Why i did this: When measuring the beat signal with the PLL and the cable-delay setup simultaneously we didn't see correlation between both techniques.
We realized that we can't measure the current noise level using the (short) cable. However the estimated "noise floor" for the setup was much below.
So something must be wrong:
So i checked the noise floor of the cable setup again - it looks like i don't know how to determine the real noise floor for the setup:
I've set up a simple cable delay measurement setup using a Marconi, a power splitter and a +7dBm mixer. Marconi power is +10dBm, so LO is +7dBm.
Cable is 3 feet or so. Mixer signal is filtered (LP 1.9MHz) and terminated (50Ohm) before amplified with the battery-powered FET preamp.
I measured the noise floor by terminating the RF input of the mixer. Changing the input range and so the phase noise of the LO does not change the my "noise floor".
Then i measured the output for different input range settings of the Marconi starting from 1MHz down to 1kHz.
Here the result (not calibrated to frequency noise, only absolute voltage noise levels i measured, but does not matter for what i wanted to see):
For 1MHz and 500kHz input range everything looks OK at high frequencies. The characteristic bump at high frequencies is highly visible.
The low frequency noise looks different to what we measured before using the PLL, but that's not the problem right now.
Starting with 100kHz the measurement limit is reached but still above the "noise floor" with a terminated input.
So how do i measure the "real" noise floor without using a super low-noise oscillator?
Here the frequency noise of the same Marconi measured using the PLL:
will try tomorrow using the long cable...