IEEE
Transactions on h’uclrar Science,
Vol. NS-32. No 5, October 1985
RF POWER AMPLIFIER FOR THE CERN SPS OPERATING AS LEP INJECTOR
W. Herdrich, H. P. Kindermann
European Organization for Nuclear Research
(CERN),
CH --1211 Geneva 23, Switzerland
Summary
To permit electron-positron acceleration in the
SPS,
a new 200 MHz RF system will be installed. The
system consists of 32 single-cell, high Q accelerating
cavities,
providing a total peak
accelerating
voltage
of 30 nv.
Each cavity will be equipped with an RF
power amplifier, capable of delivering 60 kW CW or
110 kW pulsed power (pulse length up to 1 second, duty
cycle up to 20 “I.).
The power amplifier will be
mounted directly on top of the cavity. Since the
space in the existing SPS tunnel is very limited, a
compact amplifier is needed, requiring special RF
circuitry design.
Also the choice of material is
restricted due to the presence of ionizing radiation
in the tunnel.
This paper describes the design of the
amplifier and presents the test results, obtained on a
prototype version.
Introduction
A total peak accelerating voltage of 30 HV at 200
MHz is required to accelerate electrons and positrons
in
the
SPS up to the LEP injection
energy of
20 GeV 1.
Since 32 accelerating cavities will be
installed, approximately 1 MV per cavity is required,
taking into
consideration a bit of redundancy. To
achieve this voltage a power of about 60 kW must be
delivered to each cavity. Disregarding the costs,
there is not enough space available in the SPS access
pit and tunnel (60 m underground) to install 32 high
power transmission lines. The power amplifier will
therefore be mounted directly on top of the cavity.
For the SPS working as LEP injector, only pulsed
RF operation is needed.
However,
the new RF system
will also be used in the proton-antiproton collider
mode, where CW operation is required. The power
amplifier is therefore rated for CW or pulsed output
power of 60 kW.
Since the RF power tetrode is capable
of
delivering
higher
output power under
pulse
condition, the amplifier is designed in addition for
110 kW output power with a pulse length of up to
1 second and a duty factor of up to 20 X.
The amplifier will be mounted on the cavity via a
short coaxial transmission line.
This line transforms
the cavity input impedance
to a parallel resonance
circuit at
the amplifier output and permits the
installation of
directional couplers for measuring
forward and reflected power. A coaxial ceramic window
with a coupling loop is installed between cavity and
transmission line.
A characteristic
impedance of
10 R has been chosen for this window, resulting in a
minimum
electrical field strength under matched
condition (a characteristic impedance of 30 R in air
corresponds to
10 B with a
dielectric constant
of 9).
The cavity resonance impedance of about 9 MS2
is transformed via the coupling loop to the low values
required, i.e. 16 R for 60 kW and 8 R for 110 kW
operation. These low impedances keep the electrical
field strength at the window small and allow a compact
amplifier design.
Since
the drive
sys tern
will be installed on the
surface, i.e. up to 300 m away from the final
amplifiers, low drive power is appreciated to minimize
size and losses of the RF drive cables.
Amplifier design
The design of the amplifier is mainly determined
by space limitations, reliability and fast exchange in
case of a breakdown.
To get a compact amplifier the RF output circuit
is
positioned around
the tube socket. In this
arrangement, the RF input line together with the DC
bias,
the filament and cathode cables and the hoses
for the control grid - screen grid circuit water
cooling pass through the RF output circuit and are
decoupled via folded h/4 stubs.
Neither the output
coupling nor the input matching are adjustable.
Only
tuning of the amplifier is foreseen.
To permit reliable operation in the radioactive
environment of the SPS tunnel, care must be taken in
the choice of materials of the amplifier.
Teflon must
be avoided. Where organic materials are required,
only
Polyimide (Kapton, Vespel),
Polystyrene,
Polyamide-imide (Torlon), Polyphenylene
oxide
and
Silicone are employed, whenever possible.
Fast exchange of a
faulty
amplifier
is
made
feasible by using
connectors
for all electrical
connections (RF drive, HV anode,
bias, cathode,
filament,
interlock),
quick disconnect type couplings
for water and air cooling hoses and a quick flange RF
connection between amplifier and cavity. A small fork
lift is used to transport the amplifier and to place
it onto the cavity flange.
The time needed by one man
to exchange an amplifier completely is less than five
minutes.
A photo of the amplifier, mounted on a coaxial
line, is shown in Fig. 1. One can recognize three of
the four output circuit feed-throughs, i.e. RF input
on the right, filament in the middle and control grid
- screen grid circuit cooling on the left. The
cathode feed-through is identical to the one of the
filament,
but placed opposite to it (hidden by the
tube housing).
The tube housing contains the tetrode,
the anode blocking capacitor, the anode cooling hoses
and the anode filter. The housing is simply lowered
into the socket and kept in place by its own weight
and a number of RF and high current contacts. Water
inlet and outlet for the anode water cooling and HV
anode
connector
are
mounted on top of the tube
housing,
whereas the cooling air inlets are installed
on a manifold at the bottom of the amplifier.
The
bias connector cannot be seen on the photo, since it.
is mounted behind the RF input.
The total height of
the amplifier is 75 cm and its maximum diameter 45 cm.
A watercooled,
metal-ceramic tetrode (Siemens RS
2058 CJ) is employed, rated at 90 kW anode dissipation
and capable of delivering a peak cathode current of
100 A.
The tube is operated in the grounded screen
grid mode and in class AB as a compromise between high
efficiency and low drive power.
OOl&9499i85/1003-2794301.OOO 1985 IEEE
© 1985 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material
for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers
or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
2195
conductor
and a brass
outer
conductor,
both
silver-plated.
Assembly of the capacitor is made by
winding the Kapton foil around the inner conductor and
by sliding the outer conductor, after being heated to
about
ZOO"C, over
the
inner conductor. At room
temperature a force of about
1.5 tons is required to
separate the two conductors again.
When during high
power operation the capacitor warms up, this force is
even higher,
since due to the higher dilatation of
aluminium compared to brass, the inner conductor will
be pressed against the outer conductor.
The capacitor
has
a low frequency capacitance of about 10 nF. At
200 RH%
it is
approximately an open-ended h/4
transmission line, representing a very low impedance
to the output circuit current.
The tuning of the amplifier can be adjusted by
four
tuning
nuts, placed on the fOUr output
Circuit
feed-throughs.
By tuning these nuts, the screen grid
base plate together with the tetrode and its housing
can be
moved.
Fig. 1
601110
kw 2.00 MHz power amplifier
In the following the main parts of the power amplifier
will be described briefly.
Eoutput circuit
The load impedance seen by the amplifier, i.e.
16 G and 8 $2, respectively, must be transformed to
the optimum anode impedance of the tetrode for high
efficient operation.
At a DC anode voltage of 10 kV
the optimum anode impedance is about 600 $2 for 60 kW
and about 300 R for
110 kW output
power. In
principle, two h/4
transformations are
used to
achieve this task (see Fig. 2).
Cavity ,
;:i: ($-y--~---q
(Ii;;; ?lane
I
w Principle of amplifier output circuit
In practice a more
complicated
transformation is
obtained as can be seen in Fig. 3.
The 3 R line is
made of a
170 mm ~1 coaxial line, followed by a
conical line and a 390 mm 4 coaxial line.
The 18.5
R line is approximated by a combination of coaxial
and
radial lines,
followed by the
transformation
inside the tube.
The approximation is not valid for
all output impedances, but works very well in the 8 to
16 R range.
The main advantages of this type of
output circuit are a compact layout, since no special
output
coupling
is
needed and
consequently a
homogeneous field distribution.
The impedance of the control grid - screen grid
circuit
should be low for the fundamental
and
its
harmonics at the plane of
the active system of the
tetrode.
A lossy radial line (Fig. 3), terminated
with a ring of capacitors
(32 capacitors of 2 nF,
each) provides
this performance.
Eccosorb W 124
(Emerson + Cuming) has been choosen as dielectric for
the radial line.
This material is machinable and has
adequate electric and magnetic properties over a wide
frequency range.
Again,
a Kapton foil is inserted
between Eccosorb and control grid and screen grid
plate, respectively,
to reduce the risk of a voltage
breakdown.
DC and filament feeding
The anode DC voltage is fed via an anode filter
and a feed-through to the tube (Fig. 3).
The filter
consists of an
LC network and a lossy cable to
suppress the fundamental and its harmonics.
Higher order modes
are
SUppreSSed by coupling
loops terminated in 50 R.
The loops are orientated
in such a way that they couple to the axial magnetic
field of the
spurious mode,
but not to
the
circumferential magnetic
field of the fundamental and
its harmonics. Tuning of these loops to the frequency
of a special mode is not required.
The control grid bias voltage is connected to the
outer conductor of the RF input line.
This connection
is not shown in Fig.
3, since it is mounted behind the
RF input part. Also here an LC-filter and a lossy
cable are employed to suppress RF on the DC line.
The anode blocking capacitor consists
of a low
impedance
coaxial line,
using 5 layers
of 50 )J
Kapton film as
dielectric, an
aluminium inner
RF input circuit
The low input impedance of the tetrode in grounded
grid
operation is
transformed via
a
15 R coaxial
line,
a radial line and a 25 R coaxial line to an
impedance
approaching 50 R
(Fig. 3).
Then a
T-junction,
terminated on one side with
an
open-ended
25 R coaxial
stub, is used to
compensate the
reactive part of the admittance.
The real part is
transformed via a x/4
coaxial line
into 50 Q.
Since the inner conductor of the input line is on
cathode potential and the outer conductor on control
grid potential,
a ceramic coupling capacitor of 3.3 nF
and four ceramic blocking capacitors of 3 nF, each,
are employed to permit DC decoupled RF feeding. A
Kapton foil is inserted between outer conductor and
ground to avoid DC voltage breakdown.
The mounting of
coupling and blocking capacitors is designed in
such a
way, that matching is realized when
the 50 R coaxial
input line is terminated with 50 R.
Control grid - screen grid circuit
No separate input is needed for the screen grid,
because it is grounded.
The cathode connection is
made via one of the two filament cables.
The filament conductors
pass through
the RF input
circuit near to
the socket of the
tube.
Decoupling
from RF is achieved by mounting each conductor in a
A/4 line configuration. short- circuited via a
ring
of ceramic capacitors (6 capacitors of 3 nF, each).
Amplifier cooling
The water hoses for the anode cooling are placed
around the anode blocking capacitor (Fig. 3).
They
have a length of 3.5 m, each, to keep electrolytic
effects
small.
Screen grid and control grid bases are
also water-cooled to maintain the Eccosorb in the
radial line at low temperature and to minimize the
amount of warm air in the SPS tunnel.
Air cooling is
required for the electrode rings and the ceramics of
the tube
and for the other
components of the
amplifier.
A manifold
is placed around the 3 R
output circuit line. The air passes through holes in
the
separating wall into this line and is then
directed paT;tly via the RF output to the tube socket
and partly via the upper part of the output circuit to
the anode ceramic. A “Vespel” disc
(glass fiber
reinforced Kapton) guides the air to the lower part of
the ceramic,
near to the screen grid electrode ring,
where good cooling is needed.
The main air outlet is
through the RF
contacts of the
anode blocking
capacitor and through holes in the cylinder around
these contacts.
A small fraction of the air passes
through
the four output circuit feed-throughs and
through the
tube
housing, cooling RF
input
and
filament lines and coupling and blocking capacitors.
To avoid overheating at the inside of the filament
ring at the bottom of the tetrode, additional air flow
is required at this spot,
which is achieved by two
small hoses,
connected directly to the cooling air
manifold.
Fig. 3 Cross section of the 601110 kW
200 MHz power amplifier
Test results
-_--
A prototype
amplifier
was first tested on a
terminating load. The 50 R load is transformed via
A/4 coaxial lines to 16 R for CW and 8 R for
pulsed
operation.
The following
results have been
obtained at 200 KHz:
pulsed operation
CW operation
pulse length 1s
duty cycle 20%
anode voltage
10 kV
10 kV
anode
quiescent current
0.5 A
0.5 A
anode current
9.4 A
17.0 A
screen grid voltage
900 v 900 v
screen grid curent
320 m A
405 m A
control grid voltage -200 v -200 v
control grid current 105 InA
780 n-IA
filament voltage
12 v 12 v
filament current 180 A 180 A
RF output power 62
kW 114
RF drive power 1.8 kW 5.2 kW
gain
15.4 dB
13.4 dB
anode efficiency
64%
64%
The amplifier proved to be stable at all possible
RF and DC power levels.
No higher
order modes or
other spurious oscillations were observed.
RF
input matching has
been
checked for cw
operation with five different tetrodes RS 2058 CJ from
the series production.
The reflection coefficient was
less than 5% in all five cases.
After these tests on a terminating load, the
amplifier was mounted on the cavity. Only CW
operation, with the coupling loop adjusted to 16 R
cavity input impedance, has been tried so far.
Also
under these conditions, higher order modes were not
observed;
however,
a spurious oscillation occured at
about
198 MHz. At this frequency , the low
off-resonance cavity input impedance is transformed to
a very high impedance at the anode plane of the
tetrode. To suppress the oscillation good screening
of the driver system is required, to avoid external
feedback. If
necessary, an optimisation of
the
coupling loop - transmission line configuration could
be envisaged to shift the frequency of the spurious
oscillation
further away
from
the
operating
frequency. In the present test set-up configuration
the transmission line is shorter than required to
transform the cavity input impedance to a parallel
resonance circuit at the amplifier output plane. This
results in
a compact layout,
but as a consequence,
within
the cavity bandwidth
the
anode impedance
reaches its maximum not at cavity resonance, but at a
higher frequency and the spurious frequency appears
near the operating frequency.
On the other hand the
system will in
the
definitive installation be less
sensitive to this type of spurious
mode due
to the
long distance between driver and final amplifier and
the RF drive cable attenuation.
After suppression of the spurious mode in the test
set-up, full CW cavity power of 60 kW was reached.
Acknowledgements
We would like to thank all the members of the SPS
who participated
in the realisation of this power
amplifier,
especially C. Zettler for many valuable
suggestions during the development, R. Gueissaz for
the design and P. Griessen for the assembly.
Reference
[ll
LEP Design Report, Vol. 1, CERN, Geneva. 1983.