An error from IEEE, so try again. Empirical measurements using type T and K thermocouples, welded in inert noble (argon) gas, as read by an HP 34970, and compared to thermocouple junctions formed by mechanical twist, solder, and ambient-air weld; and buried in separate wells, approx 1cm depth x 2.1mm width, in an extruded Al bar, where temperature dwells were at -10 and +75 C, at 28% RH, each dwell period = 2.5 hours.
For the initial 28 hours of operations, peak diff, between all thermocouples was 0.7C, after which the twisted t/c junctions became erratic. The experiment ran for approx 100 hrs. If the twisted t/c junctions are ignored, peak diff for the duration of the test was 0.9C, which resulted from a difference in response time between the soldered and welded thermocouples, between dwells. And because I weld a fresh set of thermocouples for each series of product safety Type Tests, I do not consider oxidation to contribute to error, and I do not consider it in my uncertainty calcs. I cannot justify the time and expense of welding in argon envelope; but I do attempt to perform precise and consistent welds. luck, Brian
