Lessons learned from machine studies on existing rings Laurent S. Nadolski Accelerator Group Synchrotron SOLEIL

Contents What performance have been reached in modern light sources? What are the main showstoppers and possible improvements? Linear optics restoration Orbit, tune stabilities BBA techniques Orbit(s) feedback, feedforward systems Top-up and injection (See Also L. Emery’s talk) Coupling control Stable low coupling value, re-alignment, robustness Non-linear dynamics (see Also R. Bartolini and T. Garvey’s talks) On and off momentum transverse beam dynamics Tuneshift control ID characterization (see Also J. Bahrdt’s talk) Physical aperture Measurement Beam loss location Energy measurement Impedance and instability (See last task of the session)

Due to lattice compactness Linear optics modelling with LOCO: SOLEIL case Linear Optics from Closed Orbit response matrix J. Safranek et al. Hor.  - beating Modified version of LOCO with constraints on gradient variations Due to lattice compactness (see ICFA Newsl, Dec’07)  - beating reduced to 0.3% rms Results compatible with mag. meas. and internal DCCT calibration of individual power supply Ver.  - beating Hor. dispersion Quadrupole gradient variation Quadrupole gradient variation

Comparison model/machine for linear optics Model emittance Measured emittance -beating (rms) Coupling* (y/ x) Vertical emittance ALS 6.7 nm 0.5 % 0.1% 4-7 pm APS 2.5 nm 1 % 0.8% 20 pm ASP 10 nm 0.01% 1 pm CLS 18 nm 17-19 nm 4.2% 0.2% 36 pm Diamond 2.74 nm 2.7-2.8 nm 0.4 % 0.08% 2.2 pm ESRF 4 nm 1% 0.25% 10 pm SLS 5.6 nm 5.4-7 nm 4.5% H; 1.3% V 0.05% 2.8 pm SOLEIL 3.73 nm 3.70-3.75 nm 0.3 % 4 pm SPEAR3 9.8 nm < 1% 5 pm SPring8 3.4 nm 3.2-3.6 nm 1.9% H; 1.5% V 6.4 pm * best achieved Courtesy R. Bartolini

Linear Orbit Restoration Beta-beating, tune shift Compensation Static (e.g. LOCO) or Dynamics (e.g. feedforward)  Individual power supplies for quadrupole magnets Local or global compensation for perturbations induced by insertion devices (IDs) IDs are freely controlled by users, many different combinations: the storage ring is alive! Small residual effects X many IDs = not so small perturbation At SOLEIL, need of a global tune feedback Impedance induced tune-shift with current variation  Improve injection efficiency  Necessary step for going to low coupling value and fine resonance correction Is it possible to get an online measurement during user operation? Tracking beta beating for all ID configurations? How to get LOCO precision with turn by turn data? How to get enough turns while a TFB is running? For small emittance rings, multipole specs have to become more tighter? At SOLEIL, focusing tolerance is 1/10 of the natural focusing of the IDs More exotic insertion devices to come with linear and nonlinear perturbations (specific multipole correction patterns cf. J. Bahrdt)

Coupling Achievement: close to 0.01% (almost 1 pm.rad) Minimum emittance close to natural limit (factor 5 at SLS) Diffracted limited in V-plane already. No push from users Vertical emittance control Betatron coupling suppression necessary for a fine correction of resonance widths. Correct for low coupling and control with a vertical dispersion wave (ALS, SOLEIL, …) for few bunch filling pattern. Local control of beam sizes (Upgrade project for very low emittance ring: local round beam with solenoid B-field, skew quads?) Some beamlines want round beams, other flat beams: contradictory needs Lessons LOCO is an efficient tools. How to get a quick, on the fly method while keeping similar precision with turn by turn measurements? Increase the number of individual skew quadrupoles Difficulties to measure low coupling values (Vertically polarized SR by Anderson et al. SLS, …) Difficult to maintain low coupling value over time  BBA on quadrupole and sextupole centers is the next step to reduce further more vertical emittance ID skew quad have to be fully included in the shimming process

Low value: high sensitivity to tiny disturbance

Is it possible to maintain ultra-low coupling value other time with ID motions? L. Farvacque reporting of ESRF experiment, ESLS 2009

Sources of Perturbations Stability requirements Beamlines Integration time: If FPERTURBATION > 1/TINTEGRATION → Emittance growth, Lower photon flux in a stable way If FPERTURBATION < 1/TINTEGRATION → Noise on the measurement Beam stability should be better than 10% of the beam sizes 10% of the beam divergences Thermal effects Insertion Devices Cycling Booster operation Experimental hall activities Ground vibrations mains +harmonics Frequency (Hz) 1 10 102 103 10-1 10-2 Time Period (s) 10-3 N O I S E U R C 1/Ti Noise Emittance growth only An orbit feedback is needed to stabilize the beam position from DC up to ~100 Hz Identification of perturbation sources + passive techniques to damp oscillation amplitudes

Vertical position shift due to moving crane AT SOLEIL FB OFF. Vertical noise amplitude over one week: cultural noise 0.5 mm

Orbit stability Beam based aligned on quadrupole centers Dead band approach not suitable FOFB efficiency is suppressed at low frequencies (< 0.1 Hz) ID motion frequency band Beam based aligned on quadrupole centers Precision reached is a few micro-meters (10µm) Orbit feedback systems Slow and/or Fast feedback system How to cope with slow/fast correctors? Feedback efficiency at 100 Hz reached: reduction factor 2 to 3 Challenging part: temperature stabilization < 0.1°C Orbit stability sub-micrometric, a few tens of nm for ultra-low coupling values Increase the correction bandwidth (new fast switching magnets) Trends XBPM in feedbacks For Dipole: V-plane only For IDs how to cope with gap motion dependency XBPM response? Frequency (Hz) 1 10 10-1 10-2 DC SOFB FOFB D E A B N

Overview of fast orbit feedback performance Summary of integrated rms beam motion (1-100 Hz) with FOFB and comparison with 10% beam stability target FOFB BW Horizontal Vertical ALS 40 Hz < 2 μm in H (30 μm)* < 1 μm in V (2.3 μm)* APS 60 Hz < 3.2 μm in H (6 μm)** < 1.8 μm in V (0.8 μm)** Diamond 100 Hz < 0.9 μm in H (12 μm) < 0.1 μm in V (0.6 μm) ESRF < 1.5 μm in H (40 μm)  0.7 μm in V (0.8 μm) ELETTRA < 1.1 μm in H (24 μm) < 0.7 μm in V (1.5 μm) SLS < 0.5 μm in H (9.7 μm) < 0.25 μm in V (0.3 μm) SPEAR3 60Hz  1 μm in H (30 μm)  1 μm in V (0.8 μm) Trends on Orbit Feedback restriction of tolerances w.r.t. to beam size and divergence higher frequencies ranges * up to 500 Hz ** up to 200 Hz Courtesy R. Bartolini 12

FOFB Efficiency (1-350 Hz) FOFB efficiency 100 Hz 50 Hz 3 Hz Measurement on a BPM outside the feedback loop HORIZONTAL VERTICAL FOFB efficiency 100 Hz 50 Hz 3 Hz

Top-up operation Increases average current, brilliance, Beam-line resolution, resolving power Allow us to accept lower lifetime but there is still a limitation from radiation shielding, activation of component inside tunnels Requires to high injection efficiency, large enough beam lifetime Mandatory for sub-micrometric stability both for machine and beam-lines components Orbit distortion during injection time With the standard injection scheme, it cannot be cancelled out completely: Fine kicker tuning, stray field (septum magnet) SPRing8: use of pulse corrector Pretty large beta-function required at injection point Other injection scheme are heavily pursued both for on and off axis injection (pulse multipoles, swapped beams, See L. Emery’s talk)

Residual closed orbit distortion injection bump A. Loulergue, SOLEIL

Probing non-linear beam dynamics Goals: 100% injection efficiency Large beam lifetime (off-momentum aperture) Small footprint in the tune-space A good non-linear model of the storage ring Measurement of tune shift with amplitude Frequency contents of turn by turn data What to include in the model Lessons: Details multipole errors from magnetic measurements Thick sextupoles Quadrupole fringe fields Details B-field map for insertion devices Vertical chromaticity is not well modeled

Non-linear-beam dynamics Means/tools FMA on/off momentum Touschek lifetime computation with local energy acceptance (6D tracking) SLS resonance correction Non-linear LOCO see R. Bartolini’s talk  Still difficult but very promising Trends Symmetry broken by insertion devices in very large number Introduction of damping wigglers Local modification of focusing with additional quadrupole (double low beta)  lattice may becomes very sensitive to tune, phase advance variations, etc … BPM requirements for machine studies (analysis of fundamental lines, beta-functions, phase advances, driving terms analysis) High resolution even at low current Turn mixing, timing errors, non-linear response, sensor tilt, channel cross talk, decoherence  challenging

Frequency Map Analysis: ALS and BESSY-II ALS linear lattice corrected to 0.5% rms -beating FM computed including residual -beating and coupling errors BESSY-II with harmonic sextupole magnets, chromaticity, coupling ALS measured ALS model BESSY-II measured BESSY-II model A very accurate description of machine model is mandatory fringe fields: dipole, quadrupole (and sextupole) magnets systematic octupole components in quadrupole magnets decapoles, skew decapoles and octupoles in sextupole magnets Courtesy C. Steier (ALS) P. Kuske (BESSY-II)

Beam Loss Top-up operation, low gap insertion devices, narrow chamber give more stringent limits for beam losses Losses have to be located in shielded areas to keep radiation dose below the 0.5 µSv/h regulation limit. Off momentum particles may be lost either in the H-plane in dispersion region or in V-plane through non-linear diffusion. In machine like SPRing8 or SOLEIL, non-linear dispersion has to be taken into account. Moreover, changes of chromaticities modify the loss process. Lessons Keeping the same location of losses for all machine conditions is not always possible. Slight misalignment of a vacuum chamber may change loss location. Need a regular survey with beam position/angle bumps, … Mechanical gap of in-vacuum insertion is often smaller than magnetic gap because of shim, copper shield protecting magnet blocks (liner) Precision vertical centering with e-beam is mandatory Save operation can allow gap reduced down to 3-4 mm

Loss process SOLEIL case by varying chromaticity value xix = xiz = 0 2003, SPRing8: vacuum leakage at injection section during low emittance optics EPAC 2006, Tanaka et al., pp3369-3361 Losses in short straight section xix = xiz = 2 Losses in medium straight section

Conclusion Lots of progress have been done during the last two decades. Linear optics in well understood Beam energy measurement by spin depolarization is not a easy task: success in ALS, ANKA, BESSY II, SLS, … not for the last build light sources! Still a lot of work for tuning complex lattices with many sextupole families, trends to introduce octupole magnets to control tune shift with amplitude Non-linear optics correction et measurement based on turn by turn data is still challenging and requires improved BPM systems  individual sextupole PSs? Other challenging parts Maintaining performance with many insertion devices freely controlled by users. Top-up operation means beam delivered over many day period of time: it required the development of on-line continuous tools to measure beta-beat evolutions, local coupling, … performance degradation Going to low emittance lattices makes requirements tighter for multipole tolerances of insertion devices. Fast switching devices, low beam sizes drive orbit feedback improvement Local control of beam sizes

References CERN LER2010 workshop, 2010, http://ler2010.web.cern.ch/ler2010/ 2nd Non-linear Beam Dynamics Workshop, Diamond 2009, http://www.diamond.ac.uk/Home/Events/Past_events/NBD_workshop.html Top up workshop, 2009, Melbourne, Australia. http://www.synchrotron.org.au/index.php/news/events/australian-events/event/4-accelerator-physics-top-up-workshop ESLS 2009, 2009, DESY, Germany https://indico.desy.de/conferenceDisplay.py?confId=2325 THANK you