Facility Highlights

96

Facility Highlights

Precise Alignment of Magnetic Field Centers of Quadrupole Magnetson the Girder

Taiwan Photon Source (TPS) project was proposed 3 years ago to build a 3GeV synchrotron light source. To achievethe designated ultra-low emittance of this new light source requires high precision positioning of storage ringmagnets. Among all, the alignment of quadrupole magnets is of most importance since it directly affects the closedorbit of electron beams. Conventional on-site alignment of quadrupole magnets mainly relies on the theodoliteperformance. Most of theodolite measurement errors come from human-eye resolution and target precision. Thecumulated errors could be in the order of 0.1 mm. In this report, a new alignment scheme is proposed to enhance theon-site alignment of quadrupole magnets for TPS project. To achieve the high precision requirements, a devicepossessing the advantages of both Vibrating Wire Method (VWM) and Position Sensing Detector (PSD) is suggested.The development of this alignment device is anticipated to provide a mechanism to better align the quadrupolemagnetic centres on girder with positioning errors less than 30 μm.

Taiwan Photon Source has been planned for several years inNSRRC and is in its final designing phase. Currently this 3GeV,ultra-low emittance synchrotron light source has a 486m storagering with 24 periods and hundreds of magnets. The requirementfor high accuracy magnetic field and magnets alignment areessential aspects for the machine performance. Many methodshave been attempted to exam the magnetic field of magnets,which include Hall probe measurement, pulsed wire technique,vibrating wire field measuring technique, moving wire tech-nique, moving probe technique, and rotating coil technique, etc.These techniques were able to identify individual magnetic fieldwith high accuracy. However, when positioning magnets ongirders, these methods were usually inapplicable. Technicianshave to rely on theodolite to check the alignment of magnetsbased on mechanical targets. The cumulative alignment errorsbetween magnetic centres and girder could be larger than 100μm.

To improve the alignment accuracy of quadrupole magnets,a reliable magnetic field measurement system has been pro-posed in NSRRC. This proposed method utilizes the advantagesfrom both Vibrating Wire Method (VWM) and laser-PSD (positionsensing detector) method. The fiducialization concept is to per-form on-site magnetic field measurement using VWM then examthe misalignment between magnets using the laser-PSD system.

Vibrating Wire Method (VWM)The vibrating wire method has been known for its high accu-

racy for locating quadrupole magnetic centres. It is composed ofa stretched wire in a magnetic field, as shown in Figure 1. By

Authors

L. Tsai, D. J. W ang, S. Y . Perng,and T. C. FanNational Synchrotron RadiationResearch Center, Hsinchu,Taiwan

Facility Highlights

97

sending a sinusoidal current through the wire, thevibration of the wire induced by the Lorentz Force couldbe observed. When the vibration frequency is close to thenatural frequency of the wire, the vibration amplitudewould increase greatly hence enhance the sensitivity. Thevibrating amplitude of the wire is directly proportional tothe magnitude of the surrounding magnetic field. Whenthe wire is positioned closer to a quadrupole magneticcenter, the measured wire vibration amplitude woulddecrease accordingly, Figure 1 (b). A great advantage ofVWM is that it could be operated in a relatively smallerspace and is easier to set up on girder. Combined withlaser-PSD system, the alignment of quadrupole magneticcentres on girder is much more accurate.

PSD Alignment SystemThe position sensitive detector (PSD) has been widely

used in many engineering areas to precisely align multi-ple targets. It is composed of large area photodiodes todetect and record the position of an incident light beam.The PSD system can measure position movement with

high resolution (to the sub-micron level depending onthe active area of PSD). Figure 2 shows the configurationof the VWM and PSD system. Two PSDs were mountedclose to the ends of the stretched wire. Front PSD wasmounted on a slider in order to provide a laser light pathfor Back PSD. The tilt of PSDs was monitored and adjustedusing precision level and adjusting base, respectively. PSDtilt was controlled to within 0.1mrad. The natural frequen-cy of the wire is adjusted by changing the applied tensileforce using XYZ stage. The optical detector (phototransis-tor-LED assemblies H21A1) measured the vibrating ampli-tude of the stretched wire in both transverse and verticaldirections.

Experiment Procedures and ResultsTwo quadrupole magnets were pre-aligned on a

girder using theodolite and precision level, as shown inFigure 3. The magnetic centres of the quadrupole mag-nets were located and their misalignments were checkedthrough following procedures:(1) The vibrating wire was tensioned around the ideal

centerline of the quadrupole magnet Q1. (2) HP 33120A function generator was connected to the

wire to provide sine current. The frequency of the ACcurrent was carefully adjusted to match the funda-mental mode frequency of the Be-Cu wire.

(3) The vibration amplitude was recorded by a pair ofoptical detector H21A1, as shown in Figure 2. Themagnetic center of the quadrupole was found by mov-ing the vibrating wire along the transverse and verticaldirections. Figure 4 shows the experiment result alongtransverse direction. The location of lowest vibratingamplitude implies the magnetic center of the quadru-pole magnet.

Fig. 1: Vibrating Wire Method.

Fig. 2: VWM and PSD system setup.

Facility Highlights

98

(4) The driving current was then tuned to match the sec-ond harmonic frequency of the wire in order to findthe magnetic center along pitch and yaw directions.Figure 5 shows the wire vibrating amplitude measure-ment along yaw direction. The approximated yawmeasurement error is 0.015 mrad.

(5) A fixed laser, as shown in Figure 3, was pointing to thetwo PSDs to identify the position of the wire. A micro-scope with 50 times magnification was utilized tomeasure the position difference between PSD and thewire.

(6) The Be-Cu wire and PSDs were then moved to quadru-pole Q2.

(7) By repeating step (1) through (4), the magnetic centerof Q2 was identified. The misalignment of magneticcentres between Q1 and Q2 can be determined by thelaser and PSD system.

Overall System Error EstimationA summary of all the estimated errors is listed as fol-

lowing:

．Vibrating Wire Method: 7μm. The location with lowestvibration amplitude denotes quadrupole magneticcenter, and it was found by line fitting through the datapoints. The error came from the approximation is about7 μm.．Wire positioning: 15μm. In this research, PSD sensors

were used as the coordinates of the wire on girder,hence precise positioning of PSD and wire is very criti-cal. To measure the location of PSD and wire, a micro-scope with 50 times magnification was used. However,because of human eye resolution limit, error generatedduring this process is in the range of 10~15μm. Also,5μm error is expected when the wire is demountedand remounted between different magnets.

Fig. 4: Vibration amplitude vs. transverse location. Fig. 5: Vibration amplitude vs. yaw angles.

Fig. 3: Experiment setup.

Facility Highlights

99

．PSD repeatability: 5μm. The drift of the laser was about2μrad plus 3μm repeatability from repositioning thefront PSD during the experiments.．Sag difference: 5μm. All wire methods suffer from sag

of wire. The sag-generated vertical position offsetswere about 25~100μm depending on the applied ten-sion force. To estimate the sag of a vibrating wire, a sim-ple relation between sag and the fundamental modefrequency has been proposed by Temnykh:

S = ( )-2

In which S represents the sag of the stretched wire, g= 9.8 m/s2, and w1 represents the fundamental mode fre-quency of the wire. In this project, the fundamental modefrequency of the vibrating wire was precisely controlledby means of carefully adjusting the tension force in thewire. The sag difference of the wire between differentlocations was about 5μm.

ConclusionsThe proposed VWM and PSD method is still under

development and will focus on eliminating or minimizingthe existing error sources. Another VWM setting, whichutilizes a stretched long wire through two or morequadrupole magnets, has also been planned to verify theaccuracy of this proposed technique.

w1

2πg

32

Tab. 1: Estimated Errors of the VWM and PSD system.

References

．N. A. M orozov, Review of the magnetic measurementtechnique (experience of the SLC, LEP, CEBAF),Workshop on the TESLA spectrometer, Dubna, 13-14October 2003.．J. G. M elton, M. J. B urns, and D. J. H onaberger, IEEE

Particle Accelerator Conference Preceeding, 2944(1993).．Z. Wolf, A V ibrating Wire System For Quadrupole

Fiducialization, LCLS Document TN-05-11, M ay 2005.．Temnykh, Nuclear Instruments and Methods in Physics

Research A , 399 (1997).．S. G. A rutunian, N. M. D obrovolski, M. R. M ailian,

I. G. S inenko, and I. E . Vasiniuk, V ibrating Wire for BeamProfile Scanning, Physical Review Special Topics-Accelerators and Beams, Vol. 2 (1999).．H. J. T sai, H. P. Chang, P. J. C hou, C. C. K uo, G. H. L uo, and

M. H. W ang, Closed Orbit Correction of TPS StorageRing, Proceedings of EPAC 2006, Edinburgh, Scotland.

Contact E-mail

djwang@nsrrc.org.tw