Maybe the sensors have a bit late compare with the system sensor.
@rgl, @solidus1983 (+ others) – did any of you upgrade to the most recent firmware (and EC) versions – and if so did the changes have an impact on how the fan is operating?
(The most recent at the time of writing appears to be SD-BS-CJ41G-300-101-F. https://files.seeedstudio.com/wiki/ODYSSEY-X86J4105864/Documents/SD-BS-CJ41G-300-101-F.zip)
I am also experiencing this issue; fan does not work at all when in “Always On” mode, even when all sensors reach over 90C (194F), 5 degrees centigrade below critical point (tested with MPrime). When set to “Normal” or “Positive”, it only creates a very annoying ticking/scratching sound, and the fan is constantly “revving” up and down, without doing any real work.
Since there have been reports of firmware upgrades bricking the Odyssey, it would be extremely helpful to know for sure whether this is actually a software issue or hardware-related before attempting a fix.
Thanks for any input.
Yes and the fan seems more responsive to temperature changes. Unfortunately there is no way to monitor the speed of the fan so we cannot be sure but it seems to spin more than before and it is not stuck.
Thanks for confirming, @milasx. My main issue was that the fan was either completely switched off (when, ironically, it was in “Always On” mode), or did its constant “revving-up” thing while in “Normal” or “Positive”. I’m normally not a fan of fan noise (pun intended), but the scratching and ticking sound really became bothersome, even when it was in the other side of the room.
I’ll try a firmware upgrade, but obviously wanted to make sure if there were adverse effects in doing this beforehand, since some users reported that the Odyssey was left in an unbootable state after doing so (which is – shall we say – suboptimal)
btw. did you also remove the CMOS battery as the guide suggested (presumably to set it back to factory defaults)?
Be easy to upgrade the firmeare. The EC first, then BIOS. Upgrade the BIOS may take severial miniutes, donot remove the power or force restart the machine unless you see the boot logo showed in the screen.
Fingers crossed for you!
Ignore the post about the sha256sums, I see now that they are based on the actual zip file, not the BIOS files (which is more common practice).
It’s probably a good idea to mention upgrading the EC firmware prior to the BIOS on the Wiki though, based on the comments above, as it sounds like a critical step.
Waitting your good news.
Finally did the EC + BIOS upgrade – happy to report that both went without a hitch. Now both my Odyssey-based systems are running the same BIOS: (I have one J4105 + the new J4125)
dmesg -T|egrep ‘Celeron|SD-BS’
[Fri May 7 20:51:11 2021] DMI: Default string ODYSSEY-X86J4125/ODYSSEY-X86J41X5, BIOS SD-BS-CJ41G-300-101-F 04/08/2021
[Fri May 7 20:51:12 2021] smpboot: CPU0: Intel® Celeron® J4105 CPU @ 1.50GHz (family: 0x6, model: 0x7a, stepping: 0x1)
[Tue May 4 20:25:40 2021] DMI: Default string ODYSSEY-X86J4125/ODYSSEY-X86J41X5, BIOS SD-BS-CJ41G-300-101-F 04/09/2021
[Tue May 4 20:25:41 2021] smpboot: CPU0: Intel® Celeron® J4125 CPU @ 2.00GHz (family: 0x6, model: 0x7a, stepping: 0x8)
Initially thought the EC hadn’t changed, as I could not identify its version in the BIOS (completely different from the filename). But I can confirm that the fan behaves correctly now, in that “Always On” does actually mean what it says, and Normal has the fan running (in a low-rpm, so it is very silent, basically inaudible unless you put your ear next to it. Good stuff.)
For the upgrade I did a complete power drain for both EC & BIOS. EC went very fast. The BIOS took longer than expected (before the reboot). After it finished I turned it off, unplugged the DC and let the voltage drain. I did not remove the CMOS battery connector for either EC or BIOS.
The subsequent bootup took around 3-4 minutes, with 2 reboots in between.
btw. – it took some time to realize that the EFI System Shell for some reason did not support an exfat filesystem, which was a little strange (don’t usually have older FAT-based filesystems lying around, so a stick had to be formatted for it). For the record - many EFI Shells also support Linux filesystems now (ext4/ext3) as well, and it would be nice if Seeed supported this also, considering how many users buy it in order to run a Linux variant or pfSense etc.
Also did some testing with memtest95+ and MPrime (Prime95 for the W*n crowd) using its torture test (#1, small FFTs). It’s the most CPU and heat-intensive app that I know of.
Where the previous BIOS managed to get it up to 95C (203F) – where it had to be manually terminated for safety reasons, the new BIOS (in ‘Normal’ setting) hovers around 71-73C (157-163F); all while running completely silent.
So for my part at least the issue is resolved – kudos to the Seeed techs for fixing the fan issue. It would as I mentioned be nice if the EFI shell supported more filesystems, so feel free to take that as a suggestion for a future BIOS version.