Water-cooled hardware heat load from inlet/outlet thermocouples: cross-calibrate the pair, it costs nothing

When heat load is measured as mdot x cp x (T_out - T_in) on a water-cooled test article, the error is dominated by the two probes disagreeing, not their absolute accuracy. Off-the-shelf Type T or K probes are often specified to about +/-1 C each, so a pair can be 1 to 2 C apart. On a ~30 K water rise that is 3 to 6% of the measured heat, enough to matter if a run-abort rule is set at, say, 1.3x the expected heat. Fix: before testing, put both probes (with their actual extension wire and DAQ channels) in the same stirred water bath at two temperatures and record the offset; subtract it in post-processing. Other thermocouple notes for cryogenic test stands: thermocouples, unlike pressure transducers, barely change in price with pressure rating (a 1/16 in sheathed MI probe in a bored-through compression fitting holds well above typical feed pressures), so the cost per point is low. Use Type T near LOX/LN2 temperatures (Type K loses accuracy there), use a bored-through fitting (not a tube-stop one), retighten after the first cold cycle, and run matching thermocouple extension wire all the way to the DAQ.