I have two DAQ cards along with a number of others for a distillation system. Some of my PT100s require calibration as they are off spec. I have been trying to calibrate my sensors but am finding that none of the attempts are taking hold.
Both cards are below:
Mega RTD firmware version 2.03
Vin 5.131V, Vrasp 5.112V, CPU Temp 51C
Mega RTD firmware version 2.03
Vin 5.200V, Vrasp 5.166V, CPU Temp 51C
As the sensors themselves are off spec, I have been using hot and cold baths to calibrate. My cold bath is an ice water solution around 0c and my hot bath is an oil solution around 140c.
My calibration procedure has been the following:
1. Submerge sensor into the cold bath
2. Run something like "watch -n 1 rtd 1 readres 1" on the given sensor and wait for resistance to stabilize.
3. Enter correct ohms for the stabilized temperature in the first calibration: "rtd 1 cal 1 100" for example.
4. Place sensor in hot bath. Wait for resistance from #2 to stabilize.
5. Enter correct ohms for the stabilized temperature in the second calibration: "rtd 1 cal 153.58".
However, once this is completed, the temperature does not change. Am I doing the calibration incorrectly? I have tried this with and without calrst as a precursor and neither seems to make any difference.
The calibration was made to compensate for the measurement errors but it might do the same for the sensor but I am doubt the sensor error is linear with the temperature.
Anyway, I notice that for point 5 of your description, there is an error in the command; it must be "rtd 1 cal 1 153.58", not "rtd 1 cal 153.58". I am guessing you did the typo only here. The second problem I see is that the two points of the calibration are too close.
Thanks for that great info.
You are right about #5. I just typod it.
So that explains the recommended calibration only using short/resistor connectors. It's really about calibrating the thermistor and contacts.
What sort of error should one expect from a commodity sensor? I've been seeing +/- up to 2c - which is why I was trying to calibrate them. Incidentally, I also use your 8 port TC cards and I see similar errors with commodity K type probes on them. Could that be due to some wider system issue possibly?
Cheers Alex
Theoretically, if you calibrate the RTD card, you will get under +/-1c but probably this depends on the sensor also.
The thermocouple is used for a high measurement range, not for precision. More, we compensate for cold junction by measuring the connector temperature with thermistors, which also introduce some errors (+/-2c °C is what you expect from the thermocouples).
Since you use a Raspberry Pi to read the temperatures, you can do a multipoint calibration (mapping) to get more precise temperatures.
I just invested in some high quality sensors so will see if they perform any better. I've been troubleshooting this for a while and suspecting cheap sensors are of poor quality. I was expecting a precision of 0.1c or so for TC and 0.01c for RTDs. Hopefully I will at least get linearity with any errors that come with the good sensors.
I haven't heard of multipoint calibration. Where can I find some info on that? I'd love to see more. Is there any sort of all-encompassing documentation on calibration for either the TC or RTD cards?
Thanks again Alex
What I meant by multipoint calibration was to record multiple points of read and real temperatures, then find the polynomial to apply to the measured data to match the real temperature, so you can apply that polynomial function to the measured data. You can do that with any data analysis program if you use fit algorithms. I remember many years ago, I used Origin, but you might have similar results with Excel.
Oh you meant to do the transform in software. I've been mulling over writing a generic implementation for other data sources and controllers. After I get these new sensors in, I'll see how they do and figure out how soon I need to write it.
Thanks a bunch Alex!
