+A  Click here to enlarge/reduce to/back from full screen 
Paul Scherrer Institut PSI 3rd International Workshop on Beam Orbit Stabilization - IWBS2004

Paul Scherrer Institut
5232 Villigen PSI, Schweiz/Switzerland
Tel. +41 56 310 21 11
Fax. +41 56 310 21 99



Updated:
25.01.2005
E-Mail: iwbs2004@psi.ch


Printer Friendly Printout without Logo or Navigation Elements available here... just click and print

IWBS2004

next up previous
Next: CONCLUSION Up: SUMMARY OF THE 3RD Previous: USER EXPERIENCE

STABILITY REQUIREMENTS IN 4TH GENERATION LIGHT SOURCES

The closing talk of the ``Orbit Measurement/Correction'' session, on beam stability improvements in the SPring-8 linac necessitated by ``top-up'' operation stability requirements in terms of bunch energy, intensity and purity already highlighted some of the challenges of 4th generation light sources. In 2004, the linac achieved an RMS energy/current stability of 0.01/1.9 %, starting from $>$1/20 % in 1998, mainly by means of careful source suppression (ambient/klystron temperature, RF amplitude/phase, synchronization with ring RF), implementation of an energy compression scheme (ECS) and feedback based stabilization of the beam position to $\approx$60 $\mu$m.


It was followed by a report on beam stability issues covered at the Free Electron Laser (FEL) Conference FEL2004 [11]. The talk was meant to introduce the auditorium to the specific stability problems of FELs. Different FEL types (storage-ring FELS, high average-power devices, single-pass devices) were discussed highlighting the critical issues followed by a comparison of the stability requirements. Storage-ring FELs impose serious constraints on the longitudinal (RF frequency/multi-bunch modes) and transverse stability ($\mu$m level) of the circulating beam. The average output power is only $<$1 % of the intra-cavity power. In general, resonator based FELs need high average current whereas single-pass FELs need high peak current. It can be distinguished between ``start-up from noise'' (SASE) and ``controlled startup'' (SEED) single-pass FELs. SASE FELs (e.g., LCLS, TTF, XFEL) provide high output power, are flexible in wavelength tuning but suffer from a ``spiky'' photon beam spectrum. In return SEED FELs (e.g., BESSY, FERMI, TTF) feature an improved spectral purity (e.g., HGHG: $\approx$0.23 nm FHWM) and intensity stability.


Compared to a storage ring which is like a ``spinning top'' (intrinsically stable: steady state, closed orbit, radiation damping, beam clearing through dynamic aperture limitation, small kHz noise BW, $\Delta$E/E$<$10$^{-5}$ in ``top-up'' mode) a linac is like an ``archery'' (potentially noisy: pulsed, open trajectory, no damping, no beam clearing, large GHz noise BW, $\Delta$E/E$>$10$^{-2}$). Among the sources of noise in single-pass devices are AC line fluctuations, switch-tune pulse-to-pulse jitter, switching noise fluctuations, AD/DA digitizing noise, temperature fluctuations and ground motion. Feedbacks are mostly not possible necessitating the use of feedforwards between micro-/macro-pulses instead. The required transverse orbit stability when applying the ``10 % of a $\sigma$'' stability rule is typically of the order of a few $\mu$m (fs-as for the longitudinal motion). Within the undulator the orbit must be controlled over a few gain-lengths in order to ensure the overlap between optical field and the electron beam, which is a prerequisite for reaching saturation.


The SASE FEL to be built at DESY (XFEL), featuring an 1400 m long 20 GeV SC linac, requires low emittance ($\epsilon_N$=1 $\mu$mrad), low energy spread ($\Delta$E/E $<$0.02), extremely high charge density (the XFEL will accelerate 3200 bunches of 1 nC charge and 80 fs length per pulse at a rate of 10 Hz) and a long undulator (L=700 m). The ``bullet'' like beam, with 20-30 $\mu$m beam size, requires sub-micron (or fs) stability in all three dimensions following the ``10 % of a $\sigma$'' rule. Experiments at the end of the 800 m long beam-lines demand a position stability of 0.1$\sigma$ in the last part of the undulator and a pointing stability of $\approx$$\mu$rad.


The XFEL will have the need for feedforwards between micro-/macro-pulses. Up to now the single bunch BPM resolution is only $<$50 $\mu$m for cold button and cavity BPMS which is not sufficient to fulfill the stability requirements. A prototype of a vertical ``feedback'' system for the VUV-FEL at DESY employing the same linac structures as the XFEL has been installed in the TESLA test facility (TTF). It has been demonstrated that the digitally controlled system, which utilizes 2 BPMs/kickers, is capable of confining 180 bunches within $\pm$50 $\mu$m, compared to 400 $\mu$m peak-to-peak without ``feedback''.
next up previous
Next: CONCLUSION Up: SUMMARY OF THE 3RD Previous: USER EXPERIENCE
Michael Boege
2005-01-25