WeatherShield reporting incorrect Humidity/Temp

[mhaggerty]

Not sure if I have bum unit or if there is some weird power issues going on that could affect readings...  I set up a Moteino with a WeatherShield using the code from the lowpowerlab github repository.
The issue I am having is that the readings are high - even reporting humidity levels over 100% (as high as 107% [avg about 10-15% higher than other sensors and local weather stations] - and no the unit was not under water!).  Temperature values also high by ~2-5 degrees F.
Input voltage is supplied from 3 NiMH AA batteries that are charged via an attached solar panel (voltage runs ~4.2V in bright sunlight, ~3.6V at night). 
I've tried resetting the Moteino by powering it down, waiting a few minutes and the powering backup. Any suggestions on troubleshooting? Would it be worth trying another power source?  or is there a possibility that the sensor is bad?
Any input is appreciated!

[Felix]

To really tell if a sensor is bad there are some considerations.
The sensors are all read at least 1 time after assembly, and their values have to be in a certain range and consistent with other shields being read in sequence at the same time in the same conditions, that's how I know I am shipping good units to everyone.
If between 2 weather shields you'd get significantly different readings then yes I would assume one has somehow gone bad (assuming you used good practices and the sensors are not abused electrically or mechanically).
Also comparing to "local weather stations" is not really a good test. Because other local stations could be Joe on the next street with a DHT11 or LM35 and even if they are good stations you cannot really control the conditions in which they take samples. So a few degrees there im not concerned about at all, i see the same thing on my end.
The way you're powering should be OK.
You should hardly hit 100% Humidity ever, I've never seen it happen.

Could you please try this sketch which I added recently? See if it makes a difference?

[mhaggerty]

Thanks for the updated sketch!!!  I uploaded it, but unfortunately it did not change the values, I agree that the sensor humidity reading should rarely get to 100%, that is why I was concerned when I was getting readings of 107% (not even possible!). I really would like to validate the readings - you think a 2nd Moteino/WeatherShield would be useful in troubleshooting?  I also have a DH22 sensor that I could hook up and compare readings in the same area (yeah I know, DH22's are not very accurate :-( ).   
It may be unremarkable, but the the voltage being broadcast from 3 AA 1.2V NiMH batteries is 4.07V - although there is a 4.5V, 1.5W solar panel connected with a diode to prevent bleed into the panel, the reading is at night so the panel should not have any impact.  The voltage reading is consistent with a meter reading from the battery pack.
Thoughts?

[Felix]

The voltage is fine.
Yes a second WShield would definitely be a good test case against the other.

[mhaggerty]

Well I'm sure I'll find uses for another WeatherMote or two, I'll go ahead and order some.  On a side note... the WeatherShield reported humidity of 103% using the code you shared yesterday (thanks for that!).

[Felix]

103% Humidity is wrong. So either the sensor or the code is giving bad readings.
It's up to you, if you get more please try the same code and see where the difference is, let us know what you find...

[mhaggerty]

I deployed a 2nd WeatherMote (with the same code, just different Node #!), it is positioned right next to the one with suspected bad data... After letting it sit for a few hours to ensure that the sensor were climatized, I found that the humidity on the original is significantly higher (80% vs. 69%).  Since both sensors are using the same code, I would expect them to be almost equal.  Temperature is off a bit between the two but close (71.97 vs 73.34), pressure is also close (30.03" v 30.01").  Any other troubleshooting you can think of?  Or is the original a bad sensor? 

[Felix]

Mike, is the original having the white filter on it by chance?

[mhaggerty]

no... neither weathermotes have the white filter...

[Felix]

Mike, contact me at felix@ this domain.com to get this resolved further. Thanks