>6.5 digits of resolution on the 34465?

[new299]

I’ve been watching Dave’s reviews and tear downs of the current Keysight meters. In particular it seems clear that the only hardware difference between the 34465A and 34470A is that the 34470A opts for the higher stability LTZ1000A reference.

However, Dave also noted that it’s possible to extract more resolution from the instrument via the software and I believe mentioned that these values have about 8 digits of resolution.

So.. if I’m making measurements where the absolute measurement isn’t critical, can this additional resolution be informative? I’m in particular interested in looking at the performance of a high resolution/linearity DAC, but I’m not so concerned about there being a DC offset... some longer term drift in the measurement.

Does that make any sense?

[beanflying]

There was a thread in the Metrology section I saw it discussed with Agilent/Keysight meters. The basics of it was the last digit was unreliable and if used would tend to contribute to errors not reduce them. Will see if I can find it (was a voltage reference thread).

Not the thread I read but well worth a read in particular Dr Frank's reply https://www.eevblog.com/forum/testgear/the-mysterious-'7th-digit'-(hp-3457a-dmm)/msg385219/#msg385219

[tszaboo]

Use ratio measurement.
https://cdn.testequity.com/documents/pdf/keysight/DMM-simultaneous-measurements-ab.pdf

Connect your Vref to one input, the DAC output to the other. Maybe this way you have less error.
The 34465A has 1ppm of reading + 1ppm of range linearity error, so anything above 6 digit is false hope.

[Dr. Frank]

I’ve been watching Dave’s reviews and tear downs of the current Keysight meters. In particular it seems clear that the only hardware difference between the 34465A and 34470A is that the 34470A opts for the higher stability LTZ1000A reference.

However, Dave also noted that it’s possible to extract more resolution from the instrument via the software and I believe mentioned that these values have about 8 digits of resolution.

So.. if I’m making measurements where the absolute measurement isn’t critical, can this additional resolution be informative? I’m in particular interested in looking at the performance of a high resolution/linearity DAC, but I’m not so concerned about there being a DC offset... some longer term drift in the measurement.

Does that make any sense?

I made an elaborate analysis on both instruments, which will for sure answer your questions:
https://www.eevblog.com/forum/testgear/keysight's-new-34465a-(6-5-digit)-and-34470a-(7-5-digit)-bench-multimeters/msg889217/#msg889217

[kj7e]

By using the statistics you can display an additional digit, but this is an averaged display of the buffered readings.  You can also use the trend chart to extract data beyond 6.5 digits.  GPIB and logging data also display beyond 6.5 digits.  While the accuracy beyond 6.5 digits is not there, the precision can be useful in some cases.  The biggest issue is keeping the 34465a thermally stable, within + or - 0.2 deg, when trying to measure beyond 6.5 digits. When measuring a 10 volt reference for example, my 34465a readings will increase by 4-6 uV per deg C rise.  But I have found some limited use for extra resolution.  Some say the 34470a while having a more stable long term reference, the internal tempco of the meter limits its usefulness as a 7.5 digit meter as well. Other than the reference, the 34465a and 34470a are the same and also share some of the same tempco limitations.

Edit, I found the ACAL is not really usable beyond 6.5 digits either.  I found it best to use the ACAL when the internal temp is at or near the calibration temp and try to take critical measurements all at the same internal temp.  That is, if the meter's internal temp is -2.5 deg C from last ACAL for the initial measurement, take all subsequent critical measurements at the same internal temp.