T43/T43p Fan Control

[CoolDragon]

A lot posts here are arguing about the T43/T43p fan noise issue. It seems to be pretty chaos, since everyone has a different configuration, room temperature, noise tolerance level......

So I suggest that we should include the following information when talking about this problem.

1. Hardware configuration
2687D5U, 2G Hz, 2G RAM, V3200, Fuji 5400RPM 80G

2. Software configuration
Windows XP SP2, BIOS: 1.23, Embedded Controller: 1.03

3. Power scheme ( Powersource: CPU, FAN, Brightness )
AC: Adaptive, Balance all, 4
DC: Lowest, Balance, 0

4. Using mobmeter, I observed:
Just after resume ( I seldom shutdown, only suspend) for a long time, CPU is below 40C, so the fan is off.
After a while, when the CPU reaches 46, the fan will be on, but at the first level, so barely hearable.
When the CPU arrives 48/49C, the fan reaches second level, and is quite noisy in a quiet room.

5. INTERESTING OBSERVATIONS
1. By accident, I found out that when my CPU is at 49C and fan is at second level, if I switch to "Maximum Performance" scheme which is built in, the fan will down to level 1 for about 5 minutes, while the CPU temp is still at 49!

2. If I use Powerplay to set the video card to be battery optimized while using AC, it can help to reduce the CPU temp down 1-2C, and delaying the triggering of level 2 fan a little bit.

3. It seems that the harddisk temperature might be another source for triggering the fan, though I haven't found any relation, I hope others might be able to report. It seems that the Fuji 5400RPM 80G is pretty hot, mine is around 48C under normal operation, while I saw that the latest 7200RPM is at 40-42 under normal load!

4. Setting the CPU in the power scheme to be "maximum" will not make it stick with the highest freq of your CPU! Strange, but true on my computer.

6. INITIAL CONCLUSION
The fan is all controlled by this "IBM Power Manager" stuff with regard to whatever temperature they get, if we can trace and debug this dude, we might be able to do something. I wish I knew about related knowledge such as ACPI, windows driver debugging etc. So what I can do now is switching power scheme while it gets noisy, this stupid workaround seems to be working!

All in all, this is definitely a FIRMWARE/SOFTWARE bug, they just don't want to fix it since it is not on the critical list.


Added at 11:07AM, Nov 9: This leads me thinking the difference between my power scheme and the "Maximum Performance" scheme, it should be "Fan control option", by setting it to be "Maximize Performance", you are telling it to tolerate higher system temperature. This actually has been reported by somebody before. So I modified my power scheme to be:
AC: Adaptive, Max, 4
DC: Lowest, Max, 0
And now, the fan will remain at level 1 while the CPU temp is at 48! Under normal system load, such as text editing and web browsing, my temperature will not exceed 48C, which means the fan will stay at level 1!

Still, it would be nice to know how to modify this power driver and change its threshold.

Added at 12:33PM, Nov 9: I tested the battery option, and interestingly, for battery scheme, you should set the fan control to be "balance all", and this will tolerate higher temp than setting to "maximize performace". So now my scheme is:
AC: Adaptive, Max, 4
DC: Lowest, Balance all, 0

Added at 6:07PM, Nov 9: Interestingly, with my latest setting, and if I put my laptop on my Port Replicator II, 48C will still trigger the fan to level 2, but it will not without the PRII. It seems that the fan control program is really COMPLICATED!

Added at 6:07PM, Nov 9:Obviously, there is an ambient temp sensor and the Power Manager will take that into account. With my current power scheme, if I am in the lab, which is pretty cool (20C), the fan will not go to level 2 at 49C. But if I am at home, which is pretty warm (25C), the fan will go to level 2 at 49C.

Added at 6:07PM, Nov 13: It is almost for sure that the Power Manager will reset something when you change/re-select a power scheme. My computer now always acts this way: (AC: Adaptive, Max, 4)
- When CPU reaches 46C, the fan will turn on at level 1, which is not even as loud as the harddisk.
- When CPU reaches 49C(Warm room temp)/51C(Cool room temp), the fan will upgrade to level 2, which is quite noisy in a very quiet room.
- When CPU get down to 48C, the fan is still on level 2, but if I reselect the profile, it will go to level 1 immediately!

THOUGHTS
After searching this forum and other BBS, here are some thoughts:

1. Changing the Power Scheme helps only a little bit unless we hack the Power Manager.

2. Using CHC(NHC)/RMClock can reduce the CPU and GPU temperature and hopefully will not trigger the fan to be at level 2 (We all agree that level 1 is just humming and acceptable). But I am quite a conservative person and I don't want to do this undervolting thing.

3. Using Arctic Silver 5 might help reduce 2-4 degrees of the CPU. I am planning on buying some and see if it really helps.

4. Hacking the Power Manager and/or even the BIOS/Embedded Controller. I guess starting from the Power Manager makes more sense since I think most of the "control decision" is made by this dude.

The shortcut to this Power Manager is: C:\WINDOWS\system32\rundll32.exe C:\PROGRA~1\ThinkPad\UTILIT~1\PWRMGR.OCX,StartPowerManager, so diassemble this .OCX file should be a good start.

I found this Neuron PE Disassembler v.1.0 b3 (it is a shareware), and might work.

All sorts of feedbacks and comments are welcome!

[christopher_wolf]

Yup; I have noticed similar things here. One note about the CPU temperature on Battery though. It usually approaches aroun 55 Degrees Celsius for me before the fan comes on...If you are on Mains Power, it will come on much sonner unless you select the "Balance all Parameters" option in the IBM Power Manager. Again, as you said, it does indeed depend on the IBM Power Manager settings.

Here are my settings on Battery:

Display Brightness: Lowest
Maximum CPU Speed: Lowest
Fan Control: Optimize to Balance all Parameters


Basically, take everything down to the lowest level possible; works like a charm for me. Later on, I will post other things that I noticed help out, as per my previous post. This includes any features from CHC/NHC that helps out...Also, HDD noise is at least slightly noticable during the time when the fan is not on, actually turns out to be the first hour I am on Battery. I will try the silent option in CHC/NHC as well as changing the *,ini configurations files as well. I will post as soon as I can with new information.

[ruthlessbrad]

2668-75U, 2 GHz, 1G RAM, X300, Hitachi 7k60 60GB
Windows XP SP2, BIOS: 1.23, Embedded Controller: 1.03
AC: Adaptive, Maximum Performance, 7
DC: Adaptive, Maximum Performance, 5

At work, my CPU temp starts at around 40 degrees and levels off around 45 degress. I think the fan has 3 speeds, and it reaches the top speed after about 20 minutes at 44 degrees. At home, the CPU temp usually starts at around 40 degrees and levels off at about 48 degrees. It reaches the top speed after about 20 minutes, and the fan usually doesn't even kick in until the temp is at 44-ish. It seems the fan speed is time dependent, not temperature dependent.

Anyway, the really weird thing is this... if I hold my hand over the vent to block the fan exhaust, you can hear the fan speed increase immediately. This is due to the fact that the air cannot go anywhere, not because the laptop is increasing the voltage. After holding my hand there for about 5-10 seconds, the fan speed slowly decreases to whatever it was before I put my hand on it. Keep in mind my hand is still covering the fan exhaust. When I remove my hand, the fan speed decreases even further since the air can now exhaust freely. It decreases to a level below the max, I'd say fan speed level 2. However, after a few minutes, it increases back to level 3. I think this proves that the laptop adjusts the RPMs, not the voltage, according to some algorithm (a freakishly screwed up algorithm at that).

I had planned on using a resistor to lower the fan speed, but I think the laptop will just compensate for this until it reaches the desired RPM level.

And why does the fan rev to like level 10 when booting up? I think level 3 is too loud, but when it boots up, it sounds like I'm at a Nascar race!!!

[nrj45]

Same conclusion as all of you. My specs are in the signature.

I made some tests in classrom and found a setting that kept my t43p's fan at level 1 during 4 hours.

Cpu downvolted to 800MHz@0.700Vcore and blocked to this speed. Power manager set to maximize performance (to allow higher temps without the fan to run like a jet engine). This setting will let my firegl run at full speed. So after this setting is applied, i have set powerplay to min perf.

The problem is that if i remove and reput the ac/dc adapter, powerplay will be reset to max perf due to the max perf setting of the power manager.

room temp was about 20°C. Cpu temp stayed at 42°C during 4 hours and fan at level 1.

Finally i have to profiles "perso" and "perso2". one with balanced mode and the other with max perf.

BTW : i don't think hdd temp is really an issue. Mine is at 36°C after 2 hours and the fan is at level 2 for a long time...

[CoolDragon]

Cpu downvolted to 800MHz@0.700Vcore and blocked to this speed. Power manager set to maximize performance (to allow higher temps without the fan to run like a jet engine). This setting will let my firegl run at full speed. So after this setting is applied, i have set powerplay to min perf.

The problem is that if i remove and reput the ac/dc adapter, powerplay will be reset to max perf due to the max perf setting of the power manager.

Yeah, I did this test on my machine, and found out the same thing. The Power Manager setting will override the ATI Powerplay setting!

room temp was about 20°C. Cpu temp stayed at 42°C during 4 hours and fan at level 1.

Wow, the downvolting really reduces the CPU temp that much!

Finally i have to profiles "perso" and "perso2". one with balanced mode and the other with max perf.

Why do you need these two profiles?

BTW : i don't think hdd temp is really an issue. Mine is at 36°C after 2 hours and the fan is at level 2 for a long time...

Seems like you are right, I think it is just an execuse for me to buy the long waited 7200RPM ones.

[nrj45]

...
Why do you need these two profiles?
...

Lol... it's just to have a shorter way to switch the powerplay settings in other situations than when i want my t43p to be the quietest possible

I also used the 2 profiles to be able to compare the difference between balanced mode and max perf.

[fran1451]

Have you ever taken a look at the values the power manager stores for each power profile in the registry? Have a look here:
HKEY_CURRENT_USER\Control Panel\PowerCfg\PowerPolicies

There are two interesting facts:

1. The keys OPTI_FANCONTROL_AC and OPTI_FANCONTROL_DC represent the chosen fan control value - either "Maximize Performance" (value=1) or "balance all" (value=2). But: what about the values 3 and 4? In my case, they are not selectable through the power manager but if you edit them directy in the registry, the energy manager will display the two new values with descriptions like minimize fan noise (value=3) and minimize temperature (value=4). But it seems that it does not have any effects on my computer What does it have for effects on your computer?

2. Does anybody have a clue what the values "RRATE_IDLE_TIMER_AC", "RRATE_IDLE_TIMER_DC", "RRATE_WHEN_IDLE_AC" and "RRATE_WHEN_IDLE_DC" stand for?

[nrj45]

I also tried to modify the registry (to value 3) but i didn't see any change.

[CoolDragon]

1. The keys OPTI_FANCONTROL_AC and OPTI_FANCONTROL_DC represent the chosen fan control value - either "Maximize Performance" (value=1) or "balance all" (value=2). But: what about the values 3 and 4? In my case, they are not selectable through the power manager but if you edit them directy in the registry, the energy manager will display the two new values with descriptions like minimize fan noise (value=3) and minimize temperature (value=4). But it seems that it does not have any effects on my computer What does it have for effects on your computer?

Seems like they have some "hiden" and "unimplemented" feature in this Power Manager. So it makes more sense to hack this stupid Power Manager.

[CoolDragon]

2. Does anybody have a clue what the values "RRATE_IDLE_TIMER_AC", "RRATE_IDLE_TIMER_DC", "RRATE_WHEN_IDLE_AC" and "RRATE_WHEN_IDLE_DC" stand for?

I will guess this means "Rotation Rate"?

[christopher_wolf]

If it was Rotation rate; it would have to be in pretty funky units...why not just use RPM instead?

[vpn-user]

No, hacking the Power Manager makes NO sense! FAN control is done via the Embedded Controller. It must be in hardware because it has to work even when no Windows OS with Power Manager is installed, for example Linux.

The linux FAN control solution works by changing some registers in the Embedded controller. Should work for windows, too if someone codes it.

[CoolDragon]

No, hacking the Power Manager makes NO sense! FAN control is done via the Embedded Controller. It must be in hardware because ist has to work even when no Windows OS with Power Manager is installed, for example Linux.

The linux FAN control solution works by changing some registers in the Embedded controller. Should work for windows, too if someone codes it.

Here is my guess:

The embedded controller should only be responsible for reading the temperature sensors and setting the fan RPM control data. It must have some default value set, in case there is no OS and driver installed. Of course you can change some value in it as long as you know the exact memory address of some control register. And I guess the Linux hack does exactly this.

And the Power Manager Driver should be the bridge between the Embedded Controller and the Power Manager software. The embedded controller interface should be just a block of memory and you can either write raw HEX number such as 0x70, and the Power Manger Driver should wrap the raw write ability and provide an API such as setFanRPM(int RPM).

The policy or inteligent should be in the Power Manager which has some algorithm such as based on all the input information(temperature, power scheme.....) and dynamically determine/call setFanRPM().

[frogla]

I guite agree that it makes no sense.It is no rational explanation to why manipulating between two differently named profiles with the exact same settings in the IBM Power Utility should make the fan speed go down. But after tesing this on two different T43 s for nearly half a year now I have experienced that it works, combining 1) Ibm Power utility and 2) NHC and using the settings as mentioned earlier in this link ..

http://forum.thinkpads.com/viewtopic.ph ... 856#105856

Hope this could be of help..

No, hacking the Power Manager makes NO sense! FAN control is done via the Embedded Controller. It must be in hardware because it has to work even when no Windows OS with Power Manager is installed, for example Linux.

The linux FAN control solution works by changing some registers in the Embedded controller. Should work for windows, too if someone codes it.

[CoolDragon]

I guite agree that it does it makes no sense.It is no rational explanation to why manipulating between two differently named profiles with the exact same settings in the IBM Power Utility should make the fan speed go down. But after tesing this on two different T43 s for nearly half a year now I have experienced that it works, combining 1) Ibm Power utility and 2) NHC and using the settings as mentioned earlier in this link ..

http://forum.thinkpads.com/viewtopic.ph ... 856#105856

Hope this could be of help..

No, hacking the Power Manager makes NO sense! FAN control is done via the Embedded Controller. It must be in hardware because it has to work even when no Windows OS with Power Manager is installed, for example Linux.

The linux FAN control solution works by changing some registers in the Embedded controller. Should work for windows, too if someone codes it.

Yeah, switching between two identical profiles DO lower the fan speed, this is weird ......

But unfortunately, using CHC does not help me lower the CPU temperature, which is really strange, since I set the x6@0.7v and x15@1.132v.

[DIGITALgimpus]

Yeah, switching between two identical profiles DO lower the fan speed, this is weird ......

But unfortunately, using CHC does not help me lower the CPU temperature, which is really strange, since I set the x6@0.7v and x15@1.132v.

Not that wierd. It supports my theory that the problem lies in the software not monitoring correctly. When you switch profiles, you likey re-init the settings to the driver, causing it to do a first check, and adjust accordingly.

[ddutta]

Nothing like that happens on my T43 which is a Sept CTO model (2668AJU).

[ruthlessbrad]

2. Does anybody have a clue what the values "RRATE_IDLE_TIMER_AC", "RRATE_IDLE_TIMER_DC", "RRATE_WHEN_IDLE_AC" and "RRATE_WHEN_IDLE_DC" stand for?

I think RRATE stands for Refresh Rate, since the power manager can adjust the screen refresh rate when idle.

[CoolDragon]

Nothing like that happens on my T43 which is a Sept CTO model (2668AJU).

Can you describe in detail?

[Navck]

Heres irony
I finally hear the fan noise (No, not this)
I went from my adaptive profile to the maximum preformance (IBM provided defaults)
The fan suddenly stops/lowers down a lot.
It stays that way untill it reaches about 47C, then it goes onto a higher level, it stays there and gets louder
I swap back to my adaptive profile (Fan goes WAAAAAY louder than max preformance, but drops temperature to 42C)
Set back to maximum preformance, fan goes away...
Weird, yes?

Irony - Maximum preformance is MORE silent vs adaptive profile.

[frogla]

Here is a " trick " that might be helpful. When changing to "Max.performance" in the IBM Power Utility, do not forget to change to " Dynamic switching" in NHC. Otherwise the default setting in NHC will be " Max.performance " as well , resulting in rising CPU temperature and eventually fan RPM. In other words : You silence the fan for a while , but have to " pay" later on as the temp. goes up and the fan cicks in at a higher pace . Setting the NHC to " Dynamic switcing " seems to prevet this ( but changing fan/temp/CPU parameters in BIOS does not seem to work for me ).

This " trick " will also work if you choose the Windows based power utility in combination with NHC - at least is does for me.

For details about these findings and settings you might want to take a look at this link..

http://forum.thinkpads.com/viewtopic.ph ... 856#105856

Heres irony
I finally hear the fan noise (No, not this)
I went from my adaptive profile to the maximum preformance (IBM provided defaults)
The fan suddenly stops/lowers down a lot.
It stays that way untill it reaches about 47C, then it goes onto a higher level, it stays there and gets louder
I swap back to my adaptive profile (Fan goes WAAAAAY louder than max preformance, but drops temperature to 42C)
Set back to maximum preformance, fan goes away...
Weird, yes?

Irony - Maximum preformance is MORE silent vs adaptive profile.

[frogla]

I have now completely disabled the IBM Power Utility ( by deinstalling with the IBM Software installer ) and I have to admit that the my former settings / profiles made with the IBM Power Utility does NOT work when using only the Windows Power Utility in combination with the NHC .

So having the IBM Power Utility installed seems essensial to be able to manipulate the fan RPM ..

(So when I thought I was manipulating beetween Windows Power Utility based- and "home made" profiles from the IBM Power Utility earlier on the IBM based system seemed to be working "underneath it all" .. )

Here is a " trick " that might be helpful. When changing to "Max.performance" in the IBM Power Utility, do not forget to change to " Dynamic switching" in NHC. Otherwise the default setting in NHC will be " Max.performance " as well , resulting in rising CPU temperature and eventually fan RPM. In other words : You silence the fan for a wile , but have to " pay" later on as the temp. goes up and the fan cicks in at a higher pace . Setting the NHC to " Dynamic switcing " seems to prevet this ( but changing fan/temp/CPU parameters in BIOS does not seem to work for me ).

This " trick " will also work if you choose the Windows based power utility in combination with NHC - at least is does for me.

For details about these findings and settings you might want to take a look at this link..

http://forum.thinkpads.com/viewtopic.ph ... 856#105856

Heres irony
I finally hear the fan noise (No, not this)
I went from my adaptive profile to the maximum preformance (IBM provided defaults)
The fan suddenly stops/lowers down a lot.
It stays that way untill it reaches about 47C, then it goes onto a higher level, it stays there and gets louder
I swap back to my adaptive profile (Fan goes WAAAAAY louder than max preformance, but drops temperature to 42C)
Set back to maximum preformance, fan goes away...
Weird, yes?

Irony - Maximum preformance is MORE silent vs adaptive profile.

[CoolDragon]

I have now completely disabled the IBM Power Utility ( by deinstalling with the IBM Software installer ) and I have to admit that the my former settings / profiles made with the IBM Power Utility does NOT work when using only the Windows Power Utility in combination with the NHC .

So having the IBM Power Utility installed seems essensial to be able to manipulate the fan RPM ..

(So when I thought I was manipulating beetween Windows Power Utility based- and "home made" profiles from the IBM Power Utility earlier on the IBM based system seemed to be working "underneath it all" .. )

Here is a " trick " that might be helpful. When changing to "Max.performance" in the IBM Power Utility, do not forget to change to " Dynamic switching" in NHC. Otherwise the default setting in NHC will be " Max.performance " as well , resulting in rising CPU temperature and eventually fan RPM. In other words : You silence the fan for a wile , but have to " pay" later on as the temp. goes up and the fan cicks in at a higher pace . Setting the NHC to " Dynamic switcing " seems to prevet this ( but changing fan/temp/CPU parameters in BIOS does not seem to work for me ).

This " trick " will also work if you choose the Windows based power utility in combination with NHC - at least is does for me.

For details about these findings and settings you might want to take a look at this link..

http://forum.thinkpads.com/viewtopic.ph ... 856#105856

Which makes it meaningful to understand what exactly this stupid "IBM Power Manger" does!

[vpn-user]

I still think it' s just a nice GUI for "powercfg.exe" with meaningless status bars and nothing more.

[christopher_wolf]

OK, Have you actually *TRIED* it out, vpn? I have used it and it has proven very useful for me...As an exmaple, my T43 has now been on for 2 hours and has only switched the fan on in the past 30 mins. It is very quiet and has remained at that level for quite some time, no jumping around in speed, no sudden bursts of noise; nothing annoying whatsoever.

[nrj45]

christopher : lucky guy

I made all the tweaks makable on a t43p (undervolted to 800Mhz@0.700Vcore, powerplay set to minperf,...)... now my t43p can wait 1/4 hour before switching the fan on (before it was 5 minutes).

For me it's impossible - even in dream - to make my t43p be quiet for 1h30 like you...

room temp 20°C, cpu temp 40°C - 44°C, fan level on 2nd speed (the one that is quite quiet for all places except for the library or classroom)

[christopher_wolf]

Hmmmm, This isn't the only T43 I have seen with this behavior, one of my Profs has one and it does the same thing....I have looked at all the settings I can't find anything else that would make this that different. I have almost the same settings as you do. Next time, I will compare that T43 with mine, I really have no clue as to what the difference could be.

[CoolDragon]

I guess the P in T43p really makes some difference.

But I need to compile programs all the day and running CAD tools all the day, no other choice.

[frogla]

Unluckely I have not been able to make my own profiles in the IBM Power Utility or find the IBM default settings after de-and reinstalling it with the IBM Software installer earlier today
I have done the de-and reinstalltion prosess a number of times now and I am stuck with the default settings / profiles from the Windows Power Utility in the IBM Power Utility diagram!!
The fan is now blowing all the time, and I am not able to manipulate the fan RPM ..
Is it a registry failure
Must I do a full recovery to fix this problem ?
Anybody who can help?

[CoolDragon]

Unluckely I have not been able to make my own profiles in the IBM Power Utility or find the IBM default settings after de-and reinstalling it with the IBM Software installer earlier today
I have done the de-and reinstalltion prosess a number of times now and I am stuck with the default settings / profiles from the Windows Power Utility in the IBM Power Utility diagram!!
The fan is now blowing all the time, and I am not able to manipulate the fan RPM ..
Is it a registery failure
Must I do a full recovery to fix this problem ?
Anybody who can help?

I can send you my exported registry setting, but I cannot upload files here, and I don't see your email address.