Humidity sensors very inaccurate

Hi,

I would like to know the humidity in a few rooms, but the sensors are very inaccurate.

My mechanical hair hygrometer is accurate.
The DTH11 shows values of 34 to 36, which is way too low.
The AM2302 (DTH22) has a value of 52, while it should be 75. I tested this outside to be able to compare it with actual values from wheater channels/stations.

This is the same problem: http://forum.arduino.cc/index.php?topic=96470.0

The value of the DTH11 is way too low, but when I wrap it in a damp towel, the value rises to almost 100%.
I did use some glue with the DTH11, perhaps the vapors could have caused it.
To calibrate it, it need 50 to 60 degrees Celsius with humidity <10%. I don’t know how to get that dry air.

How can I get an accurate humidity sensor ?

Does a hair dryer get you close to calibration point, i.e. hot and dry?

I tried a hair dryer, the temperature was between 50 and 55 degrees Celsius, and the measured humidity was low. But after that (after a few hours) they return to their old wrong value. I might have to do that again in the winter, when the air is dry. The outside humidity is 86% at the moment. I also tried to make dry air, with an air container in the freezer, but that didn't work.

Did you know you can get predictable humidity with saline solutions see this document and maybe it will suggest a way?

Sorry forgot the link http://www.omega.com/temperature/z/pdf/z103.pdf

Cheers, Rob

You may wish to have a read of this.
http://rumkin.com/reference/desiccant/

Its possible to reduce humidity with dessicant crystals.
You can buy them , or make your own as per the above article.

Thanks robwlakes and mauried.

It would be very nice to have a calibration, but I don’t understand how I can use those salts.
Using silica gel to get dry air to calibrate the DHT11 and AM2302 seems to way to go.

As I understand the idea would be to make up a jar of sodium chloride and water (reagent quality sodium chloride and distilled water would be more accurate I guess but may not be necessary with the overall accuracy of the Arduino based detectors not being all that accurate) such that there was solid salt crystals in the base of the jar. This means that the salt is saturated in the water and no more can dissolve, however what the table shows is that the humidity of the air and water above that solution is also in equilibrium at certain repeatable humidity/temperature combinations. I am using a DTH22 on my weather station and it has a built in temperature sensor, so the sensor could be put in the jar in the air and humidity/temp readings taken to see how they compare with the table. The Sodium Chloride would give you about 20 degrees = 75% humidity. If you could get you hands on the other salts you could get other readings, and maybe interpolate the translated values? Once they were set up you just have them stored for recalibration at any other time. Potassium Carbonate would give a lower reading, eg 20 degrees = 43%. Some of the other chemicals look a bit harder to get hold of.

I would be curious myself now that you have raised this issue to test my DTH22, I assumed it was just much better than the sensor in the Bios WS console. They looked physically similar though. Though I am not sure what humidity has to with the likelihood of catching fish, so I doubt if my viewers will be that fussed about its absolute accuracy.

Cheers, Rob

PS is the DTH22 temp sensor a digital one capable of sub centigrade resolution or a thermistor with fairly low accuracy? I believe the library uses both to moderate the sensor readings, so maybe that needs checking as well.

Thanks for the explanation. I happen to have a few bottles of distilled water from another experiment.

The AM2302 is the DHT22. In the documentation they warn about certain chemicals, like anti-static colored plastic bags. If the sensor is stored in such a bag, the humidity value is no longer accurate.