MARCH 1951 SYNCHROCYCLOTRON (DESCRIPTION) 247

n. THE OSCILLATOR AND THE MODULATOR

left to "cool down" between blocks of concretefor a certain length of time (depending on the "life"and the concentration of the radioactive isotypesformed) before being used again.

The materials are usually irradiated until aradioactivity of some tens of millicuries 5) has beengenerated in them. The neutron radiation emanating

5) With phosphorus,for instance, an activity of 150 micro-curiesper (LAh is obtained, thus in this caseour beam of20 IJÀ yields3 millicuriesper hour. The unavoidablelossesof radioactive phosphorus in the chemical processingrequiredbeforethe productcanbe suppliedto the "custom-er" are already discounted in this figure for the yield.(The loss of activity during the chemical processingisnegligible consideringthat the half-value time of thisatom is about 14 days.)

Principles underlying the construction of the oscil-lator

Owing to the phase stabilization of the particlestravelling round in the synchrocyclotron these canbe allowed to describe a large number of revolutionswithout risk of too many being lost. Consequentlyfor a certain final energy a small gain per "loop"suffices. This is the reason why the alternatingvoltage between the dees in a synchrocyclotron ofthe dimensions chosen by us need not be greaterthan, say, 15 kVpeak, whereas for a classical cyclo-tron of the same dimensions a voltage of certainly100 kVpeak would be required between the dees.

Much smallervoltages than 15 kVpeakare also possible,but then there is the objectionthat the permissiblemodulatingfrequencyand thus the strength of the beam current likewisebecomemuch smaller;this willbe reverted to later.

This so much lower voltage has two very import-ant advantages. In the first place there is muchless risk of a breakdown of the insulators anddisruptive discharges through the gas in theaccelerating chamber, and this is an advantagethat can be turned to good account in various ways.A higher gas pressure can be permitted in theaccelerating chamber, thus making it possible to getgreater ion concentrations and therefore morepowerful beam currents. Further, one of the twodees can be earthed (thus the full dee voltage, in-stead of half of it, comes to lie between the otherdee and the earthed chamber walls), thereby makingit much easier to mount the ion source, the target,etc.

from the target (and other parts) is being con-tinuously measured with a boron counter chamber 6)mounted in a fixed position close to the acceleratingchamber (13 in fig. I).The photograph in fig. 4 finally gives a view of

the cyclotron as seen in the direction of the arrowin fig. lb. This and the photograph in fig. 3 alsogive some idea of the extensive accessory apparatusthat is needed. In the foreground on the right offig. 4 is the modulator, which will be discussed indetail in the article now following.

6) This contains a solid or gaseousboron compound fromwhichalphaparticlesare releasedbyneutrons. Theaverageionization current produced by the alpha particles ismeasured.

The second advantage of the lower dee voltageis that a much smaller high-frequency poweris required than is the case with the classical cyclo-tron. If the dee circuit hás a quality factor Q anda capacitance C (it is imagined for a moment asbeing replaced by a lumped L-C circuit), then apower

wepp = ___::__-

2Q(1)

is needed to maintain, in the circuit, an oscillationwith peak voltage V and angular frequency w.(The index E used with w in article A will henceforthbe omitted.) The capacitance C is determinedmainly by constructional requirements; in our caseit could not be kept below 400 pF. Careful construc-tion allowed of the quality factor Q being raisedto 1500. With w = 2:n;X 10.7 X 106 and V= 14,kVwe therefore get P ~ 1.6 kW for the dee circuit.The modulator circuit takes up an equal amount ofpower, so that the high-frequency power requiredfor our synchrocyclotron totals somewhat over3 kW, as compared with about 100 kW for a classic-al cyclotron of the same dimensions.

Infig. 1 a very schematic representation is givenof the oscillatory circuit. It may be regarded as atransmission line with two, large, lumped eapaci-tances at the ends: the (non-earthed) dee in theaccelerating chamber at one end, with the capaci-tance Cl = 400 pF, and at the other end an equallylarge capacitance C2 (the modulator M). Withthe aid of an oscillator the transmission line iscaused to oscillate in such a way that a standing

248 PHILIPS TECHNICAL REVIEW VOL. 12, No. 9

wave occurs with at each end a voltage anti-nodeand in the middle a voltage node (see the sketch).Thus the system forms a half-wave line. Owing tothe influence of the lumped capacitances, however,the transmission line. need not be 15 m long, which

..II:II

to

~ __ --ilV=14kV64&15

Fig. 1. Very schematic representation of the high-frequencyoscillatory circuit of the synchrocyclotron. Lel' Le2 coaxialtransmission line,D dee, M modulator, 0 oscillator valve withcoupling loop. The other dee (Do) being earthed, it can inprinciple be regarded as a narrow strip connected to thewalls of the accelerating chamber and placed opposite theedge of the dec under high tension. Where in future we speakof "the dec" this is to he taken as meaning the non-eartheddec. P is a device for measuring the dee voltage. Below thisdrawing a curve has been plotted showing the voltage distri-bution along the system.

would correspond to the wavelength of 30 m, butonly about 7 m. By periodically varying the capa-citance C2 of the modulator, the frequency is madeto fluctuate periodically between about 10.7 and10.3 Mc/s (variation of 4%, see I).In order to minimize radiation losses the trans-

mission line has been made in the form of two co-axial conductors, the inner one with a diameterof about 15' cm (of course this inner conductoritself is made in the form of a hollow pipe, sincethe high-frequency currents flow only along thesurface), the other one with a (average) diameterof 60 c~. The peculiar configuration of the trans-mission line made it impossible to compute preciselyin advance the dimensions required for the desiredfrequency of lq.7 Mc/s. The dimensions of thetransmission line and the -modulator were there-fore determined experimentally with a model(the dimensions of the dee are governed byother considerations, viz. the need to avoid dis-ruptive discharges and to provide as much spaceas possible for vertical deviations of the travellingparticles), for which purpose a full-scale roughmodel was made which could easily be altered andwith which the resonant frequency was determinedafter every modification. The :final form of themodel served as pattern for the actual construction,

which of course differed from the model in itsprecise finishing: all parts are made of copper,perfectly fitting and soldered with silver, so thatthe dissipative resistance of the whole line is nomore than 0.05 ohm. In this way the quality factor1500 was reached.The necessity of keeping the dissipative resistance

so small is evident from a calculation ofthe currentstrengths occurring in the system: to charge the400 pF capacitance Cl (the non-earthed dee) to apeak voltage of V = 14,000 volts at a frequencyw = 2n·l07 a current is needed with the peak value

I=wC1V,

i.e. around 350 amperes. With the small total dissi- 'pative resistance mentioned, we thus arrive atthe previously mentioned total dissipated powerof somewhat more than 3 kW. Since the streamof particles impinging on the target generatesabout 560 watts (it may be possible to increase thisfigure still further by improving upon the ionsource), we may say that about 15% of the high-frequency power is utilized, which for a cyclotronis a remarkably high "efficiency".

The circuit diagram of the oscillator is given injig. 2. A Meissner circuit has heen used, i.e. theanode circuit of the transmitting valve B is in-ductively coupled to the oscillatory circuit(drawn here as a lumped L-C circuit for the sakeof simplicity), whilst also the reaction to the gridcircuit (excitation) is inductive. The anode and grid

Fig. 2. Circuit of the oscillator. The oscillatory circuit of thesynchrocyclotron can be regarded as an L-C circuit with theinductance mainly concentrated in the transmission line (Le)and the capacitance mainly in the dcc (Cl) and in the modu-lator (C2). This circuit is inductively coupled to the anode andgrid circuits of the transmitting valve B (Meissner circuit).Undesired but unavoidable impedances are shown in brokenlines.

coupling coils each consist of only one loop, placedheside the inner conductor of the transmission linein such a way that the magnetic lines of force ofthe loop envelop the conductor, and vice versa.

MARCH 1951 SYNCHROCYCLOTRON (OSCILLATOR AND MODULATOR) 249

The Meissner circuit hád to he chosen because Set-up -öfthè- ~scillatorit is not possible to connect the inner conductor ofthe transmission line at the point of coupling toa part that has to have a defined (high-frequency)voltage. It is true that at that point, in the middleof the transmission line, there is a voltage node(cf fig. I), but, owing to the periodical variation ofthe capacitance C2 of the modulator, the positionof the node is not fixed but shifted a little to andfro along the line. With the inductive couplingsaccording to fig. 2 no trouble is experienced fromtros.

In order to avoid random oscillations with fre-quencies determined not by the transmission linebut by other L-C circuits occurring in the system,it is necessary that there should he no direct coup-ling between the anode and the grid circuits. There-fore the inductive coupling between the anodeand grid coupling loops was made as weak as possible(see the next section) by giving these loops a suit-able shape. Further, also capacitive coupling viathe valve is prevented as far as possible by adequate-ly earthing the grid of the valve, for high fre-quencies, with the large capacitance Cg (in whichpracticallyall self-inductance is avoided). Thus theexcitation voltage is applied to the cathode, andit would be better to speak of a "cathode couplingloop". The grid bias required is taken from asupply unit with safety device; as soon as the oscil-lations deviate from the right frequency the gridbias is increased to such an extent that the valveis cut off.

The oscillatory condition of the circuit reads, toa first approximation:

where Ml and M2 are the coefficients of mutualinductance between the coupling loops and thetransmission line (fig. 2), r is the circuit resistancealready mentioned, and S is the mutual conductanceof the valve. Since cois fixed and rand Scannothe chosen 'arbitrarily low and high respectively,it was necessary to give (Ml - M2)M2 the highestpossible value. Thus Ml has to he as large as possibleand M2 = !Ml' which is actually the case. Since,however, Ml and M2 could only approximately becalciilated in advance, it was again necessary toresort to the experimental method with a full-scalemodel to see whether (Ml - M2)M2 would be largeenough. By employing two transmitting valvesof the type TA 12/20 connected in parallel, it wasfound possible to get a mutual conductance highenough to fulfil amply the oscillatory condition.

Fig, 3 gives a rather more complete and morerealistic hut geometrically not yet quite exactrepresentation of the oscillator. The two parallel-connected, water-cooled, transmitting valves (hereonly one is drawn) are mounted in an earthed metalcasing divided, by a horizontal partition at grid

!I

,,(/,mmW~~WW7W«/(L7(/P'PP'////LWW'W"«W"_P'LV/L(/,o/(/m(/m/!LV//P'1'/F'((//P'/\P

Lr: :\ Od Le2

64811

Fig. 3. Detailed circuit of the oscillator with the anode couplingloop MI and the cathode coupling loop M2• For B two trans-mitting valves, type TA 12/20, connected in parallel are used.Cg capacitors used for by-passing the grid of each valve toearth, Ca capacitors for by-passing the anode coupling loopto earth, Rg grid resistor, K supply for the cathodes, W coolingwater for the anodes.

(2)

level, into two entirely separated parts, the anodespace and the cathode space. The anode couplingloop, a hollow copper conductor with two right-angled bends, passes out of the casing at the side(in the photograph in fig. 4 it can just be seen)and then runs partly inside a channel of the innerconductor of the transmission line. In this way thedesired tight coupling Ml is obtained, since prac-tically all the magnetic lines of force envelopingone conductor now envelop in this part of the loopalso the other conductor, while at the same timethe lines of force are kept as short as possible. Atits free end the anode loop is earthed for high fre-quencies by means of a circle of vacuum capacitors(Ca in figs 2 ánd 3) which introduce hardly any self-inductance and have a very small loss angle, whilstbeing sufficiently proof against breakdowns towithstand the anode voltage of 12 kV; they can beseen in fig. 4 011 the left. The cooling water for theanode is passed in and returned at this point throughinsulating hoses in the hollow conductor of the loop.

The cathode coupling loop, a copper pipe some-what shorter and thinner than the anode couplingloop, is led out of the earthed casing and approaches

250 PHILlPS TECHNICAL REVIEW VOL. 12, No. 9

Fig. 4. The central part of the oscillator, opened. The two transmitting valves are seenwith the anodes downward. The two hoses for the supply and return of the cooling waterpass through the hollow anode coupling loop led through the wall on the left of the valves.Farther to the left is the end of the anode coupling loop connected to the earthed casingby a circle of vacuum capacitors. The connection is made with wide strips so as to keepthe inductance low. Also the connection to the water supply pipes (disappearing at thetop) is clearly seen. In the lower half of the photograph the undercarriage can be seenthat supports half of the transmission line with the dee and runs on rails, so that theseparts can easily be run out for any servicing work.

the inner conductor from the top. This could notbe drawn in fig. 3, but it is illustrated in the sketchjig. 5 and can partly be seen in the photographof jig. 6. In this way the coupling between anodeand cathode coupling loops is kept small. Withthe same object in view, the legs of the loops, whichcontribute towards the mutual coupling but notto the coupling with the transmission line, havebeen made as short as possible. The useful part ofthe cathode loop is about half the length of theanode loop, in accordance with the requirement

M2 = tM1• The free end of the cathode loop,returning upward, is earthed, and at this pointthe filament voltage for the cathode is carried throughthe hollow conductor of the loop.

The grid of each transmitting valve is by-passedto earth by a circle of capacitors (Cg in figs. 2and 3); here simple mica capacitors could be used,because the requirements as to the breakdownvoltage are not so severe.Although, as stated, the oscillatory condition

was amply fulfilled, in the beginning difficulties

MARCH 1951 SYNCHROCYCLOTRON (OSCILLATOR AND MODULATOR) 251

were still encountered in the starting of the os-cillator. The cause was found to lie in the oscillatorycircuit being overloaded by discharge currentsflowing in the accelerating chamber from the deeto earth and by electrons reaching the dee from theion source. By applying a negative direct voltageof some hundreds of volts to the dee, via a chokeand the inner conductor of the transmission line(the choke may be seen in the middle of fig. 6),the extra load was sufficiently reduced for the systemto be easily brought into oscillation.It was then found possible to generate an alter-

nating voltage up to 25 kVpeak on the dee withoutany trouble from disruptive discharges in the gas,overheating of parts or overloading of the valves. Al-ready with 5 kVpeak, however, the apparatus workedsatisfactorily, yielding a reasonable beam of deuter-ons of 28 MeV. With the normal working voltage of14 kVpeak on the dee the measured input powerto the two valves was 5.7 kW and the anode dissi-pation 2.2 kW, so that the high-frequency outputwas 3.5 kW and the efficiency of the valves 61%.The value of the quality factor Q calculated fromthis output and the given values of V, wand C(see equation (1)) agrees well with the value deter-mined from the width of the resonance curve ofthe oscillatory circuit. The voltage on the deeis measured by means of a diode voltmeter and acapacitive voltage divider (P in fig. I).

64010

Fig. 5. Configurationof the anode couplingloop (M]) and thecathode couplingloop (M2). The inner conductorLe] of thecoaxial transmission line has a channel in whichthe anodecouplingloop is laid and an elevation where the cathodecouplingloop is brought down close to the line. With thisconstruction the mutual coupling between the two loopsis kept as small as possible.

The modulator

It has already been mentioned that the required4% periodical variation of the oscillator frequencyis brought about by a variation of the capacitanceC2 (fig. I). Since approximately all the capacitance

Fig. 6. Viewinsidethe riddle part of the coaxialtransmissionline, showingpart of the anode couplingloop (at the bottom)and part of the cathode couplingloop (top right). At thebottom on the left is the end of the anode couplingloopwithconnections for the circle of earthing capacitors and withwater pipes.

of the oscillatory circuit is lumped in Cl and C2,

the total circuit capacitance is given by:

C

Since in our case C2 Cl it follows that

L1e

CL1C2.1 __

2 C '2

whilst further

L1w L1C_.1 __2 C 'co

so we find for the required relative variation of C2:

Therefore, to reduce co by 4%, we have to increaseC2 by 16%, i.e., .in the case where C2 = 400 pF,by 64 pF.

252 PHILIPS TECHNICAL REVIEW VOL. 12, No. 9

This capacitance variation is effected with theaid of a capacitor consisting of a toothed disc ro-tating between two sets of fixed vanes; see fig. 7.When the teeth of the disc are exactly oppositethe vanes the capacitance is maximum, and whenthey are opposite the' spaces between the vanesthe capacitance is minimum. The disc has 40 teethand makes 50 revolutions per second, so that themodulating frequency of the dee voltage is 2000 cis.

Fig. 7. The modulator comprises a periodically varying capa-citance obtained with a toothed disc T rotating between twosets of fixed vanes U. The difference between maximum andminimum capacitance is 64 pF.

Connected parallel to this capacitor is a second onewith a constant capacitance, such that the totalcapacitance in the minimum position of the rotatingcapacitor is approximately the desired value C2 .

400 pF. This fixed capacitor can be slightly adjustedwith the aid of a servomotor, during operation,to get exact resonance between the frequency ofthe oscillator and the travelling particles in theaccelerating chamber. Resonance can he judgedby the occurrence' of a maximum in the currentflowing to the target, or by the concentration ofneutrons around the accelerating chamber.

Both the rotating capacitor and the fixed capa-citor are mounted in an evacuated drum (10-5 mmHg), thus making it possible for the clearancesbetween the rotating and the fixed capacitor partsto be kept relatively small (6 mm) in 'spite of thehigh voltages prevailing, and thus to obtain therequired large capacitances with a, comparativelysmall plate surface. In this way it has been possibleto keep the modulator small enough to be able towork with the high speed of 3000 r.p.m, and thusreach the high modulating frequency of 2000 cis 1).

1) In this connection it is important that a capacitance varia-tion of only 64pF is needed. If the frequency sweephas tobe much greater than. 4% (thus for a higher final energy ofthe particles, and for protons instead of deuterons; see I),then rotating capacitors of a much higher 'value are needed,and it is then not so easy to reach' high modulating fre-quencies. .

Fromthe circuit diagram offig. 8 it is seen that.one plate of the fixed capacitor (C") and thefixed vanes of the variable capacitor (C') are connec-ted to the inner conductor of the transmission line.The other capacitor plate and the rotatingtoothed disc have to be earthed. The earthingof the rotating disc could be achieved, in principle,by means of a sliding contact (brushes), butthis was found to be unsatisfactory owing tothe too high brush resistance for high-frequencycurrents and on account of arcing arismgfrom the sparking in vacuo. The disc is thereforeearthed by means of a large capacitor consistingof a number of fixed, earthed plates and, rotatingbetween them (with a spacing of 2 mm), a numberof plates mounted on and electrically connectedto the spindle of the rotating capacitor disc C'.The earthing of C' obtained in this way is not, itis true, complete, since the capacitances of C' inseries with those of the earthing capacitor form avoltage divider, so that the rotating disc and therotating plates of the earthing capacitor continueto be at a small high-frequency voltage, of about0.5 kV. This necessitated insulation of the rotatingpart of the modulator in the manner indicated infig. 9. The disc and the plates are mounted on abush turning in two ball bearings about a fixed,water-cooled, shaft insulated with blocks of a cera-mic insulating material ("Kers\ma"). The set-up issuch that the, capacitance between the shaft andearth is small, so as to avoid any appreciable high-frequency currents flowing across the ball bearings.

Le,

64UO

Fig. 8. Circuit of the modulator. The periodically varying. capacitor C' is shunted by a constant capacitor C" for makingup the desired total capacitance. The rotating disc of C' has

• :to be earthed.. ,

These currents could be further reduced in a well-known way, by shunting the capacitance just men-tioned by a coil to form a circuit tuned to 10.7 Mcls,which thus has a maximum impedance for thatfrequency.

The rotating part of the niodulator is driven bya motor also placed in vacuo. It would have beenpossible to drive the modulator from outside the

MARCH 1951 SYNCHROCYCLOTRON (OSCILLATOR AND MODULATOR) 253

vacuum drum but then it would have meant havingto pass through the wall of the drum a motor shaftcontinuously rotating at a speed of 3000 r.p.m.while still maintaining a vacuum of 10-5 mm Hg,which would not have been so easy to ensure.

C'

a

Fig. 9. Construction of the modulator, with the rotatingcapacitor C, the fixed (adjustable) capacitor Cu and theearthing capacitor cm (schematically drawn). Lel and Le2inner and outer conductors of the transmission line; b bushcarrying the rotating capacitor parts and rotating about theshaft a insulated with two sets of "Kersimu" blocks ke; alsomounted on b is the rotor (an iron bush) of the inductionmotor m.

Our solution is very simple, only the three-phasevoltage supply cables for the motor having now tobe led in through the wall of the drum. Therotor of the induction motor has the form of aniron bush slipped onto the bush merrtio ned above,which carries the rotating parts of the modulator.The three-phase stator producing the rotating field,is placed around the bush with a gap of a fewmm in between. An essential condition is that invacuo the motor must not give off too much gasor impurities, and this condition is very difficultto fulfil with field coils of insulated wire. The motorhas therefore been built for a vcry strong current(each of the three phases 185 A, voltage 2.1 V;the starting current is 300 A) and a very smallnumber of windings. Thus the windings couldbe made in the form of free, bare copper tubesthrough which cooling water can flow and whichare mounted on two "Pertinax" plates (the onlyinsulating material contained in the motor). Thisand other details of the construction are shown infig·lO.As the motor has only a small torque and the

system has to start quickly in order to pass smoothlyover some critical speeds, the bearings of the systemhave been very carefully designed. The friction isso small that it takes more than an hour for thesystem to run out to a standstill (it is to be remem-

bered that there is no air resistance). The bearingscan be lubricated from the outside with specialvacuum oil, but this proves to be hardly ever neces-sary.It has been mentioned that the dimensions of

the parts of the oscillator, including the modulator,had to be determined experimentally with modelsin order to make sure of getting the right frequency.Measurements taken show that as a result the fre-quency fluctuates between 10.46 and 10.11 Mc/s.This is a variation of slightly less than 4%, but itproved to be more than sufficient. From the reson-ance condition w = qB/m it follows that at themaximum frequency resonance occurs when B =1.36 Wh/m2. The actual flux density in the middleof the accelerating chamber is a little more thanthis, so that resonance is not reached until thefrequency has dropped slightly. Consequently duringa cycle of the modulating frequency accelerationtakes place roughly between the instants M andN in fig. 11, i.e. during somewhat less than half1/2000 sec, thus about 1/5000 sec.The shape of the frequency variation curve,

which is related to the chosen form of the teethand blades of the rotating capacitor, is not critical,owing to the phase stability of the travellingparticles.

Here something has to be added about the modu-lating frequency. It has been pointed out that thehigh value of 2000 cis is favourable for obtaininga high average beam current on the target. The

Fig. 10. The modulator (without the parts of the capacitorsC' and Cu fixed to the transmission line; see fig. 9). Fromright to left: the slightly adjustable plate of C" (guided bythree pins), the toothed disc of C, the large earthing capacitorC", and the stator coil of the driving motor with the peculiar"windings" for the three phases made up of copper water,"!pipes standing free.

254 PHILlPS TECHNICAL REVIEW VOL. 12, No. 9

question arises whether a still higher modulatingfrequency would be desirable.

In our synchrocyclotron the deuterons have aperipheral rotational speed of roughly 107 "loops"per second (viz. about equal to the frequency of10.7 Mc/s of the voltage on the dees). The acceler-ation of a particle to tbe final energy of 28' MeVhas to take place in about 1/5000 sec, in which timethe particle describes 107/5000 = 2000 loops, sothat it is accelerated by the dee voltage 4000 timesand has to gain 7 keV in energy each time. At apeak value of 14 kV of the sinusoidally alternatingdee voltage, the particles coming under the phase-stabilizing action and accelerated to the final energyare those which pass the gap with a phase anglecp = are cos 7/14 R:::I 60° with respect to the peak. ofthe dee voltage (see A). If a higher modulating fre-quency is chosen then a smaller angle cp is neededto allow the particle to reach the final energy of28MeV in the shorter time available. This, however,makes the orbit less stable, the phase tolerancesbecome small~r and there are fewer ions in a group.Consequently the mean beam current will increaseless than proportionately with the number ofcurrent surges per second and ultimately even de-crease.

164611

Fig. 11. Variation of the oscillator frequency with time.Modulation frequency 2000 cis. Acceleration of the deuteronstakes place during the interval MN of each modulation cycle.

Indeed it is seen that during the starting periodof the modulator the mean current on the targetincreases first roughly in proportion to the modu-lation frequency and then more slowly. At 2000 cis(and 14 kV dee voltage) the expected maximum isnot yet reached. Constructional difficulties havefor the time being prevented any further increaseof the modulation frequency. It is particularly themechanical stresses in the rotating toothed discthat then constitute a serious problem: at a peri-pheral speed of no m/sec the material has to with-stand very great forces. That is also why, insteadof copper, duraluminium was chosen for this disc.

We can now also understand the influence of the dee voltageon the choice of modulation frequency as already referred toin passing. If instead of 7 kV the gain in energy per "loop"were to be only 3 kV then the aceeleration to the final energywould take 7/3 times as long and the modulation frequency(for the same cp) could not be more than 850 cis. The Harwellsynchrocyclotron does in fact work with 3 kV gain per loop,so that for a final energy of 180 MeV the particles have tomaké 30,000 loops, and, as the mean oscillator frequency ofthat cyclotron is about 23 Mcls, the modulation frequencycannot he made any higher than about 300 cis.

Set-up of the transmission line with modulator~d dee

Taking a final look over the whole assembly ofoscillatory circuit with dee and modulator, we findthere are still some other constructional problemstypical for the synchrocyclotron.

The inner conductor of the transmission line mustbe insulated. In order to avoid excessive dielectriclosses in 'the supporting insulators the latterhave preferably to be placed where no high-fre-quency voltages occur. As we have seen in fig. 1,this is the case near the middle of the line. Thus,roughly speaking, we get the picture of a copperpipe about 15 cm in diameter and about 5 metreslong supported only over a short distance near themiddle, carrying at one end the weight of the deeand at the other end the vanes of the variable capa-.citor between which the toothed disc is rotating at~ speed of 3000 r.p.m.Since the copper pipe would not be able to bear

this mechanical load it has heen provided withan internal frame of brass.Another complication is the fact that the pipe, at

least at the ends, has to be in vacuo. The simplestsolution would appear to he an arrangement wherebythe whole transmission line is in vacuo, thus joiningup the modulator drum. and the accelerating cham-ber into one continuous vacuum. In that case theports through which the coupling loops of theoscillator are passed would have to be vacuum-tight,whilst in the event of any changes having to hemade to these couplings or in the rather complicatedconstruction of the modulator the vacuum of thewhole cyclotron would of necessity be disturbed andhave to he restored. We therefore, preferred toseparate the vacuum ofthe accelerating chamber andthe modulator and to give each its own pumpingsystem (see I, fig. I). The two oscillator couplingloops are now easily accessible and come to lie inair. On either side of the anode coupling loop arethe supporting insulators carrying the inner con-ductor of the transmission line.

Since this conductor has to be led into the twovacuum chambers via vacuum-tight insulators it

MARCH 1951 SYNCIIROCYCLOTRON (OScILLATOR AND MODULATOR) 255

Fig. 12. Set-up of the oscillatory circuit. The approx. 5m long copper inner conductor of thetransmission line, with at the ends the dee and the vanes of the modulator (all drawn inheavylines), is supported by the ceramic insulators k. ,which, in order to reduce their electricload, are placed not far from the voltage node in the middle of the line. To give the widecopper pipe the necessary strength it is fitted with a.brass frame on the inside (not showin the drawing). For eentering the system the supporting insulators can be shifted bymeans of screws (and vacuum-tight metal bellows). The dee and the modulator are eachin separate vacuum chambers, 50 that the middle part of the line with the coupling loopsis readily accessible. The ceramic insulators kv provide for vacuum-tight ports throughwhich the line can be passed. Also part of the cooling water pipes is shown in the drawing.

would havê been obvious to use the supportinginsulators for this purpose. In fact, also the portinsulators must not be too far away from thevoltage node. Accordingly, the vacuum chamberof the modulator and also that of the acceleratingchamber is in fact extended close to the voltagenode by a cylinder enveloping the copper outerconductor of the line. Nevertheless it was foundnecessary to keep the functions of supporting andport insulators separate. These insulators already

vacuum

have to answer rather exceptional requirements: thedielectric losses at a frequency of 10 Mcfs have tobe low, in connection with which also the dielectricconstant has to be small, and they must stand upto a voltage of at least several kilovolts (they arestill that far away from the actual voltage node).Furthermore, the port insulators have to make avacuum-tight seal of rather large dimensions (thepipe to be passed through is 16 cm in diameter),and the supporting insulators have to be strongenough to bear the weight of the pipe with dee andmodulator vanes. Only with the functions separatedwas there enough freedom in the shaping of th~insulators for answering the requirements in both.cases. As insulating material "Kersima" 2) wasagain used, a ceramic material with very goodelectrical and mechanical properties, in particularexceptionally high compressive strength (porcelainwould bè quite unsuitable). Fig.12 shows the ulti-mate set-up.

2) Cf. Philips Techn. Rev. 10, 205, 1948.

To ·complete the picture it is to be mentioned thatin the conductors of the transmission line, with deeand modulator, with high-frequency currents ofabout 250 Arms in permanent working a power ofover 3 kW is dissipated. Owing to the insulatedmounting and the fact that the greater part of thesystem is in vacuo, only a small part of this heatcan he carried off by conductance or by convection.Neither can radiation help much in cooling, sinceowing to the high frequency the surface cannot

vacuum 84623

have a properly radiating layer but rather has tobe bright. This accounts for the extensive use ofwater cooling, which to the uninitiated may atfirst sight he astonishing. Water pipes run to allparts of the transmission line, the dee and the modu-lator. Of course the cooling water inlet is in themiddle of the line, at the point of the voltage node.Since, however, in connection with the frequencymodulation this node swings somewhat to and fro,the pipe system is connected to the municipalwater mains with insulating hoses of polystyrenerubber (see fig. 6 at the bottom on the right),which have only very low dielectric losses.

Finally, by way of demonstrating with an exam-ple the very close relationship of all constructionaldetails of the synchroncyclotron, it may be pointedout that it is in fact due to the properties of theceramic insulating material that two transmittingvalves are needed in the oscillator: if the propertiesof the insulating material were such that the sup-porting and port insulators could be placed fartheraway from the middle then more space would be

256 PHILIPS TECHNICAL REVIEW VOL. 12, No. 9

available for the anode coupling loop, higher valuesof the inductances M1 and M2 could be reachedand the oscillatory condition (2) could be fulfilledwith a smaller mutual conductance, thus with onlyone valve.

Summary. I. The Philips synchrocyclotron at Amsterdamis producing in continuous operation deuterons with an energyof 28 million electronvolts, which are being used for manyforms of nuclear transmutations. The fundamental consi-derations to he met in the designing of the installation were,apart from the choice 'of the particles to be accelerated andthe final energy desired, that the beam current should be ashigh as possible and the working as reliable as possible. Atthis moment a eurrent of 20 microamperes is already beingreached with the energy mentioned. This energy of 28 MeVmay just lie within the scope of the classical cyclotron, butthe application of the principle of the synchrocyclotron makesthe construction very much simpler and the installation moreeconomical ID working. In the general description the mostimportant parts of the installation are mentioned and an ideais given of the extent of the accessory apparatus required.

11. In a synchrocyclotron the particles to be acceleratedcan without objection he made to describe a large numberofloops, sothat in this casea deevoltage of only about 15kVpeakis needed. For resonance with the travelling particles the fre-quency of this voltage has to he 10.7Mcls (wavelength about30m). The oscillatory circuit in which this frequency is gener-

ated consists of a coaxial transmission line Wiili at each enda high capacitance (about 400 pF) formed by one of the dees(the other dee is earthed) and the modulator respectively.Owing to these large capacitances, high-frequency. currentsof about 350 A (peak value) are flowing in the system, butby very careful construction the dissipative resistance hasbeen limited to about 0.05 ohm, so that the dissipation isno more than about 3 kW. Of this about 560 watts is dissi-pated as "useful energy" by the beam of deuterons on thetarget. All parts of the oscillatory circuit are cooledwith water.The transmission line is made to oscillate by means of aMeissner circuit, comprising two TA 12/20 transmittingvalves connected in parallel. The mode of vibration is suchthat a voltage node occurs in the middle of the line. Parasiticoscillations are suppressed by by-passing the grid of the valvesto earth and giving the anode and cathode coupling loops asuitable configuration. The frequency is periodically variedby 4% by means of the modulator at the free end of thetransmission line; the modulator comprises a capacitor with.a rotating toothed disc in vacuo. The driving motor is likewisein vacuo, this being made possible by a special construction(fewwindings, field current 185A, voltage 2.1 V). The modu-lating frequency is very high, viz. 2000cIs.The vacuum spacesof modulator and accelerating chamber are separated. Tho,insulated, about 15 cm thick, inner conductor of the transmis-sion line is led into these spaces via port insulators made of"Kersima", which material is likewiseused for the supportingInsulators bearing the inner conductor, the (non-earthed)dee and part of the modulator. The supporting and the portinsulators are situated close to the anode coupling loop, wherethe high-frequency voltage is low.