Further Optimisation of the Diamond Light Source Injector

Optimisation of the Diamond Light Source injector has continued since user operation started in January 2007. Beam losses have been minimised by injector tuning and stabilisation, and high level software has been developed to characterise booster and injection parameters. Operation in top-up and single bunch storage ring fill modes are reported, and low-energy linac fault-mode studies are detailed. THE DIAMOND INJECTOR The DLS pre-injector is a 3 GHz, 100 MeV linac delivering up to 3 nC in a bunch train of up to 1000 ns, or a single bunch of up to 1 nC. It includes two identical accelerating structures and klystrons, a three stage bunching section and a thermionic DC gun [1]. The fullenergy booster has a single five-cell copper cavity driven by an IOT amplifier [2]. The linac and booster are cycled together at 5 Hz. The booster has a single kicker for onaxis injection, and a preseptum, septum and fast kicker for extraction. Injection into the storage ring is through four identical kicker magnets and one septum [3]. LINAC AND BOOSTER STABILITY Thermal drifts in the linac LLRF, multipacting in the pre-buncher and sensitivity of the booster dipole power supply to drifts in the mains frequency were dealt with in the first year of operation [4]. The 50 Hz linac gun heater was synchronised to the 5 Hz booster dipole cycle using a timing system oscillator to eliminate the effects of the mains frequency drift. A small linac energy variation arising from the beating of the mains-driven filament heater with the oscillator was eliminated by locking the klystron filament heaters to the same oscillator. Figure 1: Comparison of klystrons with heater powered by mains (blue) and locked to the timing oscillator (red). Slow power beat is absent for the synchronised heater. Stabilisation of the linac and booster injection components has allowed work to progress on injection efficiency into the booster, in particular for low-charge single bunch operation, where injection efficiency into the booster from the LTB has increased from below 50% to over 70% in the last year. The booster RF is cycled from 1 kW to 55 kW every injection cycle. The sine wave RF ramp used initially has recently been changed to a sine dependence on booster dipole current to follow ramp losses. This has had no observable effect on the beam, and has the advantage of reducing the total RF power demand by 45%. TURN-BY-TURN BOOSTER ANALYSIS Booster tune has been routinely measured since initial commissioning using the FFT of turn-by-turn BPM or stripline data [5]. In the last year, the turn-by-turn analysis has been extended to measure booster chromaticity by establishing betatron tune evolution through the ramp over several different measurements, each taken at a different master oscillator frequency. Booster chromaticity calculated from tune measurements across a range of master oscillator frequencies from 499.648 MHz to 499.656 MHz are shown in figure 2, together with the design values of chromaticity calculated including the natural chromaticity and eddy current effects in the booster vessel. Sextupoles were turned off for this measurement. Figure 2: Booster chromaticity in the horizontal plane (top) and vertical plane (bottom). Agreement between design and measured values is generally good, but there are some differences in the early part of the ramp. This may explain the fact that the current settings of the sextupoles, which are based on theoretical settings to achieve a chromaticity of +1, +1, still result in some beam losses with high charge single bunches, and this is the subject of ongoing work. The Proceedings of EPAC08, Genoa, Italy WEPC070 02 Synchrotron Light Sources and FELs T12 Beam Injection/Extraction and Transport