DMM ACAL function design/implementation detials.

[miro123]

I was wondering how the ACAL function is implemented.

Does somebody know documentation describing detials of ACAL function?

I'm interested in implementation in vintage devices like 3458a as well as in modern one like 34465/34470.


BR,
Miro

[Kleinstein]

There is rather long thread with repair and teardown on the R6581 DMM that at the end includes a quite detailed description of the ACAL process in that meter.
For the 3458 the HP journal 1989-4 about the 3458, though not that many details there.
The HP 3455 is one of the first meters with some ACAL (HP journal 1977-2).

The process should be about the same: measure the same voltage / current in different ranges and calculate the gain ratio from that.

The Datron 1281 is a little different: it uses a factory calibrated precision transformer as a stable divider and uses this as additional reference.

[alm]

I think it's important to distinguish what Datron called Autocal, which was their name for closed-case (purely digital/NVRAM) adjustment, as opposed to adjusting trimmers as was common at that time, and artifact calibration as in the HP 3458A. Fluke also wrote some papers about artifact calibration in their 5700A multi-function calibrator, for example Artifact Calibration: An Evaluation of the Fluke 5700A Series II Calibrator and Artifact calibration: theory and application.

To understand artifact calibration, it might make sense to first understand how the manual process of calibrating a 10 MOhm resistor from a 1 Ohm standard resistor: Find a way to do a 1:10 transfer (e.g. a bridge setup), then transfer up from 1 Ohm to 10 Ohm, 10 Ohm to 100 Ohm etc until you reach 10 MOhm. If you can get the ratio really accurate, you can get a quite good uncertainty this way. See the diagrams in TiN's article on CalFest/RepairFest 2023 or The Fluke paper "THE EFFECTIVENESS OF ARTIFACT CALIBRATION IN COMPUTING INTERNAL RESISTANCE VALUES" what the uncertainties can look like if you have the right equipment and procedures / skill.

The HP 3458A uses its highly linear by design ADC to do these accurate ratio transfers. When doing artifact calibration, it will measure the null, 10 V DC relative to its internal LTZ1000 voltage reference, and 10 kOhm relative to its internal 40 kOhm reference resistor. Then when running ACAL ALL, it will transfer up and down from its 10 VDC range to the other DCV ranges using its ADC, do the same for resistance and current with clever switching, transfer DCV to ACV and use that to calibrate the ACV ranges.

[Mickle T.]

A highly simplified INTCAL algorithm of the Advantest R6581 multimeter.
In reality, most procedures during the calibration process are encoded in a kind of bytecode, and are decoded and executed by a unified interpreter. The procedures include numerous calls to pre-calculated delays, as well as references to about fifty constants in the EEPROM, the values of which are flashed at the factory. In this way, numerous sources of static and dynamic errors during artifact calibration are compensated. Of course, manuals and beautiful advertising magazines/booklets don’t write about this