If I remember right from reading the American forums, the Ecoda 22kHz switches have both diseqc through on both ports, and permanent power through on both ports.
So if there is a voltage issue, I would look at the ecoda switches, and maybe change them.
Does the problem also exist, if you check the two halves of your system, without the Ecoda switches?
Just a thought: If the problem is just at certain frequencies, I wonder if the ecoda switches also cause impedance mismatches at the
unused and unterminated inputs at the diseqc switch at the 'not-chosen' port of the ecoda switch. Not sure if that could happen with these Ecoda switches, though.
Could you check and connect a DC-blocked terminator resistor there? (Or, for testing, a resistance of 1K, or a satellite beeper or so, and does the problem alter in that test?)
BTW. If in fact terminators are needed and used, in this setup you don't need the Ecoda switches anymore, but could change to simple splitters.

(See for the parallel use of diseqc switches:
https://www.sat4all.com/forums/topic/343724-parallel-schakelen-van-diseqc-switches/?p=2277533 , needed terminator resistors not shown in the pictures though)
If the Ecoda switches are responsible, I'm thinking about other alternatives as well, but that depends on the cabling possibilities to your two dish locations, and what switches you allready have lying aroud maybe....?
Edit: changing the Ecoda's for simple 2/1 diseqc1.0 switches would do the job, that is: if your receivers do diseqc repeat....
greetz,
A33