Humidity Sensors

Mathew Cohen says:

Try using a dew sensor from a VCR. these devices are used to detect condensation in a VCR to avoid damage. They are a small device and I dont think they should cost to much.

Jim Paul says:

...a piece of PC board with a couple of small traces on it. Use thin stuff and mount it directly to the [surface] so that it takes on about the same temperature. If [moisture] starts [to form on the surface], it also will start on the pc board. You could read it digitally as part of a voltage divider or you could read it with an A/D converter input on one of the PIC's. Or if the application permits, put two traces of adhesive backed copper tape directly on the [surface] to be monitored. Read it the same as the PC board sensor. I used a sensor similar to this for detecting rain and it worked like a champ.

Darran Logan says:

Once upon a time... I worked with a particular water detector sensor, which was basically a small ceramic tile (size of a finger nail), coated with near touching tracks. When water droplets landed on the tile, the water acted to bridge the tracks and hence lower the resistance between the tracks.
However, what I noticed was, after the initial application of a water doplet, the resistance would quite rapidly increase (over minutes not hours). It started off at something like a few tens of kOhms, and ended up a few hundred kOhms after a few minutes... even though the doplet remained the same size. [..so use AC vice DC to prevent this effect]

Annie Ogborn says:

Put a collimated beam of light through [a glass] at right angles. Put a light detector at an off angle. (obviously all this needs light sheilding from the outside world).
When the glass is clear the beam will pass straight through. When the glass is fogged, the beam will be reflected at random angles off the little spherical dew particles.
Another way -
put a beam of light through the glass to a detector. Where it passes through the glass, put four resistors in a square as a heater loop.
At intervals (I don't know how often you need to do this measurement) measure the light transmitted, turn on the heater for a minute, and measure again.
If the light has increased, you've heated a local area and cleared it. You can do your measurement by bumping a reference voltage up until it matches the output voltage, then doing a differential compare.
Why not just pass light through and compare to a reference? Probably not sensitive enough considering the long term noise fluctuations.

PIC Temperature Sensing

Humidity

Tom Handley of New Age Communications says:

Low-cost humidity sensors behave like capacitors that vary with humidity to change the frequency of an external oscillator. Typical relative humidity ranges from 10 - 90%. Philips makes (or use to) a low-cost sensor (P/N 2322 691 90001). General Eastern also have low-cost sensors in their G-CAP line. Their sensors cover 1 - 100% RH.
For my PIC-based weather station, I needed better accuracy and I wanted to reduce the complexity of the support circuitry. I ended up using HyCal sensors which provide an output within 0 - 5V and require minimal support. The outdoor sensor is an IH-3602L which comes in a T0-39 can with a slotted cap and for the indoor sensor, I used the IH-3605 hybrid element. Both sensors operate from 1 - 100% and provide an output from around 0.8V - 3.9V with a 5V supply. They should be buffered and require a simple low-pass filter and, as with most sensors of this type, need to be shielded from sources of bright light. You do need to factor in temperature compensation so you need to measure ambient temperature near the sensor. You normally combine both in a package for your outdoor sensors.
There are more sophisticated sensors that provide a voltage, current, or pulse output and include the temperature sensor giving humidity and/or dew point. [see links below]
As I mentioned above, I used the Honeywell/HyCal sensors. ... you could use the IH-3605 hybrid element. I supplied +5V to the sensors and buffered the outputs with an LMC660 quad Op Amp using a +12V supply. While the LMC660 has a rail-rail output, it's input common mode range is much less and would not be suitable for a +5V supply in this application due to the output range of the IH-3605. If you are only using one sensor, then a LMC662 would suffice. The sensor first goes to a voltage-follower followed by a simple R-C filter and through another follower to the A/D. Though I used a MAX186, an 8-Bit A/D is fine for this application.
For "Comfort level", Dewpoint is actually more relevant. You will need to install a temperature sensor with the humidity sensor. In general, a Dewpoint above 60 starts to feel `sticky'. The downside for a PIC is the Dewpoint calculation... There are a lot of methods to find the Dewpoint. In my weather station, I already had a large SRAM so I used some rather complicated but very accurate equations to generate a data table. The following is a simplified method which still provides good results. Also, you are probably only interested in a small range of values so you might still want to generate a data table. Note, I broke the equation up into smaller pieces for convenience.
      Tc  = (Tf - 32) / 1.8
      a   = (7.5 * Tc) / (237.7 + Tc)
      Pw  = 6.11 * 10^a
      Pwp = Pw * (RH / 100)
      b   = log10(Pwp) - log10(6.11)
      DPc = (237.7 * b) / (7.5 - b)
      DPf = (DPc * 1.8) + 32

      Where:
         RH  = Relative Humidity (%)
         Tf  = Temperature (F)
         Tc  = Temperature (C)
         DPc = Dewpoint (C)
         DPf = Dewpoint (F)
- Tom

To calabrate: Place in jar suspended above Potassium Carbonate. Rh is

44% at 15 degrees C
44% at 20 C
43% at 25C

or 75% (sat. NaCl) and 90+% (e.g., sat. KNO3). If its for human environments, you'd want 11% and 75% (sat. LiCl, sat. NaCl).

PIC 16F876A based Temperature (DS1620) / Humidity (HS15p) Display (Hitachi LCD) by Jody Wisman