Changing the set temperature on the A9 board would allow for new aging in getting used to the new, lower set temperature. This would mainly be an issue if the meter was powered on near 24/7. Powered on only a small part of the time, like 12 hours a week chanches are ref. is not stabilized to the high setpoint anyway.
A first point would be checking the TC of the meter. Not all units are the same and one may not need an extra improvement. Another point is using ACAL, that can compensate for the drift part, except for the reference. So the 2 parts of the DVM gain (the reference and the ADC itself) would ideally be treated separate. Trimming the reference can not really compensate for a high TC in the ADC part.
Thanks!
I see some indication of TC drift in the reference. But so far no deeper investigations that to see that after changing ambient/internal temperature and an autocal is done, there can be quite some movement in measured value of a known stable reference (not affected by ambient temperature). So plan is to take out the A9s from my 3458as and run temperature sweeps and if needed do some compensation on these.
Hello,
I guess I was the very first one to do that modification inside a 3458A, based on the idea from an old paper from P J Spreadbury (1990): "The Ultra Zener .. is it a portable replacement for the Weston Cell?"
It describes, that the annual drift rate is mainly determined by the oven temperature.
As the oven set point is changed only, but no soldering is done on the LTZ1000A, no "new aging" is to be expected.
As an example, all my stand-alone LTZ1000 run continuously on about 50..55°C (12k:1k), and all show a drift after a few months of less than -1ppm/yr.
The drift rate very slowly decreases on most references to -0.7 ... -0.5 ppm/yr., or stays constant.
My 3458A is running intermittently only, and shows a drift of maybe -1ppm/ 6yrs.
It runs on about 65..70°C (13k:1k), therefore it makes most sense to modify the oven temperature when the instrument is running continuously.
I see practically no relaxation / hysteresis effects of its reference voltage, as otherwise described in one of the Service Notes, between the regular measurements I do, which are happening on a monthly basis.
It makes sense anyhow, to determine the T.C. of the A9 assembly, the T.C. of U180 by monitoring CAL? 72 vs. TEMP?, and the overall T.C., by cooling / heating the whole instrument.
I also planned to reduce the ~ 0.25ppm/°C of the reference to zero, as the usual change of room and interior temperature of the 3458A of 0.2 .. 1.0°C during my measurements, caused by operator (100W) and fluorescent lamp (36W) is sometimes visible in the comparison measurements of my 9+ references, mostly on the order of a 0.1 .. 0.2ppm drift.
I assembled two LTZ1000A a few years ago, and zero T.C. was achieved in both cases without that 200k resistor.
This resistor is also not intended in the original datasheet for the A version, but HP made this additional mistake to assemble it on the A9 board, anyhow.
Therefore, if you determine its initial T.C., please have a look on the table I published a few years ago in the LTZ Ultra reference thread, might be around p. 65, or so. There I documented the T.C. trimming process, I think, i.e. which values lead to which T.C. That might give very rough estimates, how to change this said resistor for a found T.C. value.
The removal of the 200k will change the reference value, but again will very probably NOT re-iniate the drift rate.
Have fun, and please give a feedback on your experience.
Frank
Home
Help
Search
Login
Register
