ACPI EC Hacking

Hi,

I'm sure some of have already heard of the eeectl program which allow fsb/fan/brightness/cpu voltage control for the EEE netbooks.

While we can already control at some point the fan and brightness on the A1 using the ACPI driver, there's some limitation and on windows requesting data directly from the ACPI EC generate some errors.

It seemed that the EC access with eeectl is done by communicating directly with the EC, without talking to the ACPI driver, thus not generating errors.
This was made possible because the EEE EC chip was identified and its firrmware dumped and dissassembled. This also allowed to find how to control the fsb speed and change the processor voltage.

If anybody with some electronic skills could find which EC is present in the A1 and dump its firmware, I may be able to dissassemble it and try to find those informations.

WIth those infomations, it would be possible to adapt the eeectl program for the A1, and do some other great things maybe.

 

I've looked into this as well... I figured if its an atom processor, how hard would it be to adapt it to overclock the AAO?

Turns out things aren't quite so simple. At least from what I gathered from the source code of the program.

It actually does run through acpi in a way. What I think happens is it uses acpi mods specific for the Eees (not sure who developed them but they're part of the OS so it could be part of the acpi-support package). There is a modification to the OS specific to the Eees that creates a set of links to devices in /proc/devices/eee (or something to that effect) which allow the app to control the device.

In a way it's a driver issue. But you're probably spot on in how to deal with it, I just wish I knew more about such low level programming.

Maybe bushmills can help out somewhat if he sees this thread..

 

I found this thread by searching for fan speed control. I need to turn my fan up! Every time I leave the One in a hot room overnight, it fails bad.

I have an eeepc and the eee kernel module was made possible by the discovery of a PLL chip on the motherboard that generated the frequency for the fan and cpu. There was a multiplier and a divisor setting for the clock and a speed setting for the fan. Someone identified the PLL as having oscillator output pins connected to the CPU and fan. They looked up the PLL and found it was controlled through the I2C bus and had memory that was mapped. Writing to it would change or read what was happening. That's what I remember from the module.

Now if I can find out if this has been done yet to the One, I'd be happy...

 

Actually, what we need to add support for complete & precise fan control to my a1ctl soft (viewtopic.php?f=44&t=6063) is to identify the PLL chip that control the fan.

If anyone has to open his A1, please take some time to take Hi-res pictures of every square/rectangle black chips so we can check the references to find out which one it is.

Then we can find the references docs to see how it's working.

 

EC - Nuvoton (Winbond) WPCE775LA0DG

http://202.39.229.124/hq/enu/ProductAnd ... CE775x.htm


PLL - SLG8SP513v
Datasheet - http://www.silego.com/resources/pdf/xSL ... 132007.pdf

The datasheet wasn't public but the files have similar names and I was able to guess it The service manual says the clock generator can be a ICS9LPRS365 as well but I've only seen SLG. This is all listed in the manual though so I'm sure it doesn't help There are some closeup pics in this persons gallery

http://picasaweb.google.com/FFurcas/AA1#

 

Nice find!

I'll check the docs, thanks for your work

Ow, the PLL pdf link return me an error 404 :/

 

I think they just dont allow hotlinking, if you go to the website then paste the address for the pdf it should work. Uploaded it to rapidshare just incase.

http://rapidshare.com/files/160382694/x ... 7.pdf.html

 

*bump*

any progress made on this? it could defently help to find a way to over clock the one!

 

there is some progress. You can change, now, only the voltage of the cpu while running on 1600mhz to 1.01v, giving about 30mins of battery life, this is using crystal cou id ad its speed step option, while setting some windows settings in power management as well. I think that finding the right hex codes overclocking is also possible,

 

"up"

 

incase anyone is still looking, yes softfsb supports the pll now and can do 100 or 166mhz fsb

 

Any instructions on how to overclock?

 

just google softfsb :lol:

 

What I meant was which pll clock generator do I select as the previously mentioned pll clock generator wasn't on the list.

 

oh sorry, I said it a few posts up,
PLL - SLG8SP513v

 

Which softfsb version are you using? The 1.7final version that i have doesn't have this pll clock generator.
Or you can just transfer the pll data file to me.

 

I gave up on softfsb.
I used SETFSB and it supported SLG8SP513v pll

SETFSB detects the pll, but can't change it. :twisted:

 

Ahhhh I'm sorry, it's SetFSB not softfsb I meant to say!

http://www13.plala.or.jp/setfsb/

 

Here's what I did,
-opened up the program, selected the pll
-either slected or not selected the "ultra" checkbox
-click get FSB
-after making sure the FSB/DDR/AGP/PCI frequencies are all correct, I started to adjust the FSB.

There's 6 steps and the default is 1/6 or 133mhz. When i select 100mhz or 166mhz and click "set fsb" button, it tries to change the fsb, but quickly reverted back to 133mhz. It doesn't work at all. Also the 2nd slide, I assume it's PCI or PCI E frequency is greyed out. The auto button is greyed out as well.

So how did you get it to work?

 

mine didn't work either, plz don't tell me that this is a hoax.
I've tried other Clock Generators, and they either said PLL error, or my machine freezes completely.