Slow Global Orbit Feedback at Pohang Light Source
Slow Global Orbit Feedback at Pohang Light Source (PLS) Heung-Sik Kang Pohang Accelerator Laboratory Pohang, Korea 1
Aerial View of PAL Brief History of PLS l Project started Apr. 1 1988 l Ground-breaking Apr. 1 1991 l 2 -Ge. V Linac commissioned Jun. 30 1994 l Storage ring commissioned Dec. 24 1994 l User’s service started Sept. 1 1995 l Energy ramping to 2. 5 Ge. V Sept. 1 2000 l 2. 5 -Ge. V injection 2002 Nov. 1 2
Pohang Light Source 2. 5 Ge. V Linac / Storage Ring § § § § § PLS Orbit Stability Requirements Beam Size Beam Energy Beam Current Lattice Superperiods Circumference Emittance Tune RF Frequency Energy spread 2. 5 Ge. V 200 m. A TBA 12 280 m 18. 9 nm-rad 14. 28 / 8. 18 500 MHz 8. 5 x 10 -4 <1% x-y coupling> Orbit Stability Horizontal Vertical Bending Magnet 230 m 24 m 23 m 2. 4 m Insertion Devices 455 m 35 m 45 m 3
March 2004, w/o SOFB Orbit Feedback q Slow global orbit feedback (SOFB) For 9 days RMS-y 200 m Ø Improvement of Power Supplies § V: 12 bit -> 20 bit resolution (22 ea) New power supply with the controller developed in BESSY II § H: 12 bit -> 16 bit (22 ea) modification of existing power supplies Ø Operational since October 2004 Frequency Spectrum of BPM reading § Feedback Speed: 4 sec § SVD algorithm, MATLAB / EPICS § Feedforward correction for ID is under test q Fast global orbit feedback (FOFB) High frequency movement ü Under consideration 4
MATLAB GUI for Slow Global Orbit Feedback § The orbit feedback algorithm uses the SVD (singular value decomposition) method. § Use the Matlab Channel Access to EPICS IOC of BPMs and correctors § The GUI displays the response matrix, the spectrum of singular values, the real time orbit, and the correction kick. 5
Correctors and BPMs for SOFB P 8 P 6 P 7 P 5 P 9 P 4 P 3 P 1 ID C 1 • • • C 6 P 2 C 4 C 5 C 3 One sector C 2 9 BPMs & 6 Correctors /sector Totally, 108 BPMs and 70 correctors in each plane BPM electronics: Bergoz MUX BPM Insertion Devices (6) – Undulator: U 7, EPU 6, U 10, In-Vacuum Revolver (min. gap: 5 mm) – Multipole Wiggler: HFMX, HFMS u SOFB uses - 2 correctors (C 1 & C 2) / sector -> 22 correctors in each plane - 6 BPMs / sector - Current dependence table for BPM electronics Horizontal plane: 16 bit resolution -> 0. 06 rad/ 1 bit Vertical plane: 20 bit resolution -> 0. 004 rad/ 1 bit 6
Orbit Variations in SOFB during USER RUN (Nov. 16 – 25, 2004) For one day Reference orbit is re-set For 10 days RMS-y RMS-x RF freq ü Reference orbit is re-set at 150 m. A after refill. 1. The number of correctors is not enough for correction ü Current dependence data of BPM electronics is referenced to the BPM reading at 150 m. A 2. To effectively compensate the BPM electronics’ current dependence 7
False BPM Reading • BPM Reading BPM reading 1) Real Beam Position change 2) BPM Electronics’ dependence on Ambient temperature 3) BPM Electronics’ dependence on Beam current 180 m. A 2 m 4) Chamber movement 120 m. A Ø Current Dependence table is used to subtract the false value from the BPM reading for SOFB Ø But, chamber movement is not compensated. - BPM electronics problem: Gain drifts and nonlinearities u Compensation of false BPM reading is absolutely necessary in order to minimize the false motion by orbit feedback. u Ambient temperature dependence can not be compensated, thus should be minimized. 8
Linear rate of current dependence [um/m. A] Beam Current Dependence of BPM electronics Horizontal vertical 188 m. A 153 m. A 112 m. A Non-linearity of current dependence BPM reading change between 120 m. A and 180 m. A (not include the bad BPMs) X – rms : 2. 9 m Y – rms : 5. 0 m § Different Change rate of current dependence low current rage: 153 – 112 m. A (blue line) high current range: 188 – 153 m. A (green line) § Current Dependence table for SOFB : red line (188 -112 m. A)9
Patterns of BPM’s Intensity dependence Chamber Motion + BPM electronics Temperature of Vacuum Chamber in straight section ( Sep. 14 – 15, 2004, Bad Orbit condition) BPM 6 -4 Y BPM 9 -2 Y 10 m After the orbit correction BPM 7 -5 Y 12 -8 Y BPM 9 -5 Y 5 m 2 m 10
BPM Chamber Movement Photon Stop BM Photon Fan One sector 2. 0 Ge. V 400 m. A 2. 5 Ge. V 250 m. A Synchrotron radiation power 115 k. W 128 k. W 161 k. W Photon power / sector 9. 6 k. W 10. 6 k. W 13. 4 k. W Ø vacuum chamber moves due to the change of synchrotron radiation heat load Ø dependent on orbit Ø Look-up table is not easy to implement. 11
Change of BPM Reading in SOFB BPM with a negligible chamber movement 7 -6 Y BPM with a small chamber movement Beam current 5 m 7 -5 Y 5 m BPM with a large chamber movement SOFB OFF 9 -5 Y SOFB ON BPM 9 -5 Y 1 m 5 m 12
Measurement of BPM Chamber Movement 4 -9 5 -1 4 -8 4 -7 4 -6 4 -5 sector number 4 -4 4 -3 4 -2 5 -1 40 m 10 m Digital Position sensor (accuracy: < 100 nm) BPM 4 -9 40 m 10 m 4 -8 10 m 4 m y 4 -7 x 4 -6 2 m 4 m Beam Current @ 2. 5 Ge. V 200 m. A 13
Variations of Corrector Currents in SOFB 3 CV 1 Beam current 2 CV 1 9 CV 2 3 CV 2 4 CV 2 ü Very similar to BPM chamber movement ü BPM electronics' current dependence looks compensated well. 14
Ambient Temperature Dependence of BPM Electronics (BPM reading oscillation) BPM 7 -4 Y BPM 8 -2 Y 5 m 2 m AGC 0. 02 V Cooling Fan OFF 18: 00 2 hours Dec. 6 ü The same oscillation was observed in the ambient air temperature in the control shed where the BPM electronics is. ü BPM electronics Must be influenced by the ambient temperature. ü One BPM electronics module shows Dependence on Ambient temperature : 1. 4 m / C ü Ambient temperature in control shed should be well controlled. 15
BPM’s intensity Dependence limits the SOFB performance. Solution is TOP-UP! BUT, Decided not to use it until …. Because 1. For Linac, Injection efficiency is not so good compared to Booster. Synchronization of RF between SR and Linac is required. 2. We will start SASE-FEL project in 2005. No. 1 Priority of Linac is changed… 16
PAL XFEL Project Period: 2005 -2009 Budget: 80 M$ 17
How to Compensate Chamber Motion in SOFB ü Real-time measurement of BPM Chamber Motion for all BPMs (108 ea) by Digital position sensor or LVDT (Budget allocated in 2005) ü Chamber position is monitored with respect to Girder, which is equivalent to Quad because Girder is very rigid. ü EPICS Database ü Compensate the chamber motion from BPM reading in SOFB (data refresh time : 1 -3 minutes) v Neglect the Girder motion with respect to ground, and the Girder to Girder differences v Quad does not move as the Beam loading changes. v Care about the orbit with respect to Quad. Chamber QM QM invar 18
SOFB for Insertion Device Due to EPU gap change Beam current BPM 4 -3 Y 2 m FB OFF FB ON 13: 41 • Vertical orbit changes up to 6 m in rms when the EPU gap is moving between 20 and 25 mm. • Feedforward correction is required. H: 16 -bit correctors V: 20 -bit correctors 19
Feedforward for EPU 6 New Correctors for Feedforward P M 2 P M 1 Q C Q Q 3 M 2 1 D 6 D D C M 1 ID C M 2 P M 3 Q 1 Q 2 C Q 3 M 3 P M 4 BM 1 P M 6 P M 5 SD Q 4 Q 5 SF C Q 6 BM 2 Q 6 D C M M 4 5 P M 7 P M 8 SFD Q Q SDD 5 4 D D P M 9 BM 3 Feed-forward correction - correctors: 1 CM 6 and 2 CM 3 just done for test New correctors for Feedforward ready for test - speed: 10 Hz 20
EPU 6 Feed-forward Correction Feedforward table Vertical correctors: 1 CM 6 and 2 CM 3 5 th degree polynomial for fitting Feedforward speed: 10 Hz 21
Summary l Achieved orbit stability by SOFB - short term (1 hour) : < 1 m - long term (12 hours) : < 3 m l BPM Chamber movement due to Synchrotron Radiation heating mainly limits the SOFB performance. Improvement Plan of SOFB in 2005 l Real-time measurement of BPM Chamber Motion for all BPMs (108 ea) l Reduction of BPM Noise – Feedback speed : 4 sec 2 sec l 70 correctors in vertical plane 20 -bit resolution 22
- Slides: 22