My friend Gianni IW1EPY discovered a tricky problem with the ADALM-PLUTO used above specifications. Here is his report:
This info is for ADALM-PLUTO users who experience this strange behavior.
During the metal boxing of an ADALM-PLUTO Rev C of my friend, I discovered a problem.
Connecting it to an LMX2595 to test the 5 harmonics.
Left a small box containing the LMX2595 board with the double coax transitions in WR42 to cancel the fundamental, and on the right, the Pluto under test.
A terrible notch appeared.
Without signal, Pluto noise is around -70, so the notch kills all around 22.3 GHz.
After the Pluto board investigation, bad SMA soldering was discovered.
The central pin was soldered on the pad, but a length of around 1.5 mm starting from the PCB edge was not soldered.
The missing soldering is mainly due to a non-planarity of the SMA vs the PCB.
The RF enters the PCB, not at the PB edge, and sees two lines, one of which is the correct path to the balun and a second line that extends for 1.5 mm to the edge of the board.
The second line acts as a stub with an open end, applying a short at the frequency of lambda-quarter resonation.
Resoldering the full length of the SMA central pin, the problem went out.
This is the normal behaviour of an ADALM-PLUTO without calibration, giving you a little more than 20 dB of dynamics in the 24 GHz band.
73, Gianni IW1EPY