The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: Steve Turner
Subject: Re: Humidity sensor - peculiar behavior
References: <email@example.com> <firstname.lastname@example.org>
X-Newsreader: Forte Agent 1.92/32.572
Date: Sun, 12 Jan 2003 02:56:16 GMT
NNTP-Posting-Date: Sun, 12 Jan 2003 02:56:16 GMT
Organization: AT&T Worldnet
>If the effects you measure ( up to 40% change) are due to temperature changes
>in the air caused by heat dissipated by electronics as you postulate, the
>sensor is badly out of spec. See the last graph on the spec sheet.
Marc, as I see it there are two very distinct temperature effects
here. One is the normal temperature coefficient of the sensor --
i.e., the variance in output voltage as a function of temperature *at
constant humidity*. The "constant humidity" qualification is
crititcal. This effect, the sensor tempco, is what's graphed in the
datasheet referenced above. This variance, which represents an error
term, is best measured under conditions in which the temperature
variation is slow relative to the dynamic equilibrium which maintains
the constant humidity.
The second temperature effect is the fact that the relative humidity
of air will change as a function of temperature *at constant moisture
content*. This will be the case with a proper functioning humidity
sensor, even with a tempco of zero. And this effect will be seen for
changes in temperature which are fast relative to the humidity
equilibrium. This effect is not an error term but rather a real
physical effect. The sensor is doing exactly what it's designed to
I believe that the latter effect was at least partially responsible
for the sensor drift that I was seeing, as evidenced by the
experiments that I did (posted earlier in this thread). However,
Spehro Pefhany correctly pointed out that I should at least ballpark
the figures to see if the sensor drift I was seeing would be
predictable based upon some reasonable figure for sensor heating.
I've not done this yet. :(
>Note that Honeywell suggests a way to measure air at near 100% relative
>humidity (which is difficult to do) by locally reducing the RH to about 90%.
>This involves setting the sensor on top of a 120W resistor. Nuff said.
Thanks for the reference. This is one I hadn't run across yet.
This app note states that a 5 deg. C temp increase should be
sufficient to bring the RH from 100% to 90%. Using this as a rough
ballpark guide would indicate that about 3 C temperature rise could
have accounted for the shift in RH that I was seeing.
>I have a bunch of HIH-3610 including several that have never been out of
>factory sealed packaging. Send me your snail-mail and I'll loan you a couple
>to test/examine. I paid about $18 each when they first began shipping from
>the Mexico factory so these are from early production. Dunno who might be
>charging the $40 you paid. If you decide that you need to replace yours, you
>can have one for what I paid.
I appreciate the offer, and I may indeed take you up on it. FWIW, the
$40 sensor came from Newark. Around $5 of the price was due to the
individual calibration which was supplied with the sensor.
>Other technical information of interest including humidity equations,
>constants, and concepts:
>Curious that none of the responders mentioned reading the product literature.
Well, reading the literature was incumbent upon me and something that
I should have done *before* asking questions. In my own defense, I'll
state that I *tried*, but, as mentioned previously, found the
Honeywell site to be poorly organized (IMO). I did eventually manage
to find the datasheet for the sensor, and had read it thoroughly
before posting originally.
>FWIW, I typically mount the HIH-3610 (and Panasonic PNA4603H for light and
>thermistor/LM34 for temperature ) on one side of a small pcb and a quad op
>amp and voltage references/regulators on the other side. This fa ciliates
>mounting the pcb in a case with the sensors in appropriate position with
>respect to case openings and would isolate the HIH-3610 from the thermal
>effects you postulate.
>I drive a HIH-3610 from a LM4040 voltage ref buffered by one section of a
>LM324. The three outputs (humidity, light, and temperature) are buffered by
>the other three LM324 sections. Fancy-antsy op amps aren't needed at room
>temperature to get adequate 0-5vdc (or 0-10vdc) signals . CAT-5 wire has four
>twisted pairs for the 13.8vdc supply (from a sealed lead battery with a
>"float" regulator/supply) and the three DC signals. You can gild the lily by
>regulating the 13.8 vdc at the pcb with a 12vdc Low Drop Out TO-92 package
>regulator. Sprinkle with capacitors "to taste".
I appreciate these descriptions.
Perhaps I should look into CAT-5 cabling. I've been using 4-conductor
phone wire for most of my remote sensing (rarely do I have more than
one sensor in one place). This is obviously not shielded. I've
relied on signal averaging at the ADC to filter out noise, and it
seems to work well (as judged by temp sensor calibration with a
precision thermometer). I send the operating voltage up one conductor
and take the output off another, usually with a common ground. Wire
runs are rarely more than 30'. I have an anemometer on a longer
cable, but that uses frequency encoding and is pretty much immune to
Interesting that you should mention using low dropout regulators; the
same thing occurred to me as a strategy for minimizing heat
dissipation in the regulator.
I'm curious what you use as a light sensor, and whether you take a
linear or logarithmic output. I have not yet set up "radiant energy"
sensing; have just begun experimenting with a solar cell for the
Very nice comments -- thanks.
Real address contains worldnet instead of spamnet
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup