It's indeed a 2 wire cable and the length is 2 meters.
To test I tried to calibrate the setup with a 100 ohm resistor (simulating a PT100) and once this calibration is complete it works perfectly (tested with a potmeter). No deviations (worth mentioning)
I'm not sure quite what this means as the problem could still be anywhere.
That means the MCP3551 is fine and that the breadboard is not causing me a problem but don't know what other conclusions to make.
I guess the problem is with Rb or the combination of Ra/Rb or maybe I do get some noise (the capacitors having a wrong value?) the further away from the calibration point. Is that possible or would noise cause more random problems?
I will try to adjust the Rb value to see where that takes me? Should I keep Ra and Rb the same?
Maybe I need to look at the library?
What should I graph? Rpt measured with my setup vs Rpt measured by another means?
Could you tell me why the Ra should be a 1% resistor and Rb only 5? edit: this doesn't matter after correct calibration
It's indeed a 2 wire cable and the length is 2 meters.
calibration
That is the problem, those 2 metres of wire.
You need a kelvin connection.
Read my earlier post 4 or 5 and try to understand the reasons for 2,3 and 4 terminal connections..
Wiki explains fairly well.
As far as Ra is concerned it probably will not matter for your application but for high precision and repeatability between units and interchangeability of ptz heads temperature effects can upset the calibration.
From the data sheet 0.1 % gives a 2 deg error but over the range you are using it should be better than that.
Using high precision resistors makes it easier to make lots of modules that match.
they are expensive though.
In large production they would be laser trimmed after soldering as its a lot cheaper that way.
To test I tried to calibrate the setup with a 100 ohm resistor (simulating a PT100) and once this calibration is complete it works perfectly (tested with a potmeter). No deviations (worth mentioning)
You carried out you calibration with a resistor which has a very different temp coefficient from a ptz.
By dunking a ptz in ice water you are also dunking the copper cables in ice water (or enviromentaly cooling them)which have a different temp coefficient.
Using 2 wire ptz in say a cupboard with the storage cylinder in it will significantly affect the readings.
I've made the following adjustments:
Using 3,3V instead of 5V
Rb = 1k
Ra = 70k
Smoothing of the returned values
and this seems to work quite well, still not perfect but Temp readings are stable within about 0,2C.
I think I will leave it like this for the moment. I've contacted the manufacturer and if they get back to me with more info I'll post it here.
As you were talking about interference I guess it would be a bad idea to put the MCP3551 into a shield (along with one or two more shields for the LCD and DS18B20 and relays to control the outputs)? Or should it be fine?