20 Hours later 56 microvolts apart and both seem to have similar Tempco drifts (1 PPM/C). Glitches are where I needed to use the 34401A for a proper job and forgot to switch it back to Hiz mode before hooking back onto the reference
Gratuitous screengrab just because
So that means either one, or both, need a calibration.....correct?
What one or both of them need an 'adjustment' but only after they have been looked at with an appropriate Calibration source. If I was to 'assume' an actual voltage was exactly between the two measurements then both would in theory still be inside their 1 year spec for accuracy either side of that mid point. So it is not a disaster by any means but if both are being used on the bench and disagree by 0.006% on what seems like most common DC voltages it is less than ideal as this will compound when accurately measuring RTD's (0.6C+ error at 100C) or other precision resistors.
The 34401A was last Calibrated in 2014 and the 34461A has never been Calibrated as far as it is clean and unstickered. So getting a baseline for both is important and as the 34401A was last Calibrated by Agilent elsewhere in the world it may be possible to get an idea of long term drift of it if their unit tracking is what it claims to be (7 years of data).
The other option is to do what you have done is set your 1, 10, 100 V as from your best meter then Cal the rest from it assuming in your case the Siglent to be that source. For 95% of what I do that method would be fine but some of the monitoring stuff I am playing with needs a NATA/NIST starting point.
When the A1 & A2 get back using their calibration data I can then go through 'all' of my other gear using the pair to reduce the measured uncertainty of what they are or are not. So Reference L,C & R's, AC and DC voltage sources and references, multiple other 6.5-7.5 digit measuring gear can be tested against their own specs to that combined uncertainty. That pool can then in turn be used to sanity check each other until I then get one or both recalibrated next time around. Basically chase my own tail
Temp is back up to 28 and the differential has closed to 52 microvolts so a rough idea is both meters drift slightly with temp against a heated reference and one or both are still 'wrong' in spite of tracking each other very closely over time and temperature.
Keysights Cal lab will be my third very best clock but for how many $ I am still waiting for a quote