I find some SMT mystery components on a few-watts RF pulse board for an NMR probe driver. Little white ceramic cylinders with gold end caps, maybe 0.1" diameter. But they're wired *across* a 50-ohm pcb trace, and driven by some 500V FETs via a resistor chain and one HV capacitor. WTF?
Huh. What if these are spark gaps? If the designer chose several different voltages, then one DC pulse could switch the 800MHz signal through and back out. A higher DC pulse would short the RF to ground through a different spark gap. But why not just modulate the RF at signal level using diode switching?