Beam based alignment in storage rings R Bartolini

Beam based alignment in storage rings R. Bartolini Diamond Light Source and John Adams Institute, University of Oxford Beam dynamics meet magnets II 02 December 2014

Outline Motivations Review of tools for beam based alignment BPM-quad centring ORM alignment Girder based alignment Conclusions Thanks to M. Apollonio, M. Boge, R. Dowd, X. Huang, L. Nadolski, J. Safranek, V. Sajaev Beam dynamics meet magnets II 02 December 2014

Motivations A correct alignment of machine component allows reaching the nominal performance of the ring Misalignments left over after installation or developing with time require corrections because: • feed down effect from magnets (et al. ) • orbit distortion • dispersion (emittance) • tune shifts + beta beating • generate linear coupling • excitation of nonlinear resonances • dynamic and momentum aperture (injection efficiency and lifetime) • misalignment with photon beamlines • … Beam dynamics meet magnets II 02 December 2014

Alignment Simulations usually assume (Diamond, NSLS-II, ASLS, …) ~ 100 um rms offset for girders ~ 30 um offset for magnets on girders ~ 200 urad rolls New trends (NSLS II, for ESRF II, Diamond DDBA, …) employ alignment in situ with vibrating wire or a stretched wire measurement. Resolution Many lightmagnetic sources use laser trackers and Magnets magnet fiducials. Magnet of finding centre is 5 -10 micron. are locked onto to magnet positioning ~ 50 um (but rely on the correct fiducialisation of girder then installed. Overall magnet to magnet positioning claimed the magnetic to the mechanical targets). better than 30 field micron. Beam dynamics meet magnets II 02 December 2014

e. g. in situ alignment “In-situ” and beam-based magnet measurement and alignment Alignment towers in place for central group SPring-8 concept based on NSLS-II vibrating wire method K. Soutome ESRF type stretched wire for Diamond II DDBA

Tools for beam based alignment If the alignment is still not good or changes occur after installation we have to rely on beam based alignment techniques BPM - Quadrupole centring (BBA) – or sextupole centring e. g. Diamond, SPEAR III, SLS, … BPM rolls e. g. Diamond, SLS, … Individual magnet alignment e. g. APS, ASLS Girder based alignment SLS Diamond Beam dynamics meet magnets II 02 December 2014

Quadrupole centring Procedure for aligning the BPM centre of the quad centre dipole corrector quad BPM BPM • scan quad gradient • if orbit moves, beam is not at centre of quad • change upstream corrector(s) to change position of the beam in quad • when orbit does not move redefine the zero of the adjacent BPM • orbit correction will put the beam at the centre of the quad • crucial in the commissioning phase (with loco and orbit correction) • routinely repeated at light sources (every run ~ every month) Beam dynamics meet magnets II 02 December 2014

Quadcentre – MATLAB middlelayer Orbit change Example of quad scan for Diamond BPMS (same at all light sources using Middlelayer) quad. centre BPM reading Orbit change due to a quadruple gradient change measured at 5 different corrector settings corresponding to 5 positions of the beam in the quad A fit allows determining the centre of the quad Beam dynamics meet magnets II 02 December 2014

BPM - Quadrupole offsets (Diamond) First runs during Diamond commissioning Quite large, but: - electrical centres of the BPMs were not calibrated - no accurate mechanical survey. Black dots are primary BPMs (different geometry) Reproducibility in adjacent runs less than few 10 s m unless some work has been carried out in CIA or in the tunnel (few 100 s m) Watch out for current dependence of BPM response Heat load on chamber matters! Procedure is slow (few h for full machine)

BPM - Quadrupole offsets (SPEAR III) Quad to BPM offset < 700 um; reproducibility 23 um after 38 h Courtesy J. Safranek Beam dynamics meet magnets II 02 December 2014

BPM - Quadrupole offsets (SLS measurements) Courtesy M. Boge

BPM - Quadrupole offsets (SLS) Courtesy M. Boge

BPM rolls from LOCO (Diamond) LOCO uses the off diagonal clock of the response matrix to fit a number of machine parameters which include the BPMs gain and coupling. This is a measure of the BPM mechanical roll and electrical cross talk between the X and Y electronics channels

BPM rolls from orbit cross talk (SLS) Similar analysis based on in-house software; Measured via corrector pattern analysis: exciting V orbit bumps at each BPM and measuring H correctors Dispersion function looks more like a betatron wave after cross talk correction Results cross checked with LOCO Courtesy M. Boge

BPM- sextupole offset A similar technique can be applied to sextupoles dipole corrector sext BPM BPM Not routinely used as the BPM-quad centring SLS used BPM to sextupole alignment in the sextupoles which are also equipped with skew quadrupoles. Assuming that skew quadrupole magnetic centre coincide with sextupole magnetic centre Beam dynamics meet magnets II 02 December 2014

Sextupole alignment from orbit response matrix (APS) At APS the orbit is deliberately sent off axis to the sextupoles to comply with users golden orbit. Need to know the offset at sextupoles. LOCO approach can be extended to measure magnet misalignment by introducing “virtual” quadrupoles at the sextupoles location to compute the feed-down effect due to sextupole misalignment quadrupole fits the H offset at the sextupole skew quadrupole fits the V offset at the sextupole Procedure: • scan sextupole by family (S 1 – S 4) • measure ORM • fit the quadrupole and skew quadrupole component • linear fit on the gradient to find the H and V offset at the sextupoles

Sextupoles alignment from orbit response matrix (APS) Offset reconstruction is in good agreement with the golden orbit provided by the BPMs (WIP) Courtesy V. Sajaev This technique was further developed at the Australian light source (ASLS) Beam dynamics meet magnets II 02 December 2014

Sextupoles alignment from orbit response matrix (ASLS) A prerequisite is a clean linear machine (with LOCO). Main target is the V offset in sextupoles individually to reduce the V emittance. Instrumental in achieving the quantum limit 0. 35 pm Used the V offset to shim each individual magnet on the girder (the sextupoles were systematically lower vertically). Shim in units of 25 um, total shim reached 150 um in many cases Courtesy R. Dowd Beam dynamics meet magnets II 02 December 2014

Sextupoles alignment from orbit response matrix (ASLS) Shimming sextupoles for V alignment at ASLS Courtesy R. Dowd Beam dynamics meet magnets II 02 December 2014

Survey data for girders: SLS found (2010) that the main source of quadruple misalignment is actually due to the girder misalignment (correlation in offset of individual quads on the girder) Aiba et al NIMA 694, 133, . (2012) Beam dynamics meet magnets II 02 December 2014

Survey data for girders: Diamond Survey August 2013 H-plane Survey January - August 2013 V-plane best fit plane Beam dynamics meet magnets II 02 December 2014

Survey data for girders: Diamond Survey March 2014 Beam dynamics meet magnets II 02 December 2014 H-plane

Survey data for girders: SPEAR III Courtesy J. Safranek Beam dynamics meet magnets II 02 December 2014

Girder shift vs corrector: SLS Such girders shift produce a characteristic corrector pattern and induce V dispersion and optics perturbation Courtesy M. Boge

Girder remote control movers: SLS Courtesy M. Boge

Beam assisted girder alignment Girder realignment based on quadrupole survey data (2011). Girder pitch and heave were fitted to the quadrupole data. Girders were moved according to these fits on survey data, with stored beam and running FOFB. Resulting corrector changes used to calculate the girder movement to confirm the proper mechanical re-alignment online Quadrupole-BPM BBA done successively Courtesy M. Boge

Beam assisted girder alignment Courtesy M. Boge Girder realignment instrumental in reducing the vertical coupling to 1. 2 pm and 0. 9 pm. New campaign of BAGA foreseen based on survey data taken in 2013

Girder movers experience at Diamond LVDT motion sensors bellow @ G 3 end Uncontrolled cam motion cracked a bellow in a mock up cell. LVDTs were installed Bearin g Camsha ft Axes of rotation Range of motion

Modelling effect of girder misalignments • AT model • MATLAB functions change magnet positions according to survey girder (sway, yaw, heave, pitch) • quadrupoles/sextupoles magnetic centres moved w. r. t. to girder according to survey data (residuals <50 um) • BPM in the model primary fixed to floor, secondary anchored to girder DX SEXT YAW primary BPM QUAD Bending secondary BPM Beam dynamics meet magnets II 02 December 2014

Realignment strategy Used both survey data and correctors pattern to devise the required re-alignment Survey data have to be used with care • absolute survey data do not reproduce the corrector pattern in the model – this is basically still unclear why! • however small changes in girder positions (~300 -400 um) can be reproduced well and generate a distinctive corrector pattern after orbit correction. Machine and model agree well. Additional (and severe) complications arise by the constrain of not moving the photon beam down the beamline (IDs and bendings) We re-aling the girder in collaboration with the beamlines: usually we introduce a golden offset that restores the beam position to exactly the same point in the beamline (transparent realignment)

Girder shift vs correctors: machine vs model HC 3 G 2 sway=+324 um G 1 G 2 C 03 G 3 new position original position - orbit variation correctly reproduced - CM variation correctly reproduced - local reduction in total CM kick angle reasonably reproduced

Girder based alignment (cell 3) MOPRO 101, IPAC 2014 VC 3 G 1 - 22/10/2013 model heave = -245 um pitch = -21. 4 urad data GO =191 um 22/10/2013 (SP, a) = (-95 um, -28 urad) VC 3 G 1 (1) (2) heave = -245 um pitch = -21. 4 urad (A) BPM offset after BBA (B) VCM @ cell-3 (A) Orbit after move (B) Effect of BBA (C) Effect of GO(3, 1) = 181 um XBPM-measured tilt: -28. 7 urad / AT prediction: -28. urad

Conclusions A collection of tools for beam based alignment is available and has been well tested at several different light sources ORM technique at ASLS shows that sextupole horizontal and vertical offsets can be measured with the beam to a accuracy of 1050 microns depending on the magnet position. Girder alignment with remote control proven at SLS and encouraging results are obtained at Diamond Beam based alignment techniques can potentially achieve measurements of the magnetic centre of magnets to precision comparable to the best bench-top techniques Beam dynamics meet magnets II 02 December 2014

Conclusions We should include the BBA and tuning technique in the definition of the tolerances for alignment and for the magnet design (e. g. supports) This is not commonly done in the design stage error study generally assume orbit corrections, tune corrections chromaticity correction but not BBA It is likely that the requirement will ease to “just allow the beam to be stored” without maxing out the correctors. Beam dynamics meet magnets II 02 December 2014

Conclusions Adapted from SLAC DLSR Workshop, Accelerator Session Close-Out December 11, 2013 Tolerances on alignment and field quality Over-specifying tolerances can be very expensive There was a general consensus that the tolerances assigned to magnet alignment and the field quality are computed with a pessimistic approach. The possibility of beam based correction should be considered in specifying the tolerances. This is true for alignment errors both for orbit and for optics corrections. Beam based correction tools like BBA, LOCO, coupling free steering can significantly relax the tolerances. It as suggested to explore setting tolerances based on first-turn just to store the beam.