I am attempting to select a temperature sensor for a controlled heating application. The sensor will need to be inserted into a chamber with a cartridge heater and steel ball bearings inside of some stainless steel tubing. I would like to meet the following specifications:
- Temperature control between 160.0 and 240.0°C
- Temperature accuracy of +/-1.0°C (in the above range)
- <250 ms thermal time constant for the temperature sensor
Everything I've read/calculated so far has led me to believe that I can accomplish what I would like to with a J or K-type thermocouple with appropriate cold junction compensation and correct thermal bonding inside the chamber while using a 12-bit ADC with signal conditioning mapping 0-4.5V to 140.0-260.0°C. Based on feedback the duty cycle of the cartridge heater will be adjusted. My questions are:
- Should I consider other types of sensors (have I overlooked something critical in my sensor selection)?
- What aspects of design will be the most critical to getting the most accurate possible temperature readings? (Guard rings, return current routing, etc.)
Best Answer
A thermocouple is probably the best available sensor for that set of requirements but it requires careful design of the cold junctions and associated CJC sensor and thermal EMFs. No guard rings are required, thermocouples are very low impedance sources.
If you can electrical connect the junction to the mass being controlled you will get faster response but you may have to float the front end etc.
1 degree C absolute accuracy is difficult to guarantee (remember that includes the wire limits of error, extension wire matching, any plug, sockets, cold junction sensor errors, as well as the amplifier and ADC and linearization errors) however, with careful design, stability of << 1’C is not so hard.
There are plenty of well designed commercial controllers that include isolation, and a lot of really crummy cheap ones on the market, you could consider using a decent one rather than going component-level for this common function.
You could also consider a thin-film RTD if you could couple that thermally and fit it. The design job for the circuitry then becomes almost intern-level.
As an aside, you should take care that the thermal coupling of the windings inside the cartridge heater is as good as possible to the sheath, otherwise you’ll degrade the control response considerably.