Two outlet thermocouples, two inlet thermocouples, of which all four are from a matched set of five thermocouples, all of which agree when the tips are all in the same spot. A third outlet temperature sensor is built into the Bosch BME280 atmospheric pressure/temperature/humidity sensor. Two other stand alone RTD sensors with digital displays plus a mechanical thermometer monitoring the air temperature around the calorimeter (but these are not recorded).
The “extra temperature” is there when no heat source is active in the box.
In other words everything is at room temperature.Easy enough to fiddle with anyways, since the two inlet and two outlet thermocouples are connected to a dedicated thermocouple datalogger with a live digital display for all 4 channels, (and then that digital TC data goes to the main datalogger along with recorded voltages, currents, BME280 data, etc.
I'm writing this without knowing exactly how you set up the measurement (its faster for me just to throw out ideas) - but here are ideas for the "extra temperature"
1. radiation cooling of inlet air (because it is near a window or a cold room) and/or radiation heating of outlet air (closer to a hot room or furnace).
2. does fan power explain it? I think you said it doesn't.
3. does the "extra temperature" only show up when fan isn't running? if so, data without fan running should not count because fan reduces thermal gradients
4. make sure you are happy with the amount of thermocouple area exposed to the air flow because it should be as much as reasonably possible.
5. if it persists, then you can zero it out and hopefully it is small and doesn't interfere with the excess heat measurement