TemperHUM temperature calibration

[brian-maloney]

Hi there! This project was perfect for a system I was building, but unfortunately I am using the TEMPer1F_H1V1.5F and it seems to be miscalibrated. Two of them so far both seem to be reading around 5.5° C above the room temperature. I've tried calibrating one using the horrible Windows software since it seems that that calibration is stored in flash, but it doesn't seem to make a difference. Do you have any advice to get it to work, or barring that would you consider adding a software calibration parameter? I've looked at the code and can't make much sense of the math operations you're applying to the value that's read from the device to get the °C value but maybe making one of those user-controllable is the way to go?

[yrro]

That's interesting, my TEMPer1F_H1V1.5F also reads about 4-5 degrees too hot. The numbers in the formula come from I took them from https://github.com/edorfaus/TEMPered/wiki/TEMPer2HumiV1.x (actually, from the SHT1x page on the same wiki). Based on your findings, it seems quite likely that the numbers are wrong.

I had a go at reverse engineering the Windows software... there's a 'get calibration' command, but for me it only returns 0, and anyway the software only seems to store the values, and never actually uses them when it calculates the temperature! As for the temperature calculation, it just seems to use the number read from the device directly, without doing the same calculations I'm doing, which is very mysterious. I wanted to use Wireshark to figure out exactly how the software communicates with the device, but I can't even get the software to run on my Windows machine so I haven't gotten any further with that. If you fancy trying it out it would be really helpful!

[brian-maloney]

I'll try to get a capture of the USB traffic. I actually wasn't even aware that USB pcap was a thing, that's pretty cool. The only note that I had was that my readings appear to be incorrect even when using the button and/or the Windows software, which is inconsistent with your experience noted in TEMPered #39.

I've also obtained a Yoctopuce Yocto-Meteo which seems to be much more reliable (but also more expensive) than the TemperHUM. When I have time I'm going to build a Prometheus exporter for that hardware (maybe for the entire family of Yoctopuce sensors if I can make the code generic enough) based on your work here. Then I'll be able to compare the two as the temperature changes and see whether the TemperHUM is consistently 5.5° C off or if it varies depending on the ambient temperature, which will determine what sort of change to the equation would be needed.

[notsureofagoodname]

Looks like there's calibration @ https://github.com/edorfaus/TEMPered/wiki/Calibration