August 11, 2021

Thermocouples in some depth

I am investigating thermocouples to monitor coffee roasting. This involves temperatures well over 400 degrees F, and the fairly common integrated circuit temperature sensors are not up to that job.

Thermocouple basics

Whenever you join two different metals, a voltage is generated at the junction. This is known as the Seebeck effect. The voltage is small and it is temperature dependent. The temperature dependence is what allows a metallic junction to serve as a temperature sensor. The temperature dependence is not linear. Critical applications will use a polynomial to convert voltage to temperature.

So there is the basic idea. There are a number of important details. One is that the voltage is very small, and a well designed circuit is needed to accurately measure it. Another is that unintended metal to metal junctions tend to pop up when you build a circuit. For example, the common K-type thermocouple is a junction between Nickel-Chromium and Nickel-Alumel wire. At some point you will want to connect this to copper wires in a circuit, and this will put Copper and Nickel-Chromium junctions along with Copper and Nickel-Alumel junctions into your circuit. Extremely precise systems will ensure that these junctions are on some block (often called an isothermal junction block) that has a known (or measured) temperature. This will allow the effects of those inadvertant junctions to be calculated and compensated for in the voltage to temperature calculation. This is called "cold junction compensation".

What is a K type thermocouple

The main virtue of K type thermocouples is that they are cheap and readily available.

As was mentioned above, a K type thermocouple is a junction between Nickel-Chromium and Nickel-Alumel wire. You typically just buy these, though you can certainly make them if you have the requisite types of wire. You can twist the two wire ends together. For ruggedness and reliability, the wire ends are usually spot welded together.

Alumel is an alloy consisting of approximately 95% nickel, 2% aluminium, 2% manganese, and 1% silicon. If you have a commercial type K thermocouple, the wires are alumel and nickel-chromium from the thermocouple junction all the way to the ends you connect to.

A type K thermocouple can measure temperatures from -200°C to 1350°C which is pretty amazing. At higher temperatures, the issue becomes not incinerating or melting wire insulation or the thermocouple housing. My application requires measuring temperatures up to 500 F (260 C) which is no big deal for a type K thermocouple.

One source says, "The type K is the most common type of thermocouple. It’s inexpensive, accurate, reliable, and has a wide temperature range. Maximum continuous temperature is around 1,100C."

At 260 C, a type K should produce 10.5 millivolts. The sensitivity is about 41 microvolts per degree C.

You can get type K thermocouples on Amazon for less than $10. I bought one and it just arrived and looks fine, but the proof will be getting temperatures from it and comparing it to yet another type K thermocouple or other known or measured temperatures.

Interface electronics

I intend to use a MAX31850 chip on a breakout board from Adafruit. This uses some kind of "one wire" interface. If I had wanted SPI (and I did, but screwed up the order) I would have bought the MAX31855 for the same price. Perhaps it doesn't really matter. The data format is similar to the DS18B20 temperature sensor, which I also have, so if you have written software for one, the other should be easy.

The device has an A/D converter and gives 14 bit readings with 0.25 C accuracy.

The Adafruit tutorial goes on about power and level shifting, I think because they expect you to use this with a 5 volt arduino. I intend to use it with a 3.3 volt ARM of some kind, so I won't need any level shifters (the MAX38150 runs on 3.3 volts) and I won't need the regulator on the breakout board. I can probably just power it via the 3.3 volt pin. We shall see.


Have any comments? Questions? Drop me a line!

Tom's coffee pages / tom@mmto.org