[hpsdr] calibration - noisefloor

[Georg Prinz]

Hello Anybody,

could anybody explain me following behaviour when calibrating:

I am using a R&S SMDU as signal source feeding into one of two mercury 
boards with diversity19 PowerSDR
(My generator is not calibrated, so this value is more or less acurate.):

Input:          -140 dBm                         -120 dBm        -70 dBm

RX1 Metr: -137 dBm                           -111 dBm        -17 dBm

Panadapter: -145 dBm                       -100 dBm        +5 dBm

Noisefloor: -150 dBm+/-1dBm         -130 dBm        -80 dBm

Why is the noise floor dependent upon the calibration signal?

To be clear, I do not understand these measurements/readings.

If I take into account, that the reading at my SMDU is not accurate, I 
agree the input with -137dBm.
The noisefloor is always 10 dB below the input signal?
The value of the Panadapter does not match anything.
If I look at the proposed calibration value of -70dBm, the readings are 
not corelated.

Any idea?

Vy 73, georg - dl2kp