The documentation says:

Quote

Note The currently released firmware (2.0.4) does not save the calibration every 12 hours. We had to disable this because of a bug in the proprietary load/save mechanism that we can't fix. We will release a new firmware that has this enabled again as soon as this bug is fixed.

My Air Quality Bricklet reports firmware_version '2.0.5'.

Does this mean this bug is now fixed?

related commit (Disable automatic calibration load/save for now)

Anybody?

This is not yet changed unfortunately, the calibration is currently not saved in the firmware.

However, i just checked the Bosch site and they have 1.4.8.0 released while we still use 1.4.7.4. Maybe it is fixed now, i wrote it on my TODO list and will check it when i have time.

I would really appreciate that.

Since March 2019 the Air Quality Bricklet no longer saves its calibration data, correct?

As it takes 28 days to fully calibrate, and the system here reboots once a month, this yields max 3 days of reliable data per month.

I am willing to work with you to solve this (and invest some time from my side), but if there is no solution in sight I will retire this Bricklet.

 

Bump

Thank you for the bump and very sorry about the delays. I will be able to work on it next week.

Great! I am pretty busy, but let me know if I can help.

Attached you can find a firmware that has the newest Bosch BSEC version and the calibration automatic save/load enabled.

To test this i will let it run over the weekend and then on Monday unplug it and plug it in again. It should retain the accuracy then.

If you want you can test it in parallel.

air-quality-bricklet-firmware-2-0-6-beta1.zbin

Flashed. Running.

I tried it and unfortunately the accuracy turned to "unreliable" again after the reset. I will add some debug wires and a logic analyzer to see if there is a bug in the communication with the BME680.

Borg, mine has been running FW 2.0.6 for 5 days uninterrupted. It started with accuracy 0. Then 1. Yesterday accuracy was 3. Today it is 0 again, without rebooting.

My point being that IAQ accuracy will vary wildly anyway. Your sample dropping to 0 after reboot may be caused by rebooting (losing calibration), or may be just "normal" behaviour (picking up calibration at some point during previous 12 hours).

That's what I think.

I have a new test running over this weekend. This time with cables soldered to it to see the communication and lots of debug output. I also found a bug in the meantime. I have high hopes that it will work now!

I did a test where i changed the state-save-time to 5 minutes and this worked already, but the accuracy was not high at that point of course.

I just released version 2.0.6.

It uses the newest BSEC version, saves the calibration state every 12 hours and loads it on startup.

When the calibration is loaded it still starts with IAQ Index unreliable for the first few minutes and then goes to low. In my tests it did however start with the old IAQ Index after it changed from unreliable to low. I was hoping that it starts with accuracy high when the saved state had accuracy high. But it doesn't seem to work like that.

Unfortunately the whole BSEC software that is used to calculate the IAQ Index is a blackbox for us and we don't get any information from Bosch about the inner workings.