I've done that and reported this first about 10 years ago on the time-nuts list. By adding a 100k, < 10ppm/K resistor in parallel to R411, 15k, on alternative assembly place X411 (see CLIP), oven temperature is lowered from about 95°C to 65°C, improving annual drift to typ. < 2ppm/year. See appended picture of my changed reference board. You need to re-calibrate your instrument because this will change the reference voltage by about +50ppm/K *(-30K) ~ -0.15% Afterwards, you also have to limit the ambient temperature to a normal cal lab temperature (18..28°C), i.e. max. ambient temperature 35°C, keep the fan always clean (dT = 13..15K), and avoid using the instrument inside a closed rack. In total, you need to monitor the internal temperature, TEMP?, which must not rise above ~55°C any more. I use my instrument in constant environment in our basement, that is mostly 21..23°C over the whole year, TEMP? is 34..38°C. Precise measurements are not possible without having such stable conditions, btw. Recently, I did a new DCV calibration of my 3458A, which is running intermittently only. The absolute drift of the LTZ 1000A after 7 years was +1.8ppm, compared to specification of 8ppm/year (probably specified for running 24/365). Frank