08-25-2006 07:06 AM
08-28-2006 04:48 AM
08-28-2006 05:04 AM
hello matthieu
let me explain my problem a little more in detail:
i'm measuring the resistances of some 100s sensors (mixture of PT100, PT500 etc.) using current sources and DMMs (voltage). i have to calculate the resistance for my own, I'm not using some hardware that does the calculation of the temperature for me.
so i'm forced to calculate the temperature out of the well known formula
t > 0 °C: R(t) = Ro * (1 + A*t + B*T^2)
t < 0 °C: R(t) = Ro * (1 + A*t + B*T^2 + (C*t - 100°C)*t^3)
this is easy for t > 0 °C, but NOT trivial for t < 0°C. so i was just wandering if someone knows a method/code/vi to calculate the temperature out of the resistance.
08-28-2006 05:26 AM - edited 08-28-2006 05:26 AM
Message Edited by GerdW on 08-28-2006 12:29 PM
08-28-2006 06:19 AM
Hi Chris,
There was a pretty good fit for Pt resistors in an old issue of Cryogenics: L.M. Besley and R.C.Kemp, CRYOGENICS August 1978, p.497
In this publication 2 polynoms were given for 2 temperature ranges: a 16th order polynom for room temperature to approx. 80-100K, and a 9th order polynom from there downward. The only values you have to enter are the resistance at 273K (that's the easy part) and the residual resistance (i.e. R at 4.2K). However, I would not recommend using a PT100 below 40-50K, as it becomes very insensitive.
I've attached a VI that contains the fit and the polynom coefficients. The constant "switch between fits" indicates at which fraction of the RT value the switch between the polynoms should take place.
Hope this helps.
Good luck,
Guenther
08-28-2006 06:36 AM
hello gerd and guenther,
thanks for your answers!
1) we're dealing with liquid nitrogen and helium, for deep freeze measurement we use so called TVO sensors (negative temperature coefficient), but we don't know the range for the PTx sensors right now, so i said down to 0K.
2.) the parameters Ro, A, B, and C are measured in a laboratory for each sensor, so i would have to calculate the polynoms for each sensor. i decided to implement a look-up table (the DMMs have <8 bit, we're using 3 ranges of the DMM, so a LUT with some 1.000s value pairs should be enough. i'm gonna calculate the LUT each time the range of the DMM changes).
thanks again for all answers.
08-28-2006 07:41 AM
I remember doing such a thing for a temp. sensor calibration stand.... more than 10 years ago...
Fist I used a step wise interation (100K step, 10K ... 1mK), worked fine but was a bit too simple
Next version used a newton (zero finding) interation with a linear calculated start value. After 3 or 4 loops the result differences where lower than 1mK.
The nice thing about both is, that you still can use the given Callendar-Van-Dusen coefficients.
Have you searched the forum, I thought some solutions where posted. Even if I find my backup they will be LV 3.1 or 4...