[G0HZU_JMR]

For what it's worth I just checked my 1980s Maplin (Precision Gold?) model DMM on a PSU and set the PSU to just trip the meter LSD to 19.00V. The Keithley meter measured 18.997V.

1.900V on the Maplin meter showed 1.8999V on the Keithley. This was one of the early Maplin PG series meters and they were very solidly built with decent resistor networks etc. This meter hasn't been adjusted in many years and I'm not sure I've ever adjusted the DC voltage range. My cheapo yellow TMK meter I bought as a £1 factory reject at a radio rally in the mid 1990s was just as good at 1.900V and measured 18.99V at 19.000V on the Keithley. It has never been adjusted since about 1995 and just needed to have the display connection cleaned to get it working back then.

The Fluke 45 measured 5.000V on the ref195 based test PSU I have here and both the Keithleys were within 10uV of each other here.

The sensible advice (given many times already on this thread) is to pursue the approach using the precision resistor plus a 4.5 DMM that can manage maybe 0.2% accuracy. Check the DMM with a new 0.05% voltage reference if there is still a doubt. But it would be extremely unlikely that a decent Fluke meter would drift this far out anyway.

Note that most DMMs tend to get bugged by uncertainty if you use them towards the lower end of their lowest ranges. Also the current ranges are often not as accurate. I always assumed the x10 resolution extension on the 87 meter came at the (same) cost of uncertainty so the LSD error magnifies by 10. Otherwise how can you claim back 0.4uA of uncertainty just by changing the resolution?

So this could be 0.01uA x 40 = the same 0.4uA of uncertainty on the 600uA range with the standard 0.1uA resolution. If so then it's hard to understand why this approach hasn't already died a death. Can anyone confirm if the 87 meter has something like 20-40LSD uncertainty on the 4.5 digit extended range on the 600uA range?