SwissFEL double bunch operation

SwissFEL has a unique capability, among the normal conducting linac-based light sources, to simultaneously serve two separate undulator lines (Aramis and Athos) up to the machine repetition rate of 100 Hz using the double bunch operation mode. It increases twice the experiments throughput of the facility with modest additional investment. Two electron bunches spaced 28 ns apart are extracted from the cathode by two laser pulses with individually controlled repetition rates. The bunches are accelerated up to about 3 GeV in the main linac using the same rf macropulse. After separation, one bunch serves the Athos soft x-ray beamline and the other is further accelerated to serve the hard x-ray beamline – Aramis. A fast and high-stability beam kicker separates the two bunches without disturbing the electron beam and consequently the x-ray lasing. The timing and control system sets hybrid machine modes utilizing independent operation of the two undulator lines with individually programmed repetition rates. Beam diagnostics and feedback systems have to operate with two closely spaced bunches where the two beams share the same machine path. The low-level rf system manipulates the rf amplitude and phase within a fraction of the rf macropulse to provide decoupling of the acceleration parameters of the first and the second bunch. This manuscript presents measurements that show that the bunch separation does not degrade FEL lasing stability.

Figure 1

SwissFEL layout.

Figure 2

3D CAD layout of the SwissFEL dual-PCDL optical setup. The photocathode drive lasers 1 and 2 (red areas) deliver temporally and spatially shaped deep-UV pulses (blue areas and green area) as well as the LH and short “probe” pulses for the two FEL lines. The two UV-PCDL and the two LH beams are recombined, respectively (green area), and sent toward the accelerator through two dedicated imaging transfer lines. The yellow area includes the LH Herriott cell used to delay the two LH pulses.

Figure 3

LLRF strategy to tune the second bunch by introducing a step in the amplitude and the phase of the input rf pulse. The right plot is the vector representation of the method for the TW structures. The relevant region of the rf pulse for the first or for the second bunch acceleration corresponds to the filling time of the structures.

Figure 4

Input pulse shape and energy gain in C-band stations (upper graph) and magnified energy gain slope and the position of the two bunches on it (lower graph). Note the different time scale for the two graphs.

Figure 5

Panel for rf pretuning.

Figure 6

Projected energy spread as a function of the gap of six dechirper modules.

Figure 7

3D model cross-section view of the SwissFEL CBPM16 (left) and CBPM38 (right).

Figure 8

Simplified block diagram of low-Q CBPM rf front-end electronics, showing only one of its three input channels (one reference and two position signal channels).

Figure 9

Absolute single-bunch position resolution as a function of the bunch charge.

Figure 10

ADC sampling principle of the SwissFEL BAM.

Figure 11

First bunch (upper plot) and second bunch (lower plot) arrival times at 100 Hz repetition rate measured by the BAM at the laser heater, downstream of the photocathode gun at $Z=13\mathrm{m}$.

Figure 12

Beam trajectory in the switchyard with the corresponding deflecting field regions. Q denotes quadrupole magnets, K1, K2-resonant kickers, D1, D2, and D3-compensating dipole magnets, and S-the septum.

Figure 13

Resonant kicker in the vacuum tank: A—vacuum capacitor (resonator capacitor), B—ceramic plate, C—tuning vane, and D—copper bar (resonator inductor).

Figure 14

Cross section of the septum magnet, with enlargement of the gap region. The upper half of the iron core with the coil winding is in air, and the lower half with beam apertures is in vacuum, with a 1-mm copper sheet as vacuum sealing. The aluminum support block for the thin copper sealing plate is not shown.

Figure 15

3D magnetostatic model of the septum. Additional triangular plates bolted on the magnet at each end add a negative stray field to reduce the total leakage field integral to $<50\mathrm{\mu }\mathrm{Tm}$.

Figure 16

Simulation (colored area) and measurement (lines) of the septum leakage field. The measurement curves show several runs overlaid, with differences mainly due to hysteresis from varying magnet cycling.

Figure 17

Longitudinal wakefields for the 8.2-mm diameter hole for stainless steel and copper and the corresponding Gaussian test pulse. The lower loss with copper gave the incentive to add copper plating.

Figure 18

Electron beam position running standard deviation of 100 consecutive pulses right before the FEL undulator section with the bunch-separation system on and off.

Figure 19

FEL photon beam pulse energy (running average of 100 consecutive pulses) measured by the gas-based monitor and its running standard deviation of 100 consecutive pulses measured with gas-based monitor and photon beam image camera with the bunch-separation system on and off.

Figure 20

FEL photon beam horizontal and vertical position standard deviation of 100 consecutive pulses with the bunch-separation system on and off.