We get a lot of questions that sound like this: “How can you say thermocouples are less accurate than thermistors? I have a Thermapen and it is very accurate! You [turkey/rascal/hooligan].*”

First let me say that Thermapen is a perfectly good product and Thermoworks, the company who makes it, seems to be exclusively employ very nice people. They make a wide variety of industrial and scientific thermometry tools, and have the skills to do a good job. But Thermapen was designed to be fast above all else; its accuracy is in spite of its thermocouple, not because of it. There are a few reasons for this to be true, but the biggest and easiest to understand has to do with a technique known as “cold junction compensation”, or CJC.

I mentioned in Part 1 of this series that thermocouples are differential temperature sensors, and by that I mean that they can only be used to measure the difference between two temperatures. For a thermocouple, those two temperatures are at the “hot” end and the “cold” end of the wires. A thermocouple by itself can tell you that the hot end is 90℉ hotter than the cold end but it can’t tell you that the hot end is 165℉. That’s just how they work. To take an absolute temperature measurement with a thermocouple, you have to know the absolute temperature of the “cold” end ahead of time. Measuring the absolute temperature at the cold end (75℉) and then adding that to the thermocouple’s differential measurement (90℉) to get the absolute temperature at the hot end (165℉) is called “cold junction compensation.”

Any type of absolute temperature sensor can be used for CJC, but all the error in CJC adds directly to total error, so doing a good job of CJC is important. If you’re willing to void your Thermapen warranty, you’ll find a thermistor on its circuit board, right between where the two thermocouple leads are attached. Nearby you’ll see the letters “CJC,” silkscreened crisply in white ink.

The accuracy of their thermocouple measurement depends on the accuracy of a thermistor measurement. All things being equal, how could a thermocouple/thermistor solution be more accurate than a thermistor-only solution? Is thermocouple error zero? Is it less than zero? No, of course not. The Thermapen design achieves good accuracy by calibrating every unit individually after they’re built, which allows system errors (from the thermocouple, CJC, amplifiers, etc) to be quantified and subtracted away. The Range and RangeOI design achieves good accuracy by avoiding all that error in the first place. It does not require calibration to be accurate. We just build them that way.

*No one has actually called us a turkey, rascal, or hooligan.