Fluke 8842a display jitter

[tomeo.gonzales]

I bought a Fluke 8842a DMM from ebay 1 month ago based on Dave's enthusiastic review (https://www.eevblog.com/2017/08/03/eevblog-1012-ebay-fluke-8842a/)
The display is very bright and the DMM pass all self tests ok.
I measure 2 reference sources, (2.5000V and 5.0000V) and I see a display jitter about plus-minus 3-4 units for the last digit.
I check the references with a fresh calibrated Keysight 34465A DMM and they are spot on and very quiet.
I log 2000 samples of the 5.0000V reference to a file using IEEE488 interface and the mean value of the measurements is 5.0000 with a minimum of 4.9995 and a maximum of 5.0005V.
Using fluke 8842A Instruction Manual I try to resolve the display jitter.
I measure the -7V ref (TP701) and +7V ref (TP702) of the voltage reference of 8842A and the values are correct (+7.0000V and -7.000V and quiet)
I don't know how to proceed from here.
Any advice will be apreciated
 

[Vgkid]

Is that totam amount of change over a few seconds , minutes , hours , or days. That could give a hint where to look. Also how are the power supply voltages looking. Are they stable , or is there ripple on them?

[lowimpedance]

Couple of  other things to check are the input terminals clean and the front/rear switch probably not had any movement in a very long time !, unless you have done so!!. Give it a a good work out and clean the inputs with IPa then a little de-oxit on cotton tips. Now what do you get when measuring the DC sources.

[Kleinstein]

The jitter is noise.  The noise of the 8842 can depend on the settings of the DMM, like reading rate and enabling an input filter. Those old Fluke/Phillips meters use a sampling ADC and thus can have higher noise if no input filter is used.

So the fist thing would be to check the settings. For manual readings it should be slow mode most of the time.

To avoid possible noise from the source, a first noise source is usually with a short.

[tomeo.gonzales]

The jitter is noise.  The noise of the 8842 can depend on the settings of the DMM, like reading rate and enabling an input filter. Those old Fluke/Phillips meters use a sampling ADC and thus can have higher noise if no input filter is used.

So the fist thing would be to check the settings. For manual readings it should be slow mode most of the time.

To avoid possible noise from the source, a first noise source is usually with a short.

All measurements was taken using slow mode.
I try a good terminal short from Pomona and after 2 hours of warming the jitter seems to go away.
My other Fluke 8842A reach thermal equilibrium in less than 15 minutes.
The 8.2 voltage is a little low 7.5V, maybe I'll try to change the Zener diode.

[Kleinstein]

Zener diodes may show quite some noise, especially the higher voltage ones if not special low noise types.

The 8842 is not that sensitive that a special high quality short is needed. Those special shorts are needed if it goes to the nV range.  Just a piece of wire is OK.

It is normal to take about 1 -2 hours for an instrument of this size to get thermal equilibrium. So there might be some drift over the first 1 or 2 hours. This can be different between instrument. At least for the input amplifier it might be possible to adjust the analog offset and this way also  the thermal drift.