Speaker: J. Scott Berg
Abstract: CBETA is an energy recovery linac with four accelerating and four decelerating passes, where all of the beams return to the linac through a single linear fixed field alternating gradient arc. I will describe the methods we used to correct the orbits for the 7 beams with four distinct energies in this return arc. Each of the 107 FFA doublet cells had a horizontal and a vertical corrector as well as a beam position monitor. Two techniques were used. The first applied a singular value decomposition to find changes in corrector settings that would minimize the orbit error in some region of the machine. The second would find combinations of correctors in one region that would correct the orbit and/or dispersion in a downstream region for a single pass while leaving the orbit and/or dispersion unchanged in previous passes. I present details of these methods, describe how we applied them, and indicate practical and theoretical difficulties we encountered.
2 replies on “Orbit Correction in the CBETA Return Arc”
Impressive work! I have a question on the machine stability. Does the orbit drift over time due to say ambient temperature changes? Is a found correction expected to be stable day to day?
Our dominant reproducibility issues appeared to arise from two factors: RF drift and hysteresis. We believe we came up with a method to address hysteresis reasonably well, but RF drift was more difficult. We had plenty of RF control difficulties, but the big issue was that our cavity field pickup seemed to be drifting for unknown reasons, meaning that the control system would think we had a stable voltage and phase when in fact we did not. We had a couple heuristic tricks to deal with this: the first was to wait for a period of time after startup for things to stabilize; the second was to adjust cavity voltages until the orbit in our first splitter line was restored to the same location that it had been on previous occasions. This gave us a passable starting point, and on a good day we would do pretty well starting with the previous day’s settings.
We were always trying to improve things: no two days really ended the same way, so corrections needed to be applied daily; however, I think we reached the point where at least the first turn was fairly reproducible, or at least we could get to a good first turn with very little effort, on most days.