I. Introduction
Radiation power pulse-shortening is found to be inherent to most high-power microwave devices based on power extraction from high-current electron beams. Considerable effort has been invested in attempts to alleviate this shortcoming in relativistic magnetrons. Relativistic magnetrons have first been studied in the 1970s [1], modelled by particle in cell (PIC) simulations [2]–[6], and experimentally investigated in various configurations ever since [7]–[18]. The accepted explanation of pulse-shortening is cathode plasma formation followed by plasma expansion toward the anode within the electromagnetic interaction region [19]. Although some modelling work has pointed out situations where pulse-shortening occurs without the presence of plasma [20], [21], the mechanism responsible for this has not yet been fully understood. In experiments performed in our laboratory, no plasma expansion was observed across the cathode–anode gap within the first ns [22], yet circumstances of pulse-shortening were detected.