Somewhat of a physics or thermodynamics question as well...

I recently calibrated a resistive-output temp sensor with reference to an=
=20
LM35DT sensor, and have a tons of data-points.  So far so good.  I used 2=
=20
DMM's and the resistive sensor was unloaded.  Yes, there are many sources=
 of=20
error here, but most of those I can't do anything about given the equipme=
nt I=20
have.

However, now I'm considering that when I put this in a circuit, I'll have=
 it=20
as part of a voltage-divider (fixed resistance calculated for max range a=
t=20
middle of voltage divider), and send the middle of the divider to a PIC=20
input.  There will be a current flowing thru the divider, and therefore t=
hru=20
the temp sensor, and the self-heating will cause the sensor to change out=
put=20
resistance.  So I'm thinking of compensating for this.  I'm still not sur=
e if=20
it will make a noticeable difference, but it seems like a fun & learning=20
exercise.

The 3 ways I can think of are...
(1) to actually place the temp sensor in the final circuit and record the=
=20
voltage at the ADC input pin for each temp setting.
(2) to actually place the temp sensor in the final circuit, have the PIC =
dump=20
the raw reading to a display, and calibrate against those values,
(3) to calculate the heating effect.
* (4) Yes, I can always send the input to an op-amp or similar instead to=
 make=20
any self-heating effects negligible, but I'm not going to do it this way.

(1) and (2) are no problem.  But I'm wondering how I would do (3).  With =
my=20
existing table of unloaded-resistance vs. temp data points, I can calcula=
te=20
the current-flow thru the sensor, the voltage output, and thus the power=20
dissipated inside the sensor.  From here I know the temp rise and then fi=
nd=20
the new output resistance from the adjusted temperature.  Of course this=20
adjusted temperature will cause the sensor to change again.  I expect the=
=20
effects are diminishing so the sensor will come to a rest value.

My 2 questions are ... how can I calculate for this recurring effect?  An=
d how=20
can I calculate/find the actual temp rise for a given amount of power=20
dissipated within the sensor?  I have absolutely zero data on the sensor,=
 so=20
hopefully it won't require any thermal conductivity or other data like th=
at. =20
Thoughts and opinions welcome.

Cheers,
-Neil.

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu