Humidity 2.0 temp readout is innacurate.

[xsherlock]

I just started using the Humidity v2.0 around house and noticed they hive consistently higher temp readout then normal temp bricklets. They were supposed to be more accurate yes?  But I just took out the the PTC bricklet and mounted it next to the Humidity v2 and it gives exact 0.85 deg of diffrence

 

ptc:26.19  temp Hv2:27.04  diff:0.85

ptc:26.19  temp Hv2:27.04  diff:0.85

ptc:26.19  temp Hv2:27.04  diff:0.85

ptc:26.19  temp Hv2:27.05  diff:0.86

ptc:26.2  temp Hv2:27.05  diff:0.85

 

 

And Just to me clear I tested 2 different Humidity V2 units  next to each other and they are exactly the same so that is not a single faulty bricklets.

 

What is your experience with those?

 

-----edit

 

So I tested 2 diffrent sensors  and a PTC concurently and the results are

 

ptc:20.37  temp Hv2:22.0  temp2 Hv2:21.75  diff:1.63  diff2:1.38

ptc:20.37  temp Hv2:22.01  temp2 Hv2:21.75  diff:1.64  diff2:1.38

ptc:20.37  temp Hv2:22.01  temp2 Hv2:21.74  diff:1.64  diff2:1.37

ptc:20.37  temp Hv2:22.02  temp2 Hv2:21.74  diff:1.65  diff2:1.37

 

--------------

So now I placed the test setup next to the old classic temp bricklet that shows 

26.0

 

ptc:26.04  temp Hv2:27.21  temp2 Hv2:26.86  diff:1.17  diff2:0.82

ptc:26.04  temp Hv2:27.21  temp2 Hv2:26.86  diff:1.17  diff2:0.82

ptc:26.04  temp Hv2:27.21  temp2 Hv2:26.86  diff:1.17  diff2:0.82

 

So it is spot on with the PTC  and the humidity temp redout is 1.2 - 0.8 off  that is way outside the specs

 

[borg]

What you are seeing there is very likely self-heating, to quote the datasheet:

 

No guarantee of parametric performance is indicated in the electrical tables under conditions of internal self-heating where TJ > TA.

 

The new Air Quality Bricklet had the same problem, but more extreme (we measured 4.8°C difference). We had to set the default measurement frequency to 0.3 samples per second to make it acceptable.

 

 

I also just found the following in the HDC1080 datasheet (Humidity 2.0):

Moreover, to minimize any self-heating of the HDC1080 it is recommended to acquire at a maximum sample rate of 1sps (RH + Temp). In home systems, humidity and the temperature monitoring rates of less than 1sps (even 0.5sps or 0.2sps) can be still effective.

 

The Bricklet currently does not have API to set the measurement frequency, i put it on my TODO list. I will also change the default measurement frequency, since the high frequency is mostly not necessary for humidity/temperature.

[xsherlock]

Good it sounds solvable,  as it is a pressing issue for me maybe if we go down to .33 Hz (that is plenty good for the environment monitoring)

 

I also  see that a cooprocesor was not such a great idea as (for a temp sensor at least) it is a significant heat source as you can see on the IR camera. about 4.5 above ambient that was 25.6 at that moment.

 

https://www.dropbox.com/s/12wnvxibm1duqod/2018-10-15%2015.36.54.jpg?dl=0

 

BTW forum Atachments do not work anymore.

 

---------

on a second thought if that was a self heating issue that would be a problem over time, and it is showing up imediately after I power a sensor device.

[borg]

I just released Humidity Bricklet 2.0 Firmware Version 2.0.3. It uses 1 sample per second as default and it has a configuration function for it. Please note that the function will only be available with the next API bindings release.

 

But if you say that the temperature difference is there immediately on startup, i would assume that the new API does not help here.

 

What happens if you run your test for an extended period of time? Do the temperatures converge?

[xsherlock]

ok I just loaded new firmware and so far so good.

Now the Hv2 is within 0.05 C  from the PTC that is siting 1cm away from it.

So far so good. will burn it over couple days

 

Thanks

 

Maciej

[Maxicko]

Hi,

 

i got the humidity V2 bricklet a few days ago (+ remote switch V2).

I've flashed the latest firmware (2.0.3) but still have the problem, that it measures 2-3 °C more than other devices.

 

Reset of brick is done. Is there any else i can try?

 

Thank You very much.

KR,

Max

[xsherlock]

@borg We really need to have the API python module updated so we can set samples per second to even lower value then the 1 sps introduced in firmware 2.0.3

 

While 1 sps made the problem much smaller I see that after 5-10 minutes the sensor will start heating up just slower then it was with 20 samples/s , in my measurements 0.8 to 1.3 deg C.

 

It is a big problem as the sensor is not reliable as it is now

[xsherlock]

Ok I made a considerable time investment including building a test device with Precise temperature probe to measure how precise is the H2 with 1s sampling and the conclusion is that you need subtract 1 deg C  will make it more or less within .2 of the real temperature

 

https://www.dropbox.com/s/sd6kd1ieb9dqems/Inked2018-11-13%2013.55.23_LI.jpg?dl=0

 

BTW attachments still do not work on this forum, please fix it.

[borg]

Attachment test.

 

Edit: The attachments work fine for me, i tried Firefox and Chrome.