![Arxiv:2107.03965V1 [Physics.Acc-Ph] 8 Jul 2021 of 100 Mm in Air (See Fig.1 Right) We Need I = 187500 A](http://data.docslib.org/img/a7a3c26cf0e8c396d33291479a90287f-1.webp)
Proceedings of the CERN–Accelerator–School course on Introduction to Accelerator Physics
Warm Magnets
Gijs de Rijk CERN, Geneva, Switzerland
Abstract Warm magnets are magnets that function in normal ambient temperature con- ditions. These types are mostly using a soft steel yoke for field amplification and either Copper or Aluminium coils or permanent magnets to generate the field. Magnets powered with such normal-conducting coils are often also called classical, iron dominated or resistive magnets. Since decades these types of magnets are the workhorse for most linear and circular accelerators and beam transfer lines.
1 Introduction: magnetic fields and basic magnets 1.1 Magnetostatics A magnet at constant field can be described by the Maxwell equations with the time dependent terms set to zero; the case of magnetostatics. Let’s have a closer look at the three equations that describe magnetostatics (in the differential form): Gauss law of magnetism: div−→B = 0 (always holds); Ampère’s law with no time dependencies: rot−→H = −→J (magnetostatics); Relation between the magnetic field −→H and the flux density −→B : −→B = µ0µr−→H (linear materials).
1.2 Iron-dominated magnets
Figure 1: left: infinite wire with circular field line; right: two infinite wires with opposite currents.
From Ampère’s law with no time dependencies (integral form), H −→H −→dl = I we can derive the c · µ0I field generated by a current carrying line (see Fig.1 left): −→B = 2πr ϕˆ with ϕˆ the direction vector as the tangent of a circle. · If we want to make a B = 1.5 T magnet with just two infinitely long thin wires placed at a distance arXiv:2107.03965v1 [physics.acc-ph] 8 Jul 2021 of 100 mm in air (see Fig.1 right) we need I = 187500 A. We see that to reach reasonable fields ( 1 T) we need large currents, moreover in such configurations the field quality will be poor. ≥ Let us consider an iron-dominated resistive magnet with a C-shaped yoke as shown in Fig.2; H Ampère’s law then becomes, with the integration path of the equation on the dotted line: c −→H−→dl = NI. When we take the average of the fields in the iron and in the pole gap, it follows that
B B N I = Hiron liron + Hair gap lair gap thus N I = liron + lair gap. (1) · · · · µ0µr · µ0 ·
Available online at https://cas.web.cern.ch/previous-schools 1 µr"as"func)on"of"B"for"low"carbon"magnet"steel" Yoke (Magne)l"BC)" 9000 8000 coil 7000 6000 µr 5000 4000 3000 2000 1000 0 0 0.5 1 1.5 2 B (T)
1 Figure 2: left: Iron-dominated magnet: the dotted line is the path-integral loop; each coil carries a current of 2 NI; right: Permeability as function of flux density for a standard steel for laminated yokes.
As µr µ0 in the iron, we get B N I = lair gap. (2) · µ0 · For a magnet with a flux density of 1.5 T in a 50 mm pole gap we will need NI = 59 683 A. This can be done with a coil of 2 30 turns, with I = 994 A. This implies a current density of 5 A/mm2 in a Cu conductor with a surface× of 14 14 mm2. Copper coils with air cooling can run with current densities of a few A/mm2, while hollow water× cooled copper conductors can be used up to a few tens of A/mm2. The permeability of the iron depends on the flux density and the best types of magnetic steel will be fully saturated close to 2 T. In Fig.2 the relative permeability µr of a standard steel type used for laminated magnet yokes is shown. When the value of µr approaches unity the approximation to derive (2) is not valid any more and we can see that the required current to reach a field B > 2 T will steeply increase. Iron-dominated magnets with B > 2 T very quickly become unpractical or impossible to build as the required current density in the coil exceeds the cooling possibilities. In Fig.3 we compare the field of a magnet with a vertical gap of 50 mm in the cases with and without an iron yoke.
8
7
6
5
4 B [T]
3
2 with iron
1 without iron iron, no saturation 0 0 100 200 300 400 500 600 700 800 900 1000 NI [kA]
Figure 3: The field of a magnet as function of the current for the cases with and without an iron yoke.
1.3 Magnetic field quality: Multipole description The beam in a particle accelerator is confined in a beam pipe around which the bending and focusing magnets have to be arranged. From beam dynamics calculations one can derive that the field quality is very demanding; for the dipole magnets that provide the bending of the beam, the typical required field
2 ∆Bz −4 homogeneity is: |B| 10 . The field quality in accelerator magnets is expressed and measured in a multipole expansion: ≤ ∞ n−1 −4 X x + iy By + iBx = 10 B1 (bn + ian) with bn, an few units. (3) R ≤ n=1 ref
With: z = x+iy; Bx and By the flux density components in the x and y direction; Rref the radius of the reference circle; B1 the dipole field component on the reference circle; bn and an the normal and skew nth multipole component. −4 In a circular accelerator, where the beam makes multiple passes, a typical demand is an, bn 10 . ≤
1.4 Magnetic length In the longitudinal dimension the typical shape of a magnetic field can be seen in Fig.4. We can define
lmag
Figure 4: Shape of the magnetic field in the longitudinal dimension (named y in this plot).
R +∞ the magnetic length lmag from: lmagB0 = −∞ B(z)dz with B0 the field at the centre of the magnet z = 0. Typically one can estimate that for warm dipole magnets Lmag = Lyoke + d with d the pole distance and for warm quadrupole magnets Lmag = Lyoke + r with r the inscribed radius between the poles.
1.5 Magnets in an accelerator: powering circuits We have to consider a magnet or a string of magnets in an accelerator as part of a powering circuit to be able to design the magnet with current and voltage values that are feasible for the magnet, the cabling and the power converter. A schematic drawing of a powering circuit can be found in Fig.5. We consider; a typical B field stability requirement in time in the order of 10−5 to 10−4; a current rise time during the
Cu (or Al) B (flux density) cables or busbars
beam power Convertor (current source) Vacuum chamber
Cu or Al coil Steel yoke
Figure 5: Typical powering circuit for accelerator magnets.
3 acceleration cycle of 1 s ; magnet resistance R in the order of 20 mΩ to 60 mΩ; magnet inductance L in the order of 20 mH to 200 mH. For the cables connecting the power converter to the magnet we take a length of 200 m with a Cu cross section of 250 mm2 for a current of 500 A without cooling. With a specific resistance of Cu of 17 nΩ m we get a resistance for the cables R = 13 mΩ. The power converter has to supply a current up to 500 ·A with a stability of a few ppm and a voltage, for the resistive part of 40 V, and the inductive part (additional to the resistive part during ramp-up) of 100 V.
1.6 Magnetic types In Fig.6 the magnets types are shown according to the main field pole they supply up to n = 4 for both normal and skew poles. One can also see the flux line patterns in a C type dipole a quadrupole and a sextupole. Magnets are designed according to certain symmetries with mechanical errors also appearing with these symmetries. As a result, certain multipole harmonics can occur, the so-called "allowed har- monics" (see Table1). Non-symmetric designs and non-symmetric fabrication errors give rise to other, "non-allowed", harmonics.
Table 1: Allowed harmonics for iron dominated magnets.
Magnet type Allowed harmonics bn n = 1 Dipole n = 3, 5, 7, ... n = 2 Quadrupole n = 6, 10, 14, ... n = 3 Sextupole n = 9, 15, 21, ... n = 4 Octupole n = 12, 20, 28, ...
fluxlines in magnets
Dipole Quadrupole sextupole GdR 2019, warm 2019,magnets, warm - 0 0.8 1.6 0 0.8 1.6 0 0.5 1.0 T T T T T T Sept -
Figure 6: Top: Magnets11 , types according to the main field pole they supply up to n = 4 for both normal and skew Tatry poles. Bottom: typical- flux line patterns in dipole, quadrupole and sextupole magnets. Vysoke
CAS CAS 14
4 2 Practical magnet design and manufacturing To design and manufacture a magnet, we can identify the following steps:
1. Specification; 2. Conceptual design; 3. Detailed design; (a) Yoke: yoke size, pole shape, FE model optimisation; (b) Coils, cross section and cooling; (c) Raw materials choice; (d) Yoke ends, coil ends; 4. Yoke manufacturing, tolerances, alignment, structure; 5. Coil manufacturing, insulation, impregnation; 6. Magnetic field measurements.
In the following section we will go into more detail for each of these steps.
2.1 Specification Before designing a magnet we have to get a precise specification from the accelerator designers. The specification should comprise:
– Field or gradient. For a dipole the minimum and maximum flux density B(T) should be specified. For a quadrupole 2 this is the gradient G(T/m). For a sextupole this will be G3(T/m ). – Magnet type. For dipoles one has to choose between C-type, H-type, window, etc.. Further one has to know whether it will be a DC magnet, slow ramped or fast ramped (AC) magnet. – Aperture. For dipole magnets one should define a rectangular "good field region" which then defines the air-gap height and width between the poles. Quadrupoles and other higher order magnets will have a circular "good field region" that determines the inscribed circle between the pole. – Magnetic length and physical length. We need to know what the required magnetic length is and the allowed physical length. The coil will always stick out of the yoke and, depending on the coil end type selected, will take some space. – Current range. The current range of the power converter needs to be specified and with it, the available voltage. – Field quality. The field quality in the "good field region" for a dipole ∆B and for a quadrupole ∆G need to be B0 G0 defined. – Cooling type. Available cooling types are: air cooling, for which cooling surface and air flow rate are important, and water cooling, for which we need to know the available water pressure, allowed pressure drop and flow rate. – Jacks and alignment features. The supporting systems of the magnet (jacks) have to be specified and how the alignment will be done (alignment targets).
5 – Vacuum chamber. The vacuum chamber and potential bake-out system will take space and need to be known at the beginning of the design.
These specification items will probably not be given as a complete list but will need negotiation with the various experts concerned and will probably be a compromise.
2.2 Conceptual design I will give as an example a first step conceptual design of a C-type dipole magnet.
lairgap·B – With Equation (2) we can get N I from B and lairgap : N I = . · · µ0 – We get from N I and the Imax from the power converter the number of turns in the coil. · – We can then decide on the conductor cross section via the current density we will be using: Jcoil = 2 2 5 A/mm for a water cooled coil or Jcoil = 1 A/mm for an air-cooled coil. – We can now define the coil cavity in the yoke. We need to allow for an insulation thickness of 0.5 mm around each conductor and a ground insulation around the coil of 1 mm. – With this we can now draw a first cross section of the magnet by estimating the required yoke thickness using the steel saturation flux density B : W = W B . In general we will have sat yoke pole Bsat 1.5 T < Bsat < 2.0 T. – Given the available longitudinal space we can decide on the geometry of the coil ends: racetrack or bedstead. – We now have the rough magnet cross section and the outside envelope as shown in Fig.7.
Yoke Wyoke
Wpole 1 NI coil 2 Wyoke B
1 NI 2
Figure 7: Left: Dipole cross section design, mid-left: coil cross section, mid-right: end view, right: top view.
2.3 Detailed design 2.3.1 Yoke The ideal poles for iron-dominated magnets are lines of constant scalar potential. For the first three orders they are:
– Dipole: y = h/2, a straight line, ± – Quadrupole: 2xy = r2, a hyperbola, ± – Sextupole: 3x2y y3 = r3. − ± To optimise the field homogeneity, small bumps, "shims", are added on the edges of the poles. I will give an example for a dipole and for a quadrupole of this pole shimming.
6 Dipole: In Fig.8 we can see a yoke lamination of the dipole of the LEP accelerator at CERN. On the edges of the poles we can see the pole shims.
Figure 8: C shaped yoke lamination from the (now dismounted) LEP accelerator at CERN.
Quadrupole: In the case of a quadrupole, very often also a small flat part that functions as alignment feature is added to the laminations to be able to assure a precise pole distance. In the left top image of Fig.9 we can see this schematically indicated. In the right top image we can see the details of a pole with all features included. To actually measure the pole distance in a magnet purpose-build instrumentation can be used. In the figure we can see such a pole distance measurement device and the resulting measurement.
Figure 9: Quadrupole pole distance measurement.
7 To design the detailed shape of the yoke, we need to use an electromagnetic (EM) Finite Element (FE) model. Using EM FE models, we can optimise the field quality by adjusting the pole shape and minimising high saturation zones in the yoke. We can then also minimise the total amount of steel to be used to reduce the magnet weight and cost. Finally the EM FE models are used to precisely predict the R flux density magnitude (Bcentre, Bdl, Lmag) and the quality (bn and an). These values are needed for the beam dynamics models in the accelerator design. Some EM FE codes employed are:
– "Opera" from Cobham: 2D and 3D EM FE commercial software (see: http://operafea.com/ ), – "Poisson" 2D EM FE open software , now distributed by LANL-LAACG (see: http://laacg.lanl.gov/laacg/services/download_sf.phtml ), – "ROXIE" 2D and 3D EM BEM-FEM software, licensed by CERN (see: https://espace.cern.ch/roxie/default.aspx ), – FE"ANSYS models: Maxwell" 2Dsteel and 3D curves commercial software (see: http://www.ansys.com/Products/Electronics/ANSYS-Maxwell ).
The EM FE models require as input a description of the B(H) relation of the yoke material. To get the correct final field values, we have to use a measured, and smoothed, curve of the material used for the yoke. In Fig. 10, a number of examples can be found of B(H) curves for magnetic steel types. Fig. 11 shows an example of a ROXIE model for a double-aperture quadrupole (MQW) in the LHC at CERN.
2.5 End slope is µ0 GdR
2
1.5 B(T) 2019, warm magnets,2019, warm
- Magnetil LAF 1.5 mm 1 Iron AC37
Sept St-Petersburg 0.75 mm - ISR 1.5 mm
, 11 , steel Ru 21848 slope is µrµ0 Si 3.25% 0.5
Tatry Magnetil LAC 1.5 mm -
Vysoke 0 0 5000 10000 15000 20000 25000 30000 35000 H (/m)
CAS CAS 1
Figure 10: B H curves for various types of magnetic steel. −
2.3.2 Coils The straight part of the coils are of rather straightforward geometry but the end of the coils will need some attention in the design. Several types of coil ends are available but in practice they are variations of the two main types: racetrack and bedstead. In Fig. 12 the two types for dipoles and quadrupoles are illustrated.
8 Figure 11: ROXIE EM model of the LHC MQW double aperture quadrupole. Top-left: cross section, top-right: mesh, bottom-left: mesh on poles, bottom-right: calculated nominal field (G = 35 T/m at I = 710 A).
Figure 12: Top-left: dipole racetrack coil end, top-right: dipole bedstead coil end, bottom-left: quadrupole racetrack coil end, bottom-right: quadrupole bedstead coil end.
9 There are essentially two types of coils: air cooled and water cooled. In the case of air cooling, the conductor is an insulated copper wire and the heat is passing by conduction to the outer surface of the coil where the air takes the heat away. It is generally applied to rather small coils for corrector magnets. Water cooled coils are applied for both small and large magnets working at higher current density. The conductors are hollow copper or aluminium bars surrounded by insulation with the water circulating in the hole where the heat is removed locally. Practically all accelerator magnet coils are impregnated with epoxy to provide a good electrical insula- tion. Depending on the radiation load by beam residues, this epoxy might need to be radiation resistant. For magnets with extremely high radiation loads, other solution than epoxy have to be employed. Already in the 1970s completely mineral solutions were employed for magnets close to the SPS primary targets on quadrupoles featuring concrete as insulation material.
The current in the resistive material of the coil will generate power that will heat up the coil. Let us first look at the power generated in the coil. At a constant current (DC case), we can calculate the power from the length of conductor in a coil N Lturns, the conductor cross section σ, and the specific resistivity of the material ρ: · P ρ = I2 (4) l σ with: −8 −8 ρCu = 1.72(1 + 0.0039(T 20))10 Ωm or ρAl = 2.65(1 + 0.0039(T 20))10 Ωm. (5) − − In the case of pulsed magnets we take the average of I2 for the duty cycle. The coil will have most of the power losses although also the yoke will experience losses due to hysteresis and eddy currents. The power losses due to the hysteresis in the yoke can be taken as (Steinmetz law up to 1.5 T): 1.6 P [W/kg] = ηfB with η = 0.01 to 0.1, ηSisteel 0.02, f[Hz],B[T] (6) ≈ Eddy currents in the yoke will cause power losses as: f P [W/kg] = 0.05(d B )2 (7) lam 10 av with dlam the yoke lamination thickness in mm and Bav the average flux density at top field in the yoke in T.
The parameters of water cooling that are to be defined are: dcooling: the diameter of the cooling hole in the hollow conductor; Pwater [bar]; ∆P [bar] : the pressure drop over the coil; Q [l/min]: the flow rate. We then:
◦ ◦ – Choose the desired ∆T .( 20 C or 30 C depending on the Tcoolingwater). – With the heat capacity of the water (4.186 kJ/kg◦C) we now know the required water flow rate Q[l/min]. – The cooling water needs to be in a moderate turbulent regime to optimise the heat transfer (laminar flow has zero flow speed on the wall) : Reynolds > 2000. d v Re = · 140d[mm]v with (8) ν ≈ −6 2 ◦ v water velocity [m/s], ν dynamic viscosity 0.7x10 m /s for Twater 40 C (9) ≈ ≈ – Get a good approximation for the pressure drop in smooth pipes from the Blasius law: Q[l / min]1.75 ∆P [bar] = 60l[m] (10) d[mm]4.75
10 2.3.3 Raw material choice Magnet yokes are nearly always laminated to reduce the eddy currents during current ramping. There- fore, the yoke raw material has be thin steel sheets from which the laminations can be cut or punched. The magnetic properties of the steel depend on chemical composition, temperature treatments and the mechanical history. Important parameters for the magnetic performance of the steel are the coercive field Hc and the saturation flux density Bsat. Low carbon steel (with a carbon content < 0.006%) is best suited for higher fields B > 1 T. The coercive field Hc has an impact on the remnant field at low excitation cur- rents, which is important for magnets that are also used at very low field for beam injection. Typically the specification is Hc < 80 A/m, and for magnets ranging down into low fields B < 0.05 T this becomes Hc < 20 A/m. Laminations can be coated with an inorganic (oxidation, phosphating, Carlite) or organic (epoxy) layer to increase the inter-lamination resistance. These coated steel sheets can be supplied by the steel manufacturers. In Tables2 and3 the properties of two types of steel can can be found.
Table 2: Specification for Low carbon magnetic steel, oxide coated, LHC separation dipoles (high field), 1.5 mm thick. H(A/m) B(T) 40 0.20 60 0.50 120 0.95 500 1.4 1200 1.5 2500 1.62 5000 1.71 10000 1.81 24000 2.00
Table 3: Specification for Silicon magnetic steel, epoxy coated, LHC transfer line corrector dipoles (low field), 0.5 mm thick. H(A/m) B(T) 100 0.07 300 1.05 500 1.35 1000 1.5 2500 1.62 5000 1.72 10000 1.82
Low resistance copper conductor for the coils can be procured from several suppliers in many formats. For water cooled magnets one can buy hollow Cu conductor in tens of meters of piece-length. Both copper and and aluminium enamelled wire, which is coated with a very thin layer of insulation, can be directly ordered from suppliers. This insulation is often a tough polymer film.
2.3.4 Yoke manufacturing The yoke of most warm magnets needs to be split into yoke-stacks so that the finished coils can be put around the poles. For an "H-shaped" vertical field dipole, we need to split the yoke at the horizontal plane in 2 yoke-stacks or half yokes. Quadrupole yokes habitually need to be split into 4 yoke-stacks.
11 Corrector dipole in TI2 and TI8 LHC injection lines A "C-shaped" dipole yoke does not have to be split if the pole distance is sufficiently large to allow for slidingMagnet in the coils. with In glued general, laminated yoke-stacks are yokes either glued,assembled using epoxy with coated bolts. laminations, welded, with the support of full length structural plates on the outside or axially compressed by tie-rods (Fig. 13). Often a combination of 2 or 3 of these are used. To stabilise the yoke-stack, thick steel end plates are used that can be either a glued stack of laminations or a solid steel plate. The end plates can have a chamfer on the pole to shape the end field.
Main parameters Name MCIA V Type Vertical correcting dipole Nominal peak field [T] 0.26
Imax [A] 3.5 N. Of turns 1014 Résistance [ 13.9 Yoke lenght [mm] 450 Gap [mm] 32.5 Total weight [kg] 300
Figure 13: left: glued yoke of a small corrector magnet; middle: welded yoke stack of a 3 m long quadrupole magnet;Introduction right: to accelerator tie-rod physics in the pole of a quadrupoleVarna, 19 September, yoke 1 October stack. 2010 Davide Tommasini : Magnets (warm)
The productions steps of a magnet yoke-stack are (Fig. 14).
– Lamination cutting or punching. – End plate machining. – Yoke-stack stacking. – Yoke stack welding or gluing (or both). – Mount tie rods to the yoke stack. – Geometry verification measurements (straightness, flatness, etc). – Cleaning and painting.
Figure 14: left: yoke-stack stacking; middle: yoke-stack welding; right: finished yoke stack before painting.
12 2.3.5 Coil manufacturing The steps for fabricating a coil with internal water cooling are (Fig. 15):
– Hollow Cu conductor. The conductor should be delivered in a format that will be directly mountable on the winding machine without further manipulations. As the Cu has to be soft ("dead soft"), it has to undergo a thermal treatment at the supplier ( 400◦C). Copper will harden when cold worked, so between the thermal treatment and the winding≈ the copper should not be bend to allow for a smooth, low force winding. – Glass fibre insulation. The individual conductor has to be wrapped half-lapped with a glass-fibre tape with a thickness of 0.25 mm giving a total thickness of 0.50 mm to get a full coverage with no holes and assured spacing between the turns. The wrapping can be done on the winding machine just in front of the coil (for thin conductors requiring a small force to wind) or after winding by spreading out the turns and then wrapping the conductor by hand (for thicker conductors requiring a large winding force that then can damage the glass-fibre). – Winding the conductor into a coil. This operation is done on special winding machines. Depending on the type of coil, such machines are commercially available or are purpose built. – Mount connections. Hard solder the electrical connecting flags and the water connections to the coil leads. – Apply ground insulation. The coil as a whole has to be wrapped half-lapped with a glass-fibre tape with a thickness of 0.50 mm giving a total thickness of 1.00 mm. – Mount the coil in an impregnation mould and impregnate with epoxy. – Electrical tests. The electrical turn-to-turn insulation should be verified. Large defects can be found by measuring the electrical resistance with a precision better than the resistance of one turn. To detect smaller insulation inter-turn defects a capacitive discharge method can be used. The ground insulation of the coil should be tested by immersion of the coil into water (with the connections sticking out) for several hours after which the resistance between the bath and the conductor should be measured. Depending of the required operational maximum voltage of the magnet in the accelerator (turn-to- turn in the coil, and coil to ground) electrical engineering standards will prescribe the test voltages (e.g. a factor 2 + 500 V) that the coil has to be able to withstand.
Figure 15: left: coil winding; middle: coil insulating; right: coil electrical testing in water.
13 2.3.6 Magnet assembly When the yoke-stacks and all the coils are finished the magnet can be assembled. Typical steps for the magnet assembly are (Fig. 16): – Mounting a coil on each pole on the yoke stacks. – Assemble the yoke-stacks. Sometimes an intermediated step is done by first making half magnets and then joining the two to- gether. Dedicated lifting girders are needed for these operations. For the magnet mounting, special tooling is needed to assure the proper alignment (e.g. assembly tables or jigs). – Connection of the coils. Once the yoke-stacks with coils are assembled, the electrical connections have to be made. This can be done by hard-soldering copper bars between the connection lugs. The water connections are then also to be made. Water cooling can be done for the coils in parallel or in series depending on the required cooling conditions. – Add instrumentation. On water cooled magnets, temperature sensors or switches are mounted on each coil to be able to switch the current off in case of over-heating. – Dimensional measurements. The straitness of the magnet and the pole distance as function of the longitudinal position have to be measured.
Figure 16: Left: yoke stack on a precision assembly table; right:half magnet handling tooling.
2.3.7 Magnetic field measurements To measure the magnetic field with sufficient precision to be relevant for the required field quality for circular accelerators, absolute field and field harmonics have to be measured with a precision of at least 10−4. Several magnetic measurements techniques can be applied to achieve this, e.g.: – Rotating coils. Used to measure multipoles and integrated field or gradient in all magnets. – Stretched wire. To determine the magnetic centre and integrated gradient for n > 1 magnets. – Hall probes. To make a field map. – Fixed pickup coils. Measure the field on a current ramp. Magnetic measurements are highly specialised technology and separate CAS schools are offered on this subject. In Fig. 17 we can find some illustrations of measurements with the rotating coil technique.
14 The radial coil Mole assembly 1
Flux in a Radial Coil: R2 Y F(q ) = NL Bq (r,q )dr òR1
B 0 + Mole Assembly q =q wt Side View R2 Gravity Pneumatic Incremental Slip Rings Harmonic Sensor Break Encoder Coils
Induced Voltage: R1 X 2016, warm magnets, GdR magnets, warm 2016, æ ¶F ö - Oct V (t) = -ç ÷ - è ¶t ø Integrated Voltage: Mole Assembly Top View CAS Budapest, 7 Budapest, CAS Cross section of a radial coil òV(t)dt º -F(q )
12 March 2003 Magnetic Measurements on MQW O. Hans 8
Figure 17: Top-left: Schematic drawing of a radial rotating coil in a magnetic field; top-right: Drawing of a ro- tating coil mole; bottom-left: photo of the components of a rotating coil mole; bottom-right: A rotating coil mole measurement in the MQW magnet.
IEEETRANSACTIONSONAPPLIEDSUPERCONDUCTIVITY,VOL.26,NO.4,JUNE2016 4101404
3 Permanent and hybridUse magnets of Permanent Magnets in Permanent magnetic materialsMultiple can be used Projectsin various layouts at CERN to produce accelerator magnets. In Fig. 18 a few examples ofPermanent layouts withMagnets permanentP. magnets A.: LINAC Thonet can be4 found. Permanent magnet blocks can be
arranged to provide anyDTL multipolartank field. Abstract—In the past, the use of permanent magnets in the CERN Accelerator Complex was marginal. In the recent years, specific attention is being devoted to fit, where possible, permanent magnets in new projects either in case of consolidation or for the upgrade of existing accelerators and experimental areas. This paper describes and analyzes the design and manufacture of some of these projects, particularly the permanent-magnet quadrupoles with adjustable gradient installed in Linac4, a permanent sex- tupole magnet for the ASACUSA experiment, and a large dipole for the nTof experimental area. Index Terms—ASACUSA, linac4, nTof, permanent magnet.
Pictured : Cell-Coupled Drift TubeI. Linac INTRODUCTION module. • Permanent ERMANENTmagnet because magnets are of more space and more between used at CERN, DTL tanks • Sm Co Ppermanentproviding advantages magnets over conventional magnets in terms 2 17of reliability, manufacturing and operational costs. Indeed, • Integratedavoiding gradient any electrical of 1.3 break to 1.6down Tesla and water leaks which are the main failure risks with resistive magnets, permanent • II. Linac4 QUADRUPOLE 15 magnetsmagnet solutions allow to reduce down time of the machines • Magnet andlength exposure 0.100 of personnel m in radiation areas. They are cost A. Description • Field quality/amplitudeefficient as no power convertor tuningnor blocks electrical and demineralized water network are required. In addition, manufacture cost of the Linac4 is a normal conducting linear accelerator operating accelerator magnet itself is often cheaper than for a resistive at the frequency of 352 MHz. It is composed of different magnet. accelerating sections, to deliver a H− beam at 160 MeV to the In the majority of CERN permanent magnet designs, PS Booster. One of these sections, the CCDTL (Cell Coupled Drift Tube Linac) accelerating the beam from 50 to 100 MeV Samarium Cobalt material type Sm2Co17 is chosen. This ma- is composed of 21 tanks of 3 cells each for a total length of Introduction toterial accelerator has high physics magnetic charact eristics Varna, 19 in September, term of 1 remanence October 2010 Davide Tommasini : Magnets (warm)25 and coercivity and contrarily to the NdFeB, good resistance to 25 meters. Fourteen quadrupoles, all with a different gradient, Figure 18:ionizingTop-left: radiation flux [1].pattern This material for presents a permanent good temperature magnetprovide dipole; beam top-right: focusing between Permanent the tanks. The magnet available dipole space for Linac4 at stability with a coefficient of magnetization dependence on being limited, a permanent magnet solution allowing a compact CERN; bottom-left:temperature of Halbach0.035%/K atarray room temperature.for a sextupole; In most cases bottom-right:design was chosen. Permanent magnet quadrupole with a steel yoke. the temperature− stability inside the accelerator or experimental The considered design [3] is an iron free PMQ (Permanent- area is good enough to not have to correct this effect. magnet quadrupole) based on a Halbach array [4], [5]. It is Accelerator permanent magnets were installed in Linac4 composed (see Fig. 1) by a non-magnetic austenitic steel yoke and in experimental areas. The paper presents the design and and 8 permanent magnet blocks. The yoke is made in a single manufacture experience of the quadrupoles with adjustable piece and acts as an accurate frame to maintain the permanent gradient installed in Linac4 [2], a permanent sextupole magnet magnet15 blocks in their position. All the blocks are oriented in a for the ASACUSA experiment and a large dipole for the nTof different direction in order to generate the magnetic flux lines experimental area. forming a quadrupolar field. The field is adjusted by moving the permanent magnet blocks to a certain distance from the center of the quadrupole. The exact radial position of the blocks is settled with non- magnetic shims inserted between the austenitic steel yoke and Manuscript received October 9, 2015; accepted January 29, 2016. Date of publication February 19, 2016; date of current version March 14, 2016. the permanent magnet blocks. A range of radial displacement The author is with the Organisation Européenne pour la Recherche Nucléaire of 6 mm of the permanent magnetblockspermitstoadjust (CERN), Geneva 1211, Switzerland (e-mail: [email protected]). the integrated gradient from 1.1 to 1.6 Tesla. The quadrupole Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. polarity is changed by rotating the PMQ by 180◦ with respect Digital Object Identifier 10.1109/TASC.2016.2532361 to the vertical axis.
1051-8223 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. 1AP2-1 2
vertical one. Given this constraint, an “8” or “two leaves” The return yoke consists of two C-shape pieces. Low structure of the magnet yoke was chosen. carbon steel AISI 1010 was chosen as a return yoke material for reason of cost, availability and ease of machining. The coils are made of square enamelled solid copper wire with the cross-section of 4 mm × 4 mm. Each coil consists of 364 turns (18 layers of 18 turns). To avoid demagnetization of the permanent magnet materials by the opposite field generated by the coils, operation of the magnet is limited to 15.4 A, that corresponds to 5000 A of current-turns product (NI). In operation at maximum current, a current density of 1.0 A/mm2 is assumed. This avoids the use of an active water cooling system and allows minimizing the vibration of the magnet. Varying the current from zero to 15.4 A corresponds to varying the field gradient between 100 T/m to 530 T/m for Sm2Co17 and 170 T/m to 590 T/m for Nd2Fe14B that permits a wide tunability of the magnet. B. Field Computation Fig. 2. Magnet mechanical layout. Magnetic field calculations of the quadrupole were The main difference in the magnet design as compared to a performed with the Opera-2D/ST and Opera-3D/TOSCA conventional quadrupole is the presence of rare earth programs [3]. The basic parameter study was performed with permanent magnet material placed between the iron poles. the 2D code. The wedge angle of the pole and permanent This material is oriented to inject magnetic flux into or blocks and yoke cross-section were optimized by this subtract from each iron pole, as it shown in Fig. 3, where the calculation in order to achieve the required field gradient value arrows indicate the direction of the easy axis of the permanent and field quality. magnet material. It should be noted that the permanent Due to symmetry only 1/4 of the magnet geometry was magnets do not directly contribute to the field gradient in the modeled (see Fig. 4). The boundary conditions were chosen in a way that the flux lines were perpendicular to the horizontal magnetIn order aperture. to make The tunable magnetic magnets flux due with to permanentthe permanent magnet materials, it is possible to make hybrid magnets is directed to cancel the part of flux produced by the and the vertical middle planes and parallel to the limiting edge solutions.coils which In Fig. does 19 not we contribute can see ato quadrupolethe field gradient designed in the for theof the interaction model. The region coils are of modeled the CLIC as linearsquare region with a collidermagnet (CERN) aperture with [2]. aThis gradient reduces of the530 saturationT/m in effects a aperture in the1AP2 of-1total8 .25 appliedmm diameter current density that is equal tuneable to the in ampere the -turns of the 2 iron pole, making it possible to reach a field gradient about 35 coil divided by the total cross section of the coil. The effect of range 10% 100%. vertical one. Given this constraint, an “8” or “two leaves” The return yoke consists of two C-shape pieces. Low % larger− than it is available with a conventional quadrupole. structurethe of spacingthe magnet betweenyoke was chosen. individual conductorscarbon was steelneglected. AISI 1010 was chosen as a return yoke material for reason of cost, availability and ease of machining. The coils are made of square enamelled solid copper wire with the cross-section of 4 mm × 4 mm. Each coil consists of 364 turns (18 layers of 18 turns). To avoid demagnetization of 1AP2-1 the permanent magnet materials by 4 the opposite field generated by the coils, operation of the magnet is limited to IV. MODEL CONSTRUCTION AND FIRST TEST RESULTS The measured results compare15.4 A,d that with corresponds the calculation to 5000 A ofare current -turns product plotted in Fig. 16. It was shown(NI). In that operation the measured at maximum integrated current, a current density of 1.0 In order to verify the conceptual design, tolerances on 2 field gradient and the integratedA/mm field is assumed harmonics. This avoidsare in the good use of an active water integrated field strength and to study the assembling cooling system and allows minimizing the vibration of the procedure, a prototype model with full-scale cross-section, agreement with the computedmagnet values.. working at nominal conditions, but with much shorter yoke Varying the current from zero to 15.4 A corresponds to varying the field gradient between 100 T/m to 530 T/m for length of 100 mm, has been designed and is currently under Sm2Co17 and 170 T/m to 590 T/m for Nd2Fe14B that permits a construction. The mechanical layout of the prototype is shown wide tunability of the magnet. in Fig. 13. B. Field Computation Fig. Fig.4. 2Magnetic. Magnet mechanical flux distribution layout. in a 1/4th segment of the magnet. Magnetic field calculations of the quadrupole were The main difference in the magnet design as compared to a performed with the Opera-2D/ST and Opera-3D/TOSCA conventionalThe quadrupole simulations is the presencewere ofperformed rare earth programsfor two [3]. Thetypes basic parameterof study was performed with The central iron part of the magnet is made of a single the 2D code. The wedge angle of the pole and permanent permanentpermanent magnet material magnet placed materialbetween the: ironSm poles2Co. 17 with a remanence Br of piece: a ring-like structure links the four poles and the This material is oriented to inject magnetic flux into or blocks and yoke cross-section were optimized by this subtract1.1 from T each and iron coercivity pole, as it shown HcB in of Fig. 820 3, where kA/m the , andcalculation Nd2 Fein order14B towith achieve a the required field gradient value permanent magnet blocks are inserted into the corresponding Fig. 15. OPERA 3D model showingand only fiel centrald quality. part of the magnet assembly. arrowsremanence indicate the direction Br of of 1.28 the easy T andaxis of coercivity the permanent H cB of 990 kA/m. slots between the iron poles. This design enhances the magnet Due to symmetry only 1/4 of the magnet geometry was magnet material.Fig. 5It gives should thebe curvenoted that of fieldthe permanent gradient versus (NI) for both mechanical rigidity, simplifies the assembly procedure andmagnets do not directly contribute to the field gradient in the modeled (see Fig. 4). The boundary conditions were chosen in magnettypes aperture. of The permanent magnetic flux magnet due to themateri permanentals. Sincea way thatthe the permanentflux lines were perpendicular to the horizontal reduces potential assembly errors and the resulting random and the vertical middle planes and parallel to the limiting edge magnetsmagne is directedt material to cancel thepreloads part of flux the produced iron by poles the , the curve becomes multipole errors. A drawback of this ring structure is the shortcoils- which does not contribute to the field gradient in the of the model. The coils are modeled as square region with a asymmetric, allowing much higher fieldtotal gradientapplied current for density the equal to the ampere-turns of the circuiting of some magnetic flux that causes a reduction ofmagnet aperture [2]. This reduces the saturation effects in the iron polepositive, making itcurrent possible tothan reach for a field the gradient negative about current.35 coil divided by the total cross section of the coil. The effect of gradient inside the magnet aperture by approximately 4% with% larger than it is available with a conventional quadrupole. the spacing between individual conductors was neglected. It is shown that at NI=5000 A the magnet core is close to compared to the non-ring configuration. In order to achieve saturation and further increasing the excitation current is not higher gradient values, the central partFig. of 13 the. QD0 magnet prototype is mechanical made layout. efficient. Another reason to limit the maximum current is a Figureof 19:“permendur”,Hybrid quadrupole a CoFe alloy for thecharacterized CLIC linear by collider a very interactionhigh area; top-left: permanent magnet part; top- The hybrid quadrupole prototype needs riskfour permanentof demagnetization of the permanent magnet as right:saturation cross section; level of bottom: 2.35 T. prototype magnet assemblies magnet. of 100 mm length; each consists of four mentioned above.Fig. 16. Measured and Calculated results of the central part of the magnet permanent blocks of 50 mm length glued together. The assembly. permanent magnets blocks made of VACOMAX 225HR (Sm2Co17) and VACODYM 655HR (Nd2Fe14B) were supplied V. CONCLUSION by Vacuumschmelze, Germany [4]. The permendur part of the quadrupoleFig. 3. Centralstructure region ofwas the magnet, Thearrows designindicate theof direction the final of focusingFig. 4. Magnetic quadrupole flux distribution magnet in a 1/4 thfor segment of the magnet. magnetization of the permanent blocks. CLIC and the status of the prototype model were presented. 4 Examples of magnetsmachined over theby wir historye EDM ofto a accelerators tolerance of ± 25 µm. The The simulations were performed for two types of The central iron part of theThe magnet magnet is made design of a singlewas verified by the FEM calculation, selected material for this part was VACOFLUX 50 by permanent magnet material: Sm Co with a remanence B of piece: a ring-like structure links the four poles and the 2 17 r Practical accelerators startedVacuumschmelze in 1929-1930 [5]. with the cyclotron, invented by Ernestwhich showed O. Lawrence that the required at1.1 theT andfield coercivity gradient HcB valueof 820 cankA/m be, and Nd2Fe14B with a permanent magnet blocks are inserted into the corresponding achieved using the permanentremanence magnet B r materialof 1.28 T andNd coercivityFe B and HcB of 990 kA/m. The permanent magnet assemblies wereslot inserteds between between the iron poles.the This design enhances the magnet 2 14 University of California, Berkeley. Cyclotrons developed very fast and machines with single and separ-Fig. 5 gives the curve of field gradient versus (NI) for both iron poles. A specific assembly device tomechanical guide the rigidity permanent, simplifies about the assembly 45 T/m procedure less than and the required value with the Sm2Co17. types of permanent magnet materials. Since the permanent reduces potential assembly errors and the resulting random ated magnets have been built,magnets like into the the sector correspondi cyclotrons;ng slots was see design Fig.ed 20 and. built. However, the Sm2Co17 couldmagne bet materialmore favorable preloads the due iron topoles its, the curve becomes multipole errors. A drawback of this ring structure is the short- The magnetic properties of the permanent magnets blocks as radiation hardness and weakasymmetric,er temperature allowing dependence. much higher The field gradient for the circuiting of some magnetic flux that causes a reduction of positive current than for the negative current. well as the B(H) curve of the VACOFLUXgradient 50 wereinside themeasured magnet aperture final by decision approximately concerning 4% with the permanent magnet material type It is shown that at NI=5000 A the magnet core is close to by the supplier and found consistent withcompare thed toproperties the non-ring as configurationwill depend. In order on finalto achieve lattice design retained for the final focus saturation and further increasing the excitation current is not higher gradient values, the central part of the magnet is made listed in the technical documentation. and on technical reasons. efficient. Another reason to limit the maximum current is a of “permendur”, a CoFe alloy characterized by a very high The production of the 4 coils is under completion at CERN risk of demagnetization of the permanent magnet as saturation level of 2.35 T. and the magnet will be fully assembled in the next months. mentioned above. The magnet was magnetically measured by stretched wire ACKNOWLEDGMENT system in the configuration with only the central part of the structure assembled as shown in Fig. 14. In order to compare The authors gratefully acknowledge E. Solodko and P. A. the measured and calculated results, the OPERA 3D model of Thonet from CERN for the mechanical design of the prototype this partial assembly was also computed (see Fig. 15). magnet and technical support. Also, we would like to thank the personnel of Vacuumschmelze for providing us with quality components and assistance with the assembly of the magnet.
REFERENCES [1] CLIC Conceptual Design Report, CERN, 2011. Available: http://clic- study.org/accelerator/CLIC-ConceptDesignRep.php [2] K. Halbach, ‘Magnet Innovations for Linacs’, Proceedings of “1986 Linear Accelerator Conference”, SLAC-R-303, Linac86-105, TH2-1. [3] OPERA-3D/TOSCA, Cobham Technical Services, Oxford, England. Figure 20: [4] Rare-earth permanent magnets, Vacuumschmelze, Germany. Available: Left: The Berkeley 184" cyclotron at Berkeley; right: The PSI sector cyclotronhttp://www.vacuumschmelze.com/index.php?id=36 of 1974. Fig. 14. The central part of the quadrupole (permendur) with the permanent [5] Soft magnetic cobalt-iron-alloys, Vacuumschmelze, Germany. magnet Sm2Co17 inserts. Available: http://www.vacuumschmelze.com/index.php?id=118&L=2
16 Synchrotrons were invented in the beginning of the 1950-s featuring a string of magnets in a ring shaped accelerator. The Cosmotron at Brookhaven (US) was commissioned in 1953 and had a beam energy of 3.3 GeV. It was a weak focussing machine and thus had a very large aperture of 20 cm x 60 cm (Fig. 21).
Figure 21: Left: The cosmotron at BNL (1953); middle: people sitting in the huge aperture of a magnet for a weak focusing synchrotron; right: schematic drawing of a magnet for a weak focusing synchrotron.
Beam size and thus aperture size could be reduced by close to an order of magnitude by employing strong focusing. The first strong focusing synchrotrons used combined function magnets. An example is the Proton Synchrotron at CERN that was commissioned in 1959. It reached 24.3 GeV at an average flux density of 1.2 T in a combined function magnet (Fig. 22).
Figure 22: Top-left: Combined function magnet for the strong focusing Proton Synchrotron (PS) at CERN; top- right: two types of focusing magnet blocks; bottom: PS magnet pole.
17 The next step in the development of synchrotrons is the strong focusing synchrotrons with separ- ated function magnets. The Super Proton Synchrotron at CERN is a nice example of this technology. In Fig. 23 one can see the SPS magnets. The dipole magnet generates a flux density Bmax = 2.05 T using a 16-turn coil with Imax = 4900 A in a 52 mm high and 92 mm wide aperture and the quadrupole provides a gradient of 20.7 T/m with a coil of 16 turns and a current of 1938 A in an aperture with a 44 mm radius. The machine accelerates protons up to an energy of 450 GeV.
Fig. 37: The two types of bending magnets of the SPS.
Figure 23: The SPS main magnets: top-left: dipole cross-section; top-right: a photograph taken during assembly of the dipoles in the early 1970s; bottom-left: quadrupole cross-section; bottom-right: detail of the SPS quadrupole.
A more recent example of a warm magnet can be found in Fig. 24. The MBW separation dipole is used in the LHC to further separate the beams by 30 mm in the collimation areas; it has a field of 1.42 T in a 52 mm pole gap.
Figure 24: Recent warm magnet for the LHC: left: MBW separation dipole cross section; right: MBW during installation.
18
Fig. 38: Extraction elements of the SPS.
51 New elegant designs have recently been done for synchrotron light sources and small storage rings. In Fig. 25 we can find a dipole and a sextupole of the SOLEIL synchrotron light source that is located near Gif-sur-Yvette south of Paris. We can see as well two dipoles for the ELENA antiproton decelerator ring at CERN. A few words about SR Sextupoles
The production of the 124 SR Sextupoles has been industrially organised by keeping only two types : normal sextupoles and sextupoles with ears Magnets produced and measured by Sigmaphi company Half rings are keeping the poles at the right location. CERN Accelerator School (CAS) – Bruges, Belgium June 16-25, 2009 - Antoine DAËL 24 Courtesy of SIGMAPHI
CERN Accelerator School (CAS) – Bruges, Belgium June16-25, 2009 - Antoine DAËL 41
Figure 25: Top-left: SOLEIL quadrupole; top-right: SOLEIL sextupole; bottom-left: ELENA ring dipole; bottom- right: ELENA TL dipole.
Acknowledgements For this lecture I used material from lectures, seminars and reports from many colleagues. Special thanks goes to: Davide Tommasini, Attilio Milanese, Antoine Dael, Stephan Russenschuck and Thomas Zickler, and to the people who taught me, years ago, all the fine details about magnets.
Bibliography Books: G.E.Fisher, Iron Dominated Magnets, AIP Conf. Proc., 1987 – Volume 153, pp. 1120-1227 J. Tanabe, Iron Dominated Electromagnets, World Scientific, ISBN 978-981-256-381-1, May 2005 P. Campbell, Permanent Magnet Materials and their Application, ISBN-13: 978-0521566889 S. Russenschuck, Field computation for accelerator magnets: analytical and numerical methods for electromagnetic design and optimization, Wiley, 2010, ISBN-13: 978-3527407699 Schools: D. Brandt, et al., CAS Magnets, Bruges, 2009, CERN-2010-004 D. Einfeld, CAS Introduction to Accelerator Physics, Frascati 2008, Magnets (Warm), http://cas.web.cern.ch/schools/frascati-2008 D. Tommasini, CAS Introduction to Accelerator Physics, Varna 2010, Magnets (Warm),
19 http://cas.web.cern.ch/schools/varna-2010 R. Bailey, at al., CAS Power Convertors, Baden 2014, CERN-2015-003 S. Turner, et al., CAS Measurement and Alignment of Accelerator and Detector Magnets, Anacapri 1997, CERN-98-05 Reports: D. Tommasini, Practical definitions and formulae for magnets, CERN,Tech. Rep. EDMS 1162401, 2011
20