Fanless Aspire One

My AA1 is ZG5 with 8GB SSD disk. I am using BIOS v0.3109 but I will right now change to v0.3114 and I will report any problem.

And about

Looks like the situation when we unconnected the fan, the system works for some time and after turns off.

And anyone knows some place were we can execute that script at boot time?

 

What is the average temperature you people get?
It seems I get around 48-52 degrees. (Not really depended on the speed off the fan).

edit, if I turn it off with "1F" it slowly rises haven't let it go above 60 degrees.

 

The problem with the flickering might've had another cause as i've been rebuilding the video driver, i forgot about that. So maybe false alarm.

 

According to another thread here, the flickering is caused by something else.

Mine was flickering last night. A few places I've read theorize that it has something to do with running your laptop at the lowest brighness setting while under battery power. When mine was flickering, that's how it was running too.

 

Hang on, I've been out at work stuff half the day, I need to get my bearings.

Looks good, I'll have a go with this later this evening.

 

I think the assumption is that the value is the temperature in celsius, it's just in hex.

For reference; before I turned the fan off the temperature recorded was 0x2e (46C), and has been rising slowly ever since; it's currenly 0x32 (50).

Oops, now 0x33

(btw, for convenience:

Code:

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

watch is nice at times like this.

Oh but the blissful silence.

I wouldn't have thought a 16 bit value would be needed either for fan speed or for temperature - we just don't need that much range. I wouldn't expect the fan to have more than 8 speed settings, and probably doesn't have that many.

I'll probably play in a bit to see if there are any possible settings between off and its normal idling speed.

now 0x34 (52)...

I wonder, does anyone know how hot is too hot?

Now 55 C...

Waiting to see where it levels off while idling before I start trying to stress it.

edit:

Mine seems to be stabilising at 55C for now, but I was being kind and elevating the underside from a soft (furniture) base. Now removing that support for a more realistic scenario and seeing what happens...

I've updated the wiki page, adding the above command and doing a quick cleanup and english-correction and code-formatting pass.

For reference, the intel datasheet at http://download.intel.com/design/processor/datashts/319977.pdf seems to be saying that the normal operating temperature range is between 0C and 85C, with a catastrophic thermal protection cutting in at 125C. We basically don't want to be exceeding 85C for long if we can help it. At the moment I'm 30 degrees below that.

 

I don't think it's possible to kill it. I found this from the official specification.

If Intel Thermal Monitor automatic mode is disabled, the processor will be operating
out of specification. Regardless of enabling the automatic or on-demand modes, in the
event of a catastrophic cooling failure, the processor will automatically shut down
when the silicon has reached a temperature of approximately 125°C.

Otherwise it will begin to throttle its speed at 90°C, but i don't know if it's enabled in the BIOS.

Also quite interesting is this.
The digital thermal sensor (DTS) accuracy is in the order of -5°C ~ +10°C around
90°C; it deteriorates to ±10°C at 50°C. The DTS temperature reading saturates at
some temperature below 50°C. Any DTS reading below 50°C should be considered to
indicate only a temperature below 50°C and not a specific temperature.

The question is it reading the internal diode, or maybe some other thermal sensor on the board.

The official specs for the chipset are here.

Hot Temperature Trip Point
This trip point is set at the temperature at which the MCH must start throttling.

Catastrophic Trip Point
This trip point is set at the temperature at which the (G)MCH must be shut down
immediately without any software support.

Thot = 110 °C ± 5°C(Tdie,max + 5°C ± Taccuracy)
TCatastrophic = 132°C ± 5°C (Tdie,max + 27°C ± Taccuracy)
Tdie,max = 105°C

 

I've went back to the store and got my unit replaced. (the irritated customer routine works everything).
But the noise still exists, slightly less irritating, but it still remains.

So lets hope this tweaking results in improvements.

 

I think we can knock together a very simple shell script that can do basic thermal management.

Let's say, set the fan back to auto if it the temperature exceeds 80C, and set it back to off if it drops below. Possibly some sophistication to prevent it oscillating between the two stupidly when it's around that boundary. That can just be launched as root during boot and have a simple 5 second sleep loop. That'll do us until someone makes something prettier.

 

The thing is, if it is the thermal sensor inside the atom, 80 degrees is a good toggle point, is it one placed on the outside the atom, 80 might be to hot.

Any ideas on how to figure this out?

 

That sounds awfully high. The CPU can handle it, but what about the parts around it?

 

exactly, thats why we need the location of the sensor.. normally the sensor directly in the core is 10-15 degrees higher then the one outside.

example output of a core 2 duo (1 core under 100% load other not)

outside sensor:
CPU Temp: +42.0°C
on the core:
Core 0: +52.0°C
Core 1: +58.0°C

 

NB: My earlier one-line temperature monitor had a bug, and would only repeatedly show the temperature as it was at the time you issued the command!

The good news is that streaming iplayer for a few minutes only pushed the temperature up to 62-64C. It's idling now at 59 C again.

The corrected one-line monitor:

Code:

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

Note also that for the hex-to-dec conversion to work, (ie: $[$HEXVALUE]), /bin/sh must be bash. This isn't the default in Ubuntu (it's a symlink to /bin/dash). I don't know if it is in Linpus. To correct it, as root:

Code:

ln -sfn /bin/bash /bin/sh

Is a good idea anyway; I always do this to my Ubuntu boxen; I just hadn't got around to it on this one yet.

 

the auto mode (0x02?) won't work when i try to re-enable it. only 0x1f and 0x20 seem to have any function.. and the 0x20 is, well, really noisy : )

 

I thought it was 0x00 to return it to auto?

I haven't tried it yet (haven't seen the need ).

Minor expected observation. It is a tad warmer to the touch now! Like any laptop you wouldn't want to rest it on bare skin for any length of time. But quiet makes it usable even if hot. before it was too noisy to be of use, so it's a win.

edit: yes, 0x00 seems to work; ie:

Code:

./acer_ec.pl 0x55 := 00

Fan came on fast and temp dropped five degrees in about twenty seconds, then the fan dropped to a lower speed, so it looks auto to me. Watching to see where the temperature stabilises.

It's all manual at the moment, but I don't care, I have a usable system!

 

found out some interesting things..

when setting 0x00, the register is automatically filled by 0x01

when high nibble is 1, low nibble stays.. otherwise it decreases automatically.. seems to be something like a mode.

Code:

0x1f     fan off
0x1e     fan high
0x1d     fan low

ok.. setting 0x20.. decrease by one -> 0x1f, fan slowly switch down and finally is off

ranges 0x40 - 0x70 seems to allow one to set more then just high, low and off.. but slows down within a few seconds

seems interesting..

 

I am now with BIOS v0.3114 and all this works as on BIOS v0.3109.

Thanks Rachel for the update on wiki - I am Portuguese, with "bad english".

It's important to do that!!! And about that thing you are asking, the hysteresis, is very simple to implement -- you turn on fan at 80ºC and turn it of at 70ºC, so you have a hysteresis made by the real physical system (you can use the own EC registers to found If fan cooling system is turned off or automatically).

Rachel, do you know how can we have the fan turned off at each boot time?

About that wiki page, I think that at first, that text about make the hack working, should be carefully explained (can you work on it?), maybe with a way to auto do it on boot time and finally with that possible script...

 

hello
with fan off, what are the temperatures while browsing, reading pdf's or watching movie?
for example will the thing melt if i put it on bed and play movie?
when will intel come up with chipset that's appropriate for the atom?

 

60°C right now, with a spike at 63°C. The laptop's been running with the fan turned off for a couple of hours. I've been browsing the web and watching a couple of videos on YouTube. You can feel the heat on the keyboard, but then again my previous laptops (with the fan on) would also get somewhat hot. I'm cautiously monitoring the CPU temperature for now, but it looks OK. One can get used to the constant buzzing of the fan, but once you turn it off, you realize what you've been putting up with the whole time.

What I'm worried about is the other components such as the IGP chipset… Also the ambient temperature here has dropped in the last couple of days (about 25°C); I wonder how hot the AAO and CPU would get if the temperature rose by 5 degrees. We'll see…

 

I'm certain it's better than my portuguese. "por favor" is literally all I know, and I've no idea where I got it from (probably a book; I've never holidayed there so not a phrasebook...)

Curiously, without reading anything about such stuff that was exactly what I was thinking, except that, having watched the temperature on mine this evening as I've used it, I'd probably set the fan-on threshold at 70 and fan-off at 60. Basically mine settles down in the low 60s and I haven't even seen it hit 70 yet.

It should be possible to knock a simple one up in a shell script, and I may do so tomorrow. Too tired tonight. I already have now a few trivial scripts as convenience wrappers for acer_ec.pl: temptop, acerfan [auto|off] (nothing just shows current state). I'm reckoning on adding a "govern" (or some better action name) to acerfan that implements the above and can be run as a daemon or in foreground.

The nice long-term fix may actually need to be to put the ec stuff into a kernel module that can supply sensors and controls to the higher-level tools to manage some stuff that's already well established. But that would be beyond me.

In ubuntu I'd currently put "acerfan off" in /etc/rc.local, although I haven't done it yet. I expect there's a similar thing in Linpus. When the "govern" thing is there I'll launch as a daemon in the same place.

Just now found how to set the default cpufreq governor to powersave too. Should help but that's a general thing. Although having done so the temperature now seems to be settling back down to about 59-60 even though I'm here using Firefox... Also just resolved a few Ubuntu issues with waking from suspend. It's all happening today.

I think that can be done a bit later; it's still very much a work in progress. Some of us are still actively probing what those ec registers do, for instance.

Yes, my observation is also that the unit does get *hot* when run without the fan. Not too hot to use, but... you won't need gloves in winter!

This was always going to be the case. There's a reason Acer put the fan there, of course, and we're interfering with that. If it hadn't been so noisy we wouldn't have. So it is a trade-off. For me it's still a clear win, as before we had a solution it was basically useless to me. I couldn't use it at *all* in the contexts for which I wanted it. Now I can. Yes, it might have a shorter life, but it's better than none. And they're not expensive. (And the state of the art will be advancing in this area quite quickly so, say, within a year we could be looking for an excuse to upgrade!)

Mine in particular has had a hard few days; opened multiple times, and mucking with the fan and heatsink assembly. It's entirely possible that the heatsink interface pads have been compromised by all that, so mine could run hotter and have a shorter life anyway. Or not; my resting temperature seems to be in line with yours. It does matter what it's on, of course. Right this minute it's propped up at about 30 degrees from the vertical against my legs, which are drawn up as I lie on the sofa typing (much better for this than the big heavy macbook!), and that's better for heat loss that flat on the sofa itself.