Study of Ka-Band Components for Future
High-Gradient Accelerators
G.G. Denisov, S.V. Kuzikov, A.A. Bogdashov, A.V. Chirkov,
J.L.Hirshfield*, A.GXitvak, V.I. Malygin, M.Yu. Shmelyov
Institute of Applied Physics, Nizhny Novgorod, 603095, Russia
*Omega-P, Inc., New Haven, CT 06511, USA
Abstract. This paper presents a concept of quasi-optical Delay Line Distribution System (DLDS) for
feeding a future Ka-band linear collider. According to this concept the DLDS is based on oversized
waveguides. In such waveguides the so-called image multiplication phenomena seem to be very
attractive for power launching, extracting, combining, and splitting of waves. Few different schemes
of DLDS mode launchers are compared. The design of other key RF components is also considered.
Recent low power tests of mode launchers, miter bends, directional couplers, RF loads, and a mode
converter are discussed.
INTRODUCTION
The known scaling of accelerating gradient with frequency, which is approximately
linear, provides strong motivation for developing high-frequency accelerating
structures and associated components. The most compelling reason for designing,
building and evaluating Ka-band components is to establish this anticipated scaling
with solid laboratory evidence; and, of course, to demonstrate the high acceleration
gradient.
The DLDS concept is considered as one of perspective solutions for feeding of highgradient accelerators at 11.424 GHz [1-3]. According to the projected schemes the
DLDS is based on relatively oversized delay lines operated with one or several modes
and mode launchers completely or partially constructed on the base of single-mode
waveguides. Scaling of the mentioned projects to Ka-band seems to be not acceptable
because of higher Ohmic attenuation as well as bigger RF power density, which causes
a threat of breakdown. More natural way is to use pure quasi-optical solutions in the
design of each DLDS component.
MODE LAUNCHERS
A key component of any DLDS project is a mode launcher, which allows feeding
consequently different acceleration sections by means of proper setting the phases of
RF pulses from different amplifiers. In the scheme, shown in Fig. 1, the mode launcher
has four input channels and the same number of output channels providing four times
CP647, Advanced Accelerator Concepts: Tenth Workshop, edited by C. E. Clayton and P. Muggli
© 2002 American Institute of Physics 0-7354-0102-0/02/$19.00
476
34 GHz RF sources
Combiners/
Mode launcher
TE01 delay lines
To the 1st
To the 2nd
acceleration acceleration
section
section
Ill
lU
To the 3rd
TO the 4th
acceleration acceleration
section
section
Figure 1. Scheme of 34 GHz DLDS.
power gaining. The delay lines behind the mode launcher are assumed to be oversized
circular cross-section waveguides operated with axis-symmetrical TE0i mode.
We considered two possible versions of the mode launcher both based on image
multiplication phenomena in an oversized waveguide [4]. In the first version (Fig. 2)
the mode launcher is shaped of a smooth oversized (waveguide size is a) rectangular
waveguide where image multiplication occurs at one coordinate only. Four TE0i
modes at the input of rectangular waveguide are combined into one of four output
channels. The position of the resulted channel depends on the mutual phases at the
input. The required length of the waveguide equals to a2A.
The mode launcher was calculated using Kirhgof's approach, and results were
published in papers [4-6]. According to these results the wider waveguide the higher
0 rc/4 -371/4 0
0-371/4 Ti/4 0
UTKXTK^T^sXTKlx^'^
71 -7C/4 -7C/4
0
0 -7C/4-7C/4 71
a)
b)
Figure 2. Mode launcher based on rectangular waveguide a - scheme, b - photo of the prototype.
477
Ohmic losses level
Ohmlc losses level
£ -10 —
§
160
166
LA
a)
b)
Figure 3. Measurements of mode launcher efficiency when incident wave is launched into a
channel near the wall (a), or the wave is launched into a channel near the center of waveguide (b).
efficiency of the launcher. For example, at 34 GHz, width a=120 mm, and length
L=1600 mm provide 99% efficiency of the power summarizing.
Preliminary tests were performed with the launcher prototype scaled to 4-times
higher frequency in order to minimize launcher sizes and simplify its fabrication. This
resulted in higher Ohmic losses (about two times) and more critical alignment of parts.
The launcher was fed by one wave and the output field structure and reflection
coefficient from specially shaped mirrors at the launcher output were measured.
The results of the measurements, carried out for two regimes, are shown in Fig. 3,
where reflection is plotted as a function of launcher length normalized on a
wavelength. The first used mirror provided phases of four reflected waves at the output
end so, that the reflected waves were combined again in the input waveguide where the
initial wave was excited. The curves marked by 1 in both Fig 3a and b show the
reflection in this case. This reflection was actually the squared efficiency of the
launcher, because all other channels were opened and the scattered radiation was free
to go away. The measured difference between the reflection and Ohmic losses level is
rather small: 0.5 dB.
The curves marked by 2 correspond just to the flat mirror reflectors. In such a case
according to image multiplication effect the reflected waves formed two identical
waves at input cross-section. One of them came back into the feeding channel, but the
second wave was radiated away. So, -3 dB extra losses, measured in this scheme, are
agreed well with a theory (Fig. 3).
The curves marked with 3 correspond to the case when reflected radiation was
combined all power in another channel (not in the feeding channel). In this case we
could measure the scattering radiation only. Its level was about -15-20 dB .
478
The second version of a mode launcher is based on image multiplication in square
cross-section waveguide with an impedance corrugation of walls (Fig. 4). The infinite
impedance allows combining TE0i modes of a circular waveguide, and thus, any mode
converters, in order to match such mode launcher with TEoi delay lines, are not
needed. The length of the launcher equals 2a2/'k. For tests we used four-times higher
frequency and selected sizes of mode launcher, which provided 93% efficiency, with
mode purity at operating frequency estimated as 96-97%. It allowed us to measure
field structures at the input and the output cross-section (Fig. 5).
The scheme of efficiency measurements coincides with the scheme chosen for the
rectangular mode launcher described above. In our tests we found out that the
a)
b)
Figure 4. Mode launcher based on square cross-section waveguide: a - scheme, b - photograph of
the tested prototype.
b)
c)
Figure 5. Measured intensity (Ex2+Ey2) at the launcher input (a) and at the output: horizontal (b)
and vertical (c) E2-components.
1:0 ——,
Estimated accuracy of
waveguide fabrication
1
0:8 -f
:'
1.000
Figure 6. Measurement of efficiency
of the TEoi mode launcher.
;t
' I ' ' ' ' I ' '
LOOS
1.010
!
« I
1:!Q:i?S
Figure 7. Calculation of efficiency for the
TE0i launcher with non-equal walls.
479
efficiency of such a mode launcher is very sensible to the accuracy of manufacturing.
In particular, difference in sizes of the walls results in an essential drop of the
efficiency in comparison with the ideal case (Fig. 7). That was the main reason, why
we measured rather high power losses (Fig.6). Nevertheless, taking into account the
actual accuracy of fabrication and purity of the incident TE0i mode, one concludes that
the efficiency of such a component at 34 GHz is able to be high enough.
The image multiplication approaches may be used for designing of mode launchers
with more than 4 channels, if higher power gain of DLDS is necessary.
COMBINER-SPLITTER
It is used for combining outputs of RF sources into one waveguide. In particular, it
is important for the 34 GHz magnicon, which is designed with four standard
rectangular waveguide outputs.
Comparing different schemes we found that the simplest and most efficient scheme
is the wave combining in oversized
rectangular waveguide like in the
second version of proposed mode
launchers. Dimensions of such a
combiner are moderate, and the
scheme of the combiner is extremely
simple. The plot of efficiency
calculated for combiner parameters
a=b=6Q transverse sizes and L=199
mm length, are shown in Fig. 8.
The opposite dividing scheme can
be used for power distribution over
different acceleration sections.
Frequency, GHz
Figure 8. Calculation of combiner-splitter
efficiency.
MITER BEND
Miter bends are used to provide a required trajectory of transmission lines. The
component is a very critical point for TEoi circular cross-section transmission lines. To
make diffraction losses in the miter bend lower a mode combination is prepared in the
place of the bend [7]. This mode mixture (75%TEoi+24%TE02+l%TEo3) provides a
local decrease of the fields near edges, which inevitably exist in any miter bend.
According to this idea the improved miter bend consists of two symmetric mode
converters and a plane mirror (Fig. 9).
480
TBW
Igll
Figure 10. The measured efficiency of the
miter bend.
Figure 9. Scheme of quasi-optical miter bend.
Such an improved miter bend was designed, manufactured for the required 34 GHz
frequency, and tested at low power level using a special reflection mirror. This mirror
provided entire backward reflection of proper TEoi mode and small reflection for any
other modes because of diffraction losses. The results of such measurements are shown
in Fig. 10, where the dependence on frequency for the reflection coefficient is plotted.
This figure demonstrates that efficiency keeps level of 99% at approximately 3%
frequency band.
WINDOW
There are different dielectric materials being under study for their use in high-power
millimeter wave windows. Parameters of some are presented in the table:
BN
Thermal conductivity W/(m K)
Density g/cm2
Specific heat capacity J/ (g K)
Permittivity
Loss tangent (10~5)
30-60
2.25
0.8
4.4-4.8
80-120
Si3N4
50-60
3.4
0.6
7.9
10-20
C (CVD)
1500-2000
3.5
0.5
5.6-5.7
1-10
The most advanced concepts of high-power windows exploit the idea of pure running
wave through a dielectric disc of the window [ 9 ]. In this case the electric field on the
disc surface can be diminished nm times (n is refractive index of the material). The
window designs are based on circular TEoi mode propagation through the window.
This mode has zero field on the waveguide walls and field pattern is well distributed
over the window surface.
481
DIRECTIONAL COUPLERS AND LOADS
To monitor and to measure
the power of microwave
such components as bidirectional couplers and
vacuum calorimeters were
developed. The components
are based on WR28-type
waveguide.
A
spiral
stainless steel waveguide
channel
with
rounded
surfaces is used for
absorption of the waves.
The
components
are
intended to operate with
high-power microwave and
high-vacuum conditions.
Figure 11. Photo of the calorimetric loads and bi-directional couplers.
TEn-TEoi MODE CONVERTER
This component is used for coupling
of standard rectangular or circular
waveguides with TE0i delay lines. The
simplest solution is a serpent-like mode
converter [6]. It is efficient and enough
compact at 34 GHz. In particular, the
tested converter (approximately 100 mm
in length) provides 98±2% efficiency at
relatively broad Af/f=7% frequency band
(Fig.12).
31
32
33
34
35
36
37
Frequency, GHz
Figure 12. TEn-TE0i converter test. The
reflection coefficient corresponds to squared
converter efficiency
SUMMARY
The main components of Ka-band transmission line for future accelerator system
were designed, fabricated and tested in low power experiments. The test results agree
well with the design parameters. Therefore, Ka-band DLDS seems to be feasible.
High power microwave tests with the 34 GHz magnicon are planned for 2002-2003.
482
ACKNOWLEDGMENTS
This work was supported by the Division of High Energy Physics, US Department
of Energy.
REFERENCES
1. H. Mizuno, Y. Otake. "A new RF power distribution system for X band linac
equivalent to an RF pulse compression scheme of factor 2N", presented at the 17th
International Linac Conference (Linac94), Tsukuba Japan, 1994.
2. C.Nantista, S. Tantawi. "Overmoded rectangular waveguide components for a
multi-moded RF power distribution system", Proc. of EPAC 2000, Vienna,
Austria.
3. S. Tantawi "Active and Passive Multimoded Components for High Power RF
Systems". 9th International Workshop on linear colliders LC2002, California,
SLAC, February 4-8, 2002.
4. G.G.Denisov, S.V.Kuzikov. "Microwave systems based on controllable
interference of paraxial wavebeams in oversized waveguides", Proceedings of the
International Workshop Strong Microwaves in Plasmas, Nizhny Novgorod, August
2-9, 1999,Vol. 2, pp. 955-960, 2000.
5. S.V.Kuzikov. "Wavebeam Multiplication Phenomena to RF Power Distribution
Systems of High-Energy Linear Accelerators". International Journal of Infrared
and Millimeter Waves, Vol. 19, No.ll, 1998. pp. 1523-1539.
6. S. Kuzikov, G. Denisov, M. Petelin. "Quasi-Optical Components for Linear
Colliders Fed by Millimeter Waves", presented at VIII International Workshop on
Linear Colliders (LC99), Italy, 21-26 October 1999.
7. V.L.Belousov, A.A.Bogdashov, G.G.Denisov, A.V.Chirkov, S.V.Kuzikov. "New
Components for TE01 Transmission Lines", Proceedings of the International
Workshop Strong Microwaves in Plasmas, Nizhny Novgorod, August 2-9, 1999,
Vol. 2, pp. 943-948, 2000.
8. O. Nezhevenko "Centimeter and Millimeter Wave Magnicons for Linear Collider
Application" 9th International Workshop on linear colliders LC2002, California,
SLAC, February 4-8, 2002.
9. W.R.Fowkes, R.S.Callin, E.N.Jongewaard, D.W.Sprehn, S.G.Tantawi, A.E.Vlieks.
Recent advances in high power RF windows at X band. AIP Conf. Proc. 474: 289295, 1999.
483