IPAC2013 - List of Authors (Bland, M.J.)

Paper	Title	Page
MOPWA074	High Voltage Converter Modulator Optimization	852
	A. Scheinker LANL, Los Alamos, New Mexico, USA M.J. Bland University of Nottingham, Nottingham, United Kingdom
	High voltage converter modulators (HVCM) are at the forefront of long pulse high voltage (100kV) technology for Klystron voltage sources. We present results of digitally implementing an extremum seeking (ES) algorithm with which we optimized the rise time of the output voltage of a HVCM at the Los Alamos Neutron Science Center (LANSCE) HVCM test stand by iteratively, simultaneously tuning the first 8 switching edges of each of the three phase drive waveforms (24 variables total). We achieved a 50us rise time, which is reduction in half compared to the 100us currently achieved at the Spallation Neutron Source (SNS) at Oak Ridge National Laboratory. The ES algorithm is successful despite the noisy measurements and cost calculations, confirming the theoretical predictions that the algorithm is not affected by noise unless it both matches exactly the frequency components of the controller's specific perturbing frequencies and is of comparable size.

Paper

Title

Page

High Voltage Converter Modulator Optimization

852

A. Scheinker
LANL, Los Alamos, New Mexico, USA
M.J. Bland
University of Nottingham, Nottingham, United Kingdom

High voltage converter modulators (HVCM) are at the forefront of long pulse high voltage (100kV) technology for Klystron voltage sources. We present results of digitally implementing an extremum seeking (ES) algorithm with which we optimized the rise time of the output voltage of a HVCM at the Los Alamos Neutron Science Center (LANSCE) HVCM test stand by iteratively, simultaneously tuning the first 8 switching edges of each of the three phase drive waveforms (24 variables total). We achieved a 50us rise time, which is reduction in half compared to the 100us currently achieved at the Spallation Neutron Source (SNS) at Oak Ridge National Laboratory. The ES algorithm is successful despite the noisy measurements and cost calculations, confirming the theoretical predictions that the algorithm is not affected by noise unless it both matches exactly the frequency components of the controller's specific perturbing frequencies and is of comparable size.