measuring temperature with PT100

[Exo]

I'm trying to measure a temperature using a PT100. Basically what i want is to get a certain range (between -10°C and 50°C) to output a voltage from 0 to Vref(5V for example) , so -10°C = 0V, 50°C = VREF. So i can feed these readings into the ADC of a pic.
Anyone got any ideas how this can be done accurately... would 0.5°C precision be possible ?

It would be nice, but not neccesairy, if the circuit included a compensation for the resistance of the PT100's connection wires, I see this done at work on some machines with 2 wires connected to 1 side of the PT100 ...

I think i need to go deeper into opamps to do this and that's not really my territory :?

 

[Russlk]

I don't know what the PT100 is, but have you looked at the LM235AH? It outputs 10 millivolts per degree kelvin (2.98 volts at 25 deg C) It operates over -40 to +125 deg C and can be +- 1 deg accurate over that range. It is a simple matter to convert the output to 0 to 5V for whatever range.

 

[Exo]

The PT100 is a RTD, it's resistance changes with the temperature.
I have to use a PT100 because the probes are already installed in the installation. I just want to replace the control box with my own so i can add options i want using the pic ...

 

[Russlk]

What is the resistance at -10 deg C and at +50 deg C?

 

[Exo]

-10°C = 96,09 ohm
50°C = 119,40 ohm

 

[Russlk]

How many wires come from the RTD? Is the resistance element accessable? I will assume 4 wires for now since you are concerned with accuracy.

 

[Exo]

No, there are 3 wires coming from the probes, 1 side is connected double, the other is not.

 

[Russlk]

Three wire is a standard way to connect the RTD, but 4 wire is better. The difference is not significant unless there are hundreds of feet of wire. If you cannot acess the RTD, connect two wires to the single one.

My schematic uses the LT1006a opamp, it has 1 microvolt/deg C drift which will account for 2% error over the 60 degree range, if the circuit is exposed to the same environment as the RTD. To avoid this error, you can put the circuit in a temperature controlled oven, at 50 deg C. The other source of error is self heating due to the bias current. There is a nonlinearity due to this, but I don't know the thermal characteristics, so can't estimate the error. If the RTD is thermally connected to a heat sink, the error will be small.

 

[Exo]

Thanx for the circuit, i'll give it a try...
Some of the errors you speak of I can compensate in software so it might just work