From: Steve Turner
Subject: Re: Humidity sensor - peculiar behavior
X-Newsreader: Forte Agent 1.92/32.572
Date: Sat, 04 Jan 2003 14:32:52 GMT
NNTP-Posting-Date: Sat, 04 Jan 2003 14:32:52 GMT
Organization: AT&T Worldnet
Winfield Hill wrote:
> Steve Turner wrote...
>> I am interested in adding a humidity sensor to a computerized weather
>> station of my own design. The sensor is a Hycal HIH-3610 003 unit,
>> ... I find that the *initial value* of the output, on first
>> powering the sensor board, is exactly where it should be. This is
>> true at three different humidities (30%, 75%, 93%). However, the
>> initial voltage output then decays in a nonlinear fashion (quickly at
>> first, then more slowly), eventually stabilizing at a new value which
>> is 10-30% lower!!
> Welcome to the unreliable world of low-cost humidity sensors!
> Steve, tell us more about your humidity chamber. Perhaps it's
> got some internal condensation going on reducing the humidity?
The humidity chambers are a series of "Drierite" jars which contain
various salts in equilibrium with their saturated aqueous solutions.
Drierite jars were chosen because they are throwaways where I work and
because the lids are fitted with very tight sealing gaskets. These
jars are something very close to hermetically sealed. I have modified
one of the lids with a vapor tight three-wire feedthrough. The sensor
board is attached to the inside end of the cable. Different jar
bodies can then be swapped in at will (these are kept well capped when
not in use).
The salts were chosen according to constant humidity tables in the CRC
Handbook of Chem & Physics. These are designed to provide a constant
humidity in a closed environment via a dynamic equilibrium between
evaporation of liquid and condensation of vapor. I am inclined to
believe that the chambers are "working" properly because of the
excellent results on initial powerup of the sensor board in each
chamber, though I suppose this could be coincidental.
I suppose I could recalibrate the sensor using the new steady-state
measured voltages, but since I paid extra for an NIST-traceable
sensor, I'd really prefer not to do this. :) I'm also just curious
what's behind this behavior. I'm a chemist, not an EE, so it's also
quite possible that I'm doing something really stupid.....
I appreciate the replies.
Real address contains worldnet instead of spamnet