A DIY humidity sensor using two DS18B20 temperature sensors

After having a couple of HTU21's die last year, I went looking for more robust ways to measure humidity. I found a references to old school Masons hygrometers, but no one seemed to post the data you actually get from them so I built a few using some cheap waterproof DS18b20 temperature sensors:


This method of measuring relative humidity appears to be working, but I have significant differences between the psychometric constants for each sensor, and I was wondering if anyone else has built these with similar results? Any advice would be appreciated as my wet bulb depressions are about 1/2 of what the textbooks say they should be.


You may have to calibrate each temperature sensor yourself, in order to get meaningful values. I'm not sure how the implementation of the wet bulbs and ventilation affects the conversion factors, have no own practical experience with such matter. All in all you'll have to deal with three different types of temperatures: static (ambient), dynamic (air flow), and dynamic wet.

A few references list the numbers from a 1984 paper by List which provides 0.00062 as the psychometric constant for sling style hygrometers which are fully ventilated, and 0.0012 for non ventilated hygrometers. With the 2-5 degree wet bulb depressions I am getting I have to use a psychometric constants up in the 0.002 range to make the curves match the SHT-11 RH sensor data I am comparing them against.

In all cases the wet bulb is kept wet so I am afraid I don't quite follow what you mean by "dynamic wet"? Are you referring to situations where ventilation changes between static and ventilated? This is a real risk for the cave deployments as barometric changes on the surface often change the air flow patterns in the cave. I still have not figured out how adjust for this without an anemometer.

Perhaps I could add some wind shielding around the wet bulb...?

As you are finding the airflow across the wet bulb makes a lot of difference to the readings. Official weather readings are measured within a Stevenson Screen which has double-louvre walls and gaps at the bottom to ensure airflow without letting the breeze in.

You dont detail what you are using for the wick that covers the wet bulb but I imagine the properties of that would have some affect too. Google turns up http://meathaccp.wisc.edu/assets/wet_bulb.pdf as a guide to making your own Wet Bulb sensor with some advice on wick materials.

As a side note most official weather sites are now using humidity sensors instead of wet bulb due to their lower maintenance requirements.

Very interesting project :sunglasses:

The SHT-11 is a capacitive sensor, it should be accurate and stay accurate over the years.

I noticed the metal case around the DS18B20, that will have an influence, unless it is completely submersed. I don't know what the mesh will do, that might also have some influence.

A bare DS18B20 (the black TO-92) collects heat from even the slightest dim sun light. A metal case will behave different with water condensation and infrared radiation heat.

Do you have the maximum resolution in software ? Do you take a few samples and average them ? The relative accuracy can be increased that way. It is even possible to replace each DS18B20 with two (or more) of them average the result. They are super cheap anyway.

Thank you very much rw950431: as that pdf you suggested links to wick suppliers and I am wondering if my shoelace wick is causing a problem by being too thick. It also did not occur to me that the food industry used this kind of method to track RH, so I can do more digging there.

Peter_n: My primary interest in this method has to do with the fact that the caves are above 90% humidity much of the time, and atmospheric pressure changes above the caves can push them into condensing environments quite often. Most of the capacitive RH sensors I found list very long recovery times (with "drying out" procedures) if the sensor is exposed to 100% humidity and I wanted something that would gracefully bounce between 90-100% without issue. Also the accuracy of the more affordable capacitve sensors fluctuate quite a bit in that range, so there is even a chance that using a couple of DS18s might actually provide better data. But that's still a big if because while the DS18b20's discriminate down to 0.0625 C my current wet bulb depressions will be quite small above 90% as well. I have not tried averaging yet, but given the lag from the thermal mass of the waterproof DS18s, I don't expect to see much change there. My previous experience with them is that they are pretty stable sensors.

WRT Screens, just getting to the deployment site often involves squeezing through some pretty small spaces, so I would need something much more portable than a full-on Stevenson. Ideally something that fits into a small kit bag wrapped around my shoulder. It also has to be rugged enough to take some knocking about, as that also goes with the territory.

Yes, you are right. The DS18B20 is pretty stable. Avering a few samples would not do much.