1. Improvements to Diamond Light Source RF Amplifiers Peter Marten On behalf of the Diamond RF Group* *also C. Christou, P. Gu, M. Maddock, S. A. Pande, A. Rankin and D. Spink

2. 1.RF Performance 2.Reliability Related Improvements 3.RF Noise on Infrared Beamline 4.Removal of Orbit Glitches 5.Installation of additional cavities: challenges and opportunities Agenda

3. IOT Performance System 2 (standby) System 1 (active) System 3 (active) Booster RF Filament hours One IOT has >55,600 hours Several IOTs >20,000 hours

4. E2v IOT Operating Hours 19 IOTs have provided >416,000 filament hours 1 x IOT >55,600 hours e2v IOTD2130 Two 4-IOT amplifiers driving cavities for normal operation One 4-IOT amplifier for test/conditioning/standby Single IOT booster amplifier One IOT >55,600 hours operation Several >20,000 hours

5. Reliability Related Improvements Filament Beam Ion Power Supplies Grid: Modified to prevent grid shorting to ground Kapton tape to installed to prevent contact Insulate edges of cover Filament: Check terminals for dry joints Inspect before installation

6. Booster RF Modification to e2v IOTs TED tube failure Spare “good ion pump current” TED IOTs would trip DCX Millennium amplifier modified to accept e2v IOTD2130 IOT All IOTs are now one type enabling IOT conditioning in standby system

7. RF noise on infrared beamline The high voltage power supply for the IOTs is a switched mode supply and as such generates noise on the voltage at the IGBT switching frequency and its harmonics Visible in HV spectrum Infra-red beamline notice the noise in their interferograms

8. Principal frequency components Internal enable signal to PWM generator is on a 12.8 MHz clock with a 12 bit counter Frequency visible to user through module rotation N is a user-defined parameter M is the number of operating power supply modules We expect one line per amplifier plus harmonics

9. Noise measured by BPM Fourier transform of BPM transverse position reveals a surprisingly rich spectrum

10. Identification of spectral features 50 Hz mains frequency is obvious

11. Identification of spectral features Other lines are not quite so obvious: Second harmonic of N = 34, M = 64 at 3418 Hz minus 6 x mains frequency = 3118 Hz Note the gap between 06:00 and 06:30

12. Voltage-dependent skipping of module rotation number Rotation number N jumps as phase voltage changes Principal frequency changes Out of our control

13. Change in module number Module number M also changes occasionally One spectral line can degenerate into four components as operating conditions change

14. All spectral lines can be identified Two amplifiers with different and erratic rotation parameter N and module number M 50 Hz mains frequency All intermodulations of fundamentals and harmonics are visible

15. Spectral noise is eliminated when module rotation is turned off Remaining features correspond to mechanical vibration Audible outside storage ring tunnel

16. Global stability is a concern IOTs use current modulation for amplification IOT current increases with power Module voltage droops with power Number of operating modules changes as power changes Dithering of module switching can lead to instability and beam loss Power depends on Beam current Top-up cycle ID configuration Module switching cannot be avoided

17. Variable dead-band in switching voltage can exclude dither in operation New user-defined parameter HYST_TH can be used to avoid module switching in operation

18. An alternative approach IR beamline routinely operates in two spectral ranges HVPS-induced noise is invisible when interferometer S/N is high HVPS-induced noise is drowned out by interferometer noise when S/N is low Select N and M such that no lines are present in sensitive region N = 39, M = 64 is stable and invisible in both operating modes No further complaints from beamline

19. Orbit glitches Glitches in beam position have been brought to our attention Glitches correspond to steps in reject power at the first combiner Looks like a small change in output power of individual IOTs One of the IOTs feeding the green reject channel is responsible

20. Which IOT and why? Monitor all IOT outputs on an eight channel oscilloscope Amplifier 1/3 here steps down and all other IOTs respond through LLRF All IOTs in amplifier have common high voltage and drive signal Issue is local to individual IOT Effect is explained by a dip in grid bias voltage Replacing power supply (toaster) stopped the orbit glitches

21. Installation of additional normal conducting cavities Normal conducting cavities to be installed upstream of IDs in straights adjacent to present superconducting cavities To be powered by IOT pairs from present four-IOT amplifiers Controlled by new digital LLRF development Coaxial transmission line 1 to run through present waveguide labyrinth Coaxial transmission line 2 to run through personnel labyrinth

22. Diamond MTBF Diamond MTBF has risen steadily since operations began in 2007 RF is still the greatest single source of faults RF faults are overwhelmingly amplifier faults Any tube fault shuts down the HVPS for four IOTs, causing beam loss Each IOT is a potential single point of failure Each amplifier power supply is a potential single point of failure

23. Resilience with two additional NC cavities Introduction of normal-conducting cavities driven by impact, time and cost of repair of failed superconducting cavity, but provide opportunities for enhancements in amplifier reliability Four IOTs combine to deliver 300 kW maximum 300 mA operation requires 450 kW Two amplifiers are sufficient Power reduced to 170 kW when one IOT is removed No longer able to support 300 mA S1 or S2 can be rephased to give 225 kW Mechanical shifter is slow, beam is lost Can save beam if IOTs can be switched individually Amplifier demand is reduced below 170 kW

24. IOT switching Beam survives when one IOT is manually switched out Kick is visible but stored beam is retained Component trials are underway Fast-acting MOSFET switch 50kV device from Behlke Switch out IOT rather than drop HVPS Possible 4 cavity operation 1.0 MV, 150 kW on superconducting cavities 0.3 MV, 100 kW on normal conducting cavities Can save beam if IOTs can be switched individually Amplifier demand is reduced below 170 kW

25. 1.RF Performance E2V IOTs continue to impress Maximum lifetimes now over 50,000 hours 2.Reliability Related Improvements Hardware maintenance and improvements continue Consistent use of one IOT across Diamond 3.RF Noise on Infrared Beamline Case study of RF response to beamline concerns Matter closed 4.Removal of Orbit Glitches Issue noted and resolved before it became a problem 5.Installation of additional cavities: challenges and opportunities Major project driven by cavities but offering opportunities for amplifiers Summary

26. Thank You!!