When two different metals are joined, a voltage is generated across the junction, and the voltage depends on the temperature of the junction. A thermocouple uses this voltage to measure temperature. The voltage is small, and the relationship of the voltage to temperature is not linear, so special circuitry is required to determine the temperature.
Some of the things that I think thermocouples work well for in the solar thermal world are:
On the down side, they typically don't have the accuracy of Thermistors -- getting +-1F takes a fair bit of attention to detail.
Thermocouples come in several basic types depending on the particular two metals that they use. There are K, J, N, T, E and probably more thermocouples. Each of these types uses a different metal pair to generate the temperature dependant voltage. Each of these types has its own strong a weak points, but for most part any of them will work for the range of temperatures and conditions encountered in solar thermal work. Since each of these types requires unique wire type, connectors, and measuring equipment, it can be expensive to use a mix of different types of thermocouples. Best to standardize on one type. The most common and best supported type of thermocouple is the K. J thermocouples are also well supported.
Thermocouples come with a wide variety of junction ends. The cheapest is just a small bulb of material where the two wires join -- these work for a wide variety of circumstances in solar thermal work. Another form has a protective metal sheath over the junction to protect it from harsh environments. Another useful configuration has a sort of flattened end that is embedded in an adhesive pad -- these are good for measuring surface temperatures -- just stick them on. Thermocouples with low mass junctions that have a lot of surface area are also available, and react very quickly to temperature changes.
Picture shows thermocouples -- CW from lower left: 1) simple bead junction, 2) adhesive pad surface mount, 3) a metal sheathed junction, and 4) another surface mount adhesive pad. These are all K type thermocouples -- you can tell this from the yellow connector color.
While the thermocouples themselves are not to expensive (5 packs for $50ish), the connectors, extension wires, and the meter with the measuring circuitry all add up -- so make sure you add up the full cost of the system before committing to thermocouples.
I use the setup shown below to measure the output from up to 10 thermocouples at a pretty low total price.
The meter is just an ordinary multi-meter that measures voltage, resistance just as all multi-meters do, but it also has a socket for a K type thermocouple and provides two scales for measuring temperature. This adds very little to the cost of the multi-meter, and many of the multi-meters made today come with the feature.
In order to easily measure temperatures from multiple thermocouples without having to plug and unplug them all the time, I use a thermocouple switch which will accept inputs from up to 10 thermocouples, and depending on the switch position, connect only one of them to the meter at a time. The switch is made specifically to switch thermocouples -- an ordinary rotary switch will introduce error. The switches can be purchased from Omega.com or other places that handle thermocouple equipment. They also seem to come up on ebay fairly often.
Many other thermocouple reading meters and loggers are available. There are a number of 1 and 2 thermocouple meters that are fairly inexpensive -- some of them have some data logging capability. There are also some electronic kits available that can be used in front of data loggers to convert the TC output to a more normal data logger AD voltage range. On the high end, you can spend many thousands of dollars on lab quality thermocouple loggers.
The extension cords to allow the thermocouple to be placed some distance from the meter are different for each type of thermocouple, and are expensive (a good reason to standardize on one TC type). These tend to come up on ebay once in a while at more reasonable prices. The same logic applies to the connectors.
If its not terribly critical to get very accurate measurements, then at the start of the project, I bring all the thermocouples to the same area, and hook them up to the meter via the switch. I tie them together, and give them time to come to the same temperature. I then see what each one is reading on the meter -- if any of them are too far off, I replace them.
If more accuracy is needed, you can use the ice bath and boiling water procedure to calibrate them. This is pretty time consuming. If you have an accurate thermometer, you can use it to calibrate the TCs at one or more temperatures.
Thermocouples and thermocouple equipment (meters etc.) are available from a wide variety of places. Electronics supply places like DigiKey sell multimeters that read thermocouples. I buy most of my stuff from Omega.com -- they have competitive prices and don't mind small orders. I'm sure that there are many other good suppliers out there. Ebay often has thermocouple related equipment, and you can find some good deals, BUT, I have seen TC stuff sell for more on ebay than they you can buy the same item for new at retail. You also need to be very, very certain about what you are getting.
Gary Updated Oct 10, 2007