Ohms to temperature formula for PT-100 RTD

[Dazza]

Hi all,

Sorry for the slightly off topic thread, but does anyone know of a formula to convert ohms to temperature for a PT-100 rtd please?

I've found a formula which converts great for around -30 to +100 Deg C but need -100 to +100.

I need an accuracy of 0.01 Deg C in the conversion!.

Oh i really want to avoid look-up tables if i can

Cheers!
Darren

[Neutone]

Look in this PDF in the section that refers to "callender van dusen"
http://myweb.lsbu.ac.uk/~khayatej/ASEE2000.pdf

The resolution you want is not very difficult but the accuracy is. You might have to use floating point math with more than 6.5 significant digits.

Using only a second order polynomial you can model the responce of an RTD from -30C to 100C with no noticable error. Below -30 and a fourth order polynomial is required to model the responce curve. If you run some numbers based on a typical probe you can see what the difference results will be with a second order formula aginst a fourth order one.

[valemike]

I've done temperature calibration in the past using the simple slope/yintercept formula:
y=mx+b

x = temperature in F
y = a/d counts

Use a micromite and apply two reference temperatures, preferably over a linear region. Okay, so you now have two sets of x and y. Solve for the slope (m) and y-intercept (b).

Everytime you take readings later on, apply the formula:
x = (y-b) / m
~~~~~~~~~~~~~~~~~~~~

Even with formulas, if you're not calibrated, the theoretical formulas can throw you several degrees off.

[TSchultz]

There are a few items you are going to have a very good handle on to get anywhere near that kind of accuracy/resolution. The biggest one will the the current excitation for the RTD, there is a trade-off between getting "resonable" signal to measure and self-heating of the element itself. If my faulty memory serves me correctly the thin film 100ohm RTD have an upper limit of 1mA of excitation current, the wire-wound elements have higher limits. You should keep well below this limit to have negligable self-heating effects.

If you are allowed to change the sensor I would suggest a 1K RTD, at least this gives you 10X the resistance change for a given temperature change. If you can't change from the 100ohm RTD, then another technique that is often used is to excite with a higher current, and low duty cycle. This give a larger signal to measure, and if enough care is taken then the self-heating effects are not a problem.

I really like the platinum RTD's but there is a great deal of "care and feeding" involved to get truly accurate measurements.

One of the best approaches I have used is the LTC2400 series of sigma-delta ADC chips running with 2 channels, one to "measure" the excitation and the other to measure the actual sensor. The great dynamic range of the ADC means you can get away from the amplifiers, and thus much of the inherent noise/errors on the front end. The self-calibration/zeroing of these ADC's then gives you really good measurments to work with. This is not a perfect solution, but one that does work very well to get good resolution and accuracy, assuming you can calibrate the actual temperature measurement to an absolute reference. The absolute calibration may not be so important depending on your final application.

[Guest]

temp=(measured resistance - 100) / .384

[asmallri]

[SLF]

The best formula I think is to calculate with the steinhart-hart equation. Perfect for unlinear Temperature Sensors.
You only have to use more Calibration Points that the accuracy is good.

[PICoHolic]

Hello,

Does anyone have a reference design for the PT-100 RTD. How to interface it?

Thanks

[asmallri]

Microchip have application notes for intefacing with these sensors.
_________________
Regards, Andrew

http://www.brushelectronics.com/software
Home of Ethernet, SD card and Encrypted Serial Bootloaders for PICs!!

[FvM]

[sirius]

[FvM]

In my opinion, the most adequate solution is to fit a polynominal for the reversed function to a Pt100 resistance table. You can choose the temperature range of the fit to get optimal accuracy for your application. The acceptable temperature error determines the required polynominal order.

A spreadsheet program as Excel can be used to calculate the fit.

[meereck]

some info can also be found in Analog Devices AN-709

[sirius]

Thank You a lot! Both of You! You were very helpful.