Paper presented at 13 th International Conference on TRI-PP-93-99 Magnet Technology, Victoria, Sep 20 - 24 Nov 1993 / A 4.7 Tesla Metre Solenoid for a Partial Siberian Snake

L. Ratner, W. Leonhardt Brookhavcn National Laboratory and A. Otter, L. Ellstrom TRIUMF, 4004 Wctbrook Mall, Vancouver, B.C., Canada V6T tAS

CA9700598

A bitract on each turn and is soon depolarized. The second type of resonance is due to errors in the vertical closed orbit We describe the engineering design of a 4.7 T-m solenoid and here Gy — k is the depolarizing condition. For the magnet which will be installed at the Brookhavcn National AGS the solenoid will be used to cancel the effect of the Laboratory AGS for a partial Siberian Snake Experiment Gy = k resonance, and the reason it is called a partial which is an interlaboratory collaboration. The magnet snake is shown by the following argument. The snake ac­ has an overall length of 2.5 m, a clear bore of 15 cm and tion can be characterised by cos(irV,,) = cos (tG7)cos |, operates at a peak field of 2 T. It is pulsed at 3 second where V,p is the spin tune (number of spin oscillations per intervals with a peak current of 9500 A dc driven from a revolution in the accelerator) and 6 is a spin angle intro­ 150 V power supply. The construction uses conventional duced by the Snake. We can see that if 6 — 180 (100% hollow copper coils but the return flux yokes are made Snake), cos(irV,p) = 0 and V,r = 1/2. Obviously the two from 1/8 inch plates bolted together. On assembly the resonance conditions, Gy = integer, Gy = integer + Vy flux yokes and endplates are clamped tightly to the coil to prevent any movement during the current pulse. The fab ­ cannot be met except for the very special case of Vy = 1/2 which never occurs in an accelerator since the beam rication experience and test data will be presented. The magnet was installed in the summer of 1993. is lost due to this betatron tune resonance. For the AGS we will only use a 5 % Snake (partial Snake) which will I. Introduction preserve the polarization by flipping the spin from 4 to — at each Gy — n resonance. This is adequate for the AGS The solenoid described in this paper is used in the Al­ where the resonance strength < is such that the necessary ternating Gradient synchrotron (AGS) as a device to help condition preserve the polarization of an accelerating beam of po­ larized protons. It has been named a “ Partial Siberian ~ > < is met. Snake". This name, Siberian Snake, has been applied to a configuration of horizontal and vertical dipoles invented i.e., -p- = .025 2* by Durbenev and Kondratenko [1] of the Siberian Institute which by a series of horizontal and vertical bends main­ The 5% or 9* Snake solenoid requires 4.7 T-m at an tain the spin orientation of a polarized beam. These up, accelerator energy of 25 Gev. The longest straight section down and sideways bends give the impression of sinuous at the AGS is 3.05 m long which dictated that the field motion which led to the “Snake " appellation. A solenoid of the solenoid would have to be 2 T. Since the solenoid can be used as a substitute for the dipole configuration in must be pulsed to maintain 9* at all energies it had to some applications. be an air core device which could be pulsed in about 600 The necessity for such a device is due to the fact that msec. The AGS vacuum chamber ’s 15 cm diameter also a vertically polarized beam being accelerated in a circular selects the solenoid inner diameter. Using existing power accelerator will be subject to depolarizing forces due to supplies giving a peak current of 9.6 kA at 150 V then two distinct factors. One, the normal focussing gradients dictated the coil parameters. necessary to keep the beam in orbit have radial compo­ II. Magnet parameters nents that will depolarize the beam when the condition Gy = kP ± Vy holds. Here, G is the anomalous magnetic The magnet parameters are listed in Table 1 and the moment of the proton, 7 is the ratio of the particles energy magnet assembly cross section is shown in Figure 1. to its rest mass, k is an integer, P is the periodicity of the Some of these parameters were specified to meet the accelerator and Vy is the vertical tune. When this con­ Snake physics requirements and some were set by physi­ dition holds, the beam is subject to a depolarizing kick cal constraints and the use of existing power supplies and COIL END PLATE

b)

Fig. 1. a) Snake solenoid longitudinal section; b) cross section showing coil interconnections at input end.

Table 1 Magnet Parameters cabling at the AGS. The resistance of the cables between the power supply and the magnet was included in esti­ Overall strength 4.7 T-m mates of the response times. Maximum longitudinal field 2 .0 T Effective length 2.38 m III. Coil Design Overall length (100 in.) 2.54 m Clear bore 0.159 m The initial coil concept [2,3] was adopted for final de ­ Maximum peak current 9500.0 A sign. The coil is formed with six layers of 30 mm square Maximum RMS current 4000.0 A copper conductor with a cooling hole diameter of 12 mm. Maximum dc voltage 150.0 V Each layer is cooled separately from a common supply. Cycle time 3.0 s Current rise time 0.65 s The number of turns per layer is determined by the Current flattop 0.15 s keystoning of the first layer and the coil current rating Current decay time 0.38 s is determined by the cooling of the outer layer. The water Inductance 6.5 mH connections arc at each end of the solenoid and in order Resistance 11.2 mohm to achieve compact connections and the correct direction Cooling water pressure drop 1.86 MPa (270 psi) of current flow the layers were wound in opposite senses Cooling water temperature rise 40.0°C Max e g. layer #1 clockwise, layer #2 anticlockwise etc. Each

2 layer was wound with a spiral wrap rather than parallel the end plates and the horizontal yokes from opening un­ with an offset at each turn. der the mechanical forces. These rods operate with a max­ The keystoning was estimated at 2.2 mm per turn which imum load of 700 kgf. increased the conductor width to 32.2 mm. After al­ The magnet is supported at two support plates which lowances for insulation thickness conductor and winding are fastened to the horizontal yokes and can be split into tolerances were considered it was determined that each an upper and lower assembly. These plates also allow layer would be wound with 67 turns. The keystoning de­ the longitudinal yokes to be clamped and held in position creases for each outer layer so it was necessary to space relative to the coil. They have lifting attachments which the outer layers to achieve a uniform overall length. allow the magnet to be traversed around the AGS tunnel. Splicing was necessary to achieve the lengths needed Due to limited headroom requiring a large sling angle on for each layer. The manufacturer was required to pre­ the lifting slings it was necessary to install spacer rods to pare samples, some of which were cut open and some were prevent deflection of the support plates adjacent to the tested to check that the tensile strength of a spliced length lift points. was not less than 80% of the original conductor strength. V. Magnet Assembly The insulation was specified as fibreglass-epoxy with a thickness of 0.5 mm on the conductor and 0.75 mm ground Conventional magnets usually have the coils firmly wrap. It was assumed that "B staged" impregnated mate­ mounted to the steel yokes but this magnet differs, the rials would be used for interturn, inter- layer and ground coil assembly is the main structure and the flux return insulation with one final cure when the insulated assembly yokes and support stand were assembled onto and posi­ was complete. tioned relative to the coil. As the coil outer diameter was The manufacturer was asked to propose suitable ma­ not determined until the coil had been impregnated and terials and procedures and to be responsible for the coil cured the yoke and support structures were mounted us­ impregnation. This proposal was considered in the bid ing non metallic shims which had to be adjusted during evaluation. the assembly. The lower yokes and endplates were assembled and the IV. Y oke Assembly coil was placed onto them and the position of its axis de ­ termined, the lower shims were then adjusted. A similar The Snake solenoid has flux return yokes to reduce leak ­ procedure was followed for the upper yokes. The coil end age magnetic fields and the coil ampere turns. Because the connections were completed and the coil was prevented magnet is pulsed it was necessary to laminate the yoke from moving longitudinally by G10 blocks epoxied be ­ plates, otherwise an eddy current induced opposing field tween the coil ends and the end plates. would reduce the solenoid field. The nominal yoke flux VI. Manufacturing Experience density was selected to be 1.25 T and the steel plate thick ­ ness was chosen to be 1/8 in (3 mm) at which level the The contract to manufacture this magnet was let in maximum opposing flux would be less than 0.04 T. The September 1992 and the magnet was shipped on April steel plates were specified as having a nominal carbon con­ 30th 1993. The manufacturer provided all materials and tent of 0.1%. processes. This was a “one-off” contract using a large The yokes consist of six longitudinal sub-assemblies and size conductor and an unusual yoke construction. During two end plate sub-assemblies. The longitudinal yokes are the construction some problems occurred but they were arranged in a hexagonal array and there are slots between quickly addressed and solved. them which allow the coil interconnections to be brought It was found necessary to cure each coil layer as it was out conveniently. The number of plates in each assembly wrapped because the tension needed to form the conduc­ was adjusted to take accountof the plate thickness to hav­ tor around the small diameter caused damage to the insu­ ing to specify a non-standard thickness tolerance for the lating material of the previous layer. The coil length from plate material or having to machine the sub-assemblies. layer to layer was difficult to achieveaccurately, so the coil The individual plates are assembled onto a stainless steel interconnections were modified and the coil overall length plate and bolted together to form a sub-assembly. The six was longer than specified. The long coil length caused longitudinal yoke sections were machined to length. They gaps between the end plates and the longitudinal yokes were then disassembled, deburred and coated with an in­ which were filled with electrical steel shims. The over­ sulating varnish, except where they mated with the end all magnet length was 3/16 in. (4.76 mm) longer than plates. specifed but this was found to be acceptable. Some dis ­ The endplates were stepped to form a joint with the crepancies in the cooling connection hardware were recog­ longitudinal yokes. Their assembly and insulation was nised and rectified during the fabrication. similar to that described above. The steel assembly onto the completed coil was as speci­ At full current the force on the end plates was estimated fied and the magnet passed the specified factory tests. No to be 22 tonne. The magnet assembly is held together by magnetic performance tests were required by the manu­ twelve horizontal tie rods which prevent the joint between facturer. 3 The magnet was shipped to Brookbaven National Lab ­ al testing will be accomplished during an upcoming AGS oratory and immediately prepared for installation. run.

VII. Magnet Performance at the AGS VIII. References

The magnet was installed in the MO straight section [1] Ya.S. Derbenev and A M Kondratenko, JETP 35, of the Brookbaven AGS and preliminary electrical testing 230 (1972), Part. Accel. 8, 115 was conducted. The magnet was operated at up to 3950A [2] G Lazarus Ratner, "A Partial Snake for the AGS”, dc for over one hour without problems. Additionally, the presented at 9th International Symposium on High magnet was powered for three hours from a source deliv­ Energy Spin Physics, Bonn, Germany, 973-13/90. ering a 6290 A peak current pulse having a rise time of BNL 44886. 800 msec and a rep. rate of 3.6 seconds. The total number of pulses is estimated to be over 2800. No problems were [3] H. Huang, e< a/., "The Partial Siberian Snake Exper­ encountered. The magnet’s effect on the circulating unpo­ iment at the Brookbaven AGS”, presented at Euro­ larised beam was as expected. The changes in orbit and pean Particle Accelerator Conference, Berlin, March tune were not deleterious and the accelerator operated at 24-26, 1992. full intensity, whether the snake was on or not. Addition­

4