Keithley 2000, 2700, 2790 AC volts Noise

[TizianoHV]

Hi,
a few weeks ago, I was testing some current transducers and I needed to measure accurately (0.1%) low AC voltages (from 1mVrms up to 1Vrms @50Hz).
As first thing I compared a few of my meters (Agilent 34970A, Keithley 2790, aneng AN8009 ) to find the best ones.
To get precise low AC volts I feed stable 100mVrms@50Hz into my trafo decade (with 100ppm resolution).
An8009 was spot-on down to 200uV (as Dave showed in the AN8008 review).
As we know Ag34970A and K2790 use a TRMS chip so they can't measure accurately below 1% or 2% of range. Anyway, agilent performed better but I noticed (confirmed) an issue with K2790: in the 100mV and 10V ranges it was showing a low frequency oscillation (Img. 2) with a period of 1 - 2 minutes and an amplitude of 0.05% of range (regardless of the input voltage). When selecting the 10V range it clicks a relay, so it probably has the same gain (and noise) as the 100mV range.
When a short is applied it shows the usual ofset (80uV) and no noise is detectable (+- a few microvolts).

I've recently repaired the DCV and ohms of this meter (I killed it with inductive kickback) but I suspect it's not related to that since it showed the some noise even before.

Can, someone with a Keithley 2000, 2700, 2790, log a few minutes while measuring a stable 100mV source with the meter in the 10V range (since 1mV in the 100mV range is a bit trickier to generate)?

Thankyou!

[TizianoHV]

*If you can't record data with a PC you can use the internal memory of the multimeter and its statistics. You set 10VAC SLOW, feed 100mVrms, store 150 readings in its internal memory (It will take 1 or 2 minutes) and then recall Standard deviation and Peak to Peak values.

If it's as bad as mine you should get a std of 1.4mV and 5mVpkpk.

(To verify if the 100mVac source is stable enough repeat the same measurement with the multimeter in the 100mV or 1V range, std should stay below 100uV)