COMPARISON OF DIFFERENT MAGNETIC MEASUREMENT TECHNIQUES. Isaac Vasserman, Shigemi Sasaki Argonne National Laboratory, Argonne, IL 60439, USA Abstract The magnetic measurement system at APS was upgraded. A stretched wire system was added to the existing rotating coil, search coil and Hall probe systems. New software was created. Step motors were replaced by servomotors that resulted in an increased signal-to-noise ratio by a factor of 10. Comparison of results obtained from measurements using different techniques was done. Some specifics of Hall probe sensors are discussed. Limitations, advantages and disadvantages of different magnetic measurement techniques are discussed as well. Introduction Different magnetic measurement techniques are used for magnetic measurements. One of the reasons is to prove the validity of the measurements. The preferable way s to use different magnetic measurement techniques and to compare the results. This is especially true for a new technique, which must be compared with well-established techniques before implementation. Another reason is the wide range of the tasks. Different geometry, different range of magnetic field, different requirements for the quality of the field mean that different measurement techniques, appropriate to a specific goal, should be used. The APS magnetic measurement facility has two benches for magnetic measurements. One of them is 3 m long, and the other is 6 m long. Magnetic measurement techniques used for the measurements include: 1. Long rotating coil for measurements of the first and second field integrals, and integrated multipole components of the field [1,2]; 2. Hall probe for measurements of a detailed field map that allow calculations of all necessary features of the radiation; 3. Stretched wire system, recently implemented at the 3 m bench for measurements of field integrals [3]; 4. Short 2-3 period moving coils for measurements of field integrals; 5. Helmholtz system for measurement of permanent-magnet magnetic moments; and 6. Calibration system for Hall sensors calibration. Recent upgrades of the system include new software and hardware, described in refs. 4 and 5. Some details of these upgrades will be discussed later. Measurements Results of measurements using different magnetic measurement techniques are shown in Figs. 1-5. The rotation coil system was upgraded in 2003. New hardware, including servomotors instead of step motors, was implemented. The number of measured points during rotation was increased from 4 to 3000 points, which allows fitting with a proper sinusoidal function. This feature reduces errors associated with the vibration of the coil during rotation. Results of measurements using the rotating coil are strongly affected by the quality of the coil. The-turn twisted Litz wire used at the APS for rotating coil is easy to use and has rather large signal-to-noise ratio. The main problem with measurements of the first and second field integrals of such a coil is deviation of the coil width in the Z-direction. This effect is proportional to the field amplitude and affects mostly results at small gaps. To eliminate such dependence (or, better to say, to make it less significant), the measurements are performed at 2 positions of the coil, separated by ½ of an undulator period. Averaged value is then taken into account. Coil vibration during rotation is another source of errors and will be discussed in more detail later. Fig. 1 shows the signal from the coil during rotation. The rather large noise seen here is associated with 60 Hz and higher harmonics of the electrical power source and vibration of the wire. 1
Fig. 1 Signal from the rotating coil. Fig. 2 Integrated signal from the rotating coil. High frequencies of electrical noise do not affect the result of integration, which was checked by implementing filters. Vibration however is important and is a reason for the difference of the integrated signal from the theoretically predicted sinusoidal function. Fitting with an appropriate function allows correction of this effect (see Fig. 2). The moving coil is the perfect tool to measure the first field integral of the device. It is linear and not critical to the geometry of the coil. Its main problem is zero drift of the flux meter, which is usually used to obtain the field in the process of coil motion. It requires large statistics and makes these measurements rather time consuming, so they are used mainly as a reference to confirm the reliability of other sensors. Linear regression correction, assuming linear zero drift during one scan is included in the calculation of the field integrals. Comparison of the results for the moving coil and the rotating coil is shown in Fig. 3. Rather good agreement between these two sensors shows the reliability of both techniques. The difference is much less than the gap-dependence requirements for the APS storage ring. 2
Fig. 3 Comparison of the results from the moving coil and the rotating coil. To check the reliability of the stretched wire technique recently implemented at the 3 m bench, another test was done using the rotating coil and stretched wire. The comparison of measurements from 2001 and 2003 shows rather good agreement, indicating that vibration is not a critical point in these measurements even in the previous version of the system. The stretched wire does not have the above-mentioned problem of the rotating coil, associated with width of the coil, because it consists of one wire moving parallel to itself, so the results of field integral measurements using the stretched wire are more reliable in principle than those using the rotating coil. Results from Fig. 4 agree with good enough accuracy and show that both techniques could be used. It is worth mentioning however that measurements of higher orders of integrated multipole components are much more sensitive to the errors, and vibration is a possible source of such errors. Due to this fact, integrated multipole component measurements using the stretched wire was not reliable. The accuracy of stretched wire measurements can be improved by decreasing the nois e and by implementing damping of the wire vibration. Fig. 4 Comparison between first field integrals obtained from the stretched wire and the rotation coil measurements. 3
Another possible use the of rotating coil is measurement of the integrated multipole components of the field. These measurements could be done in rotation mode or translation mode. In rotation mode the coil measures the first field integrals in several locations in X-direction. During translation the coil does not rotate but moves parallel to itself. Only the difference in magnetic flux through the coil area is measured in this case, so the dipole component is not available and is taken from the rotation mode. Fig. 5 Comparison of integrated multipole components measured by the long coil in rotation and translation modes. Vibration of the coil can be easily seen in the Fig. 5. Vibration-induced errors increase at the end of the scan. According to the direction of the coil plane the horizontal or vertical field change is measured on the fly in 4000 points in our case. Some aspects of Hall probe measurements A Bell Hall probe was used at the APS as one of the Hall sensors to measure the field map. We discuss here the effect that has to be taken into account using this probe. It was noticed for a long time that results of field integrals measurements using this probe were sensitive to the angle of sensor around the vertical axis of the plate when measuring the vertical field. Fig. 6 shows the first field integral dependence on the angle measured by this probe. Sensitivity of this dependence is about 1000G-cm/deg. It is possible to find the proper angle, Fig. 6 First field integral dependence on the angle for the Bell probe. 4
which corresponds to results obtained from other sensors, like the moving coil or the stretched wire. The angle is equal to 5.2 degrees for this particular sensor and differs from 0 to 3 degrees for other sensors. Different possible reasons of such an effect were investigated. The most reasonable explanation of such dependence is found to be the nonuniform sensitivity of the sensor area to magnetic field. Rather big dimensions of the sensor (about 2 mm in diameter) lead to a different magnetic field along the sensor area for the field with a strong gradient. Calibration of the sensor was done at uniform field and involves some average sensitivity of the probe. Measurements in a nonuniform field result in errors. Amore detailed investigation of this effect showed that there is some accumulation of errors in the periodic field. We discuss here in brief the preferable use, from our point of view, use of different magnetic measurement techniques. Only magnetic measurement techniques used at the APS are mentioned. Conclusion 1. Rotating coil. This technique could be used for measurements of first and second field integrals, especially in the case when these integrals are close to zero and the period length of the magnetic structure under measurement is much less than the length of wire vibration period. The errors increase with increase of the magnetic field and undulator/wiggler period and length of the device. Measurements of integrated multipole components are reliable and are not dependent on quality of the coil. The translation mode allows decreasing of the required time and is the preferred way for such measurements. Different types of coil design are used in different laboratories (one wire, flat coil, and so on). 2. Stretched wire measurements. It is the preferred technique for measurements of the first and second field integral, especially for the vertical field with very small gap. Stretched wire results for integrated multipole components are also reliable in the case of short wire length. The large area of the coil (in comparison with rotating coil) and wire vibration are the reasons for the strong noise, which affects the accuracy of the measurements, especially for multipole components of long devices. 3. Search coil. We define such a coil as a short (1-3 period length) moving coil that measures averaged field on-the-fly with a flux meter or without it. Results are quite reliable but require large statistics and are therefore time consuming. It is recommended as a reference for other measurement techniques. 4. Hall probe. The main tool used to measure detailed field distribution. The so-called, vertical Hall probe [6] appears to be preferable to avoid the planar Hall probe effect [7]. It also has the advantage of very small size of the sensors, located in the same area in the case of 2-axis probe. Using Hall probes for precise measurements of field integrals requires good calibration with small steps. References 1. D. Frachon, Ph.D. Thesis, ESRF, April 1992 2. D. Frachon, I.Vasserman, P.M.Ivanov, E.A.Medvedko, E.Gluskin, and N.A.Vinokurov, Argonne National Laboratory, report no. ANL/APS/TB-22, March, 1995 3. D. Zangrando, R.P. Walker, Nucl. Instrum. Methods. A 376 (1996) 275 4. O.A. Makarov, B.N.Deriy, E.R.Moog, E.M.Trakhtenberg, I.B.Vasserman, report at ICAP02, 7 th International Computational Accelerator Physics Conference, October 2002, Michigan State University 5. Y. Eideman, B. Deriy, O. Makarov, I. Vasserman, THE NEW MAGNETIC MEA SUREMENT SYSTEM AT THE ADVANCED PHOTON SOURCE, ICALEPCS-2001-TUAP051, Nov 2001 6. R.S. Popovich, 9 th International Magnetic measurement workshop, CEA/Saclay, June, 1995 7. I. Vasserman, Argonne National Laboratory, Report no. ANL/APS/TB-32, Jan. 1998 5