NUCLEAR INSTRUMENTS AND METHODS 61 (1968) 1-30 ; © NORTH-HOLLAND PUBLISHING CO.

THE 2.5 GeV ELECTRON SYNCHROTRON OF THE UNIVERSITY OF BONN K. H. ALTHOFF, K. B~,TZNER, J. DREES, A. FEBEL, O. GILDEMEISTER, G. VON HOLTEY, G. KNOP, P. LOTTER, H. NETTER, W. PAUL, F. J. SCHITTKO, A. SCHULTZ VON DRATZIG, H. E. STIER and E. WEISSE

Physikalisches Institut der Universitiit Bonn

Received 29 December 1967

From 1963 to 1967 a 2.5 GeV electron synchrotron with 50 Hz chamber is used. To achieve a high circulating beam current a repetition rate has been built at the University of Bonn. Design, special rf amplitude regulation system has been built to reduce the construction and first operation of the machine are described. influence of beam loading. The maximum circulating beam Elc~ctrons are injected by a 25 MeV linear accelerator. The magnet intensity achieved up to now is 5 x 1012 electrons/sec. About 60% ring has a mean radius of 11.08 m and consists of 12 magnets with of the circulating electron beam can be ejected with spill times up the basic structure 0/2 F D 0/2. The frequency of the rf ac- to 1 msec. celeration system is 499.670 MHz. A pure ceramic vacuum

1. Introduction equivalent quanta/rain. The main aims of research have High energy physics at Bonn started in 1953 when it been the photoproduction of pi-mesons, the deuteron was decided to build a 500 MeV electron synchrotron. disintegration, and the structure of electron-photon- It came into operation in 1958 as the first alternating showers. field gradient synchrotron in Europe. Up to now it has In 1963, it became clear to us that we had to extend been working for more than 30000 h. The present the experimental facilities of our laboratory in order to intensity is 5 x l0 t i electrons/sec, corresponding to an keep up with the development of high energy physics. intensity of atypical collimated photon beam of 3 x l0 t 1 At that time, the Cambridge accelerator was running,

Fig. 1. General view of the 2.5 GeV electron-synchrotron before completion of the shielding tunnel (Photo K. Kiiffner).

APRIL 1968 1 2 K.H. ALTHOFF et al. the DESY synchrotron was just coming in opera- sections of the Collins type. The length of all of the tion and it was decided to build the NINA syn- normal straight sections was determined to be 1.8 m chrotron. However, there was a gap between these and, for instance, two accelerating cavities, the current 6 GeV machines and the existing ones around 1 GeV. strip and the ejection magnet and the inflector had to be On the other hand, we wanted the new machine to be a positioned between the different magnets. Therefore a university machine in which the students* would suitable compromise was a simple magnetic structure participate in designing and constructing the acceler- with N= 12 basic periods and an 0/2 FD 0/2 basic ator. Finally, we made a proposal for an electron lattice. At the originally designed energy of 2.3 GeV the synchrotron with an energy of 2.5 GeV at total costs of maximum magnetic field is then 10.03 kG at an orbit 12.3 million DM including the building. radius of 7.65 m. The general parameters of the machine At the end of 1963 the state government of Nord- are collected in the first part of table 1. The number of rhein-Westfalen agreed to support the new accelerator. betatron oscillations per turn is 3.4 for both directions. In 1964, the design was completed, and the major parts One reason for chosing this value is that the external were ordered. In 1965, the accelerator building and the extraction of the beam requires a shift into the half laboratories were constructed. In spring 1966, the first integer resonance at 3.5. parts of the machine were delivered, and at the end of In order to avoid corrections of the quadrupole and the year, almost all components were in place. In March higher order fields of the synchrotron magnets at the 1967, the first electrons were injected and, without time of injection it is desirable to inject at a field of at trouble, accelerated up to 2.3 GeV. In July, the least 100 G. We had chosen as an injector a 25 MeV- accelerated electrons could be extracted at an efficiency linac with an output current of 250 mA at an energy of 60%. The maximum intensity until now has been spread of _+ ½% (part 3 of table 1). 5 x 10 ~2 electrons/sec. A photograph of the synchrotron A radiofrequency of 500 MHz in the accelerating is shown in fig. 1. system seemed to be reasonable for the following reason : 2. Basic conception a. This frequency corresponds to a harmonic number The aims in designing the accelerator were mainly to of 116, which is an acceptable high value in order to obtain, as soon as possible, an efficient accelerator using reduce the amplitudes of the synchrotron oscillations. the present experience in accelerator construction. b. Furthermore, this frequency can easily be gener- However, there were some major problems to solve. ated by conventional klystron transmitters. Since we were limited in space, we had to use higher To achieve a high beam intensity the peak if-power magnetic fields than CEA and DESY. Therefore, we had been set to 100 kW, although the power to build up had to investigate how to overcome the difficulties the peak accelerating voltage of 700 kV amounts to caused by the saturation effects of iron1). Secondly, in only 25 kW. The number of the cavities should be 1 or 2 order to use the higher acceptance of the synchrotron in order to reduce the influence of the beam induced as compared to the linac emittance we decided to use voltage4). multiturn injection. For this reason, the problem of From the early beginning it had been decided to use best matching between linac and synchrotron was in- a ceramic vacuum chamber, because a similar chamber vestigated2). Finally, we considered right from the has been used successfully at the old 500 MeV electron beginning the construction of an ejected electron beam. synchrotron since 1958. This problem has been simulated on a computer to get the best parameters for the machine and the ejection 3. Linac and injection system* system3). 3.1. THE LINAC The general dimensions of the synchrotron were fixed The injector is a linear accelerator manufactured by by the fact that the machine was to be located on an Varian, Palo Alto, Calif. This linac has a zero beam area of approximately 30 x 60 m E between "Physika- energy of 39 MeV at maximum rf power of 21 MW. lisches Institut" and "Institut fiir Strahlen- und Kern- We operate the linac at 18 MW rf power, at total physik". Thereby the main diameter of the accelerator current of 500 mA, with 180 mA within an energy spread was limited to 22 m. Since the slow extraction of the of _ 0.5%, and a final energy of 25 MeV. The choice of electron beam was planned from the beginning, it a linac with this energy seemed to be a good compro- seemed not necessary to incorporate long straight mise between the conflicting requirements of high * Their names appear in footnotes of the chapters describing the injection energy and small costs. High injection energy design and measurements they were participating. * A. Febel, A. Schultz von Dratzig, P. Liitter, H. Haseroth. 2.5 GeV ELECTRON SYNCHROTRON

TABLE 1 Bonn 2.3 GeV Electron synchrotron. Parameter list.

1. General parameters Focusing type AG Injection energy 25 MeV Number of basic periods 12 Magnetic field at injection lllG Basic lattice 0/2 FD O/2 Beam pulse length 1/~s Fie]td at 2.3 GeV 10.03 G Peak beam current 250 mA Repetition frequency 50 Hz Emittance e=F/n= 3 x 10-4 Mean radius 11.077 m cm. rad Orbit radius 7.65 m Energy spread + 0.5% N umber of betatron oscillations per turn Qr = a~= 3.4 Nominal length of F-section 2.0028 m 4. rf-acceleration-system Nominal length of D-section 2.0028 m Peak voltage per turn 700 kV Length of free field region 1.7945 m Frequency 499.670 MHz Field index, n - 22.26 and 23.26 Harmonic of orbital frequency 116 Maximum fl-value 925.5 cm/rad Peak if-power 100 kW Minimum/%value 147.0 cm/rad Average if-power 40 kW Maximum of momentum compaction Number of accelerating cavities 1 or2 function 164.7 cm Cavity shunt impedance 10 M,C2/ m Minimum of momentum compaction Cavity Q-factor 30.000 function 81.0 cm Klystron, type F 2008 Revolution frequency 4.3074 MHz Energy loss of a 2.3 GeV electron Revolution time 232.16 ns per turn 325 keV

2. )~Iagnet 5. Vacuum chamber Gap height at orbit 6.0 cm Material pure ceramic Radial aperture (F-sector) 9.0 cm Width 120 mm Field gradient, max. F: 292 G/cm, Height F: 36 mm, D: 42 mm D: 305 G/cm Wall thickness 7 mm Coils, number of turns per magnet unit 36 Copper cross-section per turn 480 mmz 6. Ejection Peak current per turn at 10 kG 1380 A Efficiency ~ 60 Peak stored energy (in 12 magnets) 350 kWs Emittance hor. t Ap/po = 0 / measured ~. 1.5 mm' mrad Total weight of each of the 12 magnet vert.J values ~. 1.0 mm. mrad units with girder and coils 18.5 t Spill-out time 50 #sec-1 msec Current strip: dimension 50cm × 1.9cm × 0.08cm Current at 2.0 GeV 3500 A 3. lnjector system Septum magnet: aperture 2 cm x 0.9 cm Type Linac length 50 cm Accelerating frequency 2998 MHz (2 = 10.0 cm) Field at 2.0 GeV 1.85 kG

means a small influence of the difference in the remanent system that controls the charge voltage of the linac. The fields of the synchrotron magnets, a nearly closed orbit rise and fall times are 0.5 #s each. at a fixed radiofrequency and a better overcoming of A rectangular cavity operating in the TEl02 mode as beam loading effects in the cavities of the synchrotron. a chopper is placed adjacent to the gun anode. It is A block diagram of the linac is shown in fig. 2. The driven at a frequency of 2998.02 MHz with 10 kW gun is a bombarder type with a thoriated tungsten peak power sweeping the beam with an amplitude of cathode that produces a current peak of 2 ampere at 6 mm at an aperture to provide beam pulses with a 200 ° 100 kV. The gun filament current is regulated by a phase width. A prebuncher cavity located adjacent to transistor differential amplifier. The voltage (3 kV) this aperture produces 30 ° bunches. Two thin lenses between filament and cathode is stabilized in order to focus the beam through the linac injection system to a hold the temperature of the cathode constant. The gun second aperture at the accelerator input coupler. is normally operated as a space charged type. A pulse The accelerator has only one section with a length of forming network is employed to pulse the gun. The 255 cm. It is a disk loaded guide. The distance between voltage is maintained constant within _ 0.25% during the disks is ½2 with 2 --- 10 cm, the vacuum wavelength a pulse length of 2/~s by using an electronic feedback of the linac rf. The phase velocity equals the velocity of 4 K.H. ALTHOFF et al.

External 499MHz

Harmonic I Oscillator Generator 499Mc 2998MHz

Ktystron Master Trigger -240KV Generator • Klystron Modutator

,33db External 50Hz 20MNV

Bombar- Chopper Prebunch~ Accelerator der , Goo ; r~ Kv

Fig. 2. Schematic diagram of the linac.

x[s) [cm]

s=0 Si i 2 3.... l* 5 2--=4-~ 6So I ;/ Sm...... 8 L ~) 10 i"11S2 ] 1'2 13s[m]

]Fig. 3. Injection beam bump. Eielmholtz coils at St and S=, So point of injection. light through the whole guide. The mean group velocity The rf-source is a crystal controlled oscillator com- is (1/60)c and the filltime is 0.5/~s. The longitudinal bined with a varactor harmonic generator that pro- electrical feld, dependent on the input if-power and on duces 1 W at 499.67 MHz. This frequency is multiplied the varying diameter of the holes in the disks is in the to 2998.02 MHz. This in turn drives a 1 kW pulsed beginning of the guide very strong -170 kV/cm at travelling wave tube amplifier (TWT). The output of the P0 = 20 MW- in order to stackthe electrons in bunches TWT is fed through a variable attenuator to the of 5 ° phase width, and in the rest of the guide it is nearly klystron. The linac has only one klystron. It is a constant. The inner surface of the last 8 cavities is Thomson-Varian type TV-2014 rated at 21 MW peak, covered with an alloy of high permeability. These 30 kW average with two output arms. The klystron is cavities work as a collinear load for the remaining rf- pulsed by a modulator similar to the gun modulator. power. Beam position can be varied by steering coils The rf wave is fed into the input coupler of the acceler- located in the injection system and the accelerator ator by a rectangular guide. Two directional couplers guide. At the output of the linac a magnetic type beam in this guide are used to drive the chopper and prebun- monitor produces a signal proportional to the beam cher. intensity and the beam position. A longitudinal mag- A trigger generator produces pulses to trigger the netic field in the order of 1000 gauss is established by TWT, klystron and gun with a repetition rate of 12.5, solenoids. The accelerator guide is water cooled. The 25 and 50 Hz. The whole system is normally trig- controlled and regulated temperature is held within gered by a peaking strip pulse from the synchrotron +O.I°C. magnet. 2.5 GeV ELECTRON SYNCHROTRON 5

3.2. THE INJECTION SYSTEM holtz-coils were installed a quarter betatron wavelength Between the linac and the synchrotron ring a trans- upstream and downstream from the inflectorS). They port system is necessary to convey the beam of the linac are used to perturb the equilibrium orbit so that it into the ring and to match the beam in a manner so that passes near the inflector. During about 5 turns the field it can be accepted by the synchrotron. in the pulsed magnets is decreased gradually so that the The design of this injection system mostly depends on equilibrium orbit shrinks back to its normal position. the: fact that the electrons are injected during a time The location of the coils had been calculated from the much longer than the revolution time of the particles in condition ~(S2)-(~(81) ~- ~. ~(Si) are the phasefunc- the circular accelerator. The other possibility would be tions at the positions of the coils. One coil is fixed a single turn injection as used by all electron synchro- 12.8 cm in front of the iron edge of the synchrotron trons with a final energy higher than 4 GeV. In that case magnet M 12 and the other coil 12.8 cm behind the iron the inflector - mostly a magnetic one with a large field edge of the synchrotron magnet M 1. The ratio of the volume - is located symmetrically on the central syn- magnetic field strength of the two coils is chrotron orbit. Electrons with a momentum deviation {~(S,)I(~(S2)}~= 1.8. carl be brought to their respective closed orbit and the linac beam can be matched to the phase volume of the The shape of the calculated beam bump is shown in synchrotron. It is essential in such a system that the fig. 3. field of the inflector is switched off in a time short The inflector should be located away from the central compared with one revolution time. The requirement is orbit as far as possible in order to do not restrict the hardly to be fullfilled at this accelerator since the acceptance of the synchrotron too much. On the other revolution time is 232 nsec; that means a 20 nsec hand the electrons must not hit the vacuum chamber in switching time. the following F-sector where the bump has a maximum. In multiturn injection the inflector located between The narrower the inftector is located to magnet M 1, two synchrotron magnets is located away from the the better this condition is fullfilled. For that reason the central synchrotron orbit. Since the shrinkage of the exit of the inflector is located now 40 cm downstream equilibrium orbit due to the growing magnetic field in from the center of the straight section and the beam the synchrotron magnets is too small, two pulsed Helm- center line at the exit 3.1 cm from the central orbit. The

t.,ol

/.0

30

20.

10. inf{ector

/ central, synchr orbit

3.5 3,0 2.5 2.0 1.5 1.0 0.5 0.0 d [cm] Fig. 4. The injection efficiency'7 for different speeds of bump shrinking. Curve I: 0.83 cm/turn; II: 0.05 cm/turn; III: 0.36 cm/turn. The curves are calculated for a synchronous phase of 20°. 6 K.H. ALTHOFF el al. angle between the centerline of the downstream end of come. This result depends mostly on the synchronous the inftector and the central orbit is determined by the phase, for 3 out of 6 linac bunches are lost at a phase shape of the bump and the requirement of small excited of 20 ° . betatron oscillations. It has a value of 10.25 mrad. Once knowing that this injection mechanism works At the beginning of the operation of the machine the the design of the system could be continued. The phase inflector had been an electrostatic one. It consisted of ellipses in both directions and two electron paths with two curved plates separated by 1.2 cm with a radius of particles having a momentum deviation of --+ ~/O~o/. 5 m and a length of 40 cm. The electrical field strength starting on the injection centerline at the exit of the had been 50 kV/cm. inflector were transformed backward through the The efficiency of the injection system described so far fringing field of the foregoing D-sector of magnet M 12 had been calculated on an IBM 7090 computer. There- to a place (point C in fig. 5), 4.40 m away from the by single electrons with an energy, randomly chosen center of the field free section, where the magnetic field within the range of + 0.5%, and started at the exit of can be neglected. At this place the properties of the the inflector at randomly chosen locations and angles beam are still fixed by the structure of the synchrotron within the phase ellipse in the horizontal plane, were magnets and the chosen injection mechanism. The traced through the synchrotron ring over 100 turns. In requirements for the remainder of the injection system the Helmholtz coils the electrons get an angle deviation can be summarized: depending on the magnetic field during the shrinking a. The electrons have to be conveyed from the linac of the injection bump. In the cavity the energy and the exit to point C; specifically there must be a nearly phase of the electron were changed depending on the parallel displacement of the beam of 2.80 m; synchronous phase and the voltage of the cavity. In this b. The whole system should be an achromatic one; calculation we made the assumption that the emittance c. Moreover there should be a phase matching of the linac beam is 5 x 10 .6 m.rad and that there is a system ; linear shrinking of the beam bump. Beam loading d. The amplitude of the beam should be nowhere effects in the cavity had not been regarded. If the par- larger than 4 cm in the horizontal plane and 1.5 cm in ticle did not hit the inflector during the 100 turns it was the vertical plane; named an accepted one. In fig. 4 the injection efficiency, e. There should be a waist in the shape of the beam, that is the ratio of the accepted to the injected electrons, horizontally at a spot where the beam has a big dis- is drawn versus the position of the beam bump at the persion, in order to separate the electrons with an exit of the inflector. It should be possible to inject the energy deviation of more than lo// O by a slit system. electrons during 3 turns with a mean efficiency of 40°'o These conditions reduce the variety of possible when the beam loading effects in the cavity were over- systems very much. For instance two bending magnets

M3

o,[] o-

the linac- ....I i Im ~ I

/.,../',,,,/,.///, ",'////,,/ ~, "'/f//'/ / ////,''//// // ////' /I // / Fig. 5. The injection system. Quadrupoles Q I,...,Q7, bending magnets IM 1,... IM3, collimators S1 and $3, energy defining slit S2, synchrotron magnets MI, M12. 2.5 GeV ELECTRON SYNCHROTRON 7

] x [mm]

-2 -3 -Sl-q I i injector fringing fietd M3 Q6 Q5 IM2 $2 IM1 of M12 Fig. 6. The achromatic part of the injection system. The particle trajectories belong to momentum deviations of _+ 0.5%.

are necessary to fulfill the requirement (a), but three center the beam through the quadrupoles. The beam magnets are necessary to get additionally a waist at passes through a second pair of steering coils and enters some spot. The phase matching systems needs relatively at point B magnet 1M 1 and the achromatic system. The long drift spaces to limit the focal strength of the qua- deflecting angle of IM 1 is 41.3 °, the radius 29.2 cm. drupoles. This implies that there is not too much space The electrons then pass magnet IM2 (r=33 cm, for the achromatic system. Possible systems were cal- ~b = 90°), two quadrupoles, a magnetic pickup-station culated on the IBM 7090 computer. In fig. 5 the in- used for the measurement and the control of the beam jection system chosen is shown. The phase matching intensity and magnet IM 3 (r =20 cm, q~--90°). A system is 5 m long and consists of 4 quadrupoles. From third pair of steering coils located behind magnet IM 1 the entrance of magnet IM 1, downstream from the downstream steers the beam over the relatively long inflector exit, the system is achromatic. It consists of distance through the fringing field of M 12 to the three bending magnets, two quadrupoles, the fringing entrance of the inflector. field of the synchrotron magnet M 12 and the in- The vacuum chamber of the injection system is an flector. At point S 2, the location of the waist of the aluminium tube with a 65 mm inner diameter. In the beam, a pair of copper slits, with a 6 mm separation, is bending magnets the tube is replaced by flat aluminium installed to make the demanded separation of the chambers. Three Vacion pumps (80 1/sec) located electrons with an energy deviation of more than ½%. under the slit boxes S 1, S 2 and S 3 hold the pressure If the magnet IM 1 is not energized, the beam goes below 5 x 10-7 mm Hg. straight forward and the emittance of the linac beam can be measured with the help of quadrupole Q7, the 4. Magnets* driftspace, a pair of slits at S 3 and a Faraday cup. The A-G magnets of the synchrotron must obtain The paths of the electrons entering magnet IM 1 on two main functions. They have to produce a magnetic the centerline, but having a momentum deviation of field that deflects the electrons along an approximately _+ ½% through the achromatic part of the injection circular orbit. Moreover, they must produce magnetic system, are shown in fig. 6. field gradients, alternating in sign, so that the elec- In the following we give a more technical description trons are focused horizontally and vertically and re- of the injection system. The beam leaving the linac at K main in a small volume near the equilibrium orbit. passes a vacuum valve that separates the linac from the As well known, the tolerances concerning the mag- injection system, if required. With the help of a pair netic properties of the 12 units are extremely close. The of steering coils it is possible to deflect the beam effective bending length of all of the magnets, for in- horizontally and vertically to an angle of max. 10 mrad. stance, must coincide within 10 -3 in order to avoid The electrons pass the four quadrupoles Q1...Q4 considerable displacements of the closed orbit6). forming the phase matching system. These quadru- Furthermore, the actual field gradient in the focusing poles, as well as the other three quadrupoles, can be and the defocusing sector has to agree accurately with energized up to a field gradient of 250 G/cm. They have the theoretic value, otherwise the number of betatron a physical length of 13 cm. This system can provide an oscillations Q, and Q= may be shifted close to a exact matching of the linac emittance to the desired dangerous resonance. These conditions must be ful- phase ellipses in point C. A pair of slits at S 1 helps to * J. Drees, K. H. Becks, F. Knorr. 8 K.H. ALTHOFF et al.

filled at all field levels, that is, from the injection field, for the radial direction and where remanence effects are of importance, up to the maximum field, where saturation effects occur. AQ~ = - 1.61 {(rtae t -- rt,h)//'lth }F -~- 6.07{(nat t -- rl,h) / rtth}O , In addition to these fundamental requirements, (2) which must be fulfilled in a similar way in all A-G for the vertical direction. The indices "F" and "D" synchrotron magnets, the various following points were characterize the focusing and the defocusing sector. taken into consideration during the planning and the The theoretical values of nth are given in table 1. construction phases. In a machine with a constant gradient within the To reach a short construction time the pole profile vacuum chamber, the number of betatron oscillations and the cross section of the magnets were calculated on per turn depends on the momentum of the circulating a computer. The influence of saturation was included. particles. From the linear theory one obtains AQ/Q Thereby it was possible to start manufacturing of the =-Ap/p. This dependence can be reduced by intro- magnet blocks without checking any magnet model. duction of an additional sextupole component in the Pole face windings for the correction of the quadru- focusing and defocusing sectorT). At the same time Q pole and the sextupole contribution at injection field becomes independent on the position of the orbit at (about 100 G) were omitted, to avoid difficulties with injection. This displacement of the orbit amounts to radiation damage. -0.34 cm with respect to the ideal closed orbit at an Furthermore the complete magnet unit including injection energy of 20 MeV. The effect is enlarged when girder and supporting system should not possess me- the rf is modulated. Moreover, the appearance of chanical resonances close to the repetition frequency of satellite stop-bands due to a coupling of betatron and 50 cycles per second and its first harmonics and the synchrotron oscillations is avoided. vibration which is transmitted to the foundation should Near the orbit the field gradient then depends be as small as possible. linearly on r:

4.1. THE APERTURE AND THE POLE CONTOUR OF THE n(r) = n +S(r/p) where S = -(pZ/B~)@ZB/~rZ)]0. (3) MAGNET The aperture of the magnet is given by the amplitudes Neglecting fringing field effects one gets: of the betatron oscillations due to the emittances of the SF = -- 50 ; So = 130. linear accelerator and the "errors" of the multiturn injection, the amplitude of the synchrotron oscilla- Statistical variations of the sextupole component s tions, and the displacements of the closed orbit due to induce of course a resonance at Q = 3½. However, the imperfections and misalignment of the magnets. In this width of this resonance is only of the order 6Q ~ 10 -s, particular case one obtains for the radial direction a thus a small change of the betatron amplitude brings maximum amplitude of the betatron oscillation of the particles out of the resonance. 1.4 cm, the amplitude of the synchrotron oscillation is A field gradient due to eq. (3) can be provided by a 1.6 cm for a momentum deviation of Ap/p = 1°'o, and contour curve : the displacement of the closed orbit due to magnet z {l - ,,(,It,) - ½s(, "2 lp2)} + "(,, + s) (z3/p = errors remains smaller than 1.4 cm. Altogether this = h+~(n+s)(h3/pz), (4) requires a radial aperture of 8.8 cm. For the vertical direction one obtains 3.8 cm. that follows an equipotential line. h is half of the gap For the Bonn synchrotron a ceramic vacuum height at the orbit. chamber with a relatively high (vertical) wall thickness The pole contour of the magnet has to provide the of two times 7 mm and a tolerance of about 6 mm is desired field gradient with the required accuracy. How- used. The gap height at the orbit therefore was deter- ever, the width of the contour is not only given by the mined to be 6 cm. aperture but also by the tolerable decrease of the Within the useful field region a gradient must be function n(r) due to saturation effects at strong fields. produced which deviates from the theoretic behaviour A reduction of the saturation can be obtained by a soft at the ends of the region by less than 1%. A systematic curvature of the pole edge at the dosed side of the error of the field index Got=--(p/B~)(cOB/~r)lo magnet pole. Thus the maximum field at the iron sur- causes a shift of the working point (number of betatron face is reduced. oscillations per turn): For the design of the contour one has therefore to AQ,. = 5.81 {(nact- nth)/nth}v -- 1.68{(G,-- n,h)/nt, }o, (1) know the magnetic properties of the used steel. 2.5 GeV ELECTRON SYNCHROTRON 9 I z [mini

I I J- hyperbolic part, equation (4) ~J

N Q N ~ equilibrium ~ ~. r [ram]

• 52 - - 60 76 110 110 Fig. 7. Pole profile of tho focusing sector.

For this machine 0.47 mm thick laminations of a hot equilibrium orbit (r = 0) the agreement between the roilled transformer steel have been used (Armco Trancor measured points and the computed curves is better than A6). This material has been selected because of its favourable magnetic qualities at weak fields. In partic- n(r) ular, pole face windings for the correction at injection field could be omitted. On the other hand, the perme- !o ability at strong fields is relatively small and saturation 1.01 effects must be considered for field levels above 8 kG. ELAX (lr In this particular case it was also required that the BYL (5rnm) ceramic vacuum chamber could be inserted into the 1.00 J magnet gap radially from the outside. \I ° Bs= 5kG For the determination of the pole profile, magnetic 0.99 fie]Ld computations using the relaxation method have been carried out1). Two different computer programs wide side of mogn¢l gap (two-dimensional) complementing one another have ] i 0.98 been used: an electrostatic program (RELAX) and the 6 cm I magnetostatic program SIBYLS). The RELAX pro- F-sector, n F = - 22.26 gram enables a very accurate computation of the field within the vacuum chamber. SIBYL includes the finite permeability of the steel and the influence of coils. n(r__A ]For these calculations the theoretic values of the field n D index and the sextupole component have been used, 1.04 which have been calculated for the ideal magnetic / sector (hard-edged model). The initially unknown 1.02 / junction effects between F- and D-sector and the in- J fluence of the gap between the different identical F- or D-blocks have been corrected later on by means of a 1.00 special shape of the magnet ends. Thereby it was pos- // sible to finish quickly the design of the contour and to 0.98 start the complicated fabrication of the profile die at an .EL,X ~ r4, early time. wide side of magnet gap Fig. 7 shows the pole contour of the focusing sector. 0.96 2 /I 6 cm r in the central part the profile follows an equipotential curve according to eq. (4). D-sector, n D = 23.26 Fig. 8 shows the calculated and the measured radial Fig. 8. Calculated and measured shape of the field index n(r) at shape of the field gradient at medium fields. Near the medium fields compared with the design values. 10 K.H. ALTHOFF et al.

0.1%. The differences between the two calculations at -0.4~'/o in any way the tolerance has been set + 0.012 the narrow side of the magnet gap arise from the rough mm in the interiour part of the profile. mesh which had to be used for the SIBYL computation. For the design of the injection path and the calcula- 4.2. THE CORRECTION OF THE JUNCTION EFFECTS BY THE tion of the extracted electron beam it is of interest to MAGNETS ENDS know the radial fringing fields outside the useful field The field of the actual magnet differ s from the field of region. Fig. 9 shows the computed function B(r) for the hard-edged model especially by the fringing field at the focusing sector. The values at the abscissa mark the the magnet ends and the transition effects between the maximum difference between measurements and cal- focusing and the defocusing sector and the different culations (normalized to 1 at r = 0) for different sections. fields extend over an azimuthal distance that is small Even outside the magnet pole up to a radius of compared with the length of a magnet sector. The in- r = + 29 cm the agreement is better than 1% . fluence on the closed orbit and the working point of the The behaviour of the field gradient under the in- machine then can be described by an effective bending fluence of saturation effects can also be accurately and focusing length, that means by a normalized inte- calculated with the relaxation programs. At B S = 10 kG gration of the fringing field along the azimuthal for instance, one computes for the F-sector a decrease coordinate. of An/n=-O.82°fo while An/n =-0.74% has been E.g. the effective bending length of the fringing field measured. For the D-sector the corresponding values at the end is defined: are: computed An/n=--0.81% , measured An~n= -0.76%. At B~= 11 kG the computed values are 1.e(r) = {B,(r)}-' B(r,s)ds, (5) $1 An/n = -2.2% for the F-sector and An/n = -2.1% for the D-sector and at 12 kG An/n =-4.1% for the F- and the effective focusing length: and An/n=-4.0°/o for the D-sector have been computed. /Ge(r) = {~'B,(r)l?)r}-' {OB(r,s)/~r}ds. (6) Finally it should be mentioned that the tolerances for St the profile die have been fixed using the RELAX- At the position sl in the central part of the magnet, program. In order to avoid gradient errors greater than B 1 and #B1/c3r are independent on the azimuth s. The

[ steo~ JL excitation coi( ...... _~

B~{r)

Bs =4kO

r>0

08"

\

04-

0 z. .5 lb 15 20 25 30 35 tr [crn] ~10 ~4- ~10 -3 .I ~10 2 4 Fig. 9. Radial flinging field in the mid-plane of the focussing sector. The numbers at the bottom give the maximum difference between measured and calculated values. 2.5 GeV ELECTRON SYNCHROTRON 11 integration extends over the whole fringing field. D-sector and the different identical blocks have to be Analogous quantities can be defined for the transition compensated. This must be obtained also for electrons zone between two identical magnet blocks (at the orbit following an orbit with a mean radius r within -4 cm the gap between the blocks is 9 ram) and the transition < P < 4 cm. Of course, this is only possible at medium zone between F- and D-sector. The corresponding fields where remanence and saturation effects are of no formulas are given in9). A calculation especially of the importance. focasing lengths using a three-dimensional relaxation It is well known that the first condition can be melLhod is not yet possible. The two-dimensional calcu- precisely fulfilled in the plane case by an exponential lations give only some rough results. Therefore these increase of the gap at the end1°). effects have been measured. Z = Z0{1 + (9) The effective bending length at r = 0 can be easily cexp(½rcs/Zo) }. corrected by suitably changing the width of the two end Z0 is the distance between the iron surface and mid- blocks of the magnet. Apart from smaller influences plane in the central part of the magnet. For our three- which will not be discussed here, the shift of the number dimensional problem an increase of the exponential of betatron oscillations per turn for the radial and the form eq. (9) was kept in the planes rectangular to r. vertical direction then can be calculated from the However, the different exponential curves have been following formulas: shifted in the azimuthal direction so that condition 2 is approximately fulfilled. The final shapes of the ends 'IF have been determined after studying one end block of 'uF the F-type. =M i 'uD'EF1 (7) One additional difficulty, which has not yet been (2t mentioned, was considered at the same time. Due to the 'ID gradient of the remanent field, the field index is in- 5D creased at injection time especially in the focusing where sector. At the same time the Q-values, especially Q,, are 0.028 0.030 0.017-0.018--0.008 -0.007~. increased considerably. As mentioned earlier, the theoretic Q-values of the synchrotron have been chosen / relatively close to a half integer resonance because of -0.007-0.008 -0.017 0.017 0.031 0.029/ the slow extraction of the electron beam. Therefore, one (8) must be careful that the working point comes not too ]In eq. (7) 2~ = IGE--l~E is the excess of the effective close to the resonance at 3.5 at the time of injection. focusing length over the effective bending length of the Fig. 10 shows the azimuthal dependence of the fringing field at the ends, 2~ is the difference between gradient at the ideal orbit in the focusing sector at a focusing and bending lengths for the transitions between typical medium field of 4 kG. One can clearly see the all of the identical blocks and 2u is the same for the increase of the gradient at the end of the F-sector for a F-D transition zone9). The indices "F" and "D" mark correction of the transition effects, especially to com- again the focusing with respective to the defocusing pensate for the loss of focusing length at the transition sector. The quantities 2 are positive when effective between the 17 identical F-blocks. However, the cor- focusing length is gained. The coefficients of M essen- rection is not complete because of the remanence effects tially show the shape of the amplitude function/~(s) mentioned above. multiplied by the field index n. The numerical values of The resulting Q-values are given in table 2 for various M [eq. (8)] are given in cm-x. magnetic field levels, from the injection field to the Knowing the transition effects in the main part of the present maximum field. The values are valid for elec- magnet, the fringing fields at the ends of the magnet in trons oscillating around the equilibrium orbit with small principle can be chosen in such a way that A Q = 0. amplitudes. They have been calculated from magnetic Thus the end blocks have to fulfill the following field measurements. At weak fields the values somewhat requirements: depend on the exitation wave form. For these measure- 1. Eddy currents and premature saturation effects ments the exitation is characterized by: B~,,x = 10 kG, must be avoided, that is the magnetic induction in the Bml, = - 420 G. iron must be constant over the whole length of the At Bs = 4 kG one can see the effect of the incomplete magnet. compensation in the focusing sector, the Q-value for 2. The effects of the transition zones between F- and the radial direction is somewhat smaller than the 12 K.H. ALTHOFF et al.

n (O,S) nF F-sector, n F =-22.26 1.2. Bs =4kG

(18 rnegriet blocks

(16

0.4

0.2

0 20 40 coo 80 100 120 140 160 180 200 220 S [cm]

Fig. 10. Azimuthal variation of the field gradient n(O,s) at the centra! orbit, s = 0 marks the mid-point between the F- and D-sectors.

TABLE 2 length 1B from magnet to magnet are listed in table 3 Q-values of the central particle. (rms values). B.~ Q,- Qz The table contains the results of measurements on 5 of the 12 magnets.

90 G 3.43 3.42 4.3. MAGNET CONSTRUCTION 4 kG 3.38 3.40 9 kG 3.35 3.38 Each magnet unit is composed of 17 F-blocks and 10kG 3.31 3.35 17 D-blocks each 102 mm in length, one 113.5 mm long F-end block and one 109 mm long D-end block. The blocks are assembled on the machined surface of a theoretic value. At 90 G the Q-values are somewhat too girder and adjusted along an arc of a circle with a high. At strong fields saturation becomes important. tolerance of __ 0.1 mm. The tolerance for the vertical For electrons oscillating around an orbit with -4 cm direction is _+ 0.06 mm. Each block is straight so that for various magnetic field levels Bs. by 60 steps. The excitation of the magnets is provided by four Bs 90 G 4 kG 9 kG water-cooled coils, each containing nine turns of 0.6" 10 -a 0.26-10 -a 0.65- 10 -~ stranded conductor of rectangular cross section. Each conductor consists of four isolated cables which are 2.5 GeV ELECTRON SYNCHROTRON 13

6~5 _3

[ VJ///~d I i _a4 I

I

f

I t 3- -4- -E

\ I ,\

\\

\

\ I--_-_2L-23 L-C-Z-__-_-."t \, i [ _.1_~ \, \

y//////////////////J/////////////. l "t I ' I ' i I i t Fig. 11. Crosssection of a magnetunit. interchanged at each connection between the coils in The dimensions of the girder are: length 4.10 m, order to avoid equalizing currents. The copper cross width 1.16m, and height 0.60 m. These dimensions section of a conductor is 480 mm z. Thus the mean have been chosen after studying the mechanical pro- current density is 1.8 A/mm z at 10 kG exitation. The perties of different 1 : 5 scale models (lucite). By testing four coils of a magnet are connected in series and these models it was ensured that the actual magnet unit together they have adc resistance of 14 mfL The in- does not possess resonances close to the repetition fre- ductance of a complete sector is 30.9 mH. The magnets quency of 50 cycles per second and its first harmonics. contain four back-leg windings for the correction of the The girder is supported by four adjustable hydraulic- dipole moment at the time of injection. jacks at the nodes of the first bending mode. These 14 K.H. ALTHOFF et al.

10 tog U~o2 [db]

60] 291 Hz

50 / I I 1146 Hz // A

s01\ V v

pittQ

20 ' ~ ' 6'0 ' 8~ ~ v [Hz] Fig. 12. Mechanical resonance frequencies of a magnet unit. (u = velocity of the vibration, v = frequency of the vibrator. The vibrator was mounted on the top of the magnet).

supports contain a system of springs for reducing the malloy wire 0.05 mm in diameter which is embedded in vibration which is transmitted to the foundation. They a small glass tube and wrapped around with a 2 mm rest on 24 concrete pillars dug 6 m deep into un- high pick-up coil. This device is situated within a disturbed gravel. cylindrical biasing coil that carries some additional Fig. 12 shows the mechanical resonances of a magnet turns for compensation of the induced voltage. The unit*. The upper curve was measured on the surface of compensating turns have been arranged in such a the girder. No resonance occurs near 50 cycles per manner that errors due to a mirror effect on the iron second or at 100 and 150 cycles per second. The main surface of the magnet pole are avoided. In addition, the peak at 79.1 cycles per second corresponds to the first biasing field becomes rather homogeneous over the bending mode. The lower curve shows the amplitude of length of the Permalloy wire t. The accuracy of a B- the vertical oscillation of the concrete pillar. Between measurement is 2 x 10 -4 and the error of an absolute 60 and 70 cycles per second an additional resonance of gradient measurement is 10 -2. Fixed induction coils the concrete pillar appears. At all other frequencies the and an electronic integrator were used at medium and velocity of the oscillation is considerably reduced. At 50 strong fields, The geometrical dimensions of the cycles per second the insulation of the supports is 27 dB. cylindrical or straight coils have been chosen in such a The weight of a magnet unit including excitation way that the second order derivatives of the magnetic coils and girder is 18.5 t. field or the second order derivatives of the gradient do not contribute to the measurement. The errors are: 4.4. METHODS OF FIELD MEASUREMENT 10 -5 for a relative measurement of the magnetic field The methods of the magnetic field measurement are and 10-3 for an absolute measurement of the gradient. only briefly described here. Two methods were used*). At weak fields measurements were carried out with 4.5. MAGNET POWER SUPPLIES peaking strips. With this method both the dynamic and The 12 magnets are excited by a nearly fully biased the remanent parts of the magnetic field were mea- current: i(t) = iao-i,c cos cot, neglecting higher har- sured. A peaking strip consists of a 25 mm long Per- monics, i,~ is some percent larger than ia¢. This excita- tion is made possible by a resonant network consisting * The measurement of these curves by the "Siemens Schuckert of the magnets as well as capacitor banks + and an air AG, Dynamowerk, Berlin" is gratefully acknowledged. t The measured B-values are printed automatically on data gap choke§ for storing the energy in the intervals cards. between maximum excitation. 2.5 GeV ELECTRON SYNCHROTRON 15

a. c power $uppty ppty

2C L/2

\ .-. ?i\ //

Fig. 13. Schematic diagram of the magnet power system.

Fig. 13 shows a schematic diagram of the circuit. sources. The dc is fed into the circuit between the two Two magnets with the common inductance L are con- halfs of one of the choke windings. Each half-winding nected in series with the resonant capacitor C. Six of is connected in parallel with its resonant capacitors. The these groups are connected to a closed circuit. The network is connected to ground at one side of the dc alternating current oscillates in each of these groups input. The power to make up the ac losses is fed into the with the same phase and amplitude. The direct current circuit by 6 transformers. Their primary windings are flows through the choke coils L' (g'= 2L) which are connected in parallel and energized by a Brown-Boveri connected with resonant capacitors C' in parallel. Thus resonant circuit converter, which needs a series reso- the biased wave form is made possible. The maximum nant load. If the resonant frequency is very close to the w)ltage to earth is 4.6 kV, that is half of the group line frequency, the ac power supply can be synchronized w31tage. The choke secondary windings are distributed in phase with the 50 cycles line. Filter circuits (at the between the primary windings and are connected in exit) ensure that the original square wave voltage be- parallel in order to serve for strong coupling between comes sinusoidal. For an excitation up to Bin,x = 10 kG, tile primary windings and for strong symmetry. Similar a peak current of im,x = 1380 A is needed, the dc is circuits are used in other synchrotrons with a high 643 A and the minimum current -56 A. The total repetition frequency. energy stored in the circuit is 370000 J and the total Ac and dc power are produced by two separate power input is 550 kW. '- Manufactured by AEG. The direct current is held accurately constant (some § Manufactured by BBC, Mannheim. times 10 -4) while the ac is controlled in such a way that 16 K.H. ALTHOFF et al.

the time interval during which the field in the magnets 5.2. FIELD STRENGTH is reversed remains constant. During the acceleration cycle the rf voltage must be Differences in the earth capacitance of the connecting raised with the electron energy. This voltage increase is cables cause a phase difference of the alternating current determined by several facts: in the different magnet groups. This is especially a. At the beginning of the acceleration cycle the dangerous at the time of injection. To control this, the electric field strength must be proportional to the varia- time at which the magnetic field changes its sign is tion of the magnetic guide field with time. The energy compared in all of the magnets by means of peaking- gain AE a per turn, necessary to hold the electrons at the strips and differences are compensated by additional centerline when the magnetic field B increases, must be: earth capacities. AE1 = 2~cr~dB/dt, 5. Radio-frequency acceleration* where r = 7.6500 m denotes the bending radius and 5.1. FREQUENCY ? = 11.0772 the mean orbit radius. The acceleration of electrons is accomplished by a rf b. With increasing energy the electrons, which expe- voltage generated by cavities, which develop a high rience a continuous central acceleration, loose energy by electrical field strength when they are driven at their synchrotron radiation. This energy loss AEe per turn is resonant frequency tour. This resonant frequency is proportional to the fourth power of the electron energy made synchronous with the revolution frequency co, of W and inversely proportional 1~) to the orbit radius r: the electrons: = 8.85 × lO-32(W/e)4/F. (2)rf : ho) u. The energy for instance radiated per turn by a 2.3 GeV electron is 325 keV. This energy loss by synchrotron In our case the harmonic number h is 116 since the radiation must be compensated by an increase of the acceleration frequency is 499.670 MHz, which cor- electric field strength in the cavities, which must be responds to a wavelength ~. = 60 cm. proportional to the fourth power of the particle energy Since the electron velocity at the injection energy of or to the fourth power of the magnetic field B. The 25 MeV closely approximates the velocity of light, it is synchrotron radiation is not important until the elec- not necessary to change the acceleration frequency trons reach an energy of more than several hundred during an acceleration period of about 10 msec. The MeV. At the end of the acceleration cycle the synchro- frequency of the if-system is the 6 th subharmonic of the tron radiation is the essential factor that determines the linac frequency coL = 2998.00 MHz. The fact, that the electric field in the cavities. frequency of the linear accelerator is an integral multiple c. Other factors that influence the peak value of the of the acceleration frequency includes the possibility of electric field are the synchronous phase (& and the synchronism between these two frequencies. variation of the if-field during the passage of the elec- The required frequency stability is determined by the trons through the resonators. Taking into consideration frequency cos of the synchrotron oscillation : this time dependence of the accelerating field, the peak cos = {neU .... ~h cos %/(2rtEs)}~m~, accelerating voltage has to be raised by a factor/3: /3 = sin(½0)/( 0) = where n is the number of accelerating cavities, Um,~ the peak voltage in the cavities, c~ the momentum compac- because the transit angle 0 of the electrons in the res- tion factor, p~ the synchronous phase and E~ the onating chamber is =. equilibrium energy. The synchronous phase at injection was obtained by a With a synchronous phase of 20 ° at injection, which Monte Carlo calculation. A favourable ratio between is shifted to 45 ° at the end of the acceleration cycle, and the number of injected and the number of captured an injection energy of 25 MeV the synchrotron oscilla- electrons occurs at a synchronous phase q~s = 20° a). tion frequency cos varies from about 150 kHz at in- The variation of the accelerating voltage U~f (curve jection to 70 kHz at the final energy of 2.3 GeV. From c), which is the sum of the two terms a) and b) de- these values it follows that the short time frequency scribed above, is shown in fig. 14 during an acceleration stability should be better than 10 s, because any varia- cycle to 2.3 GeV final energy. tion of the if-phase must occur slowly relative to the Fig. 15 shows the variation of the radio-frequency phase oscillation period. voltage Ure during an acceleration cycle of 10 msec for * H. E. Stier, K. B/itzner, H. Netter. several final energies between 0.4 and 2.3 GeV. This 2.5 GeV ELECTRON SYNCHROTRON 17

voltage in the acceleration units so that the number of 75(31 Urf [kV ] electrons captured in the phase stable region is con- siderably reduced. The equivalent circuit of the beam loaded cavity, a resonating parallel circuit with inductivity L, capacity C and parallel resistance R, yields the following differ- ential equation describing the accelerating voltage U in the resonator with time dependent current excitation:

d 2 U/dt 2 + T- 1 dU/dt + co,zf =/trJcotr exp [jcotr t] + 503 + Iu(t)jr(t ) exp [j {cobt + q~(t)}] • [cob + {dqS(t) / dt}] + + {dib(t)/dt}r(t)exp [J{cob + ~b(t)}]. In this equation T = 10 .5 sec denotes the cavity time constant, Itr and co, denote the current and frequency of the transmitter respectively and I b denotes the beam current. _r(t) is 0 for t < 0 and C(t) -- 1 for t > 0, i.e. the electrons are injected at t = 0. It is almost impossible to describe the variation 250

750 2.3 GeV

Urf [

~- t [ rnszc ] Fig. 14. Increase of the accelerating voltage Urt during an ac- celeration cycle of about 10 ms. The final energy is 2.3 GeV, the 5OO synchronous phase is shifted from 20 ° to 45 °. a. rf voltage due to the increase of the magnetic field: B; b. rf voltage necessary to compensate the energy loss by synchrotron radiation ; c. resulting accelerating voltage. calculation was performed under the assumption of a negative premagnetization of the synchrotron magnets of 400 G and an electron energy of 25 MeV at injection. The peak power necessary to accelerate a circulating 250 current of 30 mA to 2.3 GeV is about 60 kW with two cavities in operation and 70 kW with one cavity only, that is when beam loading effects are taken into con- sideration and a cavity shunt impedance of 10 Mf2/m is a,;sumed. The average power has to be about 20 kW, dependent on the chosen value of the synchronous phase at injection.

5.3. THE INFLUENCE OF BEAM LOADING The circulating electron beam induces a considerable 0 5 t [mscc] = 10- voltage with a variable time dependent phase ~b(t) in the Fig. 15. Variation of accelerating voltage Urf vs time t for several high Q cavities. By this way a high circulating beam final energies between 0.4 and 2.3 GeV. The synchronous phase is current changes phase and amplitude of the resulting shifted from 20 ° to 45 ° . 18 K.H. ALTHOFF et a[.

FM - PROGRAM GENERATOR

TRANSMISSION LINE g POWER TRANSMITTER I ACCELERATION UNIT _ [__ [ AMPLITUDE I REGULATION CONTROL FEEDBACK 'i

AM - PROGRAM a. curve 1 : signal of scintillation counter installed near the target; GENERATOR curve 2: field strength without beam; curve 3: field strength with beam loading. The reduction of the rf field strength during Fig. 16. Schematic diagram of the regulation system to reduce the acceleration is clearly visible. Hor. : 1 ms/div. influence of beam loading. of the beam current Ib during the first turns after in- jection, when the beam induced voltage is built up, by analytical methods. Furthermore the differential equation of the synchrotron oscillation strongly depends on the solution of this differential equation. Therefore an exact analytical treatment of the influence of beam loading on the behaviour of particles in the synchrotron is difficult. Several approximation methods, however, have been developed to describe the beam loading effect ~2 - 14). Computer calculations are now carried out to study the particle dynamics in the machine during the first b. rf field strength during the first turns. The signals are the same turns after injection. These numerical calculations take as in fig. 17a, only the time scale is magnified: 50 ffs/div. During into consideration the variation of the phase and the whole time in which the beam induced voltage is growing up, amplitude of the accelerating voltage by the voltage, the scintillation counter indicates electron losses. which is induced by the 500 MHz component of the circulating bunched electron beam. These calculations seem to deliver the most realistic approximation of the particle behaviour during the first run. Considerable experimental effort has been made to avoid the particle losses, which are due to the beam loading effect, and we succeeded to overcome this diffi- culty in spite of a high circulating current. The first possibility to reduce the influence of beam loading was to shift the transmitter frequency during a short time after injection to a value slightly different from the cavity resonant frequency. Since the circulating electron

c. rf field strength (curve 2) during the acceleration cycle with Fig. 17. Oscilloscope traces of the rf-field strength during an amplitude regulation. No significant variation of the accelerating acceleration cycle. voltage by beam loading is visible. Hor. : 1 ms/div. 2.5 GeV ELECTRON SYNCHROTRON 19 beam accepts the transmitter frequency, the voltage There are two acceleration stations in the orbit, each induced in the cavities is lowered by a frequency modu- consisting of three tightly coupled resonators excited in latiion at injection. the TMol o mode. Fig. 18 gives a cross section of such "['he second way to reduce the influence of beam an accelerating station. In addition, the field distribu- loading, as we tried, is to compensate the breakdown of tion along the radius r, the azimuth ~o and the cavity rf voltage during the first turns after injection by a axis z is given in that picture. regulation of the rf amplitude4). A voltage proportional The length of each resonating chamber is 30 cm. to the cavity voltage is fed back to the driver station and Therefore the electron takes one half of a radio- compared electronically to the desired value (fig. 16). frequency cycle to pass a chamber. If the phase shift Any difference is compensated by a fast regulation of between adjacent cavities is 7z, the electrons are ac- the; klystron driving power. Fig. 17a shows the cavity celerated in each of the three resonating chambers. An voltage during an acceleration cycle without any regu- acceleration unit consisting of three coupled resonators lation. The breakdown of rf voltage by the circulating is represented by a 3-circuit bandpass, which has two bunched beam is clearly visible. Fig. 17b shows the resonant frequencies. The ~-mode resonance, in which voltage breakdown in a magnified time scale. As can be the fields in adjacent chambers are in antiphase, occurs seen from the scintillation counter (curve 1), during the at a higher frequency than does the O-mode resonance. time period in which the beam induced voltage is built Before installation in the synchrotron ring each resona- up (curve 3), electrons are lost from the beam. This ting chamber was mechanically tuned to the desired reduction of beam intensity is avoided with the aid of resonant frequency. amplitude regulation as can be seen from fig. 17c. The Such a cavity has an unloaded Q of about 3 x 10 4. accelerating voltage (curve 2) is held equal to the theo- The shunt impedance was measured according to the retical value. The multiplier signal (curve 1) indicates Slater-perturbation method16). The measurement yield- the rise of the accelerated beam intensity. Using this ed a value of about 10 Mf2/m. By the same method the amplitude regulation system we can increase the in- field distribution along the resonator axis was measured. tensity of the circulating electron beam significantly. During operation the cavities can be tuned by water- cooled variable tuner stubs driven by servo motors. The 5.4. ACCELERATION UNITS cavities are always tuned to a minimum of reflected The accelerating field is generated in cylindrical power. They are cooled with water, the temperature of cavity resonators, developed and constructed in a spe- which is held constant within about 0.2°C. This is cial electroforming process by DESY, Hamburg'S). necessary because a temperature variation of 2 ° C shifts

Ezz"

Fig. 18. Field distribution along the radius r, the azimuth 99 and the axis z of an acceleration unit excited in the TM010-mode. 20 K.H. ALTHOFF et aI. the resonant frequency by about one bandwidth. transconductors. Fig. 19 shows a simplified block dia- Coupling loops in each cavity control phase and gram of the AM-program generator. amplitude of the radio-frequency. To avoid ionization in the electric field no oil 5.5.2. FM-program generator diffusion pumps were applied. At each acceleration unit It is possible to vary the transmitter frequency for a two ion getter pumps keep the vacuum pressure below short time after injection according to section 1.3. This 10-6Torr. During the test measurements the field is achieved with the aid of a FM-program generator, strength in the cavities was raised up to 1.2 MV/m which makes it possible to vary the degree and the time without electrical discharge. duration of that frequency modulation. The frequency can be shifted up to about 50 kHz. 5.5. TRANSMITTER A conventional television transmitter built by Tele- 5.5.3. Limiter funken, Berlin, with a modified power amplifier is used In the case of sudden load changes, as for instance at as rf power source4). The 50 W driver station is located different beam intensities, the voltage in the resonators in the synchrotron control room. the power amplifier is must be limited to prevent breakdown by gas discharge. placed in the middle of the synchrotron ring. In the This voltage limitation is performed by a comparison following we will give a short description of some of the between the resonator voltage and the driver power. In elements, which were especially built in order to be the case of a cavity voltage being too high, the driver able to use the commercial Telefunken television trans- power is limited, which is necessary especially at a high mitter as a suitable rf power source for the synchrotron. circulating current. 5.5.1. AM-program generator 5.5.4. Amplitude regulation The amplitude modulation of the electric field: To compensate for the breakdown of rf field strength Urr = C o + Cl(dB/dt)+ C2 B4, by beam loading, a comparison is made in a differential according to the program described in section 1.2, was amplifier between the real voltage in the cavity and the achieved by means of a program generator built of desired voltage as calculated by the program generator. analog computer elements17). This program generator Any difference between these voltages is reduced by calculates the modulation voltage from the voltage regulation of the driving power. This 1egulation be- given by a back leg winding in one of the synchrotron comes very important, as mentioned above, for a high magnets with the aid of integrators and quadratic capture efficiency.

_ U ~//= tU~ B -t t ~t 1tOms t

I I

TO TRANS- TER MAGNET WINDING WINDING VARIABLEDO. i[£tretcher i t PEAKING STRIP Fig. 19. Schematic diagram of the AM-program generator built with the aid of analog computer elements. 2.5 GeV ELECTRON SYNCHROTRON 21

the cavities serves to avoid coaxial coupling to the evacuated cavities. The short waveguide is excited in the TEl0 mode by a doorknob-type transition. The wave- guide contains a variable double stub phase shifter to equalize the phases of the two cavities. The rf power is fed into the medium chamber of an acceleration unit through an exchangable copper coupling iris, which contains a 99.5% alumina ceramic disk serving as a vacuum window.

5.6. FERRITE ISOLATOR The transmission line includes a large water-cooled ferrite isolator, built by Raytheon, which protects the power klystron in the case of an electrical breakdown in the load. Additionally, the isolator decouples the Fig. 20. The ceramic vacuum chamber in the magnet gap narrow-band accelerating unit from the klystron output (photo K. KiJffner). cavity. Rf power, which travels backward when the intensity of the circulating beam changes is absorbed in 5.5.5. Power klystron the isolator. The isolation value is about 11 dB, the For a power amplifier we use the klystron F 2008 insertion loss is less than 0.4 dB. The ferrite isolator spez. fabricated by CSF, Paris. With the small band- essentially contributes to a stable operation of the rf width necessary, the power amplification is about 70 d B. system. Measurement of the output power yielded a peak power of 100 kW and an average power of 40 kW. The 6. Vacuum system* collector voltage is 26 kV; the dc power supply can The vacuum system of the synchrotron has been con- deliver up to 170 kW dc input power. structed with consideration of the following conditions: The total gas pressure should be below 10 -6 Torr and 5.56. Transmission line in the if-cavities the partial pressure of hydrocarbons is The rf power is transmitted to the cavities via 3½" to become as small as possible in consideration of the coaxial line with a characteristic impedance Z = 50 f2. ceramic windows for the rf-energy input. In the gap of If both resonators are to be operated, a coaxial trans- the magnets the chamber must be free of magnetic and formator, transforming the characteristic impedance of good conductive material. All materials and elements the coaxial line from 50 f2 to 25 f2, is installed before the should persist in their original mechanical and electrical T-junction. The coaxial line contains a compensated properties during the subjection of radiation and under variable trombone line stretcher in order to transform the influence of agressive products arising from the the load impedance to the desired value. Directional operation of the synchrotron. couplers in connection with peak voltage measuring In the gap of each of the 12 magnets, 8 straight devices serve to measure forward and reflected peak ceramic tubes each 50 cm long and consisting of porce- power. lain with 50~o AIEO3 are glued together very precisely Special simple capacitive ~2l matching units similar to with Araldit HV 123 (fig. 20). The inner size is ap- those used in matching transmitting antennas were proximately 36 x 120 mm 2 and coated by a thin film of developed which enabled us to match the cavities to the graphite, normally used for television tubes. To achieve transmission line in a very easy manner. The coaxial line a low outgassing rate this surface is dried by heating to was interrupted before the matching units and the about 200 ° C. The graphite film is connected to ground VSWR of the cavity coupled to the transmission line potential by a 50 f2 resistor and can be used to detect was measured by means of a slotted line. The VSWR electrons hitting the chamber. Each end of the ceramic was then minimized by tuning the matching units. No chamber is cemented with Araldit HV 123 to a stainless tedious mechanical handling of the coupling iris was steel flange 6 mm thick. One vacuum chamber com- necessary in order to match the shunt resistance of the prising 1/12 of the circle has a length of nearly 470 cm. acceleration unit to the characteristic impedance of the Its weight is about 50 pound. transmission line. Between the magnets there are stainless steel tubes A waveguide (type WR 1800) located shortly before * F. J. Schittko, B. Langenbeck. 22 K.H. ALTHOFF et al. each joint with a vacion pump 250 l/sec. These tubes are connected by bellows to the end flanges of the ceramic chamber. Most of the separable connections are made by clamp-flanges and sealed by indium wire. ~ AMPLIFIER TOCONTROL-ROOM The whole system works with five independent vacuum units, the injection path and the four quarters of the synchrotron ling, separated by valves 150 mm dia. (VAT-Pendelschieber). Each of these has an assembly consisting of a roughing HIGH }J RINGCORE pump (10 m3/h) with a turbomolecular pump (250 Fig. 21. Schematic diagram of beam monitor. l/sec-TVP 900); that means, we have a vacuum system free of oil. This assembly evacuates the parts of the At first a height survey had been made by means of a vacuum system and the vacion pumps which are baked Wild N 3 telescope level, and optical targets placed on by internal heater elements to gas pressure of 10 -s the magnet shoulders. This survey could be performed Torr. within an accuracy of ± 0.02 ram. Having reached this pressure, heating is stopped and The radial survey was executed with the aid of 11 m the vacion pumps are started. If we measure less than long invar steel tapes. Such a tape contains a 10 cm 10 .6 Torr in the ion pumps, the mechanical pump long ram-scale near its ends. The tape had been fixed at assembly is separated by an electropneumatical valve the central pillar on a special movable assembly in a which is driven by a pipe system with cleaned nitrogen, way so that the ram-scale was located at the centre of normally used for filling the vacuum chamber before the machine. The other end of the tapes was loaded with opening. The assembly has measuring equipment per- a 10 kg weight to achieve a defined dip. By means of a mitting it to run automaticly. For security there is an microscope and a target index, mountable in a refe- alkali halide leak detecting gauge connected to the rence groove on each magnet, the magnets were radially vacuum side of the roughing pump. adjusted within an accuracy of about _+ 0.03 mm. The In steady state the system works now with a pressure tapes were calibrated up to 0.02 mm, under the same of about 10 -7 Torr in the chamber and 10 -8 Tort in the operating conditions, at the Phys.-Techn.-Bundes- cavities. anstalt Braunschweig. The azimuthal location of the magnets was adjusted 7. Adjustment* by means of a Kern-DKM 3-theodolite, the con- The adjustment of the synchrotron magnets is made struction of which was changed for measurements at by a simple conventional method from a centerpoint short distances. The accuracy achievable in this azimu- monument, which is a concrete pillar extending ap- thal survey was _+ 0.1 mm. proximately 6 m into the undisturbed ground. The top The position of the linac and the injection elements of the pillar bears a steel plate containing a cylindrical including the inftector was also determined from the hole to receive a 12.00 mm diameter cylindrical stub. centerpoint monument with the aid of invar tape and The axis of this stub defines the centre of the synchro- telescopic level. tron. A theodolite can be fixed at this centerpoint with an accuracy of _+ 0.02 ram. Instead of the theodolite, 8. Beam monitors* an optical target can be fixed in the central cylindrical The intensity of the circulating electron beam is hole. measured with the aid of pick up coils. The beam All the magnets were adjusted from this centerpoint monitor (fig. 21) consists of a ring-core of high per- monument. After the first preliminary adjustment of the meability (/~ = 1500) bearing several turns of copper-lac 48 steel plates bearing the four hydraulic jacks of each wire (1 mm thickness). magnet, the magnets with the blocks exactly adjusted The beam passing the aperture of the core, induces a on the girder, could be installed within an accuracy of voltage between the ends of the coil, which is propor- about _+ 2 mm. When the installation of the 12 magnets tional to the time derivation of the magnetic flux in the had been completed, the final survey was made. By core. This signal is amplified and fed to the control- means of a sensitive hydrometer the machined surface room. of each magnet girder was adjusted in order to be Two different types of beam monitors are installed in completely horizontal. All further measurements were * H. E. Stier. made from the centerpoint monument. ¢ J. Drees, H. D. Wehner. 2.5 GeV ELECTRON SYNCHROTRON 23

6]X [crn] envelope of ext rocted beorr t Mll M12 M1 M2 4 + t~Bo "1/2 ABe ~ -1/2 AB°

2 b orb'l ~ _ 0

-2 F o .~ D F field-gradient D F -4 f/ lrn -6 current strip septurn magnet ~

Fig. 22. The electron extraction system, a: Trajectory of an electron two turns before jumping over the current strip; b: One turn before jumping. the ring. In the first type 20 turns are wound around the 10. External electron beam* core. The upper cut-off frequency of this system is 10.1. METHOD 6.5 MHz, its lower cut-off frequency is 12 kHz. Using For slow extraction of the electron beam we use the this device it is possible to measure the intensity during resonance method. This method, which proved success- the first subsequent 50 turns. ful for the first time at the Cambridge Electron Accel- The second type bears 360 turns around the ring core. erator~8), is able to produce an electron beam of high The lower cut-off frequency is 5 Hz so that in using this efficiency (more than 50°/o), of small emittance (about device the intensity of the beam can be observed during the same size as that of the circulating beam) and with a the,, whole acceleration time. Beyond that, one can exam- variable spill-out time (up to one msec). ine by this equipment the decrease of the intensity during The principle of the resonance method is the the', ejection. The upper cut-off frequency is 100 kHz. following: Sometimes a coupling loop in the second unexcited The number Qr of horizontal betatron oscillations is and highly detuned cavity is used as a beam intensity brought into Qr = 3.5, the half integer resonance, by monitor. Because of the relative long rise time of the a focusing element. This leads to a rapid increase of cavity (10/lsec), this device can only be used to observe betatron amplitudes so that the electrons can enter slow variations of beam intensity. the gap of a magnet in order to be bent out of the At the moment we are trying to develop an intensity machine. independent beam position monitor using if-coupling The special extraction system of the Bonn electron loops outside the vacuum chamber. synchrotron, which proved to be the best according to calculations, is shown in fig. 22 and operates in the 9. Photon beams fol!owing way: Several photon beams are used for experiments. Before the end of the acceleration cycle, the electron Rotating tungsten rods driven by small electromotors beam is brought by a closed orbit distortion into the rurming synchronous with the synchrotron magnet field of a current strip which is situated on the inner side excitation are operated as internal targets. Small varia- of straight section number 12, parallel to the direction tions of the horizontal angle of the emerging photon of the pa rticles. The direction of the current is such that beam can easily be achieved by a azimuthal shift of the the electrons are deflected radially outward. The closed rotating target in the ceramic vacuum chamber. The orbit distortion has the form of a bump and is made by phase difference between target motor and magnet exciting backleg-windings on several synchrotron excitation can be varied. Thus the energy of the brems- magnets. strahlung beam is normally chosen by changing this Passing the non-linear field of the current-strip, the phase difference. The photon beams leave the ceramic horizontal betatron frequencies of electrons are shifted chamber through small aluminium windows. towards Q, = 3.5, corresponding to the distance of the A spill time of 1 ms at a repetition rate of 50 Hz can electrons from the current strip. At the same time, easily be achieved. * E. Weil3e, G. v. Holtey, C. Pieper. 24 K.H. ALTHOFF eta/.

because of the non-linearity of the magnetic field, the It contains on the right hand side the synchrotron horizontal dimension of the whole beam is enlarged. oscillation, the distortion of the closed orbit and the When the closed-orbit bump has brought the elec- non-linear field of the current strip. Because of the tron beam near enough to the current strip, electrons last term G(s,x), which contains all parameters of come very quickly into resonance. They now are bent the current strip, we had to apply a Monte-Carlo- always at the same phase and obtain a sufficien! in- method: crease of betatron amplitudes per revolution so that We took electrons out of a phase area and a mo- most of them are able to jump over the current strip. mentum interval which were calculated from the values On the other side of the strip the electrons are de- at time of injection under consideration of adiabatic flected away from the closed orbit and are well sepa- dampling of betatron and synchrotron oscillation. The rated from the aperture of the circulating beam in the distribution of particles in that phase area and mo- next straight section number 1. mentum interval was chosen to be a gaussian. The Here is situated a septum magnet which bends the elec- number of the horizontal and vertical betatron oscilla- trons radially outwards so that they cross magnet M1 tions was 3.4 per turn. We then calculated with a butdo not come into the focusing region of magnet M2. computer, for several thousand electrons, the horizon- The electrons leave the synchrotron through an exit tal and vertical coordinates at the position of the current window of a tank in straight section 2. strip, one turn after another until they jumped over the The influence of the current strip on the vertical current strip or struck against it. betatron frequency is, by reason of its special field As a first result of such a calculation we could follow configuration and the small size of the vertical fl-func- one electron during the whole process of coming into tion, not so strong that the resonance at 3.3 is reached. resonance, as is shown in fig. 23. Here is plotted the distance of the electron from the ideal closed orbit at 10.2. CALCULATIONS the position of the current strip as a function of the For the design of the extraction system and the study number of revolutions. All points belonging to an even of the resonance process we made detailed calculations3). and to an odd number of revolutions are connected by These calculations were carried out using linear a curve. The two curves describe oscillations whose theory taking into account all points which could be frequency is strongly correlated to the betatron fre- of importance to the extraction process. We could quency Qr of the electron. The frequency of the curves neglect a coupling between horizontal and vertical becomes smaller as Q, grows. At resonance (Q, = 3.5) oscillations and consequences of synchrotron radiation. oscillation ceases and the curve grows exponentially. Thus we had to solve the following equation of motion, The two curves are oscillating around a closed orbit which in the horizontal case is: which is the sum of the bump and the momentum deviation. In detail one can see that the growth of Qr I Ap 1 AB and the jumping over the current strip takes place x"(s)+ {(I -n)Ip]}x(s) =~ -Po (s)! -)7 (s) + predominantlyTat the moment of greatest negative

x ~ lcml -Q6 .k=odd ~1 .- ," t- 1-\ /\ :\ rx ~. , - tx /-(do~d-orblt -u-zl~ ~'..L.V-~-~---~'-..~ ~ \/', :, ,-, A :~ :, :\/".~ bump

o,L/,.:,., v,, : , ..... 061 ...... k x Ccm] 2O ~ 60 80 100 I ,,j .... -3.0 x [cm] / / /f ~ \ //kz-even -201 ."-", _ ,~. I \ "- ,. F\, ~,d,,bi, : "L4------:Y'/----~. / \ I \ I "\ ...... -, ,"\ , .... / \ ~__i__ ~ .....\ I \ I:'\ I-\ I \\ I _.-k-/------b~ / / / \, 1 \: \.// \ / "k / \ / "~./ \~. ,/ \ ./ \./1 \\ / ~\ / \ ./ \kk=Odd --tO \

2850 2900 2950 3000 Fig. 23. Development of the radial amplitude of an electron trajectory at the position of the current strip during the extraction process. K = number of revolutions. The electron jumps over the current strip at K = 3028. 2.5 GeV ELECTRON SYNCHROTRON 25

x' ~_m radJ

//•K=2.800

y / K=2000 q=_020/o, K=I500 / 2- q=+OU

/K--2oo q=+0.9%0 x rmm] .... i , , ,-- , r • |0 15 2b K=800 q-- 0 q=-1'/** ~_ 1.

Fig. 24. Development of the phase ,rea at the current strip during the extraction process. K = number of revolutions, q = ZIp/po momentum deviation due to synchrotron oscillations. momentum deviation. This means a momentum focus- As another result, we could follow the development ing: effect on the extracted beam by the synchrotron of the phase area of the whole electron beam (fig. 24). oscillations. At the beginning, one has an ellipse which later on blows up and finally takes up a wide and complicated efficiency phase area form. This blowing up is of importance for long spill- i~.~rad] r,v.] out times. 80 In addition to the mechanism of resonance, we in- quired systematically into the properties of the extract- ed electron beam as a function of all the parameters for 60- 6 efficie //~~ernittance optimalisation of the system. 4O It has been found that the current and height of the current strip are of great influence on efficiency and phase area, as can be seen in fig. 25. A high efficiency is 2O ,..crn coupled with a large phase area. The spill-out time is mainly given by the rate of I000 2000 30o0 curr~t[A] moving the bump towards the current strip. The energy spread caused by synchrotron oscillations efficiency ~1 phase area leads to an increased phase area caused by dispersion of E°l.J ~ Emm rnrod] the path through magnet M12, septum magnet and 805 8 magnet M1. This can be eliminated at the target of an experiment by a beam transport system. 60--6 ~cy Parameters of the extraction system are given in table 1. 40- At a point which is far enough from magnet M2 in order to set up the first element of a beam transport

20- emillance system, we calculated the properties of the extracted current beam, as given in table 4. 2500A 10.3. ELEMENTS OF THE EXTRACTION SYSTEM 14 16 1B 20 22 24 height [cm] 10.3.1. Current strip and septum magnet Fig. 25. Calculated efficiency and emittance as a function of current and height of the current strip. A picture of the completely mounted cerurt ntsrip is 26 K.H. ALTHOFF et al. TABLE4

Emittance for 95 ~ of the electrons:

,Ip/po = 0: horizontally ~.0.7 ram. mrad, vertically ~. 0.3 ram. mrad. Ap/po = 10-3: horizontally ~. 2.0 ram. mrad, vertically ~.0.3 mm-mrad. Width for the whole beam horizontally 14 ram, vertically 3 ram. Iransformer Divergence of the whole beam horizontally 5 mrad, vertically 1 mrad. ~ shunt Efficiency 68 ~. load Fig. 27. Schematic circuit of the pulsers for current strip, septum given in fig. 26. The construction is the same as magnet and closed-orbit distortion. described by Burrtg). The current strip is machined out of a solid copper 10.3.2. Beam bump technique bar. Thin walled helical tubes of stainless steel are brazed The closed orbit distortion is produced at the end of onto the segments and provide water cooling. The the accelerating period by pulsing the back-leg windings current is restricted to the strip by many vertical slits. of four magnets. In order to avoid a disturbance of the The electrical circuit, formed in parallel by the tubes, is guiding field, the four pulsed magnets have been com- by reason of the bad conductivity of stainless steel and bined into two groups so that the induced voltages are the longer path for the current, less than 2%, making a compensated in each of them. negligible field deviation. For a peak current of 3500 A We have found best the following distortions (in (i.e. about 90 A/mm 2 current density) the temperature relative units)" remains under 70 ° C. Magnet number 11:12:2:3 = 1.0:0.5:-0.5:- 1.0. The septum magnet is a C-type with a laminated iron- Thus the backleg-windings were made, under con- core and is excited by a single turn coil within the sideration of the desired current pulse shape, with 4 aperture. The fringe-field in front of the septum was turns on magnets M 11 and M3 and 2 turns in magnets measured to 1 °/00. The conductor for the coil is pierced 12 and 2. for water cooling. They were connected in series. AB/Bo in the magnets In order to reduce heating and to prevent an in- during the extraction was. at maximum, 0.73°,/o. fluence on the electrons at injection, the current strip and septum magnet are pulsed. The pulses have a one 10.3.3. The pulsers millisecond long flat top. For generation of the high current square wave pulses Both elements are mounted on flanges of vacuum for the current strip, septum magnet and backleg- tanks and are adjustable by remote control. windings we used lumped lines switched by high power silicon controlled rectifiers, as proposed by G. A. Voss for the CEA extraction system. The basic circuit is drawn in fig. 27. The current step-up transformers needed for the current strip and septum magnet are situated close to their load, while the other components are located outside of the shielding tunnel. For the beam bump the current is fed directly into the backleg- windings. The lumped lines are charged by voltage regulated power supplies.

10.3.4. Monitors The position and width of the extracted electron beam were televised onto light screens which were situated at several places of interest. Fig. 26. The current strip. The cooling water runs through the For exact measurements of intensity distributions we helical tubes above and below the strip. exposed silver phosphate glasses2°). 2.5GeV ELECTRON SYNCHROTRON 27

28. The extracted beam and also the synchrotron radiation observed on a screen at the exit window, division width: 5 nnm.

The charge of the extracted beam was measured by a The first extraction was made at an energy of 1 GeV. lar,ge 80 radiation length Faraday cup of the DESY- The same results were found up to 2 GeV. Only the type’l). current through the current strip had to be raised more than linearly for higher energies. This was in agreement 10.3.5. Beam transport system with the deviations of the betatron frequencies from The 18 m beam transport system consists of 2 hori- Q, = 3.4, which were calculated from the field measure- zontal focusing quadrupoles, 1 bending magnet and 1 ments of the synchrotron magnets. At 2 GeV the vertical focusing quadrupole. It makes the position of horizontal betatron frequency was calculated to be the beam at the target independent of the energy, which Q, = 3.36, that means a 40% higher current. Thus at at a spill-out time of 1 msec has a spread of 0.5%. 2 GeV the current strip with 3500 A has reached its All 3 quadrupoles are of the same type having a limit. For higher energies, quadrupoles will be installed length of 0.5 m, an aperture of 45.5 cm dia. and operate in the machine, which can also prevent possible with a focal length of 2.25 m up to 2.5 GeV. The homo- difficulties with the vertical resonance at Q, = 3.3. geneous bending magnet is 1.25 m long, has an aperture The efficiency, measured by the large Faraday cup, of 19 cm x 3.5 cm and can operate up to 18 kG. was 60%. First rough measurements of the emittance were 10.4. MEASUREMENTSON THE EXTRACTEDELECTRON made by focusing the beam with one quadrupole onto BEAM the light screens. The evaluation of the light spots led After installing all elements and adjusting all param- to an emittance of the whole beam in the horizontal eters to the calculated values, the electrons were plane of about n.4.0 mmemrad and in the vertical of extracted out of the machine at the first attempt. about 71.1 mm.rad. After elimination of the dispersion In the first test, the electron beam was observed on a we obtain for Ap/p, = 0 a horizontal emittance of thin light screen located in front of the septum magnet about ~‘1.5 mm. mrad. in straight section 1 and observed through a glass The measured position of the principle axes of the window in the tank. The light spot appeared at the phase ellipse agreed with calculations. This was of predicted position. After pulsing the septum magnet, importance for the already prepared transport system. the beam left the machine through the exit window also Spill-out times were observed by a multiplier signal at the predicted position (fig. 28). up to 1 msec. 28 K.H. ALTHOFF et al.

11.3. SHIELDING cootir~ tower [ For radiation shielding, the magnet ring is inside a tunnel, 3 m wide and 2.16 m high. It is built of movable heavy concrete blocks with a density of about 3.5 g/cm 3. Since most of the radiation of the machine propagates tangentially to the circumference, the outer wall of the tunnel consists of 1 m thick blocks, while the inner wall and the top of the tunnel are built of 0.5 m thick blocks. l.inac injection The side-wall blocks are 1.08 m high and 1 m wide, devices having a convex and a concave side opposite to each other, so that by arranging the circular walls no straight cracks will arise. The tunnel roof consists of trapezoidal 4.10 m long blocks, 0.5 m wide on an average. There are three labyrinths providing a passageway to the inner part of the tunnel. The outer wall is provided with several channels, or beam-pipes, to admit collimated beams demineral.ized into the experimental areas. 12. First half year of operation synchrotron magnets, RF - transmitter, experiment(at magnets On the morning of March, 8 th 1967 the vacuum ejection devices chamber had been completely installed and had been pumped out during the day. In the evening of that day Fig. 29. Block diagram of the water cooling system. we obtained the first turns with adc excitation of the magnet. The mean life time of the electrons in the well 11. Building and technical facilities tuned machine without acceleration came out to be 11.1. SURVEY ON THE ACCELERATOR INSTALLATION 10 ms. Two hours later the magnets were energized by The main hall, including the magnet ring and the ac and dc, the cavities were tuned up and we obtained experimental floor has an area of 2500 m 2. The hall is the first accelerations. The peak circulating current had shielded against radiation from the linac-room by been measured by a current transformer to be 2 mA at means of a 1 m thick removable shielding-wall. a final energy of 2.3 GeV. The first difficulty we rec- The 24 pillars bearing the 12 synchrotron magnets ognized that night arose from the difference between protrude from a 1 m deep, 2 m wide circular cable duct, the frequencies of the magnet resonance circuit and the from which other smaller ducts radiate to the center of line frequency. The linac beam intensity oscillated with the ring and the other parts of the hall. These cable this beat frequency of about 0.5 Hz. To avoid this ducts serve for the installation of the power and control problem the magnet circuit has to be tuned to the line cables of the accelerator and the experiments, as well as frequency by changing the capacity of the resonance for the cooling water installation. circuit. Since the resonance frequency of this circuit The large capacitor bank and the 60-ton-choke, changes with time due to temperature effects in the belonging to the magnet power supply, are mounted in capacitors we will use a special power supply for the the rear of the Institute for Experimental Physics, partly linac driven synchronously with the power supplies of on top of the 500-MeV-Synchrotron hall. the magnets. At a linac intensity of only 40 mA peak current, the 11.2. COOLING or TaF ACCELEaATOR accelerated beam current in the synchrotron could be The coils of the twelve magnets and rf cavities are increased up to 12 mA. When the electrons at this high cooled by demineralized water. The water has a con- intensity were accelerated up to 1 GeV and higher, ductivity of less than 3" 106 ohm/cm; it flows in rubber repeated break-downs of the electrical field in the plated steel pipes in a closed circuit that includes two electrostatic inflector occurred due to the synchrotron heat exchangers. radiation, which prohibited a stable operation. There- A block diagram of the water cooling plant is given fore we decided to replace the electrostatic inflector by in fig. 29. In addition forced air cooling is installed in a septum magnet with similar optical properties. The the ring tunnel. septum magnet we installed is 40 cm long and has a 2.5 GeV ELECTRON SYNCHROTRON 29 bending radius of 5 m at a magnetic field of 167 G. It to 60 mA at 2.1 GeV. Until now a normal operation is excited by dc current. With this magnet the syn- value is 40 mA. chlrotron is working now quite well. During the next month the linac intensity had been 13. First experiments increased to 180 mA peak current. Nevertheless the Several experiments using the photon beams or the circulating current could not be changed in this way, external electron beam have been set up (fig. 30). because the beam induced voltage in the cavity grew 1. Photoproduction of K+-mesons. Measurements of larger. This difficulty had partly been overcome by a differential cross section and polarization of the A °- frequency modulation of the rf transmitter. During the hyperon (7 + P~ K+ + A°). first turns the frequency was shifted to a value about The K+-mesons are momentum analyzed in a three 30 kHz higher than the resonance frequency of the section strong focussing magnetic spectrometer. The res- cavity. By this means the circulating current could be olution is Ap/p = 2%, the solid angle Af2 = 1 msterad. raised to 25 mA. The velocity is measured in three threshold Cerenkov Since the beam intensity was still limited by beam counters and one differential gas Cerenkov counter. loading we tried an amplitude regulation system for the The decay of the A ° will be observed in spark chambers. cavity voltage in order to reduce the influence of this 2. Electroproduction of pions (e- +p--* e- +p+ n °, eltbct. We succeeded in getting a circulating current up e- +p~e- +n+n+).

/11 o s ~o,. jJ /16 / / /

1 I / ]L /

F f e +p+rr ° I e+p ~ ~L ..... + / \

12 7

8-__! !

E -- 1 , Fig. 30. Plan view of the 2.5 GeV electron synchrotron. 1. Linac; 2. F-section; 3. D-section; 4. acceleration unit; 5. rf-transmitter; 6. shielding; 7. current strip and septum magnet for the slow ejection; 8. magnet power supply; 9. 500 MeV-synchrotron; 10. Institut fiir Strahlen- und Kernphysik ; 11. Physikalisches Institut; 12. photoproduction of K + and A 0; 13. quantameter; 14. electroproduc- tion experiment; 15. Faraday-cup; 16. experiment on the polarization of the recoil protron; 17. pair spectrometer; 18. spectroscopic measurements using the synchrotron radiation. 30 K.H. ALTHOFF et al.

The electron magnetic spectrometer, consisting of Staatshochbauamt der Universitfit Bonn, especially to three quadrupole magnets and one bending magnet, is Reg.-Oberbaudirektor Wernicke. We thank Bauing. J. especially designed for good resolution of the center Koch for his efforts in building the accelerator hall in of mass energy of the pion nucleon system. (Ap/p = a very short time. We wish to express our thanks to O. 2.5°/0, A f2 = 0.6 msterad, A W = 10 MeV). Wieland for his valuable effort during the construction A telescope for detecting pions and protons is under of the synchrotron and the power supply. Of particular construction. significance was the help of H. Finken, H. Peschel and 3. Polarization measurement of the recoil protonfrom Mr. Offermanns (KFA Jtilich) in solving the different the reaction 7+p~ u°+p. mechanical problems. A special word of thank is given The protons are detected by a range telescope and a to K. Ktiffner for his skilful work during the con- magnetic spectrometer at higher energies in connection struction of the ceramic vacuum chamber and to H. with a counter hodoscope and several acustical spark Schneider for the construction of the electrostatic in- chambers. Hydrogen, helium and carbon are used as flector and the rotating targets. We wish to express our polarization analyzer in the different energy regions. appreciation to E. Gersing. P. Hbmke, H. O. Schultze, 4. Polarization measurement of the recoil neutron for G. Peschel, P. Haas and G. Bergheim for their aid in the reaction 7+p~ 7r + +n. setting up the machine. We thank E. Hilgermann, H. J. Hydrogen resp. helium is used as an analyzer for the Welt and H. J. K6tting for solving different electronic neutron polarization. The energy and angle of the pions problems. We are grateful to the groups of BBC, is determined in a range telescope and later on in a Siemens and Telefunken for their help during the strong focusing magnetic spectrometer. The neutrons setting up of the assembly. The help of the Institute for are detected in two large plastic scintillators. lnstrumentelle Mathematik in performing the various 5. Photodisintegration of the deuteron (7 + d--* p + n). numerical calculations on the IBM 7090 is gratefully The differential cross section of this reaction and the acknowledged. polarization of the proton is measured. Protons and neutrons are detected in coincidence. A sparkchamber We wish to thank all the other persons contributing system is used. to this project, even if it is not possible to acknowledge 6. Measurements of the photon-bremsspectrum. them all individually. As the bremsstrahlspectrum must be known with high precision for many experiments a conventional References magnetic pair spectrometer is built. The momentumres- 1) j. Drees, Proc. 2 nd Int. Conf. Magnet Technology (Oxford, olution for the electrons and positrons is Ap/p = 0.5%. 1967). ") A. Febel and P. Ltitter, Phys. Inst. d. Univ. Bonn, Report 7. Spectroscopic measurements using the synchrotron 1~06 (1966). radiation. 3) A. Febel and G. v. Holtey, Phys. Inst. d. Univ. Bonn, Report A grazing incidence spectrometer has been built to 1~010 (1966). investigate the synchrotron radiation in the extreme ~) H. E. Stier, Phys. Inst. d. Univ. Bonn, Report 1-027 (1967). 5) K. Berkelman, Cornell University, private communication. ultra-violet region and the absorption in gases and foils. 6) E. D. Courant and H. S. Snyder, Ann. Phys. 3 (1958) 1. v) K. W. Robinson, CEA-report, CEA-62 (1958). We are very much obliged to the state government for 8) p. F. Dahl, G. Parzen and R. Christian, IEEE Trans. Nucl. supporting the new accelerator, which gives us the Sci., NS-12 (1965) 408. possibility to participate actively in high energy physics. 9) K. H. Becks, J. Drees and F. Knorr, Phys. Inst. d. Univ. Bonn, Report 1-028 (to appear). We appreciate gratefully the many aids by DESY and 10) W. Hardt, ETZ-A 84 (1963) 892. the KFA Jtilich in designing and constructing the 11) j. Schwinger, Phys. Rev. 75 (1949) 1912. machine. Our thanks are extended further to the federal 12) K. W. Robinson, CEA-II (1956). government for supporting the equipment for the first 13) C. Passow, DESY 64/4 (1964). 14) K. Leibrecht, Z. angew. Phys. 22 (1967) 433. experiments. Finally, we thank all officials of the univer- 1~) H. Gerke and G. Schaffer, 5th Intern. Congress on microwave sity who were helpful to complete the work without tubes (Paris, 1964). delay. 16) j. C. Slater, Microwave electronics (New York, 1950). 17) H. Netter, Diplomarbeit (Bonn, 1967). It is a pleasure to thank Prof. Dr. H. Rollnik, Dr. K. ~s) F. W. Brasse, G. E. Fischer, M. Fotino and K. W. Robinson, CEAL-1006 (1963). Liibelsmeyer and Dr. K. Schliipmann for their help; i~) p. H. Burr, CEAL-TM-136 (1964). they played an essential part in the design of the basic ~o) V. Eckardt, DESY-Bericht 67/12 (1967). parameters of the machine. We are grateful to the ',1) A. Ladage and H. Pingel, DESY-Bericht 65/12 (1965).