THEGRAYLABORATORY’

S5Me
VLI
NEARACCELERATOR

This document was prepared by B Vojnovic and provides the following information:


A brief historical overview of the accelerator.

A description of the operation of the different sections of the accelerator.

Waveforms present in various parts of the accelerator when operating under normal/typical conditions.

Circuit diagrams of the accelerator.

Staff involved in the commissioning of the linac:

Mr KD Allen Mr BL Hall Mr C Tootel

Mr B Bloomfield Dr BD Michael Dr B Vojnovic
Mr M Grundon Mr D Nash Dr P Wardman
Mr J Draper Mr RG Newman Mr G West

Introduction and history of the accelerator.

The linac had been in use in the Radiotherapy Department of Mount Vernon Hospital from 1979 to 1993
and was used to generate repetitive photon pulses for therapy purposes. In late 1993, it was dismantled and
most of the ancillary patient and accelerator movement and dosimetry equipment was discarded. At that
time, it was decided to reconstruct the linac in such a way that it could be used to generate both repetitive
and, more importantly, single-s hotpul s
esofdur ations hortert hanthe4μspr ovi dedbyt hema chinei ni t
s
original form. More specifically, the intention was to ‘ repa cka ge’theva ri
ouspowe rs uppliesa ndpul s ing
circuits so that they were to be contained in a single assembly rather than in several separate equipment
racks and in a rotating gantry. This arrangement resulted in a system which may be conveniently screened to
prevent the intense radio-frequency interference, resulting from the generation of short radiation pulses,
from affecting relatively sensitive detection equipment used with the accelerator. The basic accelerator
structure was not altered and the only addition to the vacuum system was the installation of an output beam
line extension which incorporates an inductive pulse charge monitor used for dosimetry purposes.

The acquisition of the accelerator by the Gray Laboratory was initiated by BD Michael, in collaboration with
P Wardman and B Vojnovic. The dismantling of the original accelerator was performed by the Physics
Department of Mount Vernon Hospital in collaboration with the Mechanical Workshop of the Gray
Laboratory and the operation was supervised by Mr Godfrey Mott who had been formerly in charge of
accelerator maintenance at the hospital. G Mott also supervised the extensive mechanical modifications
needed to reshape the linac and the majority of this work was performed by B Bloomfield, D Nash, J Draper,
C Tootel and BL Hall. Some basic power distribution wiring was also completed at that time and operation
of the microwave generating circuits was confirmed. Vacuum problems were dealt with by BL Hall and G
Mott. In this initial phase of the commissioning, it was also established that the proposed methods of
interference screening would indeed be adequate for our purposes. At that time, it was also decided that all
control circuits would be housed within the accelerator so as to minimise the number of conductors which
are placed outside the shielded enclosure.

There were a number of ways by which short pulses could be generated, these included the installation of a
triode (Pierce) type of gun which can inject electrons into the waveguide by controlling the grid potential,
the installation of beam deflection circuits which sweep the beam across a slit and the installation of a
separately pulsed circuit which uses the existing (diode) gun to inject electrons for a short time during the
nominally 4 s RF envelope. All of these options were considered, but it was ultimately felt that the most
expedient way was to use the last option. This was somewhat facilitated by the availability of a pulse
transformer (just about adequate, salvaged from the Gray La bor
a t
or y’sf i
rst1. 8Me Vl i
nac )a ndbyt hef act

Copyright© Gray Cancer Institute 1995 1

that it required no modifications to the vacuum system. It was also decided that the triode gun approach
would be ultimately installed and consequently the triggering/timing electronics were designed with this
eventual purpose in mind. A prototype thyratron pulsing arrangement was constructed by KD Allen under
the guidance of BD Michael. The limitations of the available pulse transformer became evident at that time
and the circuit was modified by B Vojnovic t oincludea‘ doubl eBl uml einl ine’whi c hr educ edthes tresses
on the coaxial delay line cable. This arrangement was eventually installed in the accelerator and is in use at
this time (October 1995).

The personnel safety interlock system used on the linac is essentially the same as that used on the
labor at
or y’sCoba lt60s ourcewhi chi si nst
alledi nt he‘ DCr oom’oft heVa ndeGr a
affa ccelera tor.I
twa s
largely designed by BL Hall and constructed by M Grundon. It makes minimal use of electronic equipment
but rather relies on electromechanical relays which have been shown in the past to be adequately rugged and
reliable.I ti
st iedt ot heVa ndeGr aaffa ccelerator’si nt
e rl
ocks yste
mf ori ndicati
onpur poses,s i
nc ethel inac
is installed in one of the Van de Gr aaff’
sr a diati
onr ooms ,the‘ pul se’r oom.Thes yst
e mr e quirest he
seque nc e‘ l
ina cACpowe ron,r oom s ur vey,c l
os eout e rdoor,e nablera diati
onon’t obec ompl e tedbe fore
clos i
ngapa irofc ont actswhiche nergiset hema chine’sma i
nHT,g unf ilame nta ndfocus power supplies.
Ope ni
ngt her adiationr oom’ sdoor sope nsthe sec ont actsandde -energises these power supplies. In addition
tot his,a cti
va ti
ono fa nyofas erie
sof‘ pa ni
cb ut tons’a roundt he‘pul s
e’r oom r emove sa l
lACpowe rtot he
accelerator and to the associated pulse radiolysis detection equipment.

The final stages of commissioning of the accelerator were undertaken by G West, by RG Newman and by B
Vojnovic. This involved the wiring up and testing of all control circuits and the installation of a control
computer. First radiation beam output was obtained in mid October 1995 and the machine was almost
completely aligned in late October 1995. The linac was put into service in late November 1995.

The end result of this work was that a very stable and, more importantly, electrically quiet accelerator was
addedt ot hee xten si
vea rsenaloft hel aborato r
y’sr adiati
ons ources.Its houl dpr ovet obeama jora ssetfor
pulser adiol
ys i
swor k,essenti
allype rmitti
nga‘ turnke y
’ope r
ati
onofpul ser adiol ysise xperiments to be
achieved.

Description of accelerator.

It is assumed that the principles of operation of linear accelerators are understood by the reader. Useful
descriptions of the operating principles can be found in the documents in the appendix. The present notes are
specific to the accelerator installed in the Pulse Room of the Gray Laboratory and are intended to be updated
as required when changes to the installation are implemented. This document was started in October 1995 at
which time the accelerator was first commissioned to generate single-shot pulses required for pulse
radiolysis work. When a reference is made to a ########.SCH diagram, this refers to an EASYPC
schematic file. All diagrams are can be found in directory C:\PCB\GLLIN on the Electronics and Van de
Gr a af
fPC’ sa swe lla sonBVoj novic’ sPC.

If any work is carried out on the linac for servicing or other purposes, it is absolutely essential that
adequate safety procedures are observed. Electrical and radiation safety procedures must be adhered to
and no unqualified personnel are permitted to remove the copper enclosure shields. It is stressed that the
accelerator is a potentially lethal piece of equipment, both from the radiation and electrical points of view.
Lethal voltages are present in the accelerator enclosure even when it is switched off. These include a
permanently energised ion pump (240 V AC and ca 2.5 kV DC)as well as potentially charged delay lines
and pulse-forming networks. These items may have voltages in excess of 10 kV present. It is thus essential
that all high voltage circuits are discharged using earthing rods provided prior to working in the
vicinity of the gun, the magnetron and either of the two thyratrons. Remember that the radio-frequency
sections of the accelerator are designed to provide power levels of up to 2 MW and that equipment

Copyright© Gray Cancer Institute 1995 2

associated with this must be treated with respect. Do not work on this if you are tired and do not work on
this if you are alone. There can be no better advice than that which appears on every circuit diagram
provided with the original installation: IF IN DOUBT, ASK. In addition, remember that it is possible to
damage various items by operating them incorrectly and that there is a potential fire risk associated with
incorrect connections to items operating at high power levels.

The electronics of the linear accelerator can be subdivided into a number of distinct sections which, although
interconnected, can be operated independently. The control circuits are in fact designed to allow such
separate operation to aid and to simplify fault-finding. Of course great care must still be exercised when
powering up the separate sections; any test results must be correctly interpreted as the performance of any
section may well be affected by its coupling to the other sections.

The main sections of the accelerator are as follows:


The RF System consisting of the magnetron, input mode transformer, accelerating waveguide, output
mode transformer and waveguide load.

The magnetron pulsing system consisting of a fixed, regulated HT supply, thyratron (and associated
driving circuit and power supplies), a pulse-forming network, pulse transformer and magnetron filament
power supply.

The gun filament, its variable, constant-current regulator and associated power and isolation transformers.

The gun pulsing circuit consisting of a variable, regulated HT supply, a thyratron (and associated
triggering circuits), pulse-forming coaxial delay lines and pulse transformer.

Three magnetic beam focussing power supplies and associated solenoid coils.

Four beam steering power supplies and associated magnetic deflection coils.

A power distribution section with switching, soft-starting and overload trip circuits with inputs from the
personnel safety interlock system and logic circuits controlled by the on-board computer.

Low voltage (+/-15V regulated and +/- 24V unregulated) power supplies.

Enabling/switching logic control/indicator panel used to interlock, enable and/or bypass commands from
the computer.

A programmable pulse generator/sequencing circuit used to control repetitive/single shot operating
modes.

A control computer (386 PC) with an analogue/digital I/O card, running LabWindows instrument control
software.

An inductive toroidal resonant charge per pulse monitoring circuit.

A r ecirculatingwa terc hillerunitwhi chi sloc atedi nt heMe cha ni
c alWor kshop’
sma t
eria lstoresand
which is used to cool the power sections of the linac.

The fact that a control computer is placed within the screened enclosure results in only the following
connectionstot he‘out si
deworld’:

- Three phase power to accelerator

- Single phase power to the vacuum system
- Water cooling pipes
- Personnel safety interlock system connections
- SVGA monitor cable
- Computer mouse control cable
- Optional computer keyboard cable
- Dose monitor output

Copyright© Gray Cancer Institute 1995 3

All these cables carry relatively low frequency signals and could thus be filtered straightforwardly. Signals
of higher bandwidth are coupled optically using low-cost polymer fibre-optic connections; these include the
trigger input and output signals and possible control signals to the gun and pulsing circuits. Other
connections are needed during servicing and thus the following cables were installed:

- eighteen co-axial lines used to carry analogue input and output signals
- one 25 way cable used to carry logic and enabling signals
- one 25 way cable used to carry analogue monitoring signals

The process of generation of electrons and their acceleration will now be described so that the peculiarities
of the present installation are understood.

In its original form the machine was designed to produce repetitive pulses of 4 s duration at rates up to 300
per second. These pulses were generated by triggering the magnetron and pulsing the filament
simultaneously through a quadrifilar-wound pulse transformer. In the present modification a second pulse
transformer and thyratron are used to separate the generation of the magnetron pulse from that of the
filament gun injection pulse. In this way the injection pulse may be arranged to occur at any time during the
4 s RF pulse resulting in the injection of electrons (provided they have the correct energy to be accepted by
the accelerating waveguide) and thus acceleration of electrons for a time much shorter than the 4 s
provided by the original design. The duration of the electron pulse is determined primarily by the electrical
lengths of a pair of delay lines and is nominally fixed to 250 ns. In practice, the edges of this injection pulse
are degraded by the use of an inadequately fast pulse transformer, so that an essentially triangular pulse is
fed to the gun. The actual electron pulse duration is thus a function of the peak injection pulse amplitude. If
the gun filament is run at a low level, the electron pulse amplitude is also determined by the injection pulse
amplitude. Too high an injector pulse amplitude results in a wide and distorted pulse, distorted because only
a finite range of energies is acceptable to the waveguide input. This highly simplified description is shown in
figure 1. If the gun filament is operated at a high level such that the electron density is high, higher injection
voltages are needed so that the output electron pulse becomes more and more triangular.

Gun injection voltage Gun injection voltage

Output electron pulse Output electron pulse

Figure 1. Electron pulse shape obtained as a function of injection pulse amplitude. On the left panel, the
electron density is low and lower injection voltages are needed. On the right panel the injection voltage
needs to be higher when high beam currents are required, resulting in longer pulses.The dotted lines
represent the range of electron energies accepted by the input end of the accelerator.

Copyright© Gray Cancer Institute 1995 4

RF and magnetron pulsing system.

The heart of this system is a magnetron (EEV type M 5125) which is operated at a pulsed (4 s) HT voltage
of ca 40 kV and at a current of ca 100 A. The magnetron nominal operating frequency is 3 GHz (S band),
thea ctualf re que ncyma ybes etbyt uni ngt hema gnetr
on’sc avi
ty;t hisi spe rforme d us i
ngamot or
drive/gearbox/limit switch located above the magnetron. The magnetron is operated with its anode at ground
potential, i.e. its filament (heater/cathode) is pulsed to -40 kV. The HT pulse is derived from a bifilar wound
pulse transformer (see circuit MAGTHYR.SCH) with heater DC power applied across the windings. The
pulse transformer is actually a quadrifilar wound device but two of the windings are not used in the present
arrangement; they were used in the original machine to power the accelerator filament, which was pulsed
sync hronous lywi ththeRFg ene rati
ngpul s e.Thema g netr
on’sf i
lamentpowe rc oul dbes ett
oz ero,6. 5Vor
13.5 V in the original design; this was so because at high RF repetition rates the cathode is heated by the
circulating electrons, and hence the filament power needs to be reduced to prevent overheating. In the
present arrangement, the maximum repetition rate allowed is 50 Hz and hence the filament is operated at
13.5 V. The HT to the magnetron must not be applied when its filament is cold, the magnetron will then not
oscillate and the voltage will rise to >45 kV, at which point a spark gap across the magnetron will break
down. This condition should be avoided as current of >1 kA will flow during spark gap breakdown and this
may cause damage to nearby electronics, especially in the focus power supplies.

It is quite normal for the magnetron to misfire once in every several hundred to several thousand pulses.
Thi smi sfi
ringiss ome timeshearda sa‘ pi
ng ing’s ounde ma natingfrom t hema g netrona nd low or zero RF
out puti sobt ai
nedunde rthe
sec onditions.TheHTwi llalsor i
s eatt hatti
mebuts houldbe‘ caught’bya
reverse diode assembly (see reverse diode trip circuit in MAGTHYR.SCH). This misfiring is quite
dependant on correct tuning of the magnetron, i.e. it is more likely to take place if the magnetron is
mistuned. Magnetrons are very rugged devices and can take a fair amount of mistreatment, though this
should clearly be avoided. Eventually, when the misfiring rate becomes unacceptable, the tube must be
replaced. Take care to use non-ma gne t
ictool sfort hispurpos ea nddon’ tke epc reditcardsi ny ourpoc ket,
and remove your watches!

The microwave (RF) output from the magnetron is coupled through a waveguide and its associated vacuum
window and through a mode transformer to the accelerating waveguide assembly. This unit is the essential
element of the accelerator; electrons, injected through the mode transformer are accelerated through it and
the RF electric field imparts energy to the resulting electron bunch. During acceleration, RF power is
reduced along the accelerating waveguide and any remaining power is coupled through the output mode
transformer to a matched, cooled load.

The magnetron pulsing system consists of a fixed voltage, regulated HT supply, thyratron (together with its
driving circuit and power supplies), a pulse-forming network, pulse transformer and magnetron filament
power supply. Details of this are also found in MAGTHYR.SCH. The HT supply is considerably over-rated
for the present operation, as it was originally designed to power both the magnetron and the electron
injection system at repetition rates of up to 300 Hz. The system works as follows: a three-phase (star-delta)
transformer supplies a rectifier to charge a pulse-forming network to ca 10 kV. These items are located
below the accelerator body, under a perspex cover which should be removed only with extreme caution. The
pulse forming network is a lumped element LC delay line. It is discharged by a thyratron (EEV type CX
1140) into the primary of the pulse transformer, believed to be of 1:5 ratio. The output pulse is 4 s long.
There are a number of monitoring points in this arrangement and typical traces are found in figure 2. The
thyratron is located in the lower part of the screened cabinet at the rear end of the accelerator. It is triggered
by an SCR trigger amplifier which produces a +300 V pulse by discharging a lumped LC line through an
isolating pulse transformer. The SCR is triggered by a +10 V pulse provided by an optical receiver, shown in
c i
rcuitTHYRTRI G. SCH.Thi sr eceive sat ri
g gerpul s
ef rom apol yme rfibref e dfrom t hea cce lerat
or ’
s
timing logic (LINDEL1.SCH). This optical receiver may be bypassed for test purposes; this is achieved by

Copyright© Gray Cancer Institute 1995 5

injecting a +10 V, 10 s pulse into the thyratron trigger BNC socket on the BNC test panel at the rear of the
accelerator and by patching this through in a cable available next to the SCR trigger board, at the top left of
the screened thyratron rack. Refer to circuit diagram INTCON1.SCH to see this in detail.

Although not normally monitored, useful diagnostic outputs may be obtained from two detectors which
monitor the RF levels in two cavities, one tuned above, the other below the correct magnetron operating
frequency. These outputs are available at the BNC panel at the rear of the accelerator. These detectors are
straightforward silicon diodes and can be operated into 50  to show the RF envelope profile; the
interference levels are quite significant when operated in this mode and it is generally preferable to load the
detectors with the 1 M input impedance of an oscilloscope. In this instance the integrated output is
obtained and typical outputs are shown in figure 3.

Gun filament and regulator.

The filament power supply used in the original machine provided DC excitation of the gun filament,
supplied directly from the second pair of bifilar windings of the magnetron pulse transformer. The gun
filament in the current arrangement had to be powered through the new gun pulsing transformer; the current
rating of its windings were unknown (but probably round 1 A) and it was thus considered safer to take gun
filament excitation power at a higher voltage through the transformer and then to transform it down to the 5-
10 V required by the filament. This dictated the development of a constant current AC power supply. The
availability of a 10:1 auto-transformer (at the high voltage end of the pulse transformer) determined the
operating voltage of around 100 V. A pulse-width modulation scheme is employed, using a power MOSFET
as the switching element placed across a current-steering bridge rectifier, as shown in GUNMOD.SCH. A
current sensing resistor, 2 ,i sus edtosampl ethec ur
re ntfl
owi ngi nthepul set ra nsforme r‘low’t erminals;
the resulting voltage is a filtered, chopped sine wave which is synchronously rectified to provide a DC signal
proportional to the average AC current flowing through the filament. This DC signal is fed to an error
amplifier, along with a heavily filtered input set-point DC signal. The feedback loop is completed by using
the resulting error signal to modulate a simple triangular wave pulse-width modulator, the output of which is
used to drive the gate of the control MOSFET. The high frequency performance of the output auto-
transformer is probably poor enough to remove most of the switching frequency (nominally 10 kHz) and its
harmonics, but nevertheless a small choke is added to the output circuit to ensure that a relatively clean sine
wave is fed to the gun filament.

Figure 2. Waveforms related to operation of magnetron. Left panel: Upper trace (5 V/div, 5 s/div): pulse
voltage applied to magnetron, monitored into 50  through accelerator-installed resistor (sensitivity not
known). Lower trace (2 V/div, 5 s/div): Current pulse through thyratron, monitored into 50  from
accelerator-installed current probe (50 A/V). Right panel: Upper trace (5 V/div, 2 s/div): thyratron trigger
pulse monitor output, into 50 termination. Lower trace (2 V/div, 2 s/div): thyratron output current pulse,
as above.

Copyright© Gray Cancer Institute 1995 6

Figure 3. Waveforms obtained from the frequency tuning cavities. Left panels: Cavity detector outputs into
1 Mohm input impedance (500 mV/div, 5 us/div); upper panel: magnetron optimally tuned; lower panel:
magnetron tuned to a frequency too high. Right panels: upper, as above, but magnetron frequency tuning too
low. Lower panel shows thyratron current pulse (conditions as in figure 2) and tuning cavity output. In all
cases a 5 point software filter is applied to reduce noise on cavity outputs.

The gun pulsing circuit.

This circuit is also shown in large part on GUNMOD.SCH. Once again a thyratron is used to discharge a
pulse-forming device through the input of a 1:5 pulse transformer. This time, however, the pulse width is
determined by two equal lengths of coaxial cable; each length is made up from two 50 ohm RG 58 cables in
parallel, to make up two lengths of 25 cable. These are arranged in a Blumlein type of double line circuit
which has the desirable property that an output voltage equal to the DC charging voltage is obtained into a
matched load. This means that the DC voltage stress on the delay line cables is reduced, particularly
important in the present case because we need to generate pulse injection voltages of up to around 45 kV;
this could in principle be achieved with 18 kV DC and a conventional circuit using a 1:5 pulse transformer.
However, the high frequency performance of the available pulse transformer was quite poor and the device
behaved more like a 1:3 transformer for short pulses, probably due to excessive leakage inductance losses.
Using the present voltage doubling arrangement, we can generate in excess of 45 kV with just under 15 kV
DC line charging voltage. The pulse transformer output load resistor is made up from 80 series-connected
non-inductive resistors, each 15 , resulting in a load resistance of 1.2 k.

Copyright© Gray Cancer Institute 1995 7

Figure 4. Waveforms relating to the operation of the thyratron gun pulsing circuit. Upper panels: Gun
injection output pulse (upper traces, acquired with installed capacitive probe + attenuator on pulse
transformer, into 1 M of scope (left, 4 kV/div, 1 s/div; right, 10 kV/div, 1 s/div) using charging
voltages of 5 and 10 kV respectively. Lower traces in the upper panels are the thyratron trigger pulse
monitor output, through 50 x 10 attenuator.
Lower panels: Pulse HT supply output voltage recovery after trigger, when charging at 5 kV (left) and 10 kV
(right). In both cases the timebase is 5 ms/div.

The delay line is charged through a nominally 100 K (actual value 97 k) resistor fed from a voltage-
programmable, current limited power supply capable of providing 16 kV maximum output voltage. This was
a customised unit, manufactured by Applied Kilovolts Ltd and the reader is referred to correspondence
related to that power supply in the appendix. Typical traces of the injector waveforms are shown in figure 4.

The thyratron, of the same type used in the magnetron pulsing circuit, EEV type CX 1140, is triggered by a
+300Vpos i
tivepul sepr ovi
de dbywhatma yseema nexc essi
ve ly‘g ood’pul s
ea mpl i
fier,showni nde t
ailin
POSPAMP.SCH. This design, however, was intended to drive the grid/cathode of a triode type of gun,
discussed earlier. This assembly makes use of a fast MOSFET and a coaxial inverting transformer to
generate fast pulses (50 ns and longer) with sharp (<5 ns) edges. It is in turn triggered optically through a
polymer fibre coupled to a photodetector/comparator. Two DC-DC converter modules provide voltages of
up to +350 V for the FET (output pulse amplitude) and -150 V (thyratron grid bias (or gun control grid bias
in the case of a future triode gun).

Beam focus power supplies.

There are three essentially similar focus power supplies which feed three solenoidal focus coils along the
length of the accelerator. Although similar in design, they operate at different voltage/current levels; the
circuit is shown in FOCUS.SCH. Each supply is fed from a different phase of the three-phase supply and
eac hc ons ist
sofat ransforme rfe edingadi ode/SCRr ecti
fierbridge.TheSCR’ sa repha sec ontrolledthr oug h
a discrete 50 Hz triggered pulse-width modulator. The input to the modulator is the output signal from an
error amplifier (741 (IC1)) which compares the attenuated input control signal and the voltage developed
across a current-sensing resistor (R2, ca 0.004 ohms) in the output circuit of the supply. The full-wave-

Copyright© Gray Cancer Institute 1995 8

rectified DC output is smoothed by an output inductor and indeed by the inductance of the focus coil itself.
Nevertheless a significant degree of 100 Hz ripple is present in the output current as shown in figure 4.
Current setting control inputs are in the range of 0 to +5V, while the output current levels result in voltages
mps(
of the order of tens of millivolts across the current sensing resistors. Two other op-a I
C2,3,a lso741’ s)
are connected as comparators with positive feedback and are used to sense output overcurrent and
undercurrent conditions. This is achieved by comparing the output current, or rather voltage across the
output current sensing resistor, with preset voltages adjusted by means of VR1 and VR2. The output of the
comparators are joined together such that a negative trip output voltage (ca -12 V) is an indication of correct
operation while a 0V output is an indication of a fault condition, i.e. that an overcurrent or undercurrent
condition exists.

Two of the focus power supplies (Focus 2 and 3) are mounted behind the computer ’sFa r
adayc agea ndc an
only be accessed by removing the computer and screened cage. Focus 1 supply is mounted underneath the
output beam line. Focus 1 supply feeds the focus coil nearest to the gun, Focus 2 supply feeds the coil in the
centre of the accelerator and Focus 3 powers the coil nearest to the output. When working on the focus
power supplies take care around the rectifier/bridge heatsink: this is not grounded and you will blow the 40
A power supply fuse if the heatsink is accidentally shorted to earth.

Beam steering power supplies.

There are four similar bipolar controlled constant-current supplies which excite four sets of deflection coils,
two of which (Steerers 1,2) are located at the gun end, the other two (Steerers 3,4) being near the output end
of the machine. Each set of coils deflects the beam in an up-down or a left-right direction. The power
supplies are housed in a chassis at the rear of the accelerator and each accepts a bipolar input, nominally +/-
5V. The circuit is shown in DEFLECT.SCH. The output current from each supply is sensed using a1 
resistor (R15) in series with the deflection coil. The error amplifier is once again a 741 which feeds a x4
voltage amplifying stage, supplied from +/-24V (TR1-4). There are no trip/fault outputs. The alignment of
the waveguide and gun is such that relatively little deflection is required to thread the beam through the
magnetic axis of the focus coils.

Power distribution.

The circuit diagram of the three-phase power distribution within the accelerator is shown in PWRDIS.SCH.
Thr eepha sepowe riss uppli
edf rom ac ontactori nt he‘ cleane r
’scupboa rdroom’ne arthel iftsha f
tj ustof f
theVa ndeGr a affc onsolearea.Thi sc ontac t
ori st ri
ppe dwhe na ny‘pa nic ’buttoni sac t
ivatedi nt he‘ pulse
room’ .Thi si nf a cta l
soc utsof fsing l
epha seAC powe rtot hel i
na c’sva cuum s ystema ndt ot hepul se
radiolysis instrumentation system. Power may be restored only by activating a reset switch which is inside
the interlock system cabinet (normally locked, key inside Van de Graaff key cabinet). Under normal
operating conditions, the three phase contactor is energised by operating the front panel keyswitch on the
interlock panel, as described below in the safety interlock system description; once again the key to energise
the linac is kept in the Van de Graaff key cabinet. The operation of this key energises the water chiller unit,
loca tedint heMe cha ni
calWor ks hop’sme talstores.

Ina ddi tiont othe‘ cleaner’sc upboa rd’,thet hreepha selinacpowe rma ybec ut off by deactivating an isolator
switch in the pulse room, this is located on the rear (left) wall of the pulse room. Under normal conditions,
three-phase power is thus present in the linac enclosure, and indication of this is by means of three neons
located on the front of the power distribution panel, located at the rear front of the accelerator. These and
other indicators are not visible when the copper shielding is installed. The linac power input latch may now
be energised by operating either the red push button on the front of the accelerator, near the vacuum gauge
or a similar push button on the power distribution front panel at the rear of the machine. This latch will only
set if the safety interlock system is in the safe, unsurveyed state (see later description). Once set, three other

Copyright© Gray Cancer Institute 1995 9

neon indicators are lit and power to most low voltage systems is applied and the control computer is
activated. Provided various machine safety interlocks are made (water, vacuum etc.) and the magnetron
filament is energised, a 10 minute time delay is started as well as a 5 minute time delay (see
GUNMOD.SCH). During this time the two thyratron and the magnetron heaters are warming up. When the
timede lay se xpi
re,ma i
na ndpulsei nj
ectorHT’ swi llbee nerg
isedpr ovidedthe appropriate enabling inputs
ar epre sent( underma nua lorPCc ontr
ol)andt hepe rsonnelsafet
yi nter
loc kout putisi nthe‘ i
rradiat
e’s ta
te.
Thema inHTs up pl
yi nc orporat
esa‘ softs tar
t’,c urrentlimit
eds ystem oni tsinput.A s i
mi l
ar‘softs ta
rt’
circuit is present on the focus power supplies feed. Focus power supplies are de-energised when the main
HT is off.

The personnel safety interlock system turns off the gun filament, the main HT and the focus supplies when
in the unsurveyed mode. Other aspects of power distribution are self-explanatory in PWRDIS.SCH.
Overload trip switches are installed on the separate subsystems; if tripped they can only be reset manually. It
is advised that appropriate trip switches are de-energised during servicing of the accelerator and that power
is fed only to the circuits which are absolutely needed. There is generally no reason why the main HT supply
switch should ever be switched on at any time that the copper covers are removed.

Low voltage power supplies.

These supplies are shown in LOVPWR.SCH. There are two power supplies: a +/-15 V regulated supply and
a nominally +/-24 V unregulated but smoothed supply. They are located at the back end of the accelerator in
two19”c ha ssis.The+/-15 V chassis also contains a +5 V and a +300 V supply, although these are not used
in the current arrangement.

The regulated supply is of conventional if somewhat dated design relying on a LM305 regulator chip for the
+15 V side and a LM304 chip for the -15 V supply. This power supply is used to power the control circuits
of the three focus and four steering supplies, the vacuum interlock/trip circuit and the toroidal charge
monitor.

Enabling/switching logic.

This system is located at the back of the accelerator, above the low voltage power supplies and the BNC
panel. Its purpose is to provide convenient routing/enabling/disabling of PC and manual over-ride logic
signals. The front panel controls are shown in TESTPAN.SCH and the circuit can be found in PCINT.SCH.
The circuit is based on a number of inverting, open collector power drivers (Darlingtons), type ULN 2803,
which ultimately drive relay coils in the power distribution sections of the accelerator as well as local
indic at
orLED’ s.Thr eewa ys witche sineac hoft hec ircui
tsa llow any particular system to be permanently
disabled (centre position), permanently enabled (lower position) or to be under the control of the PC
interface. All the input and output connections to the PC are thus standardised to 5 V logic levels. Local
control of magnetron operating frequency tuning may also be performed on this panel. The circuit is self
explanatory and this panel need be accessed only during servicing or at times when the linac is controlled
manually. Remember to return the toggle switches to PC control (upper position) following servicing.

Toroidal charge monitoring circuit.

This is shown in QMON.SCH and the arrangement functions as a non-intercepting beam charge per pulse
monitor. It is located at the output end of the accelerator beam line, in a diecast box attached to the beam
line. A toroidally wound coil envelops the beam line and behaves as the secondary of a transformer where
the moving bunch of electrons in the beam pulse forms a single turn primary. A current is thus induced in
this‘ seconda ry ’
.Thewi ndingi sa rrangedt obepa rtofat unedc ircuitwhi chos cill
a tesati tsr esonant
frequency when excited by the beam pulse. It can be shown that the peak tuned circuit output voltage is

Copyright© Gray Cancer Institute 1995 10

proportional to the total charge in the beam if the period of oscillation is much longer than the duration of
the electron pulse. For calibration purposes, a single turn winding has been threaded through the toroidal
core and a current pulse can be injected through this into a 50 load. The connections to this test turn are
available on two coaxial connectors to the side of the toroid housing.

The amplitude of the triggered oscillation eventually decays to zero in a time determined by the Q-factor of
the tuned circuit. A differential, low-noise amplifier (SSM 2015) both amplifies the damped sinewave and
provides an adjustable degree of positive feedback to act as a Q multiplier and to ensure that the oscillation
decay relatively slowly. The output of this amplifier is coupled to a latched peak detector circuit and to a
comparator which detects the onset of oscillation, i.e. at the start of the next beam pulse. Its output is fed to a
monostable which re-enables the comparator after its time delay (nominally 20 ms) and to a reset
arrangement which ensures that the peak detector is reset at every electron pulse. The peak detector actually
measures the amplitude of the second half-cycle; this is essentially the same as the peak tuned circuit
amplitude as long as the Q is reasonably high. In this way a continuously updated peak detector output is
obtained without the need for any external trigger/reset signals. The output of the circuit is available for
monitoring/feeding to the PC and is separately brought out on a balanced connector so that it may be
injected into the dose monitoring arrangement of the pulse radiolysis instrumentation.

Pulse generator/sequencing circuit.

A block diagram of the timing electronics is shown in LINTRIG.SCH and detailed circuits are shown in
LINDEL1.SCH and LINDEL2.SCH. The sequencing system is constructed on two boards located in a
chassis frame within the PC Faraday cage. Two modes of operation are possible: continuous or repetitive
pulsing and single shot operation. Most parameters are programmable although the unit will work in a preset
fashion by unplugging the various data latches and replacing them with DIL switches. A layout of these is
provided in figure 5. The repetitive operation mode is self-evident while single-shot operation refers to a
repetitive RF pulsing mode with the production of a single electron pulse following the generation of a
predetermined number of RF pulses. As described earlier, the circuit details are more complex than needed
but the arrangement caters for the future addition of a triode type of gun which will be capable of producing
shorter pulse widths and which requires optical isolation of certain signals through the use of optical fibres.

LINDEL1.SCH connector LINDEL2.SCH connector

HCT
688

Q P O N M A B C

in place of latches (HCT 574) in place of latches (HCT 574)

Figure 5. Layout of manual DIL switches that could be used in place of data output latches whenever PC or
programmed operation is not required.

Copyright© Gray Cancer Institute 1995 11

The generation of all trigger pulses is synchronised with 50 Hz AC mains; this is needed because of the poor
ripple performance of some of the power supplies associated with the linac. The actual trigger point on the
ACc y cleisde t
ermi nedbya‘ pha se’de lay(chipJ )pre setonLI NDEL2boa r
d:t hispresets houl
dbea djusted
with caution as the eventual peak pulse current performance of the linac will be affected. The maximum
repetition rate available is 50 Hz, though lower rates (:2 (25 Hz), :3 (16.66 Hz, :4 (12.5 Hz) etc. down to
3.125 Hz) are possible and are achieved by passing the input trigger pulses through a programmable divider
(CD 4526B chip I on LINDEL2). The continuous or repetitive mode will now be described and the reader
should bear in mind that electron or gun trigger pulses must be produced such that electrons are injected
duringt he ma chine’s 4 s RF pulse. The component references are with reference to diagram
LINDEL1.SCH:

The mains-locked input triggers are routed through gate H to a clock synchronising dual flip-flop (chip J);
they also reset two 11 bit programmable digitally clocked delays which can provide independently settable
delays over the range 0-40.96 s in 20 ns steps i.e. using a 50 MHz clock. The synchronising circuit
g ene r
atesas ingle‘ load’pul set ot hepr o gr
a mma bl
ede lay stos tartane w de lay;t hel oad pulsei s
sy nchronouswi ththe‘ next’c ycleoft he50MHzc lock.Thepr ogramma bl
ede laysa ree ac hba sedont hree
74AC163 counters (chips A, B, C in the case of electron trigger delay, D, E, F in the case of the RF delay).
The differential delay is actually the important parameter, though for flexibility each delay is made
programmable. The delay generators actually provide pulses of width equal to the programmed delay and the
falling edges of these pulse widths trigger a dual monostable (74HCT123, chip R) The lower half is used to
generate ca 10 us wide pulses which are fed to a fast LED driver (74F5302, chip U). The optical output from
the LED ultimately triggers the RF thyratron and hence the magnetron. The other half of the monostable is
used to generate a number of electron pulse widths, though only the 1 us width is used in the present injector
implementation to trigger the pulse thyratron. A 3 bit code determines one of eight possible widths, where 2
us, 1 us, 0.5 us and 0.2 us pulse widths are selected using switch S (74HCT4051) to switch in one of four
possible presets into the timing RC circuit of the monostable. The shorter pulse widths are generated by
de layingt he200nsmonos tableout putus ingde l
ayl ineY a nd‘ anding ’as el
e ct
e dde laye dout putwi t
ht he
undelayed output in the driver U (74F5302). The selection of the delay line output is performed with switch
V (74HCT4051) which is used, along with the delay line, to generate 100 ns, 50 ns, 20 ns and 10 ns LED
out putpul ses.TheLED’ sg ene ratei nfr
a -red pulses which are guided along polymer fibre-optic cables to the
twot hyratrondr ivingc ir
c uit
s.Twof ur
the rLED’ s and drivers (chip T, 74F5302) are used to provide output
pulses, for pulse radiolysis instrumentation synchronising purposes, one of which is coincident with the
electron pulse, the other being the 40 ns trigger pulse which initiates the start of the digital delays (pre-
trigger output). The degree of pre-trigger obtained depends on the value loaded into the digital delays.

In the case of single shot operation, the same sequence of events as described above takes place except that
the electron pulses are inhibited by gate G (74AC00, pins 11,12,13) until a programmed number of RF
pulses has been generated, as shown in the timing diagram in LINTRIG.SCH. The number of RF pre-pulses
can be 2 to 64 as determined by the input to comparator M (74HCT688, LINDEL1.SCH). The single shot
sequence is initiated by an optical trigger pulse (10 us duration) which sets the bistable made from gates K
(74HCT132, pins 1-6). This allows the mains phase locked pulses to be counted by counter L (74HCT393),
the output state of which is connected to comparator M. When the required number of pulses is counted, the
bistable K is reset. The time delay between the arrival of the single shot trigger pulse and the production of
the electron pulse is thus not fixed but can vary from m.n/50 to m.(n+1)/50 seconds, where n is the number
of RF prepulses and m is the 50 Hz repetition rate divisor, the exact delay time depending on when the input
trigger pulse occurs relative to the repetition rate phase.

An independent circuit is also constructed on one of the timing boards (LINDEL2.SCH). This consists of
two 8 bit pulse width modulators, more correctly mark-space modulators, running at a frequency of
approximately 200 kHz. The pulse trains are used to modulate optically the outputs of two LEDs/polymer
fibres with the eventual intention of using these to determine DC voltages in a triode gun. DC voltages can

Copyright© Gray Cancer Institute 1995 12

be obtained by integrating the pulse train. Once again this feature is not used but the pulse train outputs are
integrated using the TLO72 operational amplifiers and the 4053 analogue switch. These provide a variable
output of up to +10 V to provide control (set-point) inputs to the gun filament supply and the pulse HT
supply. In effect an 8 bit digital-to-analogue converter is thus implemented. In the triode gun version we
ha vea lsotwoD/ A’swhi c ha r
ec apa bl
eofvol t
agei solationoft ensofki lovolt
s,de t
ermi ne dbyt heopt ical
fibre. The digital pulse width modulators are implemented using two 4 bit programmable counters (4526Bs,
chips D,E and F,G) in each channel. Every 256 clock pulses the counters are reset by counter J (4520B). As
well as providing this form of mark-space ratio control, a further facility is provided by selector switch L
(4502B) and board preset switch to allow an external pulse width modulated signal to control one of the
selected outputs. This external signal is injected through an input optical fibre and allows the option of
manual remote control of beam charge delivered the accelerator to be readily performed.

Control computer (386 PC) with an analogue/digital I/O card.

An IBM compatible 386 PC is located in a separate Faraday cage within the accelerator screened assembly,
at the front of the machine. This secondary Faraday cage also houses the timing chassis boards described
above. A data acquisition system is located on a separate circuit board (ANIN.SCH and LININT.SCH)
within this chassis and is intimately tied to a PC bus data acquisition card (Amplicon PC30A) within the
computer, the two systems being interconnected with a 50 way ribbon cable. All connections in and out of
the Faraday cage are extensively filtered using decoupled sockets; AC power to the cage is separately
fil
tereda ndc ome si nt hroug ha nI EC c onne ctoratt heba ckoft hec ag e.Thec omput e r
’sinput/output
cont r
ol sares imi l
arlyfilt
ered( se eFI LT.SCH)a nda r
ebr oug htoutt oas ma l
lSVGAdi splaya nd‘mous e ’at
the pulse radiolysis console area. In normal operation these are the only accessible controls and settings are
adjusted using a Graphical User Interface, developed using National Instruments LabWindows software. The
software is described separately.

The PC30A card provides 16 analogue input channels (12 bit resolution), four analogue output channels and
three 8-bit digital parallel I/O ports. Of these, the 16 analogue inputs are used and two of the digital ports.
The analogue inputs digitise conditioned signals from various points in the accelerator; these are listed
below. Signal conditioning is performed using sixteen instrumentation amplifiers (INA 114, described in
ANAL.SCH). The input senstivity of the card is fixed at +/- 10 V fsd, single-ended, and the inputs to the
differential INA 114 amplifiers are heavily low-pass filtered to prevent RF interference from affecting the
readings. The following signals are acquired:

Focus 1 output current Steerer 3 output current Spare channel for future quadrupole
Focus 2 output current Steerer 4 output current Main HT supply current
Focus 3 output current Pulse HT supply current Water temperature
Steerer 1 output current Pulse HT supply voltage Charge per pulse monitor
Steerer 2 output current Vacuum pressure

The digital I/O is somewhat more complex because of the number of digital output lines required and is
arranged so that 8-bit data words can be latched into sixteen registers. Eight of those registers are part of the
digital to analogue converters (AD 7524 on LININT.SCH) which are used to provide DC output voltages for
the three focus supplies, the four steerer supplies and a future quadrupole focussing lens. The focus outputs
are unipolar and can be set over the range 0 - +5 V . The steerer outputs are bipolar and can provide up to +/-
5 V. The other eight digital words are arranged in the following way: individual bits of one of the words are
used to enable/disable power supply systems:

- main HT supply - gun filament supply

- pulse HT supply - magnetron tune up
- magnetron system - magnetron tune down
- focus supplies - spare

Copyright© Gray Cancer Institute 1995 13

Two other words are used to set the mark-space ratio of the pulse width modulators described in the
previous section and are in effect used as two further digital to analogue converters which set the gun
filament current level and the pulse HT injector voltage. A fourth word is used to set the number of RF pre-
pulses during single-shot operation (6 bits), to initiate the single shot trigger (1 bit) and to set the mode of
external dose control (1 bit) . The fifth word sets the repetition rate (4 bits), the electron pulse width (3 bits)
and initiates the repetitive mode of operation (1 bit). The remaining three words are used to set the delays (0
- 40μs )be twe ent ri
g g ertoRFpul sea ndt r
igg ert oe lectr
onpul see ve ntsr e spectivelya ndt oe nable/disabl e
those events.

The final use of the digital I/O is to address a 16 bit multiplexer which monitors the status of power supplies
and fault conditions. The sixteen states are multiplexed on a single digital input line and the following states
are monitored:

- main HT supply on - gun supply on - tune limit reached - vacuum 10-5 trip
- focus supplies on - interlock system - focus 1 fault - vacuum 10-4 trip
- magnetron system on - tune down on - focus 2 fault - water fault
- pulse HT supply on - tune up on - focus 3 fault - water OK

Manual operation of the accelerator.

Dur ings e
rvicing/faultf indi ngitisde sirabletobea bl
et oope rat
et heacce l
era t
orwi thoutt hec omput er’
s
intervention and facilities have been provided to allow this. All the control voltages/set point inputs are
brought out on the BNC panel at the rear of the accelerator and are normally patched through with short
BNCl inkstot hec omput ers yste
m’ sD/ Aout puts.Whe nthe sec abl
e sa r
eunpl ugg edas etofBNCl eadsf r
om
the console area (see CABLES.SCH) can be plugged in and connected to a box of control potentiometers, as
shown in MANTEST.SCH. Excitation of the potentiometers can be from a +/- 12 V bench power supply or
the unit can be interconnected using a short DIN cable to a metering / logic controller, also shown in
MANTEST. SCH.A ‘ D-t y pe’plugt e rmi nate
d25wa yf latc ablei snow c onne ctedt ot her earl eft
-hand
soc ketont hec ont r
o l
lera ndt otheca blenor mallyplug gedi nt
o‘ D-ty pe’socke tXont hePC’ sFa r
a dayc age.
All the logic and status functions which the PC would have provided are now available on the controller,
which is powered through the flat ribbon cable. The controller also incorporates a 4.5 digit panel meter
which can be used in conjunction with a rotary selector switch to monitor various voltages around the
acc el
e rat
or.Thi s,howe ve r,isonl ya c hievedwhe nas e cond25wa y‘D-type ’c ablei spl uggedi nt
ot he
cont roll
er( rig
htha nds oc ke t
)a tonee nda ndt ot hec ablenor ma ll
yc onne ctedt os ocketY ont hePC’ s
Faraday cage at the other. This cable is a round screened cable and is connected to the cable which normally
carries analogue inputs to the PC.

In this mode of operation care must be taken with the timing circuits since they are under PC control. There
are two ways of dealing with this: either the data latches in the timing chassis boards can be replaced with
DIL switches as shown in figure 5 or the timing circuits are bypassed altogether by injecting optical trigger
signals directly into the two thyratron input amplifiers. The optical triggers can be derived from a dual
output pulse generator feeding two LEDs and polymer fibres.

Vacuum system.

The vacuum pumping system on the linac is extremely simple and consists of a cartridge type of ion pump
and associated power supply. The pump is a non-servicable item, and, as far as we can tell at the time of
writing, replacements can only be obtained from Philips. A cooling fan has been fitted behind the pump to
help with pumping efficiency. During the commissioning phases of the accelerator, the pump proved
incapable of pumping down the system when the filament or the RF were energised. Ion pumps should in

Copyright© Gray Cancer Institute 1995 14

any case not be operated at pressures above 10-4 Torr and the only way to coax the pump into action was to
couple a turbomolecular and rotary backing pump to the system through a flexible connection on the
isolating valve port next to the gun end of the accelerator. It is presumed that a similar situation will arise
when the filament will need replacing and the system is brought up to atmospheric pressure. With the ion
pumppowe rs upplyt urne dof f,theionpumpwa she atedwi thahota irguni na na t
temptt o‘ bake’it.The
whole process took several days; this is not too surprising as there seem to be severe restrictions in the
pumping layout, especially around the mode transformer at the gun end of the accelerator.

Ion pumps of course also behave as vacuum gauges and the vacuum pressure can be determined from the
power supply output current and the calibration chart in the appendix. An integral part of the current
monitoring circuit on the original machine was the vacuum indication meter. Its internal resistance was
specific to an additional non-linear network at the input to the vacuum trip monitoring system. Unfortunately
this meter was damaged in transit to the Gray Lab and no internal resistance or fsd markings could be found
on the meter. The meter also had a 270 o scale with calibrated markings, similar to meters normally used on
vacuum indicators. No replacement meter could be obtained and thus the circuit was modified to insert a
‘floati
ng ’3mA me teri ns e r
ieswi ththepump’ spowe rs uppl yout putandt het ripc omparators.Thi s
arrangement is shown in PCINT.SCH. The meter thus encompasses the low pressure end of the scale. The
comparators are located on a circuit board behind the pump power supply.

It is normal for the vacuum to rise and then to recover when repetitive RF pulses are energised. The
microwave power breaks up any high molecular weight impurities within the waveguide into lower
molecular weight material more readily pumped by the ion pump. Similarly, energisation of the filament will
produce the same effect as impurities adsorbed on the now heated filament are being expelled. The pumping
speeds are quite low so it is advisable not to energise the filament at full power at the start of the working
day, but rather to allow time for the pump to recover at lowish gun filament settings.

Safety interlock system.

This system is located in the Van de Graaff control area to the left of the pulse radiolysis instrumentation
rack. The circuit diagrams are LNINTLK.SCH and INTLKL.SCH. The radiation stop in the pulse room,
ope r
a t
edf rom t heVa ndeGr aaf
f’ss afetyinterlocks ystem,ma yonl ybemove dt ot he‘ i
rradi a
te’posi
ti
onif
the pulse room and magnet room are surveyed as far as the Van de Graaff system is concerned. There is no
possibility of irradiating personnel in the magnet room through the Van de Graaff beam line as this would
requiret heus eoft heVa ndeGr aaff‘ bypass’ke y,toe nt erthema gnetr oom.Ac tivationoft he‘bypa
ss’ke
y
drops the pulse room survey, thus rotating the radiation stop to the safe position in the pulse room.

The linac can only be set ready for irradiation (i.e. the main HT, gun filament and focus supplies energised)
if the sequence ‘
Sur veypul seroom,Cl oseout erpul ser oommaz edoorwi thinallowe dt imede lay ,Activate
re d‘irradiate’butt
onont hei nt
e rl
oc ks ystemf r ontpane l’isperf
or med.

This is achieved provided the interlock system operated correctly the previous time it was used; if not the
procedure is disabled (see contacts RLE/2, RLB/1). The sequence is as follows: When the survey button in
the pulse room is pressed momentarily, RLA is energised for a time determined by the 100 k/220 F time
constant, sounding the siren through RLA/1. Should RLA remain energised for some reason, the accelerator
cannot be switched on the next time due to contacts RLA/2 (at the bottom of the diagram). DC power (+27
V nominal) through three series-connected diode s(1N 4001)a ndt hroug htheuppe r‘c los
ed’doo rs witchi s
applied to RLB when the door is closed; the 220 F capacitor across RLB coil takes care of door
switch/relay contact bounce in the next phase of the sequence. At that point RLB/1 changes over, latching
RLBo nthr oughtheuppe rdoor‘ closed’swi t
ch.Cont a ct
sRLB/ 2nowe nergisethe‘ r
oom r ea dy’indi catoron
theinte r
loc ks ys
tem pa ne landt he‘ da
ng er’indi
c atorinthepul ser oom.Cl osingthedoora lsoe nergisesRLD
and thereafter RLE. One of the contacts of RLE, (RLE/1) allows latch RLC and RLC/1 to be turned on when

Copyright© Gray Cancer Institute 1995 15

the‘ radiat
ionon’p us hbut t
oni sa ct
ivatedont hei nterlocks ystemf rontpa ne l
.Re l
aysRLX andRLXX a
re
e nergisedt hr
oug hRLC/ 2,pr ovide dRLD/ 1ha dc ha ngedove r.RLX/ 1c ont
actsa r
ethe‘output
’oft
he
interlock and allow the HT/Gun/Focus to be energised. Contacts RLX/2 are used to illuminate the large
‘Ra diati
onHa zard’wa ll-mounted indicator outside the pulse room while RLXX/1 and RLXX/2 are used to
illuminate appropriately the small red/green indicators next to the pulse room door.

The linac keyswitch must be activated initially for the above sequence to take place. There are two contacts
on the keyswitch. The lower one is simply in series with the interlock output. The upper contact causes RLT
to be energised momentarily through a 4700 F c apaci
tor( andnor ma ll
yma dei .e
.i nactiva t
ed‘ panic’
switches). Contacts RLT/1 then latch RLT on while contacts RLT/2 activate the three phase contactor in the
‘cleaner’sc upboa rd’room.Re l
a yRLSi snormally always energised and maintains single phase power to the
accelerator and the pulse room instrumentation. Should any of the panic buttons be activated both RLS and
RLT drop out and all power is removed. The system may then be restarted only by opening the interlock
systemf rontpa ne landpr e ss
ingthe‘ panicr es et’swi tch;thisturnsonRLTa ndRLSbydi s
cha rgingthe4700
μFc apa cit
ort hroug h15ohmsbutt hekey swi tchmus tbeoni fthet hr
eepha sepowe ristober es t
ored.

In normal operation the keyswitch is turned only at the start and at the end of the working day. Should the
linac be turned off inadvertently at the key s
witch,i twi llnotbepos sibletor est
artitbeforet he4700μF
c apaci
torha sfullydi scharg ed;t hispr ocessma ybes pe ede dupbypr es
singt he‘ panicreset’but t
ona s
described above.

Water chiller unit.

The water chiller/circulator is located in the MechanicalWor kshop’ sme t
a lstores,ne xttot hec orridoroft he
‘tunn ele nt
ra nce’tothel abor a tory.Itiss upplie
df r
om i tsownl oc althr
e epha sesuppl ya ndi sa ctiva t
e dwhe n
the linac keyswitch on the interlock panel is turned on. It supplies water to the accelerator at a rate of up to
22 litres/minute, dependant on the load, at a temperature which is adjustable and which is at present set to 12
deg C. Adjustment to the output temperature can be made on the thermostat which is revealed by removing
the top front panel.

The unit is a standard closed loop system consisting of a chilling compressor and a circulating pump. This
unit can be serviced by any competent refrigeration engineer. The system does not make up for lost water in
the event of a leak and this must be done manually by removing the lower front cover and by operating the
appropriate valves in a logical sequence, one that becomes obvious once the cover is removed. There were
no circuit diagrams supplied with the unit and only partial circuit details are available on LNINTLK.SCH.

Performance of the accelerator.

It is normal for the output energy of electron pulses from linacs to be reduced when the charge per pulse is
increased. This accelerator is no different in this respect and this loading effect is shown in figure 6.

As was described near the beginning of this report, a range of pulse widths and shapes may be obtained and
some typical examples are shown in figures 7, 8, 9, 10.

In all cases the spatial beam shape was optimised by appropriate f ocussinga ndt he‘ 50Hzp ha se’wa s
adjusted optimally to yield the most intense pulse for a given set of injection conditions. This phase delay
was then not altered during the acquisition of this series of electron pulse recordings. In practice, there is
alsol it
tleva r
iationofe lect
ronpul sea mpl it
udea sthee l
ec t
ronpul seis‘ move d’wi thint heRFpul se ,exc ept
nea rther ise/
falltime soft hema gnetron’sout putpul se.The ser ecordingswe reobt ai
ne dwi tht heRFpul se-
to-electron pulse delay fixed and with the electron pulse occurring near the middle of the RF pulse. The
electron pulse was collected on a Faraday cup target which was mechanically and electrically attached to the
beam line. The output from the target was connected to a 50 termination through RG 58 50 cable.

Copyright© Gray Cancer Institute 1995 16

Figure 6. Drop of electron beam energy as a function of total charge per pulse.

Copyright© Gray Cancer Institute 1995 17

 Gun filament set-point 5.4 V Gun filament set-point 5.4 V

(a) 5.0 kV charging 10 kV/div (top), 10 mA/div (bottom) (b) 6.7 kV charging 10 kV/div (top), 10
mA/div (bottom)

Gun filament set-point 5.4 V Gun filament set-point 5.4 V

(c) 9.2 kV charging 10 kV/div (top), 10 mA/div (bottom) (d) 11.7 kV charging 10 kV/div (top), 10
mA/div (bottom)

Figure 7: Injection voltage (upper traces in each panel) acquired with capacitive probe (mounted on gun
pulse transformer) and attenuator & output beam pulse (lower traces in each panel), collected on a 50 
terminated target. Panels a-d are acquired with increasing line charging voltages. All traces acquired at 1
s/div and in single-shot operation mode with 6 RF pre-pulses at 10 Hz.

Copyright© Gray Cancer Institute 1995 18

 Gun filament set-point 6.0 V Gun filament set-point 6.0 V

(a) 5.0 kV charging 10 kV/div (top), 20 mA/div (bottom) (b) 6.7 kV charging 10 kV/div (top),
20 mA/div (bottom)

Gun filament set-point 6.0 V Gun filament set-point 6.0 V

(c) 9.2 kV charging 10 kV/div (top), 40 mA/div (bottom) (d) 11.7 kV charging 10 kV/div (top),
40 mA/div (bottom)

Figure 8: Injection voltage (upper traces in each panel) acquired with capacitive probe (mounted on gun
pulse transformer) and attenuator & output beam pulse (lower traces in each panel), collected on a 50 
terminated target. Panels a-d are acquired with increasing line charging voltages. All traces are acquired at 1
s/div and in single-shot operation with 6 RF pre-pulses at 10 Hz.

Copyright© Gray Cancer Institute 1995 19

 Gun filament set-point 6.5 V Gun filament set-point 6.5 V

(a) 5.0 kV charging 10 kV/div (top), 20 mA/div (bottom) (b) 6.7 kV charging 10 kV/div (top), 100
mA/div (bottom)

Gun filament set-point 6.5 V Gun filament set-point 6.5 V

(c) 9.2 kV charging 10 kV/div (top), 100 mA/div (bottom) (d) 11.7 kV charging 10 kV/div (top),
100 mA/div (bottom)

Figure 9: Injection voltage (upper traces in each panel) acquired with capacitive probe (mounted on gun
pulse transformer) and attenuator & output beam pulse (lower traces in each panel), collected on a 50 
terminated target. Panels a-d are acquired with increasing line charging voltages. All traces are acquired at 1
s/div and in single-shot operation with 6 RF pre-pulses at 10 Hz.

Copyright© Gray Cancer Institute 1995 20

 Gun filament set-point 7.0 V Gun filament set-point 7.0 V

(a) 5.0 kV charging 10 kV/div (top), 20 mA/div (bottom) (b) 6.7 kV charging 10 kV/div
(top), 100 mA/div (bottom)

Gun filament set-point 7.0 V Gun filament set-point 7.0 V

(c) 9.2 kV charging 10 kV/div (top), 100 mA/div (bottom) (d) 11.7 kV charging 10 kV/div
(top), 100 mA/div (bottom)

Figure 10: Injection voltage (upper traces in each panel) acquired with capacitive probe (mounted on gun
pulse transformer) and attenuator & output beam pulse (lower traces in each panel), collected on a 50 
terminated target. Panels a-d are acquired with increasing line charging voltages. All traces are acquired at 1
s/div and in single-shot operation with 6 RF pre-pulses at 10 Hz.

Copyright© Gray Cancer Institute 1995 21

Useful items required during servicing of the accelerator.

Only the most important tools and instruments are listed here; the instruments described were used to obtain
most of the information listed in this report.
High voltage meter: Brandenburg type 88 M
Oscilloscope Fluke type PM 3380A
High voltage probe Tektronix type 6015
General purpose DVM Fluke type 87
Manual single shot trigger In house constructed, drawing OPTRIG.SCH
Fibre optic receiver In house constructed, drawing OPTRIG.SCH
Manual signal source/interface In house constructed, drawing MANTEST.SCH
Linac high voltage attenuator probe In house constructed, atten 2000:1 with installed cables, capacitive
50 ohm coaxial components Two 10:1 attenuators, coaxial T pieces, six through terminations
Electron beam target Faraday cup, in house developed, attached to end of beam line

Description of software and listing of interface registers.

The interface registers and channels are organised as follows:
Data is 8 bits available on port A of PC30A interface card, data flow from PC30A to interface electronics.
Addr essl oc ations( 16of f)
,a ddressa vai
lableo nLSB’ sofpor tBofPC30Ai nt
erf
ac ec ard:

Address hex 0 Function Trigger to electron pulse least significant bits

Address hex 1 Function Trigger to RF pulse least significant bits
Address hex 2 Function 1, 2, 4: trigger to RF pulse most significant bits;
8: enable RF pulse;
16, 32, 64: trigger to electron pulse most significant bits;
128: enable electron pulse
Address hex 3 Function 1, 2, 4, 8, 16, 32: number of RF prepulses;
64: external dose control mode (0 = filament, 1 = pulse
amplitude)
Address hex 4 Function Pulse amplitude (mark-space control) value
Address hex 5 Function Gun filament current (mark-space control)
Address hex 6 Function 1, 2, 4: Electron pulse width;
8: repetitive pulsing on/off (1 = on, 0 = off);
16, 32, 64, 128: 50 Hz repetition rate divisor;
Address hex 7 Function Accelerator systems enabling:
1: enable gun filament power supply
2: enable main HT power supply
4: enable magnetron power supplies
8: enable focus supplies
16: tune up magnetron
32: tune down magnetron
64: injector HT supply on / reset
128: not used.
Address hex 8 Function Focus 1 power supply set-point value (D/A converter #1)
Address hex 9 Function Focus 2 power supply set-point value (D/A converter #2)
Address hex A Function Focus 3 power supply set-point value (D/A converter #3)
Address hex B Function Reserved for quadrupole supply set-point (D/A converter #4)
Address hex C Function Steerer 1 power supply set-point value (D/A converter #5)
Address hex D Function Steerer 2 power supply set-point value (D/A converter #6)
Address hex E Function Steerer 3 power supply set-point value (D/A converter #7)
Address hex F Function Steerer 4 power supply set-point value (D/A converter #8)

Copyright© Gray Cancer Institute 1995 22

No analogue output data from PC30A card to interface electronics is used.

Data from interface electronics to PC30A card;

Digitaldata( 1bi t
)s electedbya ddre
ssavailableonLSB’
sofpor
tBofPC30Ai
nte
rfa
cec
ard:

Address hex 0 Function Confirmation focus supply is enabled

Address hex 1 Function Confirmation magnetron system is enabled
Address hex 2 Function Confirmation main HT supply is enabled
Address hex 3 Function Confirmation gun filament supply is enabled
Address hex 4 Function 10-4 Torr vacuum trip
Address hex 5 Function 10-5 Torr vacuum trip
Address hex 6 Function Water flow present
Address hex 7 Function Water flow absent
Address hex 8 Function Confirmation magnetron tuning down
Address hex 9 Function Confirmation magnetron tuning up
Address hex A Function Magnetron tuning limit reached
Address hex B Function Confirmation injector HT supply enabled
Address hex C Function Confirmation safety interlock system enabled
Address hex D Function Focus 3 power supply fault
Address hex E Function Focus 2 power supply fault
Address hex F Function Focus 1 power supply fault

Analogue data from interface electronics to PC30A interface card

Channel 0 Function Spare / not used

Channel 1 Function Focus 1 power supply output current
Channel 2 Function Focus 2 power supply output current
Channel 3 Function Focus 3 power supply output current
Channel 4 Function Reserved for quadrupole power supply
Channel 5 Function Steerer 1 power supply output current
Channel 6 Function Steerer 2 power supply output current
Channel 7 Function Steerer 3 power supply output current
Channel 8 Function Steerer 4 power supply output current
Channel 9 Function Injector HT power supply output voltage
Channel 10 Function Main HT power supply output current
Channel 11 Function Injector HT power supply output current
Channel 12 Function Vacuum pressure reading
Channel 13 Function Charge per pulse monitor output voltage
Channel 14 Function Gun filament power supply output current
Channel 15 Function Water temperature reading

Copyright© Gray Cancer Institute 1995 23

Listing of accelerator circuit diagrams.

Total number of original accelerator drawings, only paper copies available = 50; these are available for
background information only; do not refer to them for servicing/fault finding purposes.

Total number of drawings in C:\PCB\GLLIN directory = 30

Diagram # 1 PWRDIS.SCH AC Power distribution, circuit breakers etc.

Diagram # 2 GUNMOD.SCH Gun injector, pulse amplifier and gun filament regulator
Diagram # 3 POSPAMP.SCH Pulse thyratron driver amplifier
Diagram # 4 THYRTRIG.SCH Fibre optic magnetron thyratron trigger amplifier
Diagram # 5 MAGTHYR.SCH Magnetron thyratron and RF modulator details
Diagram # 6 PCINT.SCH Manual over-ride computer interface / power supplies enabling
Diagram # 7 TESTPAN.SCH Layout of manual over-ride controls
Diagram # 8 INTCON1.SCH Internal signal & logic wiring / unit interconnections part 1
Diagram # 9 INTCON2.SCH Internal signal & logic wiring / unit interconnections part 2
Diagram # 10 LOVPWR.SCH +/- 15 V and +/- 24 V low voltage power supplies
Diagram # 11 DEFLECT.SCH Beam deflection / steering power supply (one of four)
Diagram # 12 FOCUS.SCH Beam focussing power supply (one of three)
Diagram # 13 FILT.SCH RFI filtering to outside of main Faraday cage
Diagram # 14 MANCONT.SCH Manual control units for external/non PC use of accelerator
Diagram # 15 CABLES.SCH Test cables installed in trunking to pulse room
Diagram # 16 QMON.SCH Toroidal charge / dose monitor
Diagram # 17 QUAD.SCH Details of beam flattening quadrupole
Diagram # 18 LINTRIG.SCH Timing electronics block diagram
Diagram # 19 LININT.SCH Interface to PC / Digital-to-analogue conversion
Diagram # 20 LINDEL1.SCH Timing electronics/repetition and pulse generation circuits
Diagram # 21 LINDEL2.SCH Timing electronics/mains sync/pulse width modulators
Diagram # 22 LINBRD.SCH Layout of timing/interface boards
Diagram # 23 LINPS.SCH Timing and interface electronics power supply
Diagram # 24 ANIN.SCH Interface to PC / Analogue signal conditioning
Diagram # 25 LINSOFT.SCH Details of interface electronics registers
Diagram # 26 OPTRIG.SCH Optical trigger and receiver test modules
Diagram # 27 Spare
Diagram # 28 LNINTLK.SCH Personnel safety interlock panel
Diagram # 29 INTLKL.SCH Interlock system indicators /VdG and water chiller connections
Diagram # 30 INDEX.SCH Index of drawings

Copyright© Gray Cancer Institute 1995 24

Typical circuit waveforms

In figures 11 to 14 are depicted typical waveforms in parts of the linac circuitry. Figure 11 shows the ripple
associated with the DC output of the focus supplies. Figure 12 shows AC current waveforms as monitored
across the sensing resistor at the output of the gun filament regulator with different set-point inputs applied
to the regulator, while figure 13 shows the settling time of the regulator at and following switch-on. Figure
14 shows a typical relationship in time between the RF thyratron current pulse and the electron pulse, with
the scope being triggered by the optical receiver unit.

Figure 11: (above) Output ourrent waveforms

from the three focus power supplies: focus 1:
upper trace, focus 2: middle trace, focus 3:
lower trace. DC zero for each trace is indicated
by a notch to the left of the panel, below each
trace.

Figure 12: (above right) Output ourrent

waveforms from the gun filament regulator,
measured across 2 sensing resistor (traces top
to bottom filament set-points 5.4 V, 6.0 V, 6.5
V, 7.0 V.

Figure 13: (right) Output ourrent waveforms

from the filament regulator at switch-on (top
trace, 100 ms/div) and following switch-on
(bottom trace, 200 ms/div)

Copyright© Gray Cancer Institute 1995 25

Figure 13: ( right)Out putpul seourrentfrom thel ineara ccelerat
or( t
optrace),a ndt he4μsRFt
hyr
atr
on
current pulse (bottom trace) showing how the gun is pulsed during the RF burst envelope.

Conclusion

Thi sma nualde scribest hede velopme ntoft heGr ayLa bor at
or y’
s5Me Vl inearacceleratorupt oNove mbe r
1995. It is not intended to be complete and it is hoped that additions will be described in subsequent pages.
In particular, relatively little operational experience with the accelerator has been had and a detailed log of
all faults should be maintained.

Copyright© Gray Cancer Institute 1995 26

Additions and modifications to the Gray Laboratory Linac during December 1995.

Interlock system.

The operation of the HT and interlock systems was modified in December 1995 to enable the installation of
an HT discharge relay used to bring the HT supply down to zero volts following its de-energisation. This
relay is now shown in the top RH corner of circuit PWRDIS.SCH. This relay is mounted above the HT
transformer and is accessible from the front of the accelerator following removal of the perspex protective
covers. The HT reservoir capacitor is discharged through a 15 K resistor, thereby limiting the current to
around 600 mA, discharge being complete in less than 100 ms. This is in contrast to several tens of seconds
prior to the installation of this relay.

In addition to the HT discharge contacts, two microswitches have been fitted to the assembly. One of these
operates when the relay is fully closed i.e. de-energised, the other when the contacts are open, i.e. the relay
coil energised. In order to prevent the possibility of applying HT to the closed contacts/resistor, the HT start
relaysRLCa ndRLC’a repowe re dt hrought hec oilenerg i
sedmi cros witch.To bring the HT on therefore,
the discharge relay coil is energised, the discharge short is removed, the microswitch operates and this in
turne nergisesr elaysRLC, RLC’ .

The second microswitch is used to monitor the and confirm the fact that the discharge relay contacts are
close d,i .
e.t hatt heHT i sa tzero,whe nt hea cceleratori si na‘ safe ’c ondi ti
on.Shoul dthisa dditi
onal
mi cros witchbe‘ on’int hes afecondition,awa rni
ngs i
reni sa cti
va tedt hroug hthenor ma ll
yc l
ose dc ont
ac t
s
of the interlock output relay RLS. This arrangement also tests the integrity of the contact assembly of RLS in
the following manner: at switch-on, +24V DC is applied momentarily to the siren through the normally
closed RLS contacts and the 470 mF capacitor / 39  current limit resistor across the HT discharge relay
microswitch. Should RLS contacts be open for some reason no such siren activation will take place.
Fur thermor e,whe nRLS c ontac
ti sope n,i.e
.whe nt hepe rsonne ls afetyi nt
e r
locki si nthe‘ i
rradia
te’
condition, thisc apacitori sdischargedt hroug hthepa ral
le l100Kr esist
or ,re adyf orthene xt‘safe’condition
oft hepe rsonnels afe
tyi nterlocks ys
tem.I notherwor ds ,e veryt imet hea cc eleratorisint he‘ saf
e’mode ,the
siren is momentarily activated.

The coil of t heHTdi sc harg er e

layi snotdr ivendi rectlybutr a thert hroug ha ne lect
roni c‘ boos t
’c ircuit
arrangement which provides a high pull-in force while maintaining a low holding coil current once the relay
has energised. No details of this circuit arrangement are provided since they are proprietary to the supplier of
ther elay .Anunf oresee nc ons equenceoft hisisthatthe‘ coil’currenta tt urn-on is around 10 A which loaded
the 24 AC control power to an unacceptable degree: the result was that the timer clutch would on occasions
trip out, causing the start of a new 5 minute timing cycle. This clearly unacceptable situation was remedied
by installing a separate 24 V AC power transformer to power the timer and magnetron filament control
power. No detrimental effects were noticed on any of the other 24 V AC energised coils due to the high
loading of the HT discharge relay.

Charge monitor.

The original charge monitor circuit was found to be unreliable at low charge per pulse operation. The
problem was due to exce ssiveRF i nterf
e rencea ffectingt heope ration oft hec ircui
t’spe ak de tector.
Surprisingly, the RF interference did not affect the operation of the resonant system pre-amplifier and this
was retained in its original form to provide a damped oscillatory signal, of amplitude proportional to charge
per pulse, to the pulse radiolysis dose monitoring circuits. The internal peak dose monitor circuit was taken
out of service pending reconstruction of the arrangement with improved RF screening.

Copyright© Gray Cancer Institute 1995 27

Magnetron tuning.

It was found, under certain conditions of malfunction of the control software, that it was possible for the
magnetron tuning motor to be energised continuously, causing the motor to drive to its end limit(s). This
condition was rectified in the software and in addition the PC tuning outputs were fitted with timing
capacitors/diodes (see PCINT.SCH) so that, should this condition occur in the future, at least the tuning
motor will be moved from its correct position only slightly.

Copyright© Gray Cancer Institute 1995 28