DHT22 Temperature/Humidity Sensors Outdoors?

[Thane of Cawdor]

Hi everyone,

I'm working on a side project to log the temperature/humidity of the outside air using a cheap DHT22 sensor. What's the best way to go about water/dust 'proofing' these types of devices to maintain accurate temperature/humidity readings? I believe the electronics inside these sensors are already potted in the plastic package, so I am mainly worried about the humidity sensor being constantly 'saturated' and reading 100% humidity.

Should I just mount them vertically such that rainwater does not accumulate inside the sensor and would a waterproof box with some air holes at the bottom work? Has anyone had experience in this type of device outdoors and over a long period of time?

Cheers

[Benta]

From the datasheet:
"Do not operate in dew conditions"

Translation: it's for indoor use.

[Fenichel]

  I have been using DHT22 sensors for an environment-monitoring system (http://www.fenichel.net/papers/home%20control%20project/Home%20control.htm) for several years now.  Here in British Columbia, days on end of RH > 90% are not rare.
  I have found the DHT22s to be surprisingly rugged; one of them died after 2 or 3 years of being completely unprotected outside; its replacement has been working fine for years, sheltered (see picture) under an old pill vial.

[james_s]

Look at how commercial weather stations do it. The sensor is mounted in a ventilated box, the better ones have some sort of solar radiation shield and the really fancy stuff has a fan-aspirated shield so that it measures the ambient air rather than being affected by sunlight heating the enclosure.

[Thane of Cawdor]

Nice Fenichel! Is the pill box enclosure sealed at the top and do you have any words of warning for setting up a project outdoors?

James/Fenichel, is there a considerable temperature/humidity differential with an enclosure that is ventilated on the sides and bottom as opposed to an open-air version?

[dave_k]

Look at how commercial weather stations do it. The sensor is mounted in a ventilated box...

The enclosures ia called Stevenson screen:
https://en.wikipedia.org/wiki/Stevenson_screen

DIY versions are not too difficult to make 
http://www.instructables.com/id/Stevenson-Screen-weather-station/

[Thane of Cawdor]

That looks pretty neat but I need this to be quite small (about 5x5cm) and so it might get a bit trickier

[Fenichel]

@Thane
   The pill vial was designed to be open at the top (with a snap lid) and of course closed at the bottom.  As shown in my picture, I am using it inverted, so it is closed at the (now) top and open at the bottom.

   I have done no comparisons, but I'd be surprised to find that the RH was any different in the shelter of the vial, naked in the air, or in a proper Stevenson screen.

  The temperature is a different question.  The water content of an air mass around your house (as measured by the dew point) will be pretty much constant, but there is likely to be real variation in temperature on the scale of inches or feet.  I get occasional temperature spikes when, Stonehenge fashion, sunlight comes between the trees to strike the vial for a few minutes.  You will want to display your weather data only after some simple smoothing, however you gather it.

  As I noted in my earlier post, the DHT22s turn out to tolerate a fair amount of abuse.  I started out using SHT15s, but they are fragile and in every way inferior to the DHT22s.  If you include a CdS photocell so that you can see the photoperiod of each day, be prepared to replace them every few months; they turn out to be much more dew-sensitive (but, fortunately, much cheaper) that the DHT22s.

[Fenichel]

  Typo:  My antepenultimate word should have been than, not that.

[james_s]

@Thane
   The pill vial was designed to be open at the top (with a snap lid) and of course closed at the bottom.  As shown in my picture, I am using it inverted, so it is closed at the (now) top and open at the bottom.

   I have done no comparisons, but I'd be surprised to find that the RH was any different in the shelter of the vial, naked in the air, or in a proper Stevenson screen.

  The temperature is a different question.  The water content of an air mass around your house (as measured by the dew point) will be pretty much constant, but there is likely to be real variation in temperature on the scale of inches or feet.  I get occasional temperature spikes when, Stonehenge fashion, sunlight comes between the trees to strike the vial for a few minutes.  You will want to display your weather data only after some simple smoothing, however you gather it.

  As I noted in my earlier post, the DHT22s turn out to tolerate a fair amount of abuse.  I started out using SHT15s, but they are fragile and in every way inferior to the DHT22s.  If you include a CdS photocell so that you can see the photoperiod of each day, be prepared to replace them every few months; they turn out to be much more dew-sensitive (but, fortunately, much cheaper) that the DHT22s.

You can still get hermetically sealed CdS cells that will tolerate the dew just fine. You can also use a photodiode.

[Thane of Cawdor]

Could I conformally coat the entire board, but leave the photodiode's light window untouched?

I am a little concerned about dew and moisture causing some issues to the PCB and DHT22 leads - was there any significant levels of corrosion with your setup Fenichel?

[Fenichel]

  My PCBs are all inside, connected to their respective sensors by runs of multiconductor telephone cable.  A nice thing about digital sensors like the DHT22 is that there's no need to keep their leads short.  My DHT22 leads are several feet long; they were happily able to use an existing hole (for telephone wires) in my wall.

  The argument for analog sensors like photodiodes and CdS cells is different, but as long as your graphs of light levels don't look utterly goofy, they'll be as good as anyone else's.

  My longest leads (to a wind vane and anemometer on my roof) are over 10 meters long.  Cable is cheap.

[Fenichel]

I had forgotten about this thread, but yesterday we had 94 mm of rain, and the CdS cell that is outside connected to my environment-monitoring system shorted out, as they do.  This one had been in place for a few days short of a year; that, in my experience, is a typical lifetime for one of these cells.

You can still get hermetically sealed CdS cells that will tolerate the dew just fine.

  What sort of sealed CdS cells are you thinking of?  I've been using jelly-bean cells like http://media.digikey.com/pdf/Data%20Sheets/Photonic%20Detetectors%20Inc%20PDFs/PDV-P9203.pdf.  They are claimed to be sealed in epoxy.

[james_s]

Like this: http://www.digikey.com/products/en?keywords=hermetic%20CdS

They're in a metal can with a fused glass window. Not as common as they used to be but still available.

[Fenichel]

Like this: http://www.digikey.com/products/en?keywords=hermetic%20CdS

They're in a metal can with a fused glass window. Not as common as they used to be but still available.

  Considering the cost difference and the ease of swapping in a new sensor in my application, I'll stick with the ones I have for now, but still, it's good to know about the other option.  Thanks.

[mdijkens]

I have one installed in my boat since end of november. So far works surprisingly good:http://esp8266.dijkens.com/mdSensor.html

I do wonder what it looks like in april, since it is a very wet environment and the DHT22 is just unprotected / open air (of course inside my boat, but even then very high humidity)

I am already happy with the current result, since this winter is the first test 

[lukaq]

What I have read, DHT22 when in a lot of 100%RH situations will have 'memory' and read humidity very wrong.
On other hand, Si7021 has heater inside, keeping sensor dry

[Fenichel]

What I have read, DHT22 when in a lot of 100%RH situations will have 'memory' and read humidity very wrong.

  I've not noticed any memory problems, although my location has never experienced rapid change from very high to very low humidity.  The attachment is a routine sample of my data.