A Modular Path Corrector for 4GLS

1. A Modular Path Corrector for 4GLS Peter Williams ASTeC - Daresbury Laboratory & Cockcroft Institute ERL07 Workshop, 23rd May 2007

2. 4GLS Beamlines Schematic • Dual High Peak Current XUV-FELs • High Average Current ID Loop & High Peak Current VUV-FEL Share main linac (three beams!) • Dual High Peak Current IR-FELs

3. 4GLS Beamlines: High Average Current Loop

4. 1 nC 77 pC ~9° 180° Path Length Adjustment for ERLs • Figure shows bunch placement on main linac RF waveform • We must energy recover the 77pC HACL bunches • Ensure π out of phase for deceleration • Reliably introduce path length retardation of up to 1 wavelength ~23cm • Must do achromatically, ie without introducing transverse dispersion (ηx), desirable to do this independently from longitudinal dispersion (R56), possible to achieve isochronously (R56=0)

5. Path Length Adjustment in Daresbury ERL Prototype • Nominal Gun Energy 350 keV • Injector Energy 8.35 MeV • Circulating Beam Energy 35 MeV • Linac RF Frequency 1.3 GHz • Bunch Repetition Rate 81.25MHz • Max Bunch Charge 80 pC • Bunch train 100 ms • Max Average Current 13 µA • In ERLP this will be done by physically moving first arc – a triple bend achromat. Also used in JAEA ERL, SDALINAC + others • Impractical for 4GLS – mechanical tolerances preclude accurate control when arc is wider than ~10m 11.5 cm trombone

6. Other Approaches for Path Length Adjustment in ERLs • JLab ERL uses Bates arcs / kickers • Two doglegs and a π-bend dipole arranged for isochronicity • Symmetric kicker dipoles at π entrance increase path length • Narrow width not practical for 4GLS configuration • π-bend dipole would need to be 4.2m for 1.5T field • CTF3 use variable field, one-period wiggler • Only achieves small path length difference • Difficult to compensate for proportionally large R56 variation – for δL=23cm, ΔR56=17cm • Cornell propose varying radius of path around CESR ring • Cross talk between path correction and fixed ID arc lattice in 4GLS • Douglas proposed magnetic mirror • Unusual shape and intricate edge structure make manufacture impractical

7. Progressive Bunch Compression in 4GLS HACL Final Decompression / Path Correction System VUV-FEL Beam Propagation Shortest bunch at VUV-FEL • An additional requirement – decompress the bunch for linac re-entry! • Otherwise we cannot energy recover

8. Path Length Adjustment for 4GLS HACL • Two non-dispersive doglegs girder mounted such that they move transverse to the beamline introducing extra path, coupled by set of bellows that expand accordingly. Our moving doglegs • These introduce negative longitudinal dispersion (arc-like R56) (higher energy particles retarded with respect to bunch centre) • Compensate for this with classical dispersive chicane, introduces positive longitudinal dispersion (chicane-like R56) - overcompensate to decompress the bunches. Our final decompressor • Chicane need not physically move, dispersion between centre dipoles ensures only small angle deviation compensates → achieve by ramping magnets Minimal displacement Maximal displacement

9. Engineering Layout of Doglegs Thanks to Simon Appleton, DL

10. Path Correction System Placement in 4GLS HACL Final Decompression / Path Correction System

11. Final Decompression / Path Correction System Path Correction System Placement in 4GLS HACL XUV-FEL Injector Main Linac HACL Injector Note final decompressor not quite correct in this drawing – transverse displacement of ~0.5m to doglegs ~1.0m

12. Characteristics of the 4GLS HACL Moving Doglegs Dispersion in Doglegs – Maximal Displacement Note zero dispersion in expanding section! R56 in Doglegs – Maximal Displacement

13. Characteristics of the 4GLS HACL Moving Doglegs β functions in Doglegs – Maximal Displacement Quadrupole Strength in Doglegs as a function of horizontal displacement angle We need to ramp these to cancel edge effects introduced by the movement

14. Characteristics of the 4GLS HACL Re-Entry Section Dispersion through final compressor and moving doglegs – minimal displacement Dispersion through final compressor and moving doglegs – maximal displacement

15. Characteristics of the 4GLS HACL Re-Entry Section R56 through final compressor and moving doglegs – minimal displacement R56 through final compressor and moving doglegs – maximal displacement

16. R56 Through The 4GLS HACL Final Decompressor 156° arcs Moving doglegs XUV / HACL spreader 48° arc Have you spotted the deliberate mistake? Yes – the R56 at the end is not back to zero – I have not introduced enough in the final decompressor here!

17. Characteristics of the 4GLS HACL Re-Entry Section β functions through final compressor and moving doglegs – minimal displacement β functions through final compressor and moving doglegs – maximal displacement

18. A Modular Path Corrector For 4GLS HACL: Conclusion • We have presented a novel system to introduce a continuously variable path length difference without variation of longitudinal dispersion for ERLs • Modular design ensures independence from rest of machine • Length of section (12 m) fits into 4GLS design • Need to evaluate beam disruptive effects (e.g. CSR) on energy recovery process in start-to-end simulations • Thanks to Hywel Owen (DL) and Sergey Miginsky (BINP)

19. A Modular Path Corrector For 4GLS: Summary • On re-entry to the main linac of 4GLS for deceleration we require: • The bunches to be π out of phase with respect to the accelerating bunches – introduce arbitrary path length of up to one RF wavelength • Any path length introduced to be independent of longitudinal dispersion • Decompression of short bunches to ensure small energy spread for energy recovery • The system to achieve this must fit in the existing 4GLS HACL design • We propose dedicated section with two parts, a final de-compressor chicane and girder-mounted moveable doglegs connected by expanding bellows • The dipoles in the chicane ramp slightly to compensate for arc-like R56 generated in the doglegs • The system is 12m long and would be placed just before re-entry to the 4GLS main linac