As I read it, a thermocouple (in my thermometer) works by comparing the temperature that I’m trying to determine with a set, known temperature inside the device, and figuring in the meaning of the different voltage that is generated. But how is the “known” temperature determined?
Thermocouples are inherently differential devices.
So, the “cold junction” temperature needs to be set, either by a reference temperature (like, an ice bath), or by an electronic cold-junction compensator (which generates a voltage equivalent to an actual ice bath), or by using ambient temperature (OK for high-temp applications, where the difference between 0C and 25C is no big deal).
But, the first question is - what makes you think your thermometer is using a thermocouple? Very few do - most use Silicon temp sensors, thermistors, or even a transistor to measure the temperature.
If you have a regular ol’ thermocouple readout + thermocouple sensor, then it’s the latter; the readout has a thermistor that measures the temperature of the connector (which is essentially the cold/reference junction). Higher-priced ones use a PRT to measure the connector temperature.
To answer the OP, the temperature of the thermocouple sensor is “known” based on the following:
- Alloy of the Conductor A.
- Alloy of conductor B.
- Temperature of the cold/reference junction.
For #1 and #2… the thermocouple is comprised of dissimilar metal alloys (Conductor A and Conductor B). NIST publishes temperature versus voltage tables, along with polynomial coefficients, for various combinations of alloys. The most common types include T, K, E, and J. Types S, R, and B are used for precision readouts.
So let’s say you have a thermocouple readout that can accept types T, K, E, and J thermocouples, and you’re using a type K thermocouple. When you configure the readout for reading a type K thermocouple, the microprocessor will either load this lookup table into memory or will load these polynomial coefficients into memory. When you make a temperature measurement the microprocessor will do the following:
- Measure the voltage between the hot junction and cold junction.
- Measure the temperature of the connector (on the readout) using a thermistor. This will be the cold/reference junction temperature.
- Use the lookup table or polynomial coefficients to convert the voltage into temperature.
The lookup table assumes a cold/reference junction temperature of 0 °C, so the microprocessor performs a compensation based on #2.
S, R and B are for high temperatures, not precision. They are actually less precise in general because their voltage signals are smaller.
Good question. I don’t know what led me to that assumption. I’m just using an inexpensive instant read thermometer - for air temp and for occasional grilling. What is the method it likely uses?
Oops, my bad. :o They’re expensive because platinum is expensive, not because they’re more accurate.
For ambient temperatures thermistors are usually used as the sensor. This is because the signal conditioning circuitry is simple & cheap. The disadvantage is that they have a fairly limited temperature range. A thermocouple with a bare-metal bead can be used at high temperatures that would easily destroy a thermistor.
On a engineering project for a college class for this my team made the initial recommendation of hiring a illegal immigrant to maintain the ice bath, on the second revision we replaced that with a solid state temperature sensor.