Well, I'm using a dynamic temperature change method, so the measurement are always done during ramping up/down of the temperature.
If that change is slow enough, nearly everything is in equilibrium, and the measurement will be precise enough.
My final setup is here:
https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg617211/#msg617211The aluminium block, which carries the resistor and the thermometer are additionally mounted inside an outer aluminium box so that also all the wires are on the the same temperature as the alu block and the outer case (thermal short circuit), This is essential, as most of heat transfer happens via the resistors wires.
I only have difficulties to observe very slow events correctly, like the described hysteresis and creeping effects on some of the econistor resistors.
It would be very useful to let the resistors relax on a constant temperature after a change for many hours.
I'd expect that the formerly hysteretical resistors would have a much lower T.C. than being falsified by the hysteresis effects.
General Resistors told me, that they let their DUTs rest for 24h on the edge temperatures, and they had totally different results than me on the same specimen.
In that sense, Andreas solution is much more versatile, and the temperature can be controlled in a much better manner, and the whole temperature profile (T-loop) can be achieved automatically.
B.t.w., I will soon publish T.C. adjustment experiments on my new LTZ#5 reference, where these hysteresis effects, supposedly caused by exactly these resistors, are strongly dependent on the temperature change rate, and so does the evident T.C.
For a quick selection and characterization of the PWW resistors and the LTZ circuit itself, my dynamic measurement approach was completely sufficient.. the LTZ reference after a few attempts, now has nearly zero T.C.
Frank