one track to follow, hood tco removes an error state, like that gas sensor triggered and stopping the magnetron relay(s), so maybe hood tco bypass some kind of error, for me this is the only logic explanation if pcb logic works normally, otherwise my previous post applies
[edit] i retract this, apparently that sensor only modulates cooking, it should not be 'touched' at least the sensor part (you can only test winding resistance), leave it alone for now, maybe i said something wrong so it needs to be analysed further if other tests give no answer