, and indescribably more pleasant than having the pc constantly roaring at 3000+ rpm, at 45-47 C, while you edit a text file in vi, under kde.

Anyone considering a firmware mod such as follows should take the time to fully understand what is being done, and fully appreciate the consequences if something goes amiss.. I recommend this procedure to no one, will not post code, and bear no responsibility in any way for anything which occurs as a result of anyone carrying out this procedure..

So, here are the steps I followed (under linux):


(1) Extract the ec code, and for convenience, rename the ec:

$ 7z x -o7iuj17uc.iso-files 7iuj17uc.iso
$ cd 7iuj17uc.iso-files
$ cp {\$,}01AC000.FL2


(2) Orientation:

$ hd 01AC000.FL2 | more .. etc

Here, alias hd='hexdump -e '\''"%06.6_ax " 16/1 "%02X "'\'' -e '\''" " 16/1 "%_p" "\n"'\'''

A quick look at the hex dump reveals the following:

000000 - 002000 (0-8k ) largely empty header terminated w/long 00-block

002000 - 010000 (8-64k) ff-block (part of header perhaps)

010000 - 02ffff (128k) probably data+code (ff-block at 027038)

030000 - eof loadable modules and related (module signatures like: 00 31 31 00 42 1B 05 .. are seen, ie, 27-byte LZ5-compressed modules)


(3) Pull out code+data portion of ec for dis-assembly:

$ dd if=01AC000.FL2 of=01AC000.FL2.bin ibs=1 skip=$((0x010000)) count=$((0x17030))


(4) Build binutils-2.16.1.tar.bz2 with configure option --target=h8300-coff and dis-asemble:

$ h8300-coff-objdump -Dmh8300s -bbinary 01AC000.FL2.tmp2 > 01AC000.FL2.binutils.s

I think binutils-2.16.1 is the last verison with h8s support.

After having examined the Renesas doc at http://am.renesas.com/products/mpumcu/h ... tation.jsp (REJ09B0331-0500 Sep.14.06 Hardware Reference and REJ09B0139-0400 Feb.24.06 Software Manual), we find the following from the disassembled code:

010000 - 0101bf
interrupt vector table; 010000 is the h8s program zero-base address H'00000000; 0x000013f0 is the Power-on/Reset vector => Power-on/Reset first instruction is at 0113f0

011300
code validation (zero-sum LRC); this code is straightforward; from it we find that blocks [010000, 010400) and [011000, 030000) are validated; block [010400, 011000) is very likely checksummed as well (it sums to zero), but is not validated here (a code bug !?)

and from the validation code at 011300:

0103fe - 0103ff probably block [010000, 010400) checksum
010400 - 010401 probably block [010400, 011000) checksum (but not validated at 011300 as noted above)
02fffe - 02ffff probably block [011000, 030000) checksum


(5) In the hope/belief that somewhere in the code there lies an rpm table, and, given a knowledge of what these values are, eg, by collecting some data like:

level reported rpms (acpi 0x84-0x85)
----- ------------------------------
128: 3294 3315 3298 3352

1: 2807 2851 2864 2847 2976 2936 2914 2932 3009 3005
2: 2819 2817 2786 2971
3: 3437 3438
4: 3627 3570 3435
5: 3494 3500
6: 3893 3886 3932
7: 3971

its easy enough to write some C to romp through the ec file looking for clusters of appropriate values. For example, read the the file word-by-word, looking for three contiguous words each of which is in, eg, the ranges 2700-3100, or 3300-3600, or 3700-4200. There are many ways to do this sort of thing, but amazingly, the foregoing worked like a charm (only a small number of hits, all but several being easy to rule out).

Thus, the table of fan rpms used for fan control is found to be at 0110b4 in the ec file. Each entry is 10 bytes:

$ hd -s 0x0110b4 01AC000.FL2|head -8
0110b4 04 20 0A BE 0B 54 0D DE 7F FF 04 20 0A 8C 0B 22 . .¾.T.Þ.ÿ. ..."
0110c4 0D 16 7F FF 04 20 0B EA 0D AC 0E D8 7F FF 04 20 ...ÿ. .ê.¬.Ø.ÿ.
0110d4 0B EA 0C E4 0E D8 7F FF 04 20 0B 86 0D 7A 0E D8 .ê.ä.Ø.ÿ. ...z.Ø
0110e4 7F FF 04 20 0B 86 0C E4 0E D8 7F FF 04 20 0B B8 .ÿ. ...ä.Ø.ÿ. .¸
0110f4 0D DE 0F 0A 7F FF 04 20 0B B8 0D DE 0F 0A 7F FF .Þ...ÿ. .¸.Þ...ÿ
011104 04 20 0B B8 0D AC 11 30 7F FF 04 20 0B B8 0D AC . .¸.¬.0.ÿ. .¸.¬
011114 11 30 7F FF 04 20 0B B8 0D 7A 0E D8 7F FF 04 20 .0.ÿ. .¸.z.Ø.ÿ.
011124 0B B8 0D 7A 0E D8 7F FF 08 00 22 2A 50 39 55 3C .¸.z.Ø.ÿ.."*P9U<

and we have:

fan level: ? 1,2 3-5 6-7 ?
0110b4: 0x0420 0x0ABE 0x0B54 0x0DDE 0x7FFF ~ 1056 2750 2900 3550 32767
0110be: 0x0420 0x0A8C 0x0B22 0x0D16 0x7FFF ~ 1056 2700 2850 3350 32767
0110c8: 0x0420 0x0BEA 0x0DAC 0x0ED8 0x7FFF ~ 1056 3050 3500 3800 32767
0110d2: 0x0420 0x0BEA 0x0CE4 0x0ED8 0x7FFF ~ 1056 3050 3300 3800 32767
0110dc: 0x0420 0x0B86 0x0D7A 0x0ED8 0x7FFF ~ 1056 2950 3450 3800 32767
0110e6: 0x0420 0x0B86 0x0CE4 0x0ED8 0x7FFF ~ 1056 2950 3300 3800 32767
0110f0: 0x0420 0x0BB8 0x0DDE 0x0F0A 0x7FFF ~ 1056 3000 3550 3850 32767
0110fa: 0x0420 0x0BB8 0x0DDE 0x0F0A 0x7FFF ~ 1056 3000 3550 3850 32767
011104: 0x0420 0x0BB8 0x0DAC 0x1130 0x7FFF ~ 1056 3000 3500 4400 32767
01110e: 0x0420 0x0BB8 0x0DAC 0x1130 0x7FFF ~ 1056 3000 3500 4400 32767
011118: 0x0420 0x0BB8 0x0D7A 0x0ED8 0x7FFF ~ 1056 3000 3450 3800 32767
011122: 0x0420 0x0BB8 0x0D7A 0x0ED8 0x7FFF ~ 1056 3000 3450 3800 32767

Code examination suggests that the '1056' and '32767' values are not rpms (the may be used as lower and upper rpm bounds), but this is unimportant here; the level 1-7 values are the ones to be modified.


(6) Modify the level 1-2, 3-5, and 6-7 as desired, eg,

change: ------ -1000 -700 -200

0x0110b4 1056 2750 2900 3550 32767 <-- original values
------------ ------ 1750 2200 3350 <-- modified values

0x0110be 1056 2700 2850 3350 32767
------------ ------ 1700 2150 3150

0x0110d2 1056 3050 3300 3800 32767
------------ ------ 2050 2600 3600

0x0110c8 1056 3050 3500 3800 32767
------------ ------ 2050 2800 3600

0x0110e6 1056 2950 3300 3800 32767
------------ ------ 1950 2600 3600

0x0110dc 1056 2950 3450 3800 32767
------------ ------ 1950 2750 3600

0x0110f0 1056 3000 3550 3850 32767
------------ ------ 2000 2850 3650

0x0110fa 1056 3000 3550 3850 32767
------------ ------ 2000 2850 3650

0x011104 1056 3000 3500 4400 32767
------------ ------ 2000 2800 4200

0x01110e 1056 3000 3500 4400 32767
------------ ------ 2000 2800 4200

0x011118 1056 3000 3450 3800 32767
------------ ------ 2000 2750 3600

0x011122 1056 3000 3450 3800 32767
------------ ------ 2000 2750 3600

Perhaps, all of these needn't be modified.. perhaps only subsets are used for particular T60s.. but I suspect (based on accumulating a large table of actual rpm readings over a 2 hour period) they all play a role in the underlying thermal/fan control management algorithm (and figuring this out, if even possible, would take a whole lot more work); so modify all of them for the first cut. Before I settled on modifying the rpms in this table, I spent many hours over several weeks tracing the ec code by first locating routines referencing the table base address, then routines calling these routines, etc.

By the way, ~ 2000 rpm is inaudible, and ~ 2800 is reasonably quiet (the hd is louder).


(7) Update the checksum word for the memory block containing this table (ie, block [011000, 030000).

This is the tricky (and dangerous step). Tracing the ec code from the Power-on/Reset vector onward, one quickly finds where the code is validated. It turns out that the ec uses simple zero-sum LRC validations. The code is at file offset 0x11300 and is quite straightforward. There appear to be at least three distinct checksummed memory blocks (this is guesswork, based on simply word-summing suspected blocks of memory; if the sum is zero, it is very likely that it is checksummed), but at 0x11300 only two of them are actually validated (a possible bug in the code/or perhaps the third is validated elsewhere). For zero-sum LRCs, the sum of all words in the memory block will be zero (the checksum word being included in the sum). In principle, the checksum word could be any word in the block, or could even be obfuscated (herein lies the danger: one could inadvertently modify important data or code), but it is often at the end, sometimes at the start, or in the header (if present). For block [011000, 030000) it appears to be at 0x02fffe.

02fff0 FF FF FF FF FF FF FF FF FF FF FF FF FF FF 06 2A ÿÿÿÿÿÿÿÿÿÿÿÿÿÿ.*

Re-sum the words in block [011000, 030000) of the modified ec, write the sum to 0x02fffe, and the re-sum the result to verify correctness.


(8) Take a couple of days to mull things over, and then, re-flash, reboot, and pray..

By default bios/ec updates will not re-flash the same version (only version upgrades and downgrades (with a warning) are allowed). EC versions change much less frequently than bios versions, and even if you do a bios upgrade, if the ec version hasn't changed it won't be re-flashed.

Its easy enough to make a custom bootable iso/cd with a modified lcreflsh.bat to allow re-flashing the current ec, if, eg, changing 'flash2.exe /u' to 'flash2.exe /u /s' works.. but I'm not sure this will work (haven't tried it).

An annoying, but easy way to do the re-flash is to use windows:

(a) get the .exe version of the bios update and run it to extract the files (by default to, eg,
C:/DRIVERS/7iuj17us); do not flash yet-- you only want to extract the files at this point

(b) copy in the modified ec over the original one in C:/DRIVERS/7iuj17us

(c) open a dosbox, cd to C:/DRIVERS/7iuj17us, and issue: WINUPTP.EXE /s

The /s option turns off bios and ec version checking, so both will be re-flashed. WINUPTP.EXE /s
will run silently (unless there's a verbose option /v ?), so as long as it terminates within several
minutes without any printed errors, the flash succeeded w/o issue.

(d) reboot after WINUPTP.EXE /s terminates


(9) Sit back and enjoy the singular quietude, for the first time with this pc.

[ausmike]

@julu*: wow ...... great writing and excellent steps for mods>> ** very impressed **


By any chance you coming to this ........... love to chat with you
or when next time am in BALT* (and have spare time) would love to buy u a beer !
maybe a football game night !

Thanks alot
Cheers

[RealBlackStuff]

@julukaForever:
FYI: there's a similar thread about the EC as used in the T4x/R5x series, which was already started back in 2006...
http://forum.thinkpads.com/viewtopic.php?f=27&t=20958
Heat is definitely the laptop's Arch Enemy #1.

[bmwman91]

Awesome thread!

So, I really want to do this now as the stock T60p cooler runs sort of hot (and loud). I was thinking of using the T61p cooler, but it sounds like there has been some evolution in laptop coolers since then and the T500 unit sounds great.

EDIT
From more reading around, it sounds like the 45N5492 is the only proper option for a T60p with ATI graphics. So at least that is settled. Now it just sounds like the only possible issue is that it can hit the bottom of the keyboard. At this point, I think I will just have to drop the $25 on eBay to get one and test it out myself. I will definitely be trying to bend the heat pipes to have the cooler plate sit on the GPU die as doing that with even the stock T60p HSF got me a temperature drop of like 10°C.

It seemed as though some people had put the 45N5490 HSF onto ATI-equipped T60's as well. From looking at pictures of this HSF on eBay, I don't see room on the "satellite" cooler plate for the GPU, only the southbridge. Weird.

[think_madness]

So from what I can understand, the 42X4685 42W2028 fans are for discrete graphics and the 42W2028 42W2822 are for integrated graphics? Or are both systems able to be utilized by a T60 with integrated graphics? My reason for asking is I am planning a mobo upgrade to a T61 42W7648 integrated with (hopefully) a T9300 chip to take advantage of SATA II speeds and up to 8GB RAM...

[iwishiwasinwhistler]

@julukaForever, yeah!!! That blew me away!! I've read and re-read your epic post and I still don't think I'm brave enough to try it!

@Johan, thanks, it's just good to be able to give something back, however small. I've learned so much from everyone here.

@bmwman91, from reading the German thinkpads forum thread (again and again) it seems that 45N5492 isn't a great choice for T60p's with ATI graphics. It can be made to fit but as I noted when I did my conversion (on an intel board) the construction of 45N5492 is a bit more solid and it seems, from the German thread at least, that even when it does go in, bending the heatpipes to fit, it's too much of a squeeze and pushes the keyboard causing it to bow upwards slightly. Not ideal.

The general rule seems to be that you CAN use the T500 intergrated 45N5490 HSF with X1300 and X1400 T60's and also V5200/5250 equipped T60p's. This does appear to be the "wrong" way round but I guess it's just a matter of space. In post 66 Hazze confirms successfully using a 45N5490 with V5250. I think maui_muc summarises a lot of the information on page 18 post 172 and essentially the intergrated 45N5490 can be made to work with all the ATI graphics. Bear in mind that a lot of the people posting have used a slightly thicker thermal pad to account for the tiny difference in height and also the penny mod to increase the pressure.

I'm not sure I follow you about there not being enough room on the "satellite" cooler plate on the 45N5490??? Admittedly the plate is smaller there than on the stock T60p ATI HSF but I think there is enough room. People do seem to have made it work?.... I've got my stock T60p ATI V5200 HSF (41V9932) in front of me on the desk now and if I turn it upside down the point at which it makes contact with the GPU can be seen imprinted on the old thermal pad. The point of contact is much smaller in area than the copper which sits above it. I can't see any problem using the newer 45N5490 with a smaller "satellite"...... Let me know if I've got completely the wrong end of the stick on this one but I don't think I have. I think 45N5490 with some mods could work for you.

@think_madness, the T61p fans you mention will be an easy enough replacement and as Tuus has posted work really well and are a direct fit. However, if you're competent enough to be building your own FrankenPad then please do consider a T500 fan. They are really smooth and quiet. Your planar upgrade to a 42W7648 is essentially the same as I did although with a 42W7872. They're basically the same board bar a slightly later update for the Penryn sensor. A 45N5492 with a bit of tape and removal of 3 tiny fins from the heatsink fits beautifully and is so easy to do. See my post above for a bit more on it. You may want the metal hold down bracket that I used from an ATI T60p V5200 as it works well without any bending, penny mods or extra thermal pads. I haven't posted the part no above but let me know if you need it and I'd be happy to find it for you.

[think_madness]

iwishiwasinwhistler,

That is absolutely fantastic! Thank you very much for the info. Also, if you could get that part number for that bracket, it would be greatly appreciated.

Again, thank you!

[iwishiwasinwhistler]

think_madness, I think I've worked out why I didn't have the part no for the bracket to hand earlier. I think it's available from IBM only as a bag of "System miscellaneous parts (15.0-in. LCD models)" and the part no for the complete bag is 41W6302. It looks like it comes with the DC cable, CPU support, modem cable and a few other bits and bobs you probably already have. I guess I was lucky because I already had it from my 15" T60p ATI V5200. I didn't have to file or modify the bracket in any way. It clamps down snugly on the 45N4592 HSF.

I've been using the T60p Hardware Maintenance Manual for this info but if anyone else could confirm this that would be great??... In the HMM the fan bracket can clearly be seen in the exploded view simply labelled as (b).

Do you currently have an intel T60? Maybe you already have it?.... The part you're looking for can be seen in the 4th photo from the OP here.....

http://thinkpad-forum.de/threads/92788- ... Cfterumbau

....it's the silver metal bracket with the piece of clear tape on it. It would be pretty easy to check by just removing your palmrest and keyboard.

If you don't already have it I'm sure somebody here would be able to help if you posted a "Looking For" in the Marketplace. Sure beats buying a bag of stuff from IBM when you only need one of the parts.

[RealBlackStuff]

That heat sink shield/clamp has P/N or FRU 26R9633
http://www.ebay.com/dsc/i.html?LH_Title ... 42W7872%29

[iwishiwasinwhistler]

Exactly the one! Thanks RBS.

Doh! Just noticed the ebay photo shows the p/n stamped on it. Guess I should have looked at that on mine, lol...

[Thinkpad Lover]

@ iwishiwasinwhistler : I have that part (T60 Heat Sink Heat Shield / Clamp FRU 26R9633) available at a great price and I can ship it out to England today or tomorrow. Email for full details.

You guys- always running to Feebay LOL.

I have many many parts here: http://forum.thinkpads.com/viewtopic.php?f=11&t=97374 but don't have time to list everything. Check with me, I can help you if you want.

Edit: Oh, I guess it's "think_madness" in Boise, Idaho who needs that part. Let me know, it's ready to go. (Email in signature below).

Daniel.

[bmwman91]

Hey 'Whistler, thanks for the info! I am not squeamish about hard-mods, so I will see what can be done to make the '92 HSF fit well. I agree that having it hit the keyboard is not acceptable since that can lead to failures in the TIM bondline. Is the issue that the heat pipes are too stiff to bend, or that even when bent the whole GPU/"satellite" portion is still too tall? I prefer to bend them anyway since I want to toss the gap pad & run some high quality TIM on the GPU. I actually work in the consumer electronics industry & got my start as a thermal engineer on products similar to this...so I have access to some pretty fancy materials & stuff (and am also paranoid about having a proper mechanical fit for proper thermal perofrmance).

[iwishiwasinwhistler]

Well, I can't be sure as I haven't actually done what you're thinking of. I wouldn't want to discourage you especially as it seems like you have the tools, ability and vision

I suppose I'm cautious for a few reasons. Even on my intel board the space/height available does look a bit tight. Looking at my old T60p ATI board (same as yours) it looks as if it would be even more difficult as the discrete chips are thicker. There seem to be a few people posting on the German thread who have tried this (admittedly with different ATI GPU's to yours - X1300 and X1400) and it doesn't seem plain sailing. I'll post the google translate results here for reference...... (I hope I'm not breaking any forum rules with this?)

POST 33 by agape:
Tinkers with me now a fan T500 (FRU 45N5492N) in the T60 with ATI X1300. Quite fit so he does not. The additional heat sink,
who sits on the panel for the ATI graphics chip (see photo on the 4th of mb.ing), is quite thick and has now has full-face
Contact to the metal surface of the bottom of the keyboard - advantage is now that the metal side of the keyboard now to better
Heat dissipation contributes - downside is that the keyboard is on the left side slightly elevated, and all shocks (eg by typing)
arising directly to heat sinks and thus graphics chip is passed -> increased risk Flexing?

Temperature evolution is great. RightMark CPU Clock Utility tpfancontrol and I need it since. The fan is despite
the built-in Intel T7600 Officmodus largely - if it starts sometimes with fan level 1, it does not bother at all, because this very
operates quietly and pleasantly. At full capacity over a long period of CPU and graphics card, I never got above 61 degrees Celsius values
with CPU and GPU.

I will now try tonight to remove the heat sink, which is on the sheet in the amount of graphics chips!

POST 124 by ounce:
Experience with ATI X1400 45N5492 report

Today I 45N5492 built into a 15 "T60 with ATI X1400 problem.

The graphics chip has a direct, full-surface contact with the heatsink.

Necessary modifications:
known to the three mini fins on the left air intake remain away (without tools by repeatedly twisting), the long right
to the south bridge (?) chip "south" of the GPU to the heatsink makes contact, use the thick thermal pad from the original fan and applied in two layers on the chip to the heatsink

Deviations from the original part 41V9932:

How do I remove the thread, some posters use the clip of the original fan that pushes down the heat sink on the graphics chip. My conversion waived the original part, because it has no enough space under the keyboard. ME is the junction also not absolutely necessary, since the contact pressure of the screw on the CPU is large enough to provide the contact of the heatsink to the GPU secure. I did this by developing and monitoring the impression that verifies the WLP leaves on the heatsink.

Lazy as I am, but I have to produce their own instead of braces, which depresses the area over the South Bridge (?), Built a double layer of thermal pad on the site. This seems to me to be emotionally belly enough. TPFanControl has so far found no remarkable temperature. But maybe I'll tinker also a brace to replace a layer thermal pad.

.......I know we're bumping up against the limitiations of using an online translator but it does look like it's possible. It looks like ounce was able to get direct contact with the GPU without the use of a thermal pad but had to discard the bracket as it wouldn't fit under the keyboard! That doesn't sound good to me!!

In summary, I reckon the pipes are stiffer but could be bent....bmwman91 it sounds like you've got the skills to do this, but will it all fit in properly under the keyboard?......

Go for it! It's always exciting to be the first to make stuff work.

[bmwman91]

Thanks for the additional info. Once I get back to the US (in China for tooling reviews with my employer's suppliers) the T500 heat sink assembly will be waiting in my mail box. I will take pictures & detailed notes of whatever I find when I try to jam it into my T60p. It may be a couple of weeks actually...I have a custom-built motor coming for my old BMW (my screen name's namesake) that I need to swap in shortly after I get back. T60p mods are fun, but a bullet-proof motor that will rev to 8000RPM all day long in a 2500lb car is more fun! I seem to recall there being a few E30 fans in here...they'll understand haha!

We'll see if my skills are enough for the '92 HSF. If not, I'll go with the '90 integrated-GPU T500 HSF since it still sounds like that one is vastly superior to the T6x ones and is known to fit well. Hopefully bending the heat pipes will kill two birds with one stone...eliminate the gap pad nonsense and allow it all to fit nicely. Judging by how the heat pipes were ground down to clear the top keyboard bezel, I'd say that it might be possible to buy 0.5-1mm on the satellite portion by doing some careful material removal (fancy name for "sand the s**t out of it and hope for the best").

[nandaiyo]

I replaced my T60p fan/heatsink with a T61/T61p 42X4685 42W2028 Fan Assembly from eBay. Put some AC5 and so far under heavy CPU/GPU load the temps are staying at 72/74C with the fan running at about 80% of max speed. At idle, I'm getting around 50C/60C for CPU/GPU temps with the fan silent or very slow speed.

[edit]
After a few days with the T61 fan, I've noticed that the noise level is much higher than the stock T60p fan. And it's more of a whining sound compared to the T60p fan, which doesn't sound as high-pitched. I'd rather have a quiet and warm machine than a loud and cool one. So I've put back the T60p HSF and took apart the T61 fan to see if I could make it quieter. Thanks to a guide posted here on lubricating the T60 fan, I put some mineral oil on the fan axle, and also cleaned up the magnetic plates which had some rust on them. So far, the T61 fan is much quieter.

[Backslashnl1]

So far it's MUCH quieter than the T60p in it's original state (that was with AS5, clean inside and undervolted). The T500 fan is certainly quieter. More so than when this board was in the 14" T61 with original fan. I'm aware that I've changed a lot in one go and we can expect the T9500 Penryn to be cooler than the original T60p board with Core Duo Yonah 2.16.

Bit of a wordy post that one. In conclusion T500 45N5492 works well in FrankenPad 15" T60p case with 14" 4:3 T61 intel planar. The fan rarely comes on. When it does it really isn't distracting and runs at a similar volume compared to my x200s Sl9400. The fan brings temps down fast and then goes off. The idle temp for the T9500 is around 31-33C.

I have a FrankenPad (check my signature for specs) and a T9300 is on the way. Currently the fan regularly turns down. I currently use the T61 integrated fan 42W2820 and the idle temp for my T7500 is around 39-42C. 
Do you (or anyone else) expect that the T500 fan cools even more efficient and quieter than the t61 fan?

[iwishiwasinwhistler]

Maybe it's not clear from my post but yes I think in your case it will be quieter.

My situation was similar to yours. Before I built my Frankenpad I was using the intel planar the same as yours in a complete T61 14" screen unit. It had the integrated fan on it like yours which I replaced with the T500 one when I swapped everything in to the 15" T60p chassis.

I found it quieter and more efficient. YMMV.

[Backslashnl1]

Ok interesting. I thought you compared the noise and cooling efficiency mainly with the T60p fan.
I almost read that whole German topic and it seems the T500 integrated fan (45N5490) is a straight swap in without modification, if you have Intel graphics. Interesting
However, that integrated fan doesn't look very impressive on the pictures

The fan I have now is pretty efficient but audible at 2850 rpm. I guess I'll go on the hunt for a T500 fan for discrete graphics

[dozer]

wow....one of the best tech threads I've ever read here. Posts with -complete- info and details. thanks guys!

Whistler...I was surprised to see your later posts detailing the discrete t500 hsf hitting the kbd, because it seemed to me that your first post in the thread is saying "everything fits great". What am I missing?

I can't live without this 4:3 uxga flexview...short widescreens just don't cut it for the work I do.....so I'm stuck with the T60-15" chassis. A t61-board swap was what I'd -wanted- to do originally, not a t60-intel, but I was told there was no way to run a modem in such a swap. And I -have- to have a modem, because dsl isn't even available here in the hills where we live. It's modem or nuttin'.

Have any of you doing Frankenpads come up with yet, or heard of a way yet, to keep an internal modem? If so....please post the details!

You guys who have studied that highly-tech German forum in depth, (and thanks for the ref. to that one!), did you happen to see any mention there of the 'modem issue'? I mean, of course, in terms of a -solution-...

My ideal config would indeed be a T61-Intel board, and using the integrated t500 fan probably. I doubt I need the 'ultimate' cooling-capacity of the discrete 500 fan. I don't really care about the sata-2 capability of this board, or even the >3gb RAM (altho it'd be nice)...I just want the low 12W power-rate!

Julukaforever....VERY good post...thanks very much for that. The only part I didn't like was the missing link for the finished EC fw image...