The Ukrainian 4m transverter has been installed into the final box, including IF relay and A*duino sequencer. The container is a repurposed manual KVM (Keyboard, Video, Mouse) switch and pre-drilled holes hold flanged BNCs just right. Also the VGA connector was a good a sturdy way to bring data from the FT817, pre-wired:
The smoke test did not produce actual smoke, but the current drawn is close to 1A. Considering the transverter should be in receive mode, that's not normal. Leaving it on a few seconds the Mitsubishi PA final warms up, so it is probably self-oscillating somewhere.
I have never troubleshoot a self-oscillating PA, just quickly read about it. I am thinking of two causes: either the DC supply cord is too long and lacks decoupling or the transverter board is too close to the box/ground.