Robust humidity sensor

Hello,

As a long-term project, I have a built a RasPi and Arduino-based weather recording station. Currently it records data from about 20 different sensors. With modern sensor technology, most of these work great. The exception is humidity.

Indoors, I record from both DHT22 and SHT31 modules. These have tracked very well with each other consistently for over a year. No problems there. However, the indoor environment is controlled and there is relatively low variability in both temperature and humidity.

Externally located DHT22 sensors are a different story. They read fine for a month or so but then peg at 100% RH and never again budge from there. I think the periodic high humidity (sometimes condensing) kills the sensor. From datasheets I've read, most sensors do not tolerate high humidity (especially condensing) conditions very well, although most also claim that the sensor will eventually recover.

So, does anyone have a recommendation for a robust, accurate, and reasonably inexpensive humidity sensor? (Or is it choose one of the three?) I'm not picky about RH or dewpoint readout but would probably prefer the latter. "Inexpensive" in this context means less than maybe US$25, though I could go higher if the sensor will last many years.

Thanks for any help..
Steve

The small electronic humidity sensors cannot tolerate condensing conditions and are
never going to be reliable outside.

I reckon the best relative humidity sensors out there are wet-and-dry bulb thermometers -
perhaps you can make an electronic version of one of these. Yes they need topping up
occasionally, and must be sheltered from precipitation and sunlight, but they will recover
from 100% RH conditions all by themselves and be more accurate.

Maybe take something like this and replace the glass thermometers with canned DS18B20
sensors?

Do you have the DHT22 devices located according to the data sheet specifications and in an environment withing the stated limits? Have you tried to rehabilitate the DHT22?
Paul

MarkT:
The small electronic humidity sensors cannot tolerate condensing conditions and are
never going to be reliable outside.

I reckon the best relative humidity sensors out there are wet-and-dry bulb thermometers -
perhaps you can make an electronic version of one of these. Yes they need topping up
occasionally, and must be sheltered from precipitation and sunlight, but they will recover
from 100% RH conditions all by themselves and be more accurate.

Interesting idea, since duplication of standard psychrometer seems inherently self-calibrated (within the calibration of the temperature sensors, something I have lots of experience with). I think this moves to the top of my list of more complex PITA approaches which will require some tinkering. I had also thought about measuring the reflectivity of a small mirror periodically cooled by a Peltier device .. theoretically a direct readout of dewpoint but with plenty of anticipated problems.
With the degree of rainfall here, I suspect I'd have more trouble with biological growth in the reservoir than with keeping it full.
Of course, I'd sort of like something that I can just plug into one of the remote Arduinos and use someone's pre-written library to interface. :slight_smile: But I suspect you're right about the available sensors.
Thanks for the feedback.
S.

Paul_KD7HB:
Do you have the DHT22 devices located according to the data sheet specifications and in an environment withing the stated limits? Have you tried to rehabilitate the DHT22?

As best I can. The unit is housed in a thick-walled foamed polystyrene box which has an opening at the bottom, and that protected by a fine gauge wire mesh. It is definitely protected from sunlight and rain or dripping water. But there is simply no way to keep it out of condensing conditions and still have it located outside; the two kind of go together in this climate.
Datasheets often do specify a bake cycle to rejuvenate a sensor exposed to condensing conditions. That didn't sound very appealing, so my approach was to make the sensor easy to replace. New ones are only a couple of bucks. But once a month is just not reasonable.
Thanks...
S.

I have used a BME680, outside for slightly more than a year.

The sensor is housed in a Stevenson's Shield.

The shield allows the sensor to be exposed to the elements whiles keeping the electronics protected.

Questions for Idahowalker:

Has your BME680 continued to report realistic humidity values over the time you've had it deployed?

Is your region subject to episodes of very high humidity and fog? What readings does it report under these conditions?

I do have a BME280 sensor which may utilize the same humidity sensing element as the BME680. The datasheet specifies it for 0-100% RH, which is encouraging, but also that the environment is non-condensing, which is discouraging. These sensors are a little on the expensive side but I think I'll put one outside to see how it fares.

I saw your earlier post showing the snow-covered sensor housing. I was not aware that those louvered enclosures had a name. I've always "rolled my own," and it's an interesting problem to protect both from sunlight and from IR radiated from the enclosure itself as it heats up from sun exposure. My approach has been thick Styrofoam walls between the outer shell and the inner cavity, to insulate the outer shell from the sensor (which is suspended in free space inside the box). I have some evidence of sensor heating nevertheless, I think due to convection and insufficient air exchange.

Thanks for the comments...
S.

Yes, the BME680 sensor has continued to report realistic values over the time I've had it deployed.

The area I live in is subject to various weather conditions. When their has been fog, heavy rain, snow the unit has reported reasonable values, compared to the local airport.

We use, indirectly, the humidity reading on a regular basis. The BME680 gives gas resistance that are indexed against an ideal values and humidity to produce an Air Quality Index. We use the AIQ to determine the best run times.

The BME680 sensor takes a reading every 11 seconds, 24/7.

Idahowalker:
Yes, the BME680 sensor has continued to report realistic values over the time I've had it deployed.

Great to know. I will give my BME280 a try. Thanks again.
S.

Wow, that's brilliant - they seem to have proprietry tech that works much better than
the polymer sensors in most humidity sensors. Given the speed of response it must use
something like surface adsorption.

This topic was automatically closed 120 days after the last reply. New replies are no longer allowed.