CERN EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH PERMANENT MAGNET QUADRUPOLES FOR THE CLIC DRIVE BEAM DECELERATOR

Transcription

1 CERN EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH CLIC Note 940 Cockcroft ASTeC ID 050 PERMANENT MAGNET QUADRUPOLES FOR THE CLIC DRIVE BEAM DECELERATOR Ben Shepherd, Jim Clarke, Norbert Collomb STFC Daresbury Laboratory, Warrington, United Kingdom Abstract CERN-OPEN /07/2012 STFC in collaboration with CERN has developed a new type of adjustable permanent magnet based quadrupole for the CLIC Drive Beam Decelerator. It uses vertical movement of the permanent magnets to achieve an integrated gradient range of T, which will allow it to be used for the first 60% of the decelerator line. Construction of a prototype of this magnet has begun; following this, it will be measured magnetically at CERN and Daresbury Laboratory. Geneva, Switzerland July 2012

2 Introduction The CLIC Drive Beam Decelerator (DBD) [1] transports two electron drive beams from 2.37 GeV to 237 MeV. Twenty-four sectors of 876m each contain a quadrupole every 1m, arranged in a FODO lattice to focus the beam along the length of the DBD. A total of 41,848 quadrupoles are required. The tunnel containing both the drive beam and main beam has a diameter of 4.5 m. A challenging heat budget of 150 W/m has been set across all components. For this reason, a quadrupole design based on permanent magnet technology is being investigated by STFC in collaboration with CERN. For nominal operation of CLIC, the energy decreases in a linear fashion from 2.4 to 0.24 GeV along the length of the decelerator. In this mode, the quadrupole strengths would be fixed. However, various different commissioning and operating scenarios are envisaged, where different quadrupole strengths would be required. Thus, all of the quadrupoles must be adjustable by a certain amount depending on their position in the decelerator line. This is illustrated in Figure 1. Figure 1. Variation of quadrupole strength along the DBD for different scenarios. Magnet Specifications Strength At the start of the decelerator line, the integrated gradient of the quadrupoles is 12.2T. This is the nominal figure corresponding to 100% in Figure 1. At the end of the line, the nominal strength is 10% (1.2T). The quadrupoles are required to be adjustable to accommodate various scenarios as outlined above, so the minimum and maximum strengths are 7% and 120% respectively. The range of integrated gradients is T. Stability The specification requires the gradient to be set to an accuracy of 5x10-4. Throughout the design process, we have assumed that we can control the position of the moving section to within 10 µm. This is reasonable for the motion system used in the prototype, and will be verified using encoders. Since we know how the gradient changes with position, we can calculate how accurately the gradient can be set. Dimensions The nominal space available for the quadrupoles is 391x391x270mm. However, it is obvious from studying a CAD model of the CLIC modules as they exist at the moment that there are places where parts of the magnet can protrude out of this envelope and not interfere with other components. Conversely, there are parts of this envelope where existing components intrude into this space for instance, the BPMs. Close collaboration between STFC and CERN is required to facilitate integration of this quadrupole into the module design as they both evolve. 2

3 The length available is relatively short, especially at the high-energy end of the decelerator. So it will be quite challenging to meet the highest strength needed this applies to the EM design as well as this one. For this reason, a high-strength quadrupole design was developed first, and this report concentrates on this one. Gradient quality The integrated gradient quality for the CLIC DBD quadrupoles has been specified at 0.1% over the vacuum chamber inner radius, presently fixed at 11.5mm. The beam is not expected to extend over more than half this radius. However, further simulations would be needed to show that the good field region could be reduced. There is no effort available currently for these simulations; so the GGR will be fixed at ±0.1% over ±11.5mm. This will also allow a fair comparison with the EM option being studied at CERN. Inscribed radius and dipole correction The vacuum chamber thickness will be 1.5mm. The minimum inscribed radius is therefore about 13mm. Some extra provision will be made for dipole correction, which will be achieved by moving the magnet off-centre horizontally or vertically. A correction field of 12 Tmm at 2.4 GeV is required, and correspondingly lower for lower energies. Some studies were done with separate corrector magnets, but it was obvious that there was not enough longitudinal space to accommodate them. Increasing the inscribed radius to 13.6mm provides enough extra space to be able to move the magnet horizontally or vertically by 0.8mm. This offset gives the required range of dipole correction. It was agreed that each quadrupole will move either horizontally or vertically, not both. This concept of offset quadrupoles has been used in the design of the non-scaling FFAG, EMMA [3]. Power dissipation The maximum power dissipation in air over all tunnel components is 150 W/m. This is one of the main reasons for studying the PM option there will be virtually no heat dissipation. The operating costs of this magnet will be virtually zero when considered against the electricity and water required for a conventional electromagnetic design. Magnet Design Figure 2. Two possible concepts for an adjustable PM quadrupole: a hybrid EM-PM quadrupole (left) and a rotating Halbach ring (right). Several different concepts were investigated to meet the requirements of the DBD quadrupoles. These included a standard electromagnetic quadrupole with PMs forming part of the yoke, and a magnet with rotating Halbach rings (Figure 2). However, there were drawbacks to both of these concepts. If part of a standard EM quadrupole is replaced with PMs, this creates a high-reluctance flux path, and a large current is still required in the coil to provide the necessary adjustment. The strength of the Halbach design is relatively high, but the field quality at low strength is poor. The quadrupole field is directly created and shaped by the PMs, which places tight constraints on the PM quality. The final design concept employs PMs to drive the flux circuits which create the gradient at the magnet centre. The gradient quality is set by the shape of the steel poles, as in a conventional electromagnet, so reliance on the PM quality is reduced. To adjust the gradient in the high-strength model, the PMs are moved vertically away from the centre, creating an air gap and reducing the gradient seen by the beam. It would be possible to add two more PMs 3

4 that retract horizontally, giving the magnet another degree of symmetry. This was not considered as it would have made the motion system much more complicated. A patent has been applied for, covering design of a PM-based multipole magnet with the design features described below [2]. The patent also covers some features included in the low-strength design, which will be described in a later paper. A 3D representation of the current design for the quadrupoles, generated using Vector Fields Opera [4], is shown in Figure 3. Only one quadrant is shown. The PM material is displayed in green, with an arrow showing the magnetisation direction, and all the ferromagnetic steel is blue. F A E D B C Figure 3. Opera-3D representation of one quadrant of the current design, with the magnet fully closed. The width of the pole (i.e. between the A and B lines) is 40mm. Other labels show the pole (C), the sandwich plate (D), the PM (E), and the bridge (F). Magnetic steel As a typical example of a good magnetic steel, grade 1010 (0.1% carbon) was used for modelling. The B-H curve used is the one supplied with Opera, measured by Cobb and Early at SLAC in 1989 [5]. PM material NdFeB was used in the design. Following a meeting with Vacuumschmelze, the grade VACODYM 837 AP was chosen. This grade has a good combination of properties with a reasonably low temperature variation. Information on grades can be found on the VAC website [6]. Table 1 summarises the properties of the chosen grade. The properties used in the model were the typical values listed in the table. 4

5 Table 1. Properties of VACODYM 837 AP at 20 C. Property Symbol Typical value Minimum value Units Remanence B r T Coercivity H cb ka/m Energy density (BH) max kj/m 3 Remanence temperature coefficient ( C) TK (B r ) -0.11% Coercivity temperature coefficient ( C) TK (H cb ) -0.62% Angled PM Several different models were investigated placing the PM at different angles to the horizontal. The choice affects the PM working point and therefore the efficiency with which the PM drives flux through the magnet centre. In the final design, the PM is angled at 40 to the horizontal. The PM, bridge and sandwich plate (labelled D, E, and F in Figure 3) all move together vertically upwards to vary the magnet strength. Bridge A bridge of steel connects the PMs on either side of the magnet and provides a continuation of the flux circuit. The bridge separates the PMs horizontally by 4.75mm at the narrowest point to make construction easier and to stop the PMs touching each other, which could result in large forces at the point of contact. PM dimensions The PM is 100mm across by 18mm deep. This maximises the magnet strength while keeping the flux density in the pole within the linear region of the B-H curve of the steel, i.e. less than 1.6T. The pole width was adjusted as well as the PM dimensions, in order to distribute the flux evenly through the pole. Faces A and B in Figure 3 are parallel with 40mm between them, so the pole can be cut from 40mm thick steel. Pole profile at tip The pole face was designed with a hyperbolic profile extending outwards from the 45 line, with a tangent taken from this to meet the sides of the pole. The pole has a taper to concentrate the flux in the centre. Decreasing the width of the pole at the tip increases the gradient at the expense of the gradient quality, so there is a compromise to be made. To maximise the gradient quality, the tangent point was adjusted to 17mm and the pole width set to 22mm. Chamfers To reduce the flux density in the pole i.e. to reduce nonlinear behaviour the ends of the poles are chamfered by 3mm. The flux density in the pole is reduced to less than 1.6T in most places. To further reduce the high flux density at the corners, additional chamfers have been added on the edges along the pole taper and longitudinally down the pole. Tunability The moving section can be moved vertically by up to 64mm. This allows the integrated gradient to vary between 3.6 T (30% of nominal) and 14.6 T (120% of nominal). The ratio between maximum and minimum gradient is about 5

6 4.2. This would provide enough variation in gradient for this magnet design to be usable over approximately 60% of the DBD line. Results Results from Opera-3D simulations are summarised in Table 2. Table 2. Summary of simulation results. Parameter Maximum Minimum Units Integrated gradient T Relative to DBD nominal 120% 30% Central gradient T/m Vertical force on moving section kn PM working point B = T H = 273 ka/m BH = 259 kj/m 3 Integrated gradient quality ±0.1% Good gradient region ±12.0 mm Maximum flux density in pole T The gradient profile is shown in Figure 4, along with the force profile and the change in relative gradient for a 10 µm movement. The relative change in gradient stays within the specified 5x10-4 value for the entire range. Figure 4. Variation of (a) force, (b) integrated gradient, and (c) relative change in gradient for a 10 µm movement, all plotted as a function of stroke (movement of the moving part of the magnet). Mechanical Design For the high-strength design, a fully engineered design has now been produced (Figure 5). Both of the moving sections (top and bottom) are controlled by a single motor and gearbox, reducing the cost and complexity of the design. The precision of the motion system will be 10 µm, ensuring that the magnet strength can be precisely and reproducibly set. 6

7 Figure 5. A CAD model of the high-strength quadrupole. The top and bottom moving sections, including the PMs, are attached to a ballscrew on each side. These are turned by a motor at the top via two right-angle and one T-style gearboxes. The position is measured using encoders (in the prototype) and can be controlled to within 10 µm. Sandwich plate Following discussions with a PM manufacturer, it became obvious that while the adhesive holding the PM to the steel (e.g. Araldite 2010 epoxy resin) would be strong enough, the PM itself could break under the tension induced by moving the magnet away from its nominal configuration. To alleviate this effect, a 2mm thick plate of steel was introduced underneath the PM. Steel straps (not shown in Figure 3) wrap around the moving part of the quadrupole, holding the PM in place against the tensile force. To accommodate the straps, there is a 0.5mm gap between the sandwich plate and the pole. This has the detrimental effect of slightly reducing the maximum gradient, but the beneficial effect of reducing the maximum force experienced by the moving section. A prototype is under construction at Daresbury Laboratory; procurement of the materials for this prototype began at the end of 2011 and the magnet is expected to be completed for mid

8 Future Work The high-strength prototype will be magnetically measured at STFC s magnet measurement laboratory and at CERN, and will undergo rigorous testing to ensure that the motion system performs as expected, and that the minimum and maximum gradient correspond to the values predicted by the magnetic modelling. Some tests will be carried out to verify the effect of various mechanical components on the field strength and quality. Further magnet measurements with a rotating coil system will later be made at CERN. The low-strength quadrupole magnet will be designed in detail and prototyped in 2012, and measured in the laboratory. Due to the smaller gradient in this magnet, it may be possible to install it at CLIC Test Facility 3 (CTF3) in the CLIC Experimental Area (CLEX), since CTF3 will operate at a much lower energy than CLIC itself. Conclusion A new type of quadrupole has been designed for the CLIC Drive Beam Decelerator, using movable permanent magnet blocks to adjust the gradient. The highest gradients achievable are comparable to those from a conventional electromagnetic quadrupole, and a wide adjustment range is possible with excellent field quality in the centre. Two different designs have been produced for the CLIC study; a prototype of the high-strength version will be ready in This type of adjustable PM-based design could easily be adapted for other multipole magnets and could be a useful technology for other accelerators in the future, especially those where the power consumption should be kept to a minimum. Acknowledgment The Magnetics and Radiation Sources group at Daresbury, in particular Jim Clarke and Neil Marks, contributed extensively to the magnetic design of these quadrupoles. The mechanical design was the responsibility of Norbert Collomb from STFC s Technology Department, and James Richmond, a student at the University of Liverpool s School of Engineering. All of this work was carried out in collaboration with CERN, with particular contributions from Michele Modena, Erik Adli, Alexey Vorozhtsov, Alexandre Samochkine, Guido Sterbini, and Antonio Bartalesi. References [1] Braun et al, CLIC 2008 Parameters, CLIC Note 764. [2] B. J. A. Shepherd, J.A. Clarke, and N. Marks, Improved Multipole Magnet, UK Patent Application No [3] R. Barlow et al, EMMA The world s first non-scaling FFAG, Nucl. Instr. Meth., Volume 624, Issue 1, 1 December 2010, pp. 1-19, doi: /j.nima [4] Vector Fields Opera software package, from Cobham Technical Services: [5] J. K. Cobb and R. A. Early, Some Initial Results from the New SLAC Permeameter, SLAC-PUB-5168 (1990). [6] Vacuumschmelze GmbH, 8