Simple fan speed control for the Firebox X750e

stephenw10

Try running WGXepc -t to read the temperature.
If you look at the thread I linked earlier you'll see that the SuperIO chip (which has the sensor) has a habit of erroneously reporting 125C after some time. If you haven't rebooted that box for while it could just be reporting wrong.

Steve

frosty

@stephenw10:

Try running WGXepc -t to read the temperature.
If you look at the thread I linked earlier you'll see that the SuperIO chip (which has the sensor) has a habit of erroneously reporting 125C after some time. If you haven't rebooted that box for while it could just be reporting wrong.

Steve

I checked that earlier and it was the same as the gui/widget and mbmon. Firewall has only been up 3 days but I'll give it a reboot and see what happens.

frosty

Reboot did the trick, thanks for the help. Probably should have tried that first.

Steve Evans

Just spotted the error. In my post I had the following instruction to start the fan control:

# /usr/local/etc/rc.d/fanctrl.sh

This is wrong. It should be:

# /usr/local/etc/rc.d/fanctrl.sh start

You can stop the daemon, and return the fans to full speed with the following:

# /usr/local/etc/rc.d/fanctrl.sh stop

Sorry about that.

Steve

frosty

I actually tried this

# /usr/local/etc/rc.d/fanctrl.sh stop

but didnt notice a change which makes sense because it sets the fans back to full speed where they already where.

Thanks for putting this together, I like it better than manually setting the fan speed.

stephenw10

Any idea what speed the fans average out at when the script is running?
I guess it would be very dependent on ambient temperature and cpu load but it would be interesting to hear any anecdotal evidence.

Steve

Steve Evans

Mine's sat at the minimum as per the constants at the top of the /usr/local/sbin/fanctrld.sh script

target=64
min_temp=40
gain=5
delay=2
min_fan=15

thus:

[2.1-RELEASE][root@pfsense]/root(8): /usr/local/bin/WGXepc -f
Found Firebox X-E
Fanspeed is f

It does rise during activity.

I'm nervous about setting min_fan any lower as I don't really want to stop the fans altogether.

Steve

Steve Evans

I should explain that min_fan setting. There are three fans in the box, two of which primarily cool the CPU, but both those and the third fan also exhaust warm air drawing cooler air over all the other components. I therefore thought it best to keep some airflow even when the CPU was below the target temperature on 64 degC. It doesn't take much airflow to provide noticeable cooling.

Steve

stephenw10

Ah interesting, F is really quite slow.
Does your script use decimal conversion, F=15?

Steve

Steve Evans

The lines that write the value do the conversion thus:

# Adjust the fan speed
/usr/local/bin/WGXepc -f `printf "%x\n" $fan` > /dev/null

The machine isn't under much load most of the time, but even when doing big downloads it still runs quite cool. Accessing the web interface, most notably the RRD Graphs causes a speed up of the fans.

Steve

stephenw10

Ah. I managed to mangle my hex and decimal values a few times when writing the WGXepc program (I'm no programmer!) with 'interesting' results. ::)

I was also concerned about cooling the rest of the box, the PSU seemed to run particularly hot. I measured it using a pyrometer since I'm pretty sure parts of it could kill you. The cooling provisions on the box in general don't seem that great especially if it's in a rack, I'd like to see some vents in the front panel to allow cool air in. I have my boxes set to fan speed 32, none have stopped working yet. It really is a shame they didn't spend the extra pennies required to put in a SuperIO chip that had auto fan control.

Steve

power_matz

Hi Steve,

I installed your scripts and the modified toll from Steve with the temperatue option.
After reboot I noticed that the fans slow down but after some time the temperature reads 35 C and the fans are up at speed FF.

The fans are always running at this speed. I noticed no change here. What is wrong? I also tried the corrected call with the start option

Matthias

power_matz

OK, found the error by myself. The min temperature is set to 40. Mine its 35. So the if-statement is not correct. I set the min temp to 30, now it is running!

Maybe it is a good idea to refactor the if statement…

Matthias

Steve Evans

Hi Matthias,

I saw your first post emailed to my phone and thought that looked like the script had taken the safety bailout. This is there to protect from erroneous temperature readings I'd been warned of, and too high a fan speed in the safe response to that.

Interesting that your box is running so cool. I've never seen mine below 50 degC. By the time its booted to the point the script starts the CPU is normally up to that kind of temperature.

Steve

power_matz

Hi Steve,

my box is in a network cabinet. There is not much space at the back for the air to come out (approx. 5 cm to the rear of the cabinet.
I am wondering myself why it is so cool! Now with your script it runs with speed f at temp 40 C.
It is ok, I think.

Bye
Matthias

stephenw10

What CPUs are you running? Is the cpu sensor touching the bottom 9f the CPU? I can't actually remember if it has a spring loaded sensor or not. :-\

Steve

Steve Evans

Intel(R) Pentium(R) M processor 2.13GHz

I don't have PowerD enabled though. That could well explain it. I should probably enable that as I'm sure the CPU will "wake up" to a higher power state quite quickly on demand.

The temperature diode is part of the CPU die, not an external sensor.

Steve

stephenw10

Ah, in which case the diode values can vary between CPU models so the real temperature may be different.
I seem to remember reading that they are actually an offset from the CPU maximum temperature so you have allow for that. No idea if mbmon is doing that. I don't think it is because I just read out the register in WGXepc. If you read the docs for the superio chip there is quite a bit about configuring it for the correct sensor type.

Steve

power_matz

Hi,

I am using an Intel(R) Pentium(R) M processor 1.70GHz with PowerD enabled.

Matthias

Steve Evans

That explains it then. I'm being power mad again! :D

An interesting comparison for anybody trying to build a lower power router then. I went for the more powerful option as I wanted to be future proofed should my broadband speed increase again, and also as I didn't know if I'd want to be using squid etc when I started.

Steve