[Weather] 1-wire humidity sensor for outdoor use?

[Steinar Midtskogen]

[Nafis, Christopher A (GE, Research)]

> I've been using the AAG TAI8540 Humidity sensor mounted in a Davis
> Radiation shield for an outside humidity/temperature sensor (AAG states
> "designed for indoor use"). I've coated the circuit board with liquid
> tape, but the honeywell sensor still seems to "die" after a year or so. 
> Hobby Boards also sells the HT3-R1-A Humidity / Temp sensor with a
> moisture resistant coating.

My Honeywell sensors usually don't last longer than half a year.
After a while, I get 0% readings, or readings stuck at a certain %, or
they indicate more or less random humidity.  Even new sensors using
factory calibration are a problem; they sort of work, but without
further calibration they may report more than 115% in heavy fog.

I've tried a lot of things: different shieldings, different boars (AAG
indoor, AAG outdoor, Springbok), coating and heating.  The idea with
the heating was to keep humidity well below 90% at all times and
calculate the true rh based on the assumptation that the dew point in
the heated box and in the ambient air is the same.  Perhaps the sensor
would last longer, but I didn't get a useable accuracy.  Right now I'm
trying to wrap a nylon sock (from old stockings) around the sensor
(either the AAG enclosure or the sensor itself), which is suggested in
the Weather Toys book.  So far it doesn't look too promising as it
seems to make the sensor very slow and less accurate (though it seems
to fix the problem with readings above 115%).

My conclusion is that the Honeywell sensor is fine for indoor use or
in climates where humidity is always between 30 and 80%.  For coastal
or mountain use, it's a constant calibration and maintenance
nightmare.  I just had 10 days of continous fog and rain here (i.e. rh
in the high 90's all the time and about 130 mm of rain/sleet and windy
at times) and two newly mounted sensors that were matching eachother
perfectly, were off eachother by 10 to 15% when the fog lifted and
things dried up.

If anyone has any other ideas how to fix the honeywell sensor, I'm
ready to try anything.  Or if anyone can think of a design using
another sensor, I would happily switch to those if they're likely to
need less maintenance, even if they're $200 each.  I can live with bad
accuracy above 90%, if the sensor goes back to normal accuracy below
90%.

-- 
Steinar