FEEDBACK ON NANOSECOND TIMESCALES (FONT): RESULTS FROM FIRST BEAM TESTS AT THE NLCTA AT SLAC P.N. Burrows (Queen Mary, University of London) Abstract We report on the first beam tests of the Feedback on Nanosecond Timescales (FONT) project to develop beam- based intra-train feedback for the Linear Collider. INTRODUCTION In order to achieve the design luminosity, in excess of /cm /s, the vertical beam size in the next genera- tion electron-positron Linear Collider must be of order a nanometre: 3 nm (J/NLC), 5 nm (TESLA), 0.7 nm, (CLIC). Any source of beam motion which results in relative verti- cal offsets of the two beams at the interaction point (IP) at the nanometre level will clearly reduce the luminosity from the design value. Maintaining vertical beam overlap is hence a particularly challenging goal for all these de- signs. In order to keep the luminosity LOSS below 80% the beam-beam miss distance must be kept below 40, 30 and 5 nanometres for TESLA, J/NLC and CLIC, respectively. In order to keep the loss below 10% the beam overlap must be maintained to better than 1 , i.e. below 5 nanometres in all cases. The many kinds of potential beam motion may be char- acterised in two classes: (i) slow drifts resulting from, for example, thermal excursions or component settling, with characteristic timescales varying from seconds to months; (ii) ground motion on a timescale comparable with the ma- chine repetition time. Both kinds of motion were experi- enced in the decade-long experience at the SLAC Linear Collider (SLC), and were dealt with by employing slow- and fast-feedback systems, respectively. Cultural ‘noise’ is expected to dominate the ground motion spectrum at fre- quencies above 1 Hz and, depending on the site, could be at the level of tens of nanometres (r.m.s.). If uncorrected this would cause an a priori large and unknown vertical offset between the electron and positron beams that is DIFFER- ENT on each successive machine cycle. Stabilisation of the effects of ground motion will hence be key to successful operation of the future Linear Collider; without it the luminosity may be 1-2 orders of magnitude below the design value, and the physics potential will not be realised. LUMINOSITY RECOVERY We are addressing the design of an intra-bunch-train fast- feedback (FB) system for the next-generation Linear Col- lider (LC). Other essential feedback systems that operate on longer timescales are discussed in [1]. A schematic of our system is shown in Figure 1. It comprises a fast beam position monitor (BPM) to detect the relative mis- alignment of the leading electron (or positron) bunches just downstream of the IP, a feedback loop, and a fast kicker for kicking later positron (or electron) bunches into collision. The latency of the system should be small enough so that it is able to make several iterative corrections to the beam position within the duration of the bunchtrain, i.e. the la- tency should be significantly less than 100 nanoseconds for J/NLC and CLIC. For TESLA the situation is more relaxed since there are 2820 bunches separated by 337 ns; a sys- tem with a sub-100-nanosecond latency would therefore be capable of bunch-to-bunch feedback. We have written a simulation package for the feedback system, based on MATLAB/Simulink, that incorporates the beam-beam dynamics (based on GUINEAPIG) and the re- sponses of the hardware components [2]. This allows us to study the feedback operation in ‘closed loop’ mode. We can also include arbitrary ground-motion model inputs, and multi-bunch wakefield effects, via simulation of the linac and beam delivery system using PLACET/MERLIN (TESLA) or a modified version of LIAR (J/NLC). We have used this simulation infrastructure to show that an intra- train FB system is capable of recovering more than 80% of the design luminosity, which would otherwise be lost due to ground motion. FONT1: FIRST PROTOTYPE INTRA-TRAIN FEEDBACK SYSTEM System components and beamline installation We chose to develop our first prototype FONT ex- periment at the Next Linear Collider Test Accelerator (NLCTA) at SLAC since the length of the bunch train there can be made as long as 170 ns, which is close to the 270 ns in the J/NLC design and the 100 ns for CLIC, and the train charge of up to 10 electrons is also close to the respective specifications. The time structure of the beam at NLCTA is such that bunches are spaced at X-band frequency, 11.424 GHz, corresponding to 0.08 ns between bunches. This is an order of magnitude shorter than the bunch spacing at J/NLC (1.4 ns) or CLIC (0.7 ns), which makes the test somewhat more difficult. The layout of the FONT1 components in the NLCTA beamline is shown in Figure 2. The beam enters from the left and is first steered vertically using a dipole magnet (not shown). The beam position is measured roughly 4m down- stream at the BPM. The position signal is processed and fed to a kicker that sits within the same assembly as the dipole. The kicker to BPM distance was chosen to match the separation between the IP and the FONT components