Improvement in cavity resonance after simplifying the feed antenna:

In the previous measurements I soldered a stub to the feed connector to reach lambda/4.
In this configuration, this stub is missing. The feed pin is shorter than lambda/4 and its angle is the angle of the cone wall.
There are only two peaks now, the first was also visible before, the second has a more narrow bandwidth and higher amplitude.

Probably these peaks represent two different cavity excitement modes in this frequency range.

We will track both of them in the next test session.