Though engineers often think about K-thermocouples as "40 uV / Celsius" it is true for the internal sensing element, not the gas-stream.

Even for long and thin thermocouples that tend to break and ruin the turbo (if installed before the turbo), the sensing element is somewhat lower than the gas temperature.

With the rugged thermocouple more heat is conducted away. Since the peak is thin, the speed (transient response) is still comparable, but the final established "static state" reading is effected: the sensing element is at an intermediate temperature. This makes the sensor even more durable. This is a reproducible effect (and also same between different sensors: we tested several ones from different batches and no measureable difference was found). This term is non-linear (exponent is between 1.5 and 2), so at 450C the difference-percentage is lower. But for exhaust-gas temperature one cares most at 900C and higher.

According to measurements at glowing 900C gas temperature the U/T was:

* 38.8 uV/C with sensor pushed in almost completely (beyond the recommended depth). (3% lower voltage than the naive expectation from the "K-thermocouple" curves)

* 38 uV/C with sensor pushed in 25mm (5% lower voltage than the naive expectation from the "K-thermocouple" curves.)

** That is easy to compensate (so reading is within +-1% if needed). In any case not a good idea to believe blindly in the uncalibrated textbook K-thermo curve.

Without gas-stream (that is nonsense in an exhaust, but happens in a laboratory calibrator equipment) we could measure 37.4 uV/C (that is 6.5% lower voltage than the naive expectation from the "K-thermocouple" curves).

If one wants to compensate for gas-temperature (not sensing-element temperature) we recommend to use *1.05 higher calibration value than the K-thermocouple textbooks or calibrator devices (like the one used for the calibration of the round input).

* in v3., egt1 calibration constant 80 instead of 76

* in round, egt calibration constant 162 instead of 154 This is example only, check your calibration sheet

----

Before turbo / after turbo

It is possible to place the sensor after the turbo, but the reading will be lower.

* the adiabatic isentrop expansion temperature drop can be calculated easily (see wikipedia)

** but the expansion in the turbo is no isentropic. Manufacturer data is almost never available, but 70-75% adiabatic efficiency can be a reasonable estimation. Therefore the temperature drop in the real turbo is lower (temperature after turbo is higher) than the ideal case