Discussion

LMSensors compatible A20 temperature sensor

20 6918
Edited by destroyedlolo at Sun Dec 7, 2014 23:40

Hi,

While googling, I discovered this code which expose A20's temperature thru Linux kernel.
More, it seems, it can be handled by lmsensors, yeeahhhaaaa

It is not part of LeMaker's 3.4.90, is it planned to be added in next release ?


The standard Linux program 'sensors' shows:

axp20_mfd-i2c-0-34
Adapter: sunxi-i2c.0
temp1:        +61.8°C  (low  = -144.7°C, high = +264.8°C)

temp1 fluctuates a few degrees from one measurement to the next, but can be used to used to detect overtemperature. I have currently set it at 80°C to poweroff my system.

Asking for plans: I doubt that anybody does know how this project will evolve

Edited by destroyedlolo at Tue Dec 9, 2014 17:07
LinAdmin replied at Tue Dec 9, 2014 10:51
Asking for plans: I doubt that anybody does know how this project will evolve


Which project ? LeMaker's  or this probe integration in the mainline kernel ?

tkaiser  
Edited by tkaiser at Wed Dec 10, 2014 05:48
LinAdmin replied at Tue Dec 9, 2014 03:51
temp1:        +61.8°C  (low  = -144.7°C, high = +264.8°C)


Do you trust these values? Really? What happens if you carefully press a finger on the A20's surface? What if you touch the AXP209? 60° is way too high.

BTW: "temp1" is AXP209. Laurent is speaking about the A20.

tkaiser replied at Wed Dec 10, 2014 05:47
Do you trust these values? Really? What happens if you carefully press a finger on the A20's surfa ...

When measuring right after switching power on then it shows room temperature.

Those 60°C occur when btrfs scrub is heavily loading the system during 6 hours.
The temperature of the disk then also goes up to about the same values.

This could only be avoided by better air circulation in the plastic box, but I was too lazy to do that and most probably 60°C once per month do not degrade lifetime of the hardware too much?

destroyedlolo replied at Tue Dec 9, 2014 10:06
Which project ? LeMaker's  or this probe integration in the mainline kernel ?

The LeMaker activities

tkaiser  
Edited by tkaiser at Sun Dec 21, 2014 07:15
LinAdmin replied at Fri Dec 12, 2014 01:36
When measuring right after switching power on then it shows room temperature.

Those 60°C occur when btrfs scrub is heavily loading the system during 6 hours.


Ok, seems you measure the AXP209's temperature and the SoC will heat up even a few more degrees. From my experiences those silly enclosures heat up the BPi really unnecessarily (came to that conclusion after several extensive tests) so I would avoid them or drill in some holes on the left and right bottom side and rotate the box by 90° to ensure just a little airflow that will really help.

BTW: Compare with this setup from Spain: 20° less with a fan. You would get the same by just throwing the enclosure away

BTW: I would never ever scrub on a device without ECC RAM (especially with btrfs-3.4). If your RAM is faulty (what you will not know unless you either experience obvious errors or data corruption or have ECC RAM with logging functionality) then scrubbing will corrupt intact data on disk. And no, it's not about statistics. It's about your machine that is not influenced by statistics but might have soft and/or hard errors regarding RAM (as every device out there )

tkaiser replied at Fri Dec 12, 2014 06:55
Ok, seems you measure the AXP209's temperature and the SoC will heat up even a few more degrees. F ...

No doubt that the plastic box is the cause for the temperature increase.
Without box or with holes there would be dust on the electronics. I still think that 60°C are harmless.

Scrubbing can not destroy data because because it is not a raid-1 and therefore it only would report errors without possibility to (mis)correct anything. My experience with 10T of data during a few months makes me believe that this part of the btrfs code is usable.

Of course I would prefer a newer kernel, and it looks like 3.18 could be okay for BaPi. However I prefer to wait for 3.20 and combine it with Jessie.


f4exb  
Edited by f4exb at Thu Dec 18, 2014 02:00

3.18 + Jessie works for a headless server

Well, so it seems it's working at least with 3.4.103 kernel.

But why is it having this strange name ? It's not ISA.
  1. torchwood ~ # sensors
  2. axp20_mfd-i2c-0-34
  3. Adapter: sunxi-i2c.0
  4. temp1:        +26.6°C  (low  = -144.7°C, high = +264.8°C)

  5. a20_tp-isa-0000
  6. Adapter: ISA adapter
  7. temp1:        +27.1°C  (low  = -144.7°C, high = +264.8°C)
Copy the Code

You have to log in before you can reply Login | Sign Up

Points Rules