Thinkpad Fan Noise Problem: Light at the End of the Tunnel

[vpn-user]

@vpn

fanbeep=0 0

works here. (Did you restart the program)? Also, it's only supposed to turn off the beep when switching fan levels was ok. Errors will still beep.

I have it at that value. Program (even Notebook) restarted. No change, still beeps. And there are no errors shown in the window...

BTW: manual value "64" pumps up my fan to highest speed while "63" is same speed as with "7".

[Shimodax]

I wrote the specs based on my T43, on which speeds between 8 and 63 are the same as 7, but maybe it's different on your R50?

Man, you really put a lot of research into this. These few pages on thinkwiki have been most helpful for me (without them I guess I wouldn't even have started this project), big KUDOS!

Btw, maybe disengaged mode is more like an emergency mode. E..g if it's really hot, fan like hell.

Also, did you have any success with the EC 0x84 register. I tried reading it but the values there are more or less random.

Is there any other useful stuff on the EC? I can't make heads or tails from the DSDT disassembly.


Markus

[Shimodax]

I have it at that value. Program (even Notebook) restarted. No change, still beeps. And there are no errors shown in the window...

Restarting fancontrol should be enough. Do you get results with other values (e.g. 2000 200 a long high beep)?


Markus

[Shimodax]

VPN,

also, are you sure you're running the new version (saying "V0.11" in the title)?


Markus

[Thinkerer]

Btw, maybe disengaged mode is more like an emergency mode. E..g if it's really hot, fan like hell.

The DSDT has critical temperature trip points for the CPU and GPU, and when that's triggered it puts the fan in speed 7. Disengaged mode is never used by the DSDT. It also raises the speed too slowly to be ideal for an emergency. So I have no idea why it's there, but it sure comes in handy for the antipulsing hack -- low speed was unbearable on my T43 due to the annoying pulsing.

Also, did you have any success with the EC 0x84 register. I tried reading it but the values there are more or less random.

Is there any other useful stuff on the EC? I can't make heads or tails from the DSDT disassembly.

There's plenty of good stuff in the EC -- look it up in ibm_acpi.c (from the dedicated project or from Linux 2.6.14 or later). The fan_read() function will show you got to decode EC 0x84.

The only things in the EC that I'm aware of and are not supported by ibm_acpi are the speed settings in EC 0x2F, "secret sensors" at EC 0xC0-0xC2, and the battery charge start threshold at EC 0x24 (see discussion here). I'm pretty desparate to learn how to control the latter under Linux!

[sebmue]

@vpn
If you're in an experimental mood, try 64 in the manual field (it should give a Set fan control to 0x40).

Ok, it works. But it is really high and I get a little bit afraid for my TP when using it Can I use this value in the ini-file as an additional fan speed (e.g. for more than 80°C?).

@Thinkerer:
for me 63 is the same as 7 is the same as 3...

[Shimodax]

Sebume,

currently you can't use it as a fan level, but I'll add that in the next release. Maybe I'll also add switching to BIOS mode at a certain temperature (under the assumption that we should not mess with the fan if the system is really hot).


Markus

[DIGITALgimpus]

I really hope this discovers something.

Fan control isn't the answer. Finding out why it's triggering when it shouldn't is the key.

[Shimodax]

inding out why it's triggering when it shouldn't is the key.

The question about the "should" and "should not" may be quite subjective, especially when talking to a technical engineer.


Markus

[DIGITALgimpus]

My only beef with relying on a program like this is that the fan is running for a reason. Altering it may not be such a wise idea.

I'd rather a real fix.

Thanks for the hard work btw. This doesn't mean I don't appreciate it.

[christopher_wolf]

Good Point Shimodax; This strange fan behavior is dependent on quite a few variables, many of them are hard to measure quantitatively (e.g. User perceptions, system build, versions of drivers, other things in Windows that are undocumented but messing around with things that are also undocumented, etc) The question of "Why it is doing what it shouldn't?" is rather open ended and doesn't give the full picture.

[Thinkerer]

Actually, we understand the basic problem quite well: it's a combination of inadequate thermal design that makes some components heat up, and too-conservative thresholds in the embedded controller firmware that make it step up the fan too soon and step it down too late. What we don't know is exactly which components are heating up and the precise decision algorithm used by the embedded controller; but neither of these are under our control anyway.

For working around the problem, it would be useful to know the sensor locations and thermal relevant thermal characteristics and specs. But if you want to "really solve the problem" without manually controlling the fan, the only information you need is IBM's phone number, since no one else can replace the embedded controller software.

A key point to note is that the laptop is not getting very hot compared to similar models; most of the problem is just the very conservative thresholds in the embedded controller. In their marketing material, IBM/Lenovo is bragging about the low temperature of the palmrest. Well, you get the picture: marketing tells management to tell engineering to make the palmrest cooler, engineering lowers the thresholds and can't hear a difference in its noisy lab environment, users suffer but can't get front-line support to get marketing to get management to get engineering to issue a firmware update. Good thing they left in the manual override.

[Aristotle11]

Anyone know what temps are OK to have without the fan running? If the T42 is OK at 70 degrees with the fan on 7, what is wrong with leaving the fan off until it reaches 68 degrees?

Thanks,
Aris

[DIGITALgimpus]

Anyone know what temps are OK to have without the fan running? If the T42 is OK at 70 degrees with the fan on 7, what is wrong with leaving the fan off until it reaches 68 degrees?

Thanks,
Aris

This is exactly my point. This is why a real IBM fix is ultimately necessary. Because if you nuke your laptop, it's your problem. If IBM fixes it, and it causes problems, warranty covers that.

[Thinkerer]

This is exactly my point. This is why a real IBM fix is ultimately necessary. Because if you nuke your laptop, it's your problem. If IBM fixes it, and it causes problems, warranty covers that.

IBM probably can't tell whether you used manual fan control or not, since a reboot erases all traces. But please read my (updated) post above - the only fix that IBM can issue (for existing units) is to increase the thresholds in the embedded controller firmware but this would make the laptop more likely to fry.

[xaverin]

it's a combination of inadequate thermal design that makes some components heat up, and too-conservative thresholds in the embedded controller firmware that make it step up the fan too soon and step it down too late. What we don't know is exactly which components are heating up and the precise decision algorithm used by the embedded controller; but neither of these are under our control anyway.

Well we can only hope that someone who likes to play around with his hardware and has an Infrared Thermometer will one day run his TP opened - so to speak without Keayboard etc. and measure temperatures for a while. The we would really know which parts cause how many heat under normal circumstances and probably select better thresholds for the Fan Control Utility

[vpn-user]

I have it at that value. Program (even Notebook) restarted. No change, still beeps. And there are no errors shown in the window...

Restarting fancontrol should be enough. Do you get results with other values (e.g. 2000 200 a long high beep)?


Markus

I really have v0.11 in the window title, but the beeps are still there even when set to "0 0". Changing these values does not seem to change anything? Even high values don' t change something. Maybe wrong build or something?

[thinktablet]

[Shimodax]

Digital,

no problem with what you say. I understand your point up to a point and it may boil down to subjectives or personal temperament. The disclaimers for my program are there for a reason.

In building a system (any system, be it hardware, software, a house, a car, a boat) there are many conflicting design goals. Some may have to do with price, others are weight vs speed, in a laptop you want it to be small but have a large display, etc. Every system in a way is a compromise. The different ways to find a compromise are called Series (A, G, T, ...) in the Thinkpad world.

In nearly every computer system you have to find a solution to the conflict between cooling, noise, weight and price. Ideally you'd use water cooling (silent (unless you use a cheap pump) and cool) but pricey and not portable.

IBM has drawn the line somewhere. Admittedly their fan isn't too loud and someone somewhere at some time has made a decision about the fan triggering levels. From what I hear they are even the same in very different models (but I'm not sure). It is sure the same on one model, no matter what parts you put in.

I think I know large organizations well enough to have an idea how this works. You make the decision, nobody compains (or just a few ... someone always complains about something) and the decision lives. At some point it will be very very hard to change it. If you change it somebody will have to take responsibility. Even if you have 100 complaints now about fan noise, a manager will ask an engeneer (or a group) is it safe to change it? Will it cause more returns? Will users compain about hot palmrest or will we be dragged through the press because someone burnt his balls? (There really was a case in the press a few years back about a user who had worked with a laptop in his lap for a few hours and burnt his balls because exposure to temperatures over 43°C may cause level 2 burns.) The answer most likely will be: Nah, why change it, don't mess with a running system.

Users are told: "Works as designed, we have our reasons, etc." The first is certainly true, it does work as designed. The reasons may be there, but nobody will tell you what they are.

Well, I mentioned temperament. If I had had a 30 days return option (I live in Gemany and bought the system through a dealer) I'd simply had given it back without reason. Would have gone on with my old T23 and probably would have tried a T50 in the future or would have waited for the new Apple notebooks. Since returning was not an option and selling it again would have lost me a couple of hundred Euros I checked what else is an option.

I have seen and verified two sensors who kick the fan on in my system (CPU and C1 at 50°C and 43°C). Nobody prevents you from running a server on a laptop or play ego shooters 8 hours a day, in both cases you would see permanently higher temperatures with the standard BIOS fan control on.

So I drew my conclusions. The patch is far from perfect for various reasons, but for me it's the only way to use this machine. I have done worse things to my car and other devices in the house ... but that's purely a matter of temperament. I'm not daring enough to open the case and put a self made copper bridge inside, but on the other hand I value silence more than playing by the rules.

As said, I can understand everybody who decides otherwise.


Markus

[Shimodax]

@thinktablet

Extreemely keewl



Markus

[Shimodax]

the only fix that IBM can issue (for existing units) is to increase the thresholds in the embedded controller firmware but this would make the laptop more likely to fry.

Well, playing a game or running a 3d renderer or fractal program will makes the laptop more likely to fry. I'm pretty sure with these you can easily achieve permanent termperatures of over 55°C with regular bios fan control.

Otoh my machine here, the way I use it, runls permanently lower than 55°C on all sensors with the fan off (in fact 46°C to 52°C).

Just food for thought.


Markus

[Shimodax]

I really have v0.11 in the window title, but the beeps are still there even when set to "0 0". Changing these values does not seem to change anything? Even high values don' t change something. Maybe wrong build or something?

Darn ... I'll post a 012 later anyway, please try with that one again.



Markus

[Thinkerer]

Well, playing a game or running a 3d renderer or fractal program will makes the laptop more likely to fry. I'm pretty sure with these you can easily achieve permanent termperatures of over 55°C with regular bios fan control.

Otoh my machine here, the way I use it, runls permanently lower than 55°C on all sensors with the fan off (in fact 46°C to 52°C).

Right. And I get roughly the same temps here.

But the thing about moderately high temperatures is that they don't instantly fry the electronics - they just induce increased wear (due electromigration, oxide breakdown, thermal paste dryup, evaporation of volatiles from plastic, etc.). So even if 55°C at idle with fan off has the same effect as 55°C running game with fan on (and this is not obvious since the thermal distribution is different), both are worse than the "typical" user who runs office apps with fan on at 45°C most of the time. And it's quite plausible that IBM speced the expected lifetime of the laptop for the "typical" user, because economically it pays to have a few heavy gamers' laptop replaced under warranty if it makes the model cheaper/lighter/faster. So by running under "heavy gamer" conditions all the time, you may find yourself in that unfortunate group.

That said, yes, 55°C should be quite OK and within specs for everything except the HDD and battery. But bear in mind that when a sensor says 55°C, the core of the component it's supposed to monitor might be at 65°C, and that's becoming unfomfortably warm. We don't know.

So my impression is that running the laptop with fan control for a few hours a day should have no noticable effect on reliability; but if you need to run it 24H/day in a quiet environment, you may have bought the wrong laptop.

[Shimodax]

Well, it's a probability game. I don't mind being in the hevay gamer group. I would not mind running fractals overnight on the machine for weeks and I wouldn't mind letting it run as a trading station 24hours a day. Otherwiise I'd probably never unpack it from the box or take it anywhere (less use is always better than more use).

The way I see it is that as long as I'm not using it under contitions that can also occur otherwise, I'm not seeing much of a problem. And heck, it's an IBM, not some cheapish no name stuff. ... you can throw a Thinkpad out of the window and it will still work (all the way down at least *grin*).



Markus

[Thinkerer]

Yeah, these ThinkPads sure are engineered very conservatively. We're talking about a laptop that sets the Pentium M core voltage to slightly higher than Intel's highest voltage profile, when it would work just as well at 75% of that volage (hence half the power draw) - just to be on the safe side. But personally, when it comes to computing fractals all night long, I rather do that on a machine that can be fixed with a $50 off-the-shelf part and a screwdriver. That's probably a bit silly, when I'm likely to drop and break the laptop as I carry it around way before something electromigrates; oh well...

[Shimodax]

@VPN

Btw, what does the log window say about the fanbeep value? Does it reed it from the file correctly? Does it read other values from the file correctly? (E.g. if you change Cycle or temp. levels)?

[28.11.2005 23:13:05] Active= 1, Cycle= 5, FanBeep= 1200 30



Markus

[meditate2001]

Shimodax, you are the hero of the day !

your program is a real ...solution for a small planet....



thanks a lot

[Jmmmmm]

Shimodax, you know the third temp is the hd temperature, right? CHC reports my hd to be at 36 degrees, which is what the third temp in your program is. The second temp is too high to be the hd (if chc is correct).

[nrj45]

...But the thing about moderately high temperatures is that they don't instantly fry the electronics - they just induce increased wear (due electromigration, oxide breakdown, thermal paste dryup, evaporation of volatiles from plastic, etc.). So even if 55°C at idle with fan off has the same effect as 55°C running game with fan on (and this is not obvious since the thermal distribution is different), both are worse than the "typical" user who runs office apps with fan on at 45°C most of the time. And it's quite plausible that IBM speced the expected lifetime of the laptop for the "typical" user, because economically it pays to have a few heavy gamers' laptop replaced under warranty if it makes the model cheaper/lighter/faster. So by running under "heavy gamer" conditions all the time, you may find yourself in that unfortunate group.

That said, yes, 55°C should be quite OK and within specs for everything except the HDD and battery. But bear in mind that when a sensor says 55°C, the core of the component it's supposed to monitor might be at 65°C, and that's becoming unfomfortably warm. We don't know.

So my impression is that running the laptop with fan control for a few hours a day should have no noticable effect on reliability; but if you need to run it 24H/day in a quiet environment, you may have bought the wrong laptop.

I think the compromise is to make the fan begin to fan later (i made mine to 55°C) but make it fan at its full speed earlier. So if you switch between game and desktop use as i do, there will be less temperature difference...

Level#0= 50°C -> Fan 0
Level#1= 55°C -> Fan 1
Level#2= 60°C -> Fan 3
Level#3= 65°C -> Fan 7

[Shimodax]

@Jmm

The second on is not the HDD, that was a misunderstanding on my part. It's a temp sensor from the Active HD Protection System, not the harddisk itself (I'll change that in the next release).


@NRJ

I'm using similar settings. Also, I think it makes sense to let the fan kick in at a high level right at the start. That way it will blow for a minute or two and bring the temperature back immediately and switch off again. (Level 1 doesn't have much cooling effect here).

Also, the next version will have an option to switch to BIOS at a certain temperature. That way you can let the system scheme take over if the whole thing is getting too hot.


Markus