Pulsed network analysis with random signal for RF feedback optimization

A system for differential transfer function measurement of a pulsed radio-frequency (RF) device has been developed. A small test signal defined in the frequency domain as a constant magnitude and a random phase signal is added to the normal pulsed signal. With this system, an RF chain can be precisely characterized before closing the loop. Then, one can optimize the compromise between open loop gain and stability margin. This has been done with the 433-MHz klystron of the ELSA accelerator. The RF loop has been nested inside an adaptive feedforward system, When both systems run, the stability is 0.05%.