Fanless Aspire One

My AAO is the same model as yours and I also can hardly hear the noise from my fan.

 

Well, you guys with the silent fans are luckier than others My fan was making such a loud noise other people in the room were asking where the whining sound was coming from. After instituting the changes by Rachel et al. my fan has only come on for about 10 minutes in 5 hours with FANOFF=50 and FANAUTO=60. The temperature is normally around 55, and feels only marginally hotter to me than when the fan used to be running all the time.

 

Yes, the fan on AA1 is rather badly designed from a thermal point of view. It moves very little air during normal operation, yet it produces an annoying whining sound signature. The only thing the fan does properly is prevent the unit from overheating when the system is under heavy load.

 

how do you run the temptop and acerfan information programs to see the current temp of processor (linux newb)

 

I've updated acerfand to work with bios 3114 and 3304.

In general, it now attempts to detect the bios version and uses acer_ec.pl values accordingly, thanks to those who teased out which values to use.

To do this it depends on 'dmidecode'. If it's not present it'll assume bios 3114 so as to continue working for people who are using it already who don't have dmidecode installed. (It may not be there by default on Linpus? Can someone confirm?)

This also means that we'll need to do an update each time a new bios version comes out so that it's recognised. Slightly inadvertently on my part it's possible to override the detected bios version in /etc/acerfand.conf, to get instant support for a new bios if the values needed haven't changed. Just add into /etc/acerfand

BIOS_VERSION="v0.3304"

for instance, to see if it still works on the next version. In a day or so after a new version I'm likely to have put up a new version of acerfand that recognises it anyway.

Same locations as before: http://www.electronpusher.org/~rachel/acerfand or http://home.strangenoises.org/~rachel/aspireone/acerfand/. (First url just points directly to the script; second one has some other stuff like temptop.)

 

Hi!

Yesterday I poked around in the EC-register 0x55 (using the acer_ec.pl script / BIOS V0.3304) and found out something interesting:

0x00 -> fan auto
0x1d -> fan off
0x10 -> fan slow (maybe temperature controlled, albeit slower than 0x00)
0x1f -> fan medium
0x1e -> fan fast

I tried all values up to 0x40 and 0x50, 0x60, ..., 0xf0, but I found nothing really usable. I.e.:
0x11-0x13 -> fan gets slower and slower until it is off, stays off
0x21, 0x40 -> fan turns on, then off

Acer Extensa 5220 notebooks have a register, which is different from the speed setting register, where you can turn off automatic fan control. I hope this EC is no different is this respect, but I did not find anything like this so far.

 

any info on how to find out information on the linux fan system like tempreture please

 

Hi!

@AspireOneChris
I'm not sure what you mean. You can find the temperature (in °C) which the EC probably uses to control the fan in register 0x58.

The temperature seems to correlate 100% with the Intel Atom's internal temperature diode. Somewhere I've read, that the TJmax of the Intel Atom Processor is 95°C, which is 10°C lower than what register 0x58 shows. I think Acer uses +10°C bias because the measurements of the internal diode are inaccurate (+10°C / -5°C in the worst case at temperatures >50°C) and they wanted to be on the save side. Values under 50°C mean only that the temperature is somewhere under 50°C.

 

Thanks for replying i just thought there would be something you could run to retrive fan information, such as current status and tempreture.

 

Hi!

I have a few kernel patches for Fedora 9:
* Intel Atom Core Temperature with TJmax=95°C / 10°C less than Acer's EC temperature
* Fan control with BIOS v0.3304. The patch adds some files in /sys/devices/platform/acer-wmi
* idle=nomwait kernel option. It seems C3 Mode does not get entered otherwise -- at least powertop shows this.

If you want to compile your own Fedora 9 Kernel (acerfanapplet doesn't work otherwise): http://fedoraproject.org/wiki/Docs/CustomKernel

Also included is a small python gnome-applet, which currently supports manual control. Automatic control would be easy to add but a (dbus)-daemon would be better in case the applet crashes.

Installation instructions:

Code:

# cp acerfanapplet.py /usr/local/bin
# cd /usr/local/bin # Important!
# chmod +x acerfanapplet.py
# ./acerfanapplet.py -p > /usr/lib/bonobo/servers/GNOME_AcerFanControlApplet.server

Restart the panel, then add the applet to the panel.

Bonus: If you own an Acer Extensa 5220 notebook then fan control is supported, too.

 

This was posted on the forum earlier I believe:

sudo watch -n 2 'echo Temp: $[$(./acer_ec.pl ?= 58 | cut -f 3 -d" ")] C'

You need to cd to the directory where your acer_ec.pl script is and copy and paste that entire line into the terminal.

This is assuming that your AA1's BIOS is covered by the acer_ec.pl script you are running then you should see it. If you updated your BIOS since you downloaded the acer_ec.pl script you will need to get a new version of the script.

 

Hi Rachel,

I haven't deliberately installed dmidecode, so unless it got installed with some other package it looks like it is there on the default Linpus AA1 installation.

Bernard

 

Hi!

The problem with acer_ec.pl and other programs that access the embedded controller (ec) directly is, that the ec doesn't get locked. Direct access might work most of the time, when the ec doesn't get accessed by the operating system, but sometimes it mixes things up. You will probably get an error in the system log. The windows tool has the same flaw. It might be easier this way and is good for testing, but it is not the best solution. A kernel module or Windows driver could properly lock the ec for access.

Maybe somebody should port my patch for Fedora 9 to Linpus (should be easy) and recompile the kernel, or just patch the acer-wmi module and recompile just the module. A patch for the coretemp module (think of sensors / sensors-applet) is included, too.

 

Setting the register to "af" seems to work as well as "1f" on my AA1 with a 3109 BIOS, maybe someone could check with a 3114 BIOS, but it most likely will work for that if it works for 3109 and 3304, so is detecting the BIOS version necessary?

 

Any reason why the kernel patch checks the DMI for exactly Aspire One A150? Won't these patches apply to to A110 too?

 

Actually it was already shown earlier in this thread that the EC doesn't read the internal diode but a probe which is probably located between the chipset and the cpu.

TJmax for the N270 is 90°C.

 

@tolyen
I do not know and I did not want to mess things up for other Aspire One users. Do both models have the same BIOS?

@macles
Sorry, I did not read the whole thread. Seems to me like a lot of duplicated effort on my side.

During my tests temperature difference was always +10°C, but I did not test at 70°C. Yes, the Intel Atom datasheet says 90°C. Any idea why the kernel patch uses 95°C? I thought the devels of coretemp knew what they were doing

Maybe the acer-wmi module should be rewritten to use hwmon. Then temperature, fan rpm, ... could be read with libsensors and other tools would work without changes.

CPU temperature seems to be no issue, maybe a heat sink put on the northbridge could help heat transfer?

Has someone tried to add a filter to smooth the pwm output a bit?

 

The patch is for an Intel Atom processor, but there are a few different models. And 90° C is just for the one used in the AA1.

It is next to the CPU, and both are under the same heatsink. All the information about it is somewhere in this thread.

 

Hi!

I thought about something like this: http://www.acer-userforum.de/thread.php?threadid=5644

The poster writes that it reduces the EC measured temperature about 10°C.

Or perhaps a real heat sink like one of these:

 

Hi guys,

i've a question: what kind of noise your fan is doing? Is it only a loud blowing out of air or (like my aao do) a grunt / vibration? My aao sounds like a bee (a little less loud). Is that a technical problem or an acer-norm?

Thanks for answering.

MrM