WeatherShield self heating? [SOLVED]

[agprimatic]

Hi,

I purchased a WeatherShield board and am running code to read the temperature and humidity every 10 seconds.

I find that the temperature slowly creeps up over many minutes.

For my test, I leave the board unpowered for a long while (tens of minutes) in the test environment.  When I first power up the board, the temperature is 28.07 degC.  (For reference, I have a DHT22 in the same location (about 6 inches away) reporting 28.6 degC over the entire test).  Every 10 seconds, the temperature increases by about 0.1 degC, until it stabilizes approximately 1.7 degC high at about 6 minutes in.

I suspect that there is a self-heating problem with the WeatherShield.  The fact that the DHT22 shows a stable temperature over the measurement interval is evidence that there is no actual temperature change in the room.

I have the WeatherShield mounted to a Moteino board with no other devices enabled (drawing power or generating heat).

The table below shows the temperature readings over the 6 minute test.

Does anyone else see this temperature rise while the board is powered?


Thanks,

Ag Primatic

---

Time	Temp
0	28.07
10	28.12
20	28.21
30	28.30
40	28.40
50	28.60
60	28.70
70	28.80
80	28.91
90	39.00
100	29.06
110	29.23
120	29.29
130	29.35
140	29.41
150	29.48
160	29.52
170	29.59
180	29.65
190	29.71
200	29.75
210	29.79
220	29.83
230	29.87
240	29.90
250	29.93
260	29.96
270	29.98
280	30.00
290	30.03
300	30.07
310	30.10
320	30.11
330	30.13
340	30.15
350	30.16
360	30.18
370	30.19
380	30.20
390	30.23
400	30.26
410	30.25
420	30.26
430	30.28
440	30.28
450	30.29
460	30.30
470	30.28
480	30.31
490	30.33
500	30.33

[Felix]

2deg is not a lot and can be explained if you're in a room where you stand near the sensor. I see this all the time in my lab where temperature goes up continually during the day while I am there - my body basically heats the room.

Does your behavior happen on each power up or after it is "heated" it starts from where it left off?

[agprimatic]

If you read my original post, I have an unpowered WeatherShield and a separate DHT22 which I'm using as my reference at my desk.

The DHT22 is reporting a steady 28.6 degC. I power up the WeatherShield, and I get the ramping behavior that I tabulated in my original post.

Yes, this is very repeatable.

I suggest you perform a similar experiment with a temperature reference that is stable, then power up the WeatherShield device under test, and I believe you will see a similar response.

And a 2 degC accuracy is not very impressive for a board whose sensor has a 0.4 degC intrinsic accuracy.


Ag

[Felix]

I can say I have been using the WeatherShield on my own with this sketch and I don't see the behavior you describe. The sensors have a heating mode where you can heat up the sensor. But I don't know what code you are running.

Not sure what you mean by 2deg accuracy, in my 1st response I was referring that your 2deg drift should not be that unusual, depending on the conditions you are testing in.

What happens if you reach the 2deg drift, then you power off and back on? Is the reading where it left off or is it back to the initial 28deg?
What code are you using for the readings?
How much does the Si7021 keep drifting? What happens after 1 hour, 2 hours, 10hours of readings (assuming you have the ability to monitor that interval)?

Sometimes these sensors can take a few minutes to stabilize. I have seen this happen before, nothing unusual. The fact that they don't behave exactly like a DHT is not totally concerning. If it keeps drifting to no end then that is concerning.

Almost beside the point - Here's a test I did with 2 WSes, the temperature and H readings are basically identical:

[agprimatic]

OK, it's not self heating, as in heating due to the Si7021. But I think it may be due to the Moteino board heating from underneath the WeatherShield.

I tried another experiment. I connected a bare WeatherShield with test clips to an Arduino Uno, and I'm comparing it to a WeatherShield mounted directly on a Moteino (please see attached picture).

I'm running the code from your git repository LowPowerLab/WeatherShield/WeatherShield_Si7021_only.ino (with the battery monitoring code removed).

The WeatherShield attached with clips reports temperatures in the range 25.35 degC to 25.63 degC.

The WeatherShield mounted to the Moteino temperatures keep rising. It started at 26.42 degC and after a few minutes is now stabilized at 28.91 degC.

I see in your picture that your WeatherShield is mounted off the Moteino by what looks like 1/2 inch(?).  Maybe I need to have more space between my Moteino and my WeatherShield.


Ag Primatic

P.S.  Now that I've been running for about 30 minutes, the WeatherShield attached with clips to an Arduino is at 25.41 degC and the WeatherShield mounted to the Moteino is at 28.91 degC.

[Felix]

I think the most accurate test would be in a temperature controlled chamber with nothing but wires attached.
I haven't been able to see worrying behavior with my WSes mounted at about 1/2" on top of Moteinos.
It does look like if they are very close to other boards, there is some influence. I think it speaks to how sensitive the sensors are.
I think your findings conclude that the WS should be left some gap beneath it and should probably not be sandwiched very closely with the Moteino or other boards, or the temperature might stabilize a few degrees higher than expected. An interesting variation would be to test this in a fridge, to see if it behaves the same at low temps (still drift up or ... ?).
Anyway thanks for the follow up. I'll mark this as solved.

[TomWS]

@agprimatic, what is your transmit power and are you using an RFM69HW?   The self-heating could very easily be from RF power since you're transmitting every ten seconds.   A more realistic test might be every 10 minutes (or longer).  I've had TH sensors on WiFi adapters self heat (and humidity drop) from the RF power and had to take the sensor measurement before connecting to the access point.

Tom