Thank you for your experiments and detailed description of your system. It could well be that we see the effect only with lower SNR when everything is more easily disturbed.
If I understand your first post right you were able to disturb the SR1 via a TV cable ON/OFF that runs apart from the EUMETCast SR1 and Spectrum Analizer cables.
So this excludes a cable to cable induction which is even less likely when the cables only run 4.5m in the same tube. This should also test negative with your ASTRA TV.
Your system is good for a couple of more combinations to nail the culprit down. Maybe Ian can repeat his experiment so we can truly exclude the TWIN LNB from consideration.
In any case it must be certain, that at the moment you do the ON/OFF switching of a second receiver, the TELLICast receiver must download data of a subscribed channel.
Looking more closely at the problem today, I even see missed packets when the second source is switched OFF. I had the TC client renamed/rearranged on receiver Europa.
For that I had to take Europa down for a couple of minutes and also unload the DVB-Interfaces in GNU/Linux which will stop tuning and powering the respective LNB cable.
This disturbed receiver Titan on HVS-1 and HVS-2 with missed packets both when OFF and ON at Europa. The basic service on Titan was not affected. See my plots below.