Compensation for Booster Leakage Field in the Duke Storage Ring
Wei Li, Hao Hao, Stepan F. Mikhailov, Victor Popov, Wei-Min Li, Ying K. Wu
aa r X i v : . [ phy s i c s . acc - p h ] M a y To be submitted to Chinese Physics C
Compensation for Booster Leakage Field in the Duke Storage Ring *Wei Li , Hao Hao , Stepan F. Mikhailov Victor Popov Wei-Min Li Ying K. Wu National Synchrotron Radiation Laboratory, University of Science and Technology of China, Hefei, 230029, China Triangle University Nuclear Laboratory/Physics Department, Duke University, Durham, 27705, USA
Abstract:
The High Intensity Gamma-ray Source (HIGS) at Duke University is an accelerator-driven Comp-ton gamma-ray source, providing high flux gamma-ray beam from 1 MeV to 100 MeV for photo-nuclearphysics research. The HIGS facility operates three accelerators, a linac pre-injector (0.16 GeV), a boosterinjector (0.16–1.2 GeV), and an electron storage ring (0.24–1.2 GeV). Because of proximity of the boosterinjector to the storage ring, the magnetic field of the booster dipoles close to the ring can significantly alterthe closed orbit in the storage ring being operated in the low energy region. This type of orbit distortioncan be a problem for certain precision experiments which demand a high degree of the energy consistencyof the gamma-ray beam. This energy consistency can be achieved by maintaining consistent aiming of thegamma-ray beam, therefore, a steady electron beam orbit and angle at the Compton collision point. To over-come the booster leakage field problem, we have developed an orbit compensation scheme. This scheme isdeveloped using two fast orbit correctors and implemented as a feedforward which is operated transparentlytogether with the slow orbit feedback system. In this paper, we will describe the development of this leakagefield compensation scheme, and report the measurement results which have demonstrated the effectivenessof the scheme.
Key words: field compensation, feedforward, storage ring, beam orbit
PACS: 29.20.db Introduction
The High Intensity Gamma-ray Source (HIGS)at Duke University is an accelerator-driven Comptongamma-ray source for photo-nuclear research. It pro-vides a very high intensity gamma-ray beam with switch-able polarization in the energy range from 1 MeV to 100MeV. The HIGS facility consists of three accelerators: alinac pre-injector (0.16 GeV), a booster injector (0.16–1.2 GeV), and an electron storage ring (0.24–1.2 GeV)[1, 2], as shown in Fig. 1. The 34-meter-long southstraight section hosts the Duke Free-Electron Lasers(FELs) with several undulator configurations, producingan FEL beam with wavelength ranging from 190 nm to 2 µ m [3]. The FEL beam is trapped in a 53.73 m long low-loss laser cavity to be synchronized with the circulatingelectron beam (2.79 MHz). Operating in the two-bunchmode with two electron bunches separated by a half ofthe storage ring circumference from each other, a highflux gamma-ray beam is produced in the middle of thesouth straight section by colliding one electron bunch head-on with the high power FEL beam produced bythe other electron bunch [4, 5]. After collimated by thedownstream collimator, the gamma-ray beam becomesquasi-monochromatic and is delivered to the target room.However, due to the proximity of the booster injec-tor to the storage ring, the magnetic field of the boosterdipoles close to the storage ring can significantly alterthe closed orbit in the storage ring when being operatedin the low energy region. The resulted beam orbit distor-tion can be a significant problem for low energy gamma-ray production, which requires consistent aiming of thegamma-ray beam, therefore, a steady electron beam or-bit in the collision area. This kind of disturbance can bea problem for similar accelerator facilities in which thebooster injector shares the same tunnel with the storagering [6, 7]. This kind of problem is also very importantfor the next generation light source, e.g. diffraction lim-ited storage ring (DLSR) [9], since electron beams in theDLSR can be sensitive to the integrated magnetic fieldvariation on the level of 10 − T · m. Therefore, it is im-portant to develop shielding or compensation schemes Received XX XXX 2016 ∗ Supported by National Natural Science Foundation of China (11175180, 11475167) and US DOE (DE-FG02-97ER41033)1) E-mail: [email protected]) E-mail: [email protected] c (cid:13) o be submitted to Chinese Physics C for the leakage magnetic field from a nearby rampingbooster [6, 8].At the HIGS facility, during routine operation theelectron beam is first injected into the booster at 169MeV, the booster is then ramped to bring the electronbeam to the extraction energy (any where between 240MeV and 1.2 GeV), the electron beam is extracted intothe storage ring, and finally the booster is ramped up tothe maximum energy of 1.2 GeV and then down to 169MeV to finish one injection cycle. In this process, it isobserved that the beam orbit in the storage ring is sig-nificantly affected by the booster leakage field, as shown in Fig. 2. For gamma-ray operation above about 15MeV, Compton scattered electrons lose too much energyso that they cannot be retained in the storage ring [10].In this electron loss mode, continuous electron beam in-jection into the storage ring is needed to compensate fora high rate of electron loss. In this mode, the amount oftime during which the collision orbit is altered due to theperiodic booster ramping becomes a significant portionof the gamma-ray production. The adverse impact of thebooster leakage field is most significant for the Comptongamma-ray production using a low energy electron beambelow 600 MeV. Fig. 1. The layout of Duke storage ring and its booster injector. The closest booster dipole magnet is about 1.7maway from the nearby storage ring beam line. M ea s u r ed r m s o r b i t [ µ m ] xy Fig. 2. (color online) The measured rms beamorbit deviations in the storage ring as a functionof time while continuously ramping the boosterwith a repetition rate 4.6 s. The storage ring isoperated at E SR = 426 MeV. To obtain a consistent collision orbit for the gamma- ray operation, an orbit compensation scheme for thebooster leakage field is developed. This scheme is devisedusing a beam based technique with two air coil correc-tors, and implemented as a feedforward system which isoperated transparently together with the slow orbit feed-back system. In this paper, we will present the develop-ment of the orbit compensation scheme for the boosterleakage field, and report the measurement results of thiscompensation in two different modes: (1) the static modeof operation while stepping the booster energies, and (2)the dynamic mode of operation with continuous boosterramping. Field leakage compensation scheme
The closed orbit in a circular accelerator is deter-mined by the magnetic field (i.e. in magnets) and elec-tric field (e.g. in the RF cavity) around the accelerator,which can be distorted by the field errors. To control theorbit distortions caused by slow and distributed magneticfield errors, a slow orbit feedback is commonly employed,2 2 2 o be submitted to Chinese Physics C in which the corrector strengths are calculated based onthe beam orbit deviations measured using beam positionmonitors (BPMs) and a pre-measured response matrix.At the Duke storage ring (DSR), the slow orbit feed-back system is composed of 55 horizontal correctors, 24vertical correctors and 32 BPMs [11]. It is operated at1 Hz, which is fast enough to correct the orbit distor-tions caused by slow varying errors, e.g. those associ-ated with temperature changes and slow power supplydrifts. For the booster injector, the full energy ramp upand down cycle takes about 1.2 s to complete, hence itis not feasible to control the orbit perturbation causedby the booster leakage field using the existing slow or-bit feedback system. The fast orbit feedback operatingat a few hundred or thousand Hz may be feasible forthe booster orbit correction [12], but it requires a signifi-cantly investment on the electronics and vacuum system.Hence, an alternative economic compensation scheme isneeded. Since the booster is located to the northeast ofthe storage ring, the leakage field is also confined to thisarea. To minimize the orbit perturbation outside of thisarea, some local correctors can be employed. Moreover,as the strength of the leakage field is repeatable and canbe pre-determined, a feedforward correction scheme canbe used. In this scheme, the corrector strengths can bedetermined in advance as a function of the leakage fieldstrength or the booster energy, and correctors can besynchronized with the booster energy ramping.In the HIGS operation, a nearly monochromaticgamma-ray beam is produced by sending the beamthrough a small collimator. The aiming of the gamma-ray beam is determined by the position and directionof the relativistic electron beam at the collision point.Therefore, to maintain the energy resolution of thegamma-ray beam, the change of the electron beam orbit(both displacement and angle) at the collision point mustbe kept small. For some electron beam displacement ∆ x c and angular deviation ∆ θ c at the collision point, the off-set of the gamma-ray beam center ∆ x coll at the locationof the collimator is given by∆ x coll = ∆ x c + ∆ θ c · L, (1)where L = 53 m is the distance between the collisionpoint and the collimator. If we require ∆ x coll to be lessthan 10% of the smallest collimator diameter (6 mm),the maximumly allowed electron beam displacement is∆ x coll,max = 0 . θ coll,max = 11 µ rad. Typically, theelectron orbit displacement is small, the main problemis the change of the electron beam angle at the collisionpoint, as its effect is significantly amplified by the largedistance between the collision point and the location ofthe collimator ( L = 53 m).The perturbed orbit in the storage ring is measured with the booster injector parked at 1.2 GeV (the maxi-mum energy), as shown in Fig. 3. It is observed that theperturbation is mostly in the horizontal direction withthe maximum offset of 179 µ m around the storage ring.In this case, the orbit displacement is about 115 µ m andangular variation is about 14 µ rad at the collision point.In vertical direction, the orbit perturbation is one orderof magnitude smaller, which is too small to be a prob-lem. Thus, the compensation scheme will focus on re-ducing the horizontal orbit distortion. Scaling this mea-surement to the lowest energy operation of the storagering (240 MeV), to achieve the tight tolerance set for thegamma-ray beam collimation established previously, ourgoal is to reduce the electron beam orbit displacementand angle at the collision point by a factor of about 2.5. O r b i t de v i a t i on [ µ m ] Collisionregionxy
Fig. 3. (color online) Perturbed electron beamorbit along the storage ring. Measured with thestorage ring energy E SR = 426 MeV and thebooster energy E BST = 1 . E SR = 426 MeV, E BST = 500 MeV. o be submitted to Chinese Physics C
With the booster ramped to the maximum energy of1.2 GeV (Fig. 3), the effective kicking angle, assumingthe orbit disturbance is localized, can be estimated to beabout θ ≈ µ rad ( E SR = 426 MeV), or an integratedfield about 2 . × − T · m [13]. This correction is small,comparable to the strength of a typical corrector in theorbit feedback system. It is also noticed that the hori-zontal phase advance in the northeast area of the storagering is ∆ µ x ≈ . π , therefore only a few correctors in thisarea are needed to minimize the orbit perturbation bothinside and outside the area of the booster leakage field.Five correctors in the northeast area of the storagering are chosen as the candidates to compensate the lo-calized booster leakage field. A beam based techniqueis used to determine the most effective combination ofcorrectors. The response matrix for these correctors asshown in Fig. 4. Using the singular value decomposition(SVD) algorithm, the most effective eigenmode is foundto be dominated by the N07FC and N02FC correctors.It is also known that these two correctors are separatedby a betatron phase difference of 1 . π , which means theyare almost equivalent but with opposite kicking angles.As the N07FC corrector is located further from the errorsource area, the N02FC corrector is clearly the betterlocal corrector. In addition, a second corrector N03FCwhich has a 0 . π phase difference from N02FC is se-lected as the secondary corrector to reduce the residualorbit errors that N02FC is not capable of correcting. C o rr e c t o r c o il c u rr en t [ A ] NAC01NAC02
Fig. 5. (color online) Strengths of NAC01and NAC02 measured with the booster injectorparked at a set of energies between 169 and 1200MeV. E SR = 426 MeV. Since a typical booster energy ramping takes only1.2 s, fast power supplies for the correctors are needed.A couple of ramping power supplies which are properlysynchronized with the booster energy ramping are usedto drive the N02FC and N03FC correctors. In the mea-surement it was found that the magnetic field produced by these correctors has a time delay due to the inducededdy current in the soft-iron yoke. To overcome thisproblem, two dedicated air coil correctors NAC01 andNAC02 were developed and installed next to the N02FCand N03FC correctors, respectively, as shown in Fig. 1.The strengths of the new correctors are determinedwith the booster injector parked at different energies(static mode), as shown in Fig. 5. It is clear that thestrength of NAC01 is much larger than that of NAC02,which indicates that the booster leakage field is mostlylocal, therefore its main effect can be corrected using onelocal corrector NAC01. The maximum integrated fieldproduced by NAC01 is estimated to be about 2 . × − T · m, which is close to previously estimated value. Itis observed that the strength of NAC01 is almost linearin the booster energy range from 300 MeV to 900 MeV,but in the low and high energy regions a nonlinear depen-dency on the booster energy shows up, which is likely re-sulted from the changed distribution of the leakage field.The compensation scheme is implemented in the EPICSbased control system as a feedforward control, where thecorrector strengths track the booster energy ramping. Compensation results
To demonstrate the effectiveness of the compensa-tion scheme, the uncompensated and compensated elec-tron beam orbits in the storage ring are measured by theBPMs with the booster operated in both static and dy-namic modes. Its usefulness is further confirmed usingthe direct measurements of the aiming stability of theundulator radiation.
Static mode M ea s u r ed r m s o r b i t [ µ m ] Booster energy [MeV] ∆ x c o ll [ mm ] ∆ x un ∆ x comp ∆ x coll,un ∆ x coll,comp Fig. 6. (color online) The rms beam orbit changein the storage ring and the estimated gamma-raycenter shift at the collimator as a function of thebooster energy with the booster being operatedin the static mode. ∆ x un is the rms beam or-bit measured without compensation and ∆ x comp is the value with compensation. ∆ x coll,un is the o be submitted to Chinese Physics Cchange of the gamma-ray beam center at the col-limator without compensation, ∆ x coll,comp is thevalue with compensation. E SR = 426 MeV. As shown in Fig. 6, the beam orbit in the storagering is measured using all BPMs around the storage ringwith the booster operated in the static modes. The beamorbit offset ∆ x c and angle ∆ θ c at the collision point canbe calculated using the two BPMs located at the eitherend of OK5-B and OK5-C undulator, respectively. Thenthe variations of the gamma-ray beam center at the col-limator can be obtained using Eq. 1. It is observed thatthe trend of the storage ring rms orbit variation withthe booster energy is very similar to the variation of thegamma-ray beam center at the collimator, which indi-cates a strong correlation between them. The measure-ment results show that in the static mode the compen-sation scheme reduces the maximum rms orbit variationby a factor of about 15, the same goes to the gamma-raybeam center variation, which is reduced from 840 µ m toabout –67 µ m, well below the stability requirement forthe HIGS operation (see in Section 2). Dynamic mode M ea s u r ed r m s o r b i t [ µ m ] Booster ramping time [s] ∆ x c o ll [ mm ] ∆ x un ∆ x comp ∆ x coll,un ∆ x coll,comp Fig. 7. (color online) The measured rms orbit dis-tortions and estimated gamma-ray center varia-tions at the collimator during a full booster rampcycle. During the cycle, the booster is parked at1.2 GeV for about 2 . s (from t = 2 . . E SR = 426 MeV. At the Duke storage ring, while the slow orbit feed-back system uses the highly filtered low noise orbit dataat a few Hz, the BPM acquisition system is capable ofproviding data at a higher rate of 30 Hz. The 30 Hzorbit data are fast enough to visualize the storage ringorbit changes during the 1.2 s booster energy ramp. Asshown in Fig. 7, the beam orbit in the storage ring ismeasured for a complete booster ramping cycle. In thismeasurement, the booster starts its ramp from 169 MeV at about t = 1 . t = 2 . x o r b i t [ µ m ] Fig. 8. (color online) Electron beam positions mea-sured using the BPM at S07 sector with boosterramping. The beam orbit is measured with thebooster ramped up in four discrete energy stepsand parked at each energy for 2.5 s, and thenramped down to complete a cycle. As a refer-ence, the uncompensated orbit at this locationwas about +179 µ m with the booster energyparked at 1.2 GeV. E SR = 426 MeV. The storage ring beam orbits were also measured withthe booster energy ramped up from 169 MeV to 1.2 GeVin four discrete steps while parking at intermediate en-ergies of 200 MeV, 600 MeV, 1 GeV, 1.2 GeV for 2.5second each, and then ramped down to finish one cycle.The beam positions measured using the BPM at S07 sec-tor (S07BPM) are shown in Fig. 8. It is observed thatthere is a small orbit peak during energy ramp-up and asmall orbit shift after each ramp, both are the results ofsmall under-correction. It is also observed that the beamorbit bump is reversed along the ramp-down process,5 5 5 o be submitted to Chinese Physics C indicating orbit over-correction. These orbit peaks, ei-ther as under-correction during energy ramp-up, or over-correction during energy ramp-down, are the results ofthe delay in the field compensation system. While manyfactors can contribute to such delay, including the finitetime response of the control system and corrector powersupplies, the most likely factor is the eddy current ef-fect in the thick storage ring vacuum chamber (3 mm,stainless steel).
Direct measurement of undulator radiation aim-ing
Beam orbit stability in the south straight section isfurther verified by directly measuring the stability of theFEL undulator radiation aiming. The OK5-B and OK5-C undulators located adjacent to the Compton collisionpoint are turned on and set to a fixed current to pro-duce synchrotron radiation. The undulator photon beam(center wavelength λ cen = 550 nm, E SR = 426 MeV) ismeasured using a beam profile monitor located about 30m downstream from the Compton collision point. TheCCD camera in the monitor is capable of taking about14 frames per second in the burst mode. Due to the longdistance between the radiation source and the camera,the variation of the undulator beam image center at thecamera is mainly caused by the change of the electronbeam angle in the undulators. In this measurement, abandpass optical filter is used to reduce the backgroundradiation. x [pix] y [ p i x ]
100 200 300 400 500 60050100150200250300350400450
Fig. 9. (color online) Measured FEL undulatorradiation beam image with center wavelength λ cen = 550 nm. The beam image centers are fittedwith Gaussian functions using the intensity distri-bution within the selected areas. The yellow dash-lines: the boundaries of the selected areas; greendot-lines: the intensity distribution of the data inthe selected areas; red lines: the fitted curve ofthe intensity distribution. E SR = 426 MeV. The image data are processed first by subtracting thebackground noise, and the image center is found by fit- ting the intensity distribution of a selected area in thehorizontal or vertical direction using a Gaussian func-tion, as shown in Fig. 9. Hence, the beam images can berecorded with the booster operated in the dynamic mode.The measurement results with booster ramping in a fullcycle, before and after compensation, are shown in Fig.10. Without compensation, the average perturbation ofthe image center is about 770 µ m (average over measure-ments between t = 4 . t = 6 . − µ m (average) with the compensation, whichis within the level of the rms noise (58 µ m, from databetween t = 0 to 2 s and t = 8 to 10 s when the boosteris operated at the lowest energy of 169 MeV). Along thebooster ramp-down process, a small but visible bumpshows up in the negative direction. The height of thispeak is about 1/4 of the uncompensated value with thebooster operated at 1.2 GeV, which is consistent withthe previous BPM measurement result. µ m1.2 GeVBooster ramping time [s] I m age c en t e r v a r i a t i on [ µ m ] Rawdata,unSmoothfit,unRawdata,compSmoothfit,comp
Fig. 10. (color online) Variations of undulator ra-diation image center measured using a beam pro-file monitor during a full booster ramp cycle, thebooster is parked at 1.2 GeV for about 2.5 s as inFig. 7. Blue stars are the measured beam centerswithout compensation; red-circles are those mea-sured with compensation; blue-line and red-lineare the smoothed curves using a 5-point movingaverage. E SR = 426 MeV. Summary and discussion
The stability of the electron beam orbit is a criticalrequirement for the HIGS operation at the Duke storagering. To keep the aiming of the gamma-ray beam stable,therefore to maintain the energy resolution of the highlycollimated gamma-ray beam, the e-beam orbit distortionin the storage ring, especially at the Compton collisionpoint, needs to be minimized. In the development of thecompensation scheme for the booster leakage field, two6 6 6 o be submitted to Chinese Physics C effective correctors are chosen using the beam based tech-nique. Faster air coil correctors without steel yokes aredeveloped and installed, and their strengths for orbit cor-rection are determined by stepping the booster througha set of energies between 169 MeV and 1.2 GeV. Thefield compensation is implemented as a feedforward inthe realtime booster control system.With this compensation scheme, the maximum per-turbation of the horizontal beam orbit and gamma-raybeam aiming is significantly decreased by a factor of 3with the booster operated in the ramping mode. In theroutine operation, the overall time-integrated distortionin the gamma-ray beam aiming is expected to be de-creased by a factor of 14 to 25. The projected maximumhorizontal gamma-ray beam center shift at the collima-tor is reduced from 1677 µ m to 447 µ m (about 340 µ min the vertical without the need for compensation) withbooster ramping and for the lowest energy operation of the storage ring (240 MeV). This value is about 7% ofthe smallest collimator diameter ( D = 6 mm), well belowour preliminary goal of 10%. We plan to improve theorbit compensation during the booster ramp-down pro-cess. Instead of using energy-indexed corrector settingsmeasured by stepping up the booster, a new set of set-tings will be determined by stepping down the boosterenergy. We would like to thank the engineering and technicalstaff at DFELL/TUNL for their support of this researchwork. This work was supported by
National Natural Sci-ence Foundation of China (No. 11175180, 11475167) andDOE Grant (No. DE-FG02-97ER41033). One of the au-thors (Wei Li) also would like to thank the China Schol-arship Council (CSC) for supporting his research visit atDuke University.
References : 224801 (2006).4 V.N. Litvinenko, J.M.J. Madey, Nucl.Instr.Meth. in Phys. Res.Sect. A, : 580-583 (1996).5 V.N. Litvinenko, et al, Phys.Rev.Lett, : 4569 (1997).6 W. Joho et al, Nucl. Instrum. Methods A, (1): 1-11 (2006).7 H.J Tsai et al, Hardware Improvements and Beam Commis- sioning of the Booster Ring in Taiwan Photon Source, Pro-ceeding of IPAC2015, Richmond, USA (2015).8 R. Keller et al, Orbit stability of the ALS storage ring, Pro-ceeding of PAC1997, :784-786 (1997).9 M. Eriksson et al, J. Synchrotron Radiat., (5): 837-842(2014).10 Y.K. Wu, et al, Performance and Capabilities of Upgraded HighIntensity Gamma-ray Source at Duke University, Proceedingsof the PAC09 (2009).11 Y.K. Wu, et al, BPM and orbit correction systems at the Dukestorage ring, Proceedings of the PAC03, : 2479-2481 (2003).12 M. B¨oge, et al, Fast Closed Orbit Control in the SLS StorageRing, Proceedings of the PAC1999, : 1129-1131 (1999).13 S.Y Lee, Accelerator physics (World Scientific, 2004).: 1129-1131 (1999).13 S.Y Lee, Accelerator physics (World Scientific, 2004).