I had an interesting experience with a customer who was utilizing an S7-1200 system. The PLC used was an S7-1200 series CPU1214C with various I/O modules, including a SM1231 RTD input module which would be used to read four (4) Pt-100 RTDs in the field. Seemed like a simple application, right? I first realized that we had a problem when I noticed that we only had two (2) leads from each temperature device. OK, I thought, these are just 2-wire RTD's. I'll just jumper where necessary. So, I jumpered the 2-wire configuration as per the SM1231 RTD datasheet and then proceeded to scale my inputs within the TIA Portal software. When I got online, the readings I viewed for the four (4) RTD input channels were not what I expected. Normally, a Pt-100 RTD wired correctly would be read by the SM1231 RTD Input module as an integer value which would be approximately units x 10 of the temperature being monitored. My readings online were approaching 32,767 - which would typically indicate an "open" in the channel. I took out my DVM and measured the resistance of one of the RTDs and discovered that my resistance was almost 3 KOhms. I thought at first that I just had a bad RTD - until I measured the resistance of the other three temperature sensors: All measured around 3 KOhms. Now, I was beginning to suspect that my temperature devices were not RTD's, but rather Thermistors. I removed one of the RTDs from the thermowell and read the P/N off the sheath. Sure enough, I had thermistors (2,252 KOhms @ 25°C). Again, I thought that this would be no problem. Surely, the Siemens SM1231 RTD input module had a setting for thermistors - or at least for resistance in the range that I could use. After reading up on the SM1231 RTD input module, I discovered that the module was not compatible to thermistors at all. The only resistance ranges the SM1231 RTD input module could read were 150 Ohm, 300 Ohm, & 600 Ohm potentiometers. The maximum input resistance that could be read without the channel detecting an open sensor is 750 Ohms.
After discovering the issue, the simple fix would have been to just requisition four (4) Pt-100 RTDs; however, the inspector was coming in the following day and the unit had to be ready. I had to find a solution that would work with the hardware that I had at the time - and I had to find that solution fast. I looked online for ways to mitigate the high resistance of the thermistor, but I could find nothing that would help. I then remembered from my basic electronics courses that if a smaller resistance were placed in parallel to a larger resistance, the total resistance measured in the parallel network would be less than the smallest resistance. Since the upper optimal range of my SM1231 RTD input module was 600 Ohms, I chose a 600 Ohm precision resistor. The characteristic curve of a 2.252 KOhm thermistor is non-linear. The programming required to extrapolate the correct temperature from a parallel 2.252 KOhm/600 Ohm parallel network would get messy - or so I thought. I took the data from the 2.252 KOhm spreadsheet and then inserted a formula for a parallel resistance network and re-charted the data. The results were astounding. My temperature curve flattened out and the parallel network actually improved the linearality of the curve. See below:
See the 600 Ohm curve below:
I built a scaling program in the TIA portal software and was able to read the real-time temperature values correctly. Our inspector came in the next day, evaluated the performance of the unit, and signed it off. We weren't required to ship the item to the final destination for another week, so I procured the correct Pt-100 RTDs and had those devices retrofitted. I installed the necessary programming to evaluate the new RTDs and we shipped the package on time. Although I didn't leave the parallel resistor/thermistor network in the package, I did prove that you can use the SM1231 RTD input modules with thermistors - with the addition of a known resistance in parallel - and that placing a fixed resistance in parallel with the non-linear thermistor device will actually make the resulting temperature curve more linear.