Energy recovery in depressed collector of klystron operating in pulsed mode Mark Kemp, Aaron Jensen, Gordon Bowdon, Erik Jongewaard, Andy Haase and Jeff Neilson Jan 21, 2016

Presentation Overview Pulsed energy recovery concept Summary of work to date Next step and conclusions SLAC/Industry collaboration

Energy Recovery Concept Traditional Modulator (DC to Pulse Converter) Klystron Power Source RF Out Energy Recovery Modulator (Pulse to DC Converter) Spent Beam Energy Recovered Feed-forward energy recovery scheme enables traditionally wasted energy to be reused Existing infrastructure is re-used Energy recovery is completely passive

Ability to Recover Rise and Fall of Pulse In addition to “flat top” efficiencies typically quoted, we can also partially recover the pulse rise and fall Typically, these are completely wasted For future systems, can have relaxed modulator specifications => less expensive

Summary of Work to-date Generation 1: Transformer-based recovery circuit Demonstrated concept. Used a few different cores. Model and experiment matched. Generation 2: Inverse Marx (no recovery) Demonstrated concept. Shown to be more efficient than transformer. Generation 3: Inverse Marx (with recovery) Used inductors as isolating elements. Recovered energy to external capacitor.

Gen 2: An “Inverse” Marx Energy Recovery Modulator Capacitors charge in series, and discharge in parallel A transformerless, solid-state topology Using resonant recovery, can passively recover energy back to the modulator During pulse In-between pulses

Gen 2

Gen 2: Comparison of Simulation to Experiment Good match between PIC/SPICE model and experiment Additionally, we can pre-charge capacitors to produce a more-square pulse

Experiments on a 25 kW Klystron Demonstrate the Concept (Gen 3) External Capacitor Voltage (V) Voltage (V) Collector Stages Time (µs) Time (µs) Cathode Potential (/3) RF Pulse Two-stage depressed collector shown to: Self-bias Rise and fall quickly in potential Recover energy passively to external storage capacitor

Geometry Optimization Wrote Matlab scripts to: Create “random” collector geometries and biasing potentials Output geometries into Magic input deck and read in Magic results Automatically calculate relevant statistics (efficiency, energy per stage, etc.) Compile and view results from many simulations Perform computational intelligence optimization routines

Geometry Optimization Constraints added to improve heat conduction and ceramic shielding Shown is 5045 result used for this presentation This “optimal” geometry is not unique

DOE Accelerator Stewardship Funding Received: Bridge Gap From Concept To Commercialization Two existing high power tubes will be modified to greatly increase efficiency CPI VKS-8262S 5.5 MW S-band tube Upgrade from 45% to 70% efficiency SLAC 5045 65 MW S-band tube Upgrade from 45% to 65% efficiency

Radiative Cooled Collector Thin electrode model (.050” thick)

ANSYS model assumptions Electrodes are solid molybdenum Fixed properties (no temperature dependence) ε = .35 Loaded on upper and inner surfaces All surfaces radiate Ground electrode copper ε = .15 All inner surfaces radiate Outer can SST ε = .39 Fixed sink temperature at 473 K

Electrode temperatures (K)

Conclusions Pulsed energy recovery concept has been experimentally proven with excellent agreement The energy recovery technology is ideally suited for ultra- short pulse (<300ns) RF sources Accelerator stewardship funding will make technology commercially available within two years

SLAC Tech Transfer to Industry (CPI) Demonstrated Active RF source tech transfer to CPI SLAC developed X-band RF source technology successfully transferred to CPI S-band source technology to be transferred soon Starting collaboration on development of pulsed depressed technology Recently teamed on proposal to DOE to develop BAC version of CPI’s VKL-7967A 300 kW CW 1.3 GHz klystron

New Model for Source Development Tube industry is risk adverse/profit driven Reluctant to develop new sources unless there is clear market Will stop production on source type with low production SLAC has opposite motivations Primarily interest in development activity, not production Designs developed at SLAC can be transferred to any RF source vendor => multiple vendors possible SLAC can act as backup production source in case industry no longer willing to produce device Combination of SLAC with subsequent tech transfer to industry could be a good model for future source developments Development of new BAC inspired X-band source, high rep rate, pulsed depressed collector?