Advantest R3272 with intermittent noise & drift of measured level at power-up

[luigiser]

I recently bought an Advantest R3272, a 26.5 GHz spectrum analyzer, and I’m testing it. I found (for now) two problems:

PROBLEM 1: NOISE ON TRACE
When I power-on the instrument the measurement trace is full of noise (see picture). Something like a superimposed noise generated internally (it is not present on the source, I verified with another SA).
If the instrument is cold (not switched on from hours) the oscillation disappears after 10-20 minutes, like it is releated to some internal heating.
If the instrument is already warm-up (i.e. switched on from 1 hour) the oscillation disappear after 10-20 seconds from power-up (so it is not only a thermal problem).
Sometimes, also after a lot of time the noise disappeared, it comes again for some seconds. Other times for 1 minute or more.

Does anyone had similar problems and can suggest me what to check?

Searching in the forum I found a user that had a similar problem on an Advantest SA of the same line. He solved replacing an electrolytic capacitor in the “RF deck”, but I really do not know what the “RF deck” is and how to identify this instrument section inside.

PROBLEM 2: LEVEL DRIFT
The measurement is instable. Looks like is temperature dependant. At a cold power-on may be 10 dB lower than real, and slowly increase to a correct value in about 1 hour. I was non able to check, up to now, if after reaching a correct measurement it continue to rise very slowly (I calibrated the instrument at a certain point, without knowing if the stability was reached).
Until the instrument have reached a certain stability, let’s say 30 minutes from power-on, I was not able to perform the self calibration without “no cal signal” errors.

A first check on the power supply outputs does not show high ripple, but the switching noise is high. Anyhow until now I do not noted any difference on the power supply outputs with or without the oscillations.
A strange thing I noted: one line at the output of the power supply present a 50Hz square waveform, synched with supply line frequency, with amplitude of about 3.8V, but is the same both with or without oscillation.

I’ve low knowledge of RF and this type of instrument. If anyone have any suggestion it’s really appreciated.

Luigi