HighVoltage,
thank you very much!
I also opened my 34465A, and they still use exactly the same TaN divider from IRC, and the same technique. I found 55 Ohm and 2,311V.
The trick is to compensate the self heating at 100V .. 1000V input voltages by applying external heat by this 55Ohm resistor.
(2,33 V over 54,9 Ohm gives 99mW, like in my instrument)
In any case, the total power dissipation is always about 100mW, so that the temperature of the ceramic array is constant in first order. (see attached table)
Changing the ambient temperature might be a 2nd order effect only.
So the ratio of this divider is always constant.
Without this heating, the ratio changes by +10ppm.
I always wondered, why the 3441x, 3446x and 34470 have no special HV calibration any more, only 500V.
The 34401A, instead, has to be calibrated at 1000V and once at 500V to compensate numerically for this effect.
The 3458A has no self heating compensation at all, and might drift by max. 12ppm @ 1000V. (my 3458A performs better, drifting only -1 .. -2ppm)
P.S.: During the measurement, I discovered a new bug in FW 2.10 (not yet released officially), which concerns this heating mechanism.
Frank