On 6/1/05, PicDude <picdude2@narwani.org> wrote:
> Hi all,
> 
> Trying to calibrate a resistive temperature sensor, but my existing method is
> producing some inconsistent results.  Pot of oil on a stove at various temps
> and comparing the sensor to an LM35DT (~0.5 deg C accuracy).  The problem
> seems to be that the resistive sensor takes a significant time to come up to
> the temp of the oil and settle to a stable resistance.  If I read the data
> points when the oil temp is gradually rising, it varies significantly from
> when the oil temp is decreasing.   It is very difficult to get the oil to
> stay still at a specific temp for any decent length of time.  However, in a
> pot of boiling water, I can get a stable and trustable measurement.
> 
> So I need to do this for a range of around room temp to a few hundred degrees
> F (say 160 deg C), and would like to select easy-to-find chemicals that have
> various boiling points in this range and would be safe for me to boil on a
> stove.  I have found a bunch on the internet (example:
> http://www.engineeringtoolbox.com/boiling-points-fluids-gases-24_155.html ),
> but I'm not sure where to find much of these, and if they're safe to boil on
> a stove.
> 
> Any ideas how I can go about this?

It seems most of the fluids on that list are flammable or will oxidize
(burn) before they boil in air.  I don't think you're not going to get
much range out of adding salts to water, either.  First, I would
mechanically attach the two sensors so they have good thermal contact.
 This will reduce error due to differing thermal time constants
somewhat.  Next, the oil should be stirred to eliminate hot spots. 
Without active control of the temperature, though, you'll get the best
results by turning off the heat and insulating the oil and allowing it
to slowly cool with stirring.  A simple time proportioning PID
temperature controller would make this easier as a large vessel of oil
has a fairly big time constant.  This would allow you to let the
temperature settle.

Other than all that, you might be able to find a very low temp solder,
but your temperature IC is probably not going to like that
temperature.  Perhaps you should think about getting a calibrated
glass thermometer.  I would use a NIST-traceable standard platinum
resistance thermometer, but I don't have to pay for the probe or the
calibration.

Bradley

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist