Why are my HTS221s getting bad Humidity values?

[M.Mardini]

Hello,

My company is using since more than one year a combination from HTS221 and stm32f407 uController , I2C is used.

The problem is that on some devices i have from the sensor different results.

also some sensors give correct output where the others give lower values, where both of them have the same circumstanses. like the same board design and the same uController firmware.

in the attachment there are the output data from two sensors.

I wonder where the problem could be?

We use the follwing adjustment :

PD = 1: active mode

1Hz

other CTRL-REG are disabled,

we have activatet the heat register for 4 seconds,but that did not solve the problem.

we are very thankful for any help 😀

[Eleon BORLINI]

Hi @M.Mardini​ ,

not many ideas, unfortunately.

Could you try compensating the sensor both with an offset and a gain (e.g. vs temperature)?

I however believe I'm missing something in the interpretation of the first graph, the M6 one: is the gray line showing the difference between the blue line and the orange line?

[M.Mardini]

Hallo, we are now trying to compensation the sensor with an offset.

Exactly the gray line shows the difference between the real value and the sensor output

the yelow line is the output plus the new offset,that we implemented by Firmware.

[Eleon BORLINI]

Ok it's clear, thank you @M.Mardini​ 

My concern about 1st graph is that the difference is rapidly varying at the beginning of the acquisition (just after the sensor turns on, I suppose), and then it gets almost constant. Which is the time scale of the x axis, seconds or acquisitions? What if you start processing the acquisition some samples after the startup, in order to avoid some "heat up" effects of the chip (I'm referring to the first part of the graph)?

-Eleon

[M.Mardini]

Hallo,

Yes the grau one shows the difference between the blue and the orange lines!

[Eleon BORLINI]

Hi @M.Mardini​ ,

my only guess is that there where some issue in the assembly / soldering process on those specific sensors.

Did you see for example a "pattern" between failures and position in the assembly line?

-Eleon