Conceptual Design Of The Optical Scheme For
A Multichannel Martin Puplett Interferometer
For Perpendicular And Oblique ECE
Measurements On JET
A.Simonetto1, C.Sozzi1, S.Garavaglia1, J.Fessey2 and JET EFDA
contributors
1
IFP-CNR, Associazione EURATOM-ENEA-CNR sulla Fusione, v. Cozzi, 53, 20125 Milano ITALY
2
EURATOM-UKAEA Fusion Association, Culham Science Centre, Abingdon OX14 3DB UK
Abstract. The single channel Martin-Puplett fast scanning interferometer existing at JET
underwent a redesign of the optical scheme, increasing the number of channels and complying
with the requirements of oblique ECE measurements. This work deals with the electromagnetic
design of the instrument.
Email: simonetto at ifp.cnr.it
INTRODUCTION
The single channel Martin-Puplett fast scanning interferometer existing at JET [1]
underwent a redesign of the optical scheme, increasing the number of channels and
complying with the requirements of oblique ECE measurements [2,3].
Radiation collected by oblique viewing antennas [4] is elliptically polarized. In the
absence of broadband elliptical to linear polarizers, at least two orthogonal linear
components need be measured for each oblique ECE antenna. And at least one
channel must be preserved for standard radial-view measurements.
DESIGN
The moving rooftop reflector consists of four spiral shaped sectors on a wheel of
150mm radius (at rooftop vertex) with a slope of 4.85 deg. The reflector is 30mm wide
in the radial direction.
The input waveguide for radial view ECE is a standard rectangular S-band WR284
(72.136 x 34.036 mm) used for the low-attenuation TE01 mode (instead of TE10).
Oblique ECE antennae use 27.75mm i.d. smooth circular waveguide. The mode at
the waveguide aperture is assumed TE11 on the ground of its lowest attenuation among
polarized waveguide modes in the long transmission line.
The band of interest is from 75 to 400GHz.
The conceptual design was made using standard Gaussian beam optics:
wz  w 0
 2z 2
1  2  ,
k w 0 
(1)
2Rz 2z
k w 02


,
(2)
k w 02
k w 02
2z
1
1
1

(3)

 ,
RIN ROUT f
where w0,w are beamwaist
 and beam radius, z is the distance from the waist, k is the
propagation constant, R is the phasefront radius and f is the focal length of the mirror.
A confocal system, made of elliptical mirrors, was designed to make performance
 The rooftop reflectors were placed at the image plane
nearly frequency-independent.
and the input waveguide on the object plane of a 2 mirror confocal telescope. The
focal lengths of the mirrors are identical, i.e. magnification is 1, given the similarity in
the circular waveguide diameter and the movable mirror width. A second confocal
telescope of magnification 0.43 (focal length of the last mirror 86.486 mm) was used
to image the reflectors on the output waveguide (12mm i.d.). The best coupling
between a TE11 and the first Gaussian mode is 86.6% for a ratio of waist to waveguide
radius w/a=0.768. The beam truncation at the input waveguide is therefore 2a/w=2.6,
and this ratio was used throughout the system to dimension the optical surfaces.
Oblique ECE Channels
A polarizer grid was placed after the first elliptical mirror to separate the two linear
components. An angle of 90° was chosen arbitrarily between the reflected and
transmitted beams. The splitter-recombiner grid of the interferometer was placed
between an elliptical mirror and the rooftop reflectors. The scheme is shown in fig 1.
These constraints fix the geometry of the oblique ECE channels. A focal length of
200mm was chosen as a compact size satisfying the requirements in the geometry
described above. Table 1 collects the beam and optics parameters.
TABLE 1. Summary of beam and mirrors parameters [mm]
Location (distance along axis)
Beam radius (frequency)
Input waveguide (z=0)
10.656
First input mirror (z=200)
26.15 (75GHz)-11.56 (400GHz)
Intermediate waist (z=400)
23.88 (75GHz)-4.48 (400GHz)
Last input mirror (z=600)
26.15 (75GHz)-11.56 (400GHz)
Moving rooftop (wheel) (z=800)
10.656
First output mirror (z=1000)
26.15 (75GHz)-11.56 (400GHz)
Intermediate waist (z=1200)
23.88 (75GHz)-4.48 (400GHz)
Last output mirror (z=1286.486) 24.32 (75GHz)-6.43 (400 GHz)
Output waveguide (z=1372.972)
4.61
Mirror curvature radius
642.473 in, 290.401 out
290.401 in, 642.473 out
642.473 in, 290.401 out
295.539 in, 122.267 out
-
In order to get a full back-reflection at the wheel rooftop, the projection of the input
raypath on the plane of the wheel must be along the tangent to the rooftop vertex. The
focal length dictates the distance along the ray (200mm), therefore the centres of the
last input mirrors, in front of the wheel, are placed at 200cos(4.85)=199.283mm in
front of the spiral mirrors' midpoints, at an angular displacement of
arctan(200*sin(4.85)/150)= 6.434 deg in the "downhill" direction with respect to the
reflection point on the wheel, and at a radius of √[1502+(200sin(4.85))2]= 150.951mm.
FIGURE 1. Condensed optical scheme for oblique and radial ECE channels.
FIGURE 2. Schematic optical components layout for oblique ECE channels. Grid at output w/g
omitted.
As a consequence, the reflected beams should point slightly "inwards". To avoid
that, one should have given up the requirement of having identical raypaths for the two
channels in a pair. The position of the mirrors along the wheel is conditioned by the
need to pair them in order to measure two linear orthogonal components from the
same waveguide: separating the polarization components somewhere else outside of
the instrument would have made it much more modular, but this configuration was
chosen because it was closer to the requirements. Therefore, the angle between
incident and reflected beams at the last input mirrors is 85.18deg for one channel and
94.82deg for the other in the pair. Since the difference is small, the mirrors were made
for 90 deg like the others, and tilted properly, accepting a small aberration [3].
FIGURE 2. Schematic instrument layout.
Radial ECE Channels
Two identical pairs of oblique ECE channels are placed in a radially symmetric
arrangement with respect to the axis of the wheel. The space between them was used
for perpendicular ECE. One channel was requested, but the space was enough for two.
The best coupling between the waveguide and the first Gaussian mode is only 60%
for w/a=0.695, where a is the long side of the waveguide and w the beamwaist. To
achieve better coupling one would need considering elliptical modes with wx/a=0.352,
wy/b=1.070, getting 88% in power. But cylindrical mirrors would be required for
matching the elliptical beams to the instrument. A waveguide up-taper to 72x72mm
was used instead, achieving 84.3% power coupling into the first Gaussian mode with
w/a=0.43. This beam is adapted to the same optical scheme of oblique ECE by
replacing the first mirror. An elliptical mirror of 583.25mm focal distance would make
the system large. A parabolic mirror of the same focal length, very close to the
waveguide (100mm, arbitrarily chosen), would do. The phasefront curvature at that
point is 11.56m at 250 GHz, so the use of a paraboloid is justified. But the beam
radius is only 31.1mm at the mirror location, giving a maximum throw of
kw2/4=505.6mm, less than the required focal length. Increasing the distance between
waveguide and mirror would make the system larger, so a focal length of 480mm was
chosen, accepting a reduced coupling (80% instead of 84%) to a Gaussian beam with a
larger radius. Coupling was evaluated by propagating the desired beam back to the
waveguide, and computing the coupling between the desired beam (25.574mm waist
there) and the waveguide mode. The truncation at the wheel, caused by the increased
waist (12.95mm waist, i.e. truncation of 2.32w), has negligible effect. The resulting
beam parameters are summarized in table 2.
Similarly to the oblique ECE channels, incident and reflected beams at the last
input mirror are not exactly orthogonal, their angle being 89.457 deg. No significant
aberrations can result from using mirrors designed for 90 deg.
The long distance between the parabolic mirror and the last input one allowed
insertion of additional components, so the optical path was folded with flat mirrors,
allowing the addition of a second identical channel for perpendicular ECE. Moving a
flat mirror and a polarizer grid to a different location, the instrument can be used to
measure both linear polarization components from one input waveguide or either
linear component from two input waveguides.
TABLE 2. Summary of beam and mirrors parameters [mm]
Location (distance along axis)
Beam radius (frequency)
Input waveguide (z=0)
25.574
First input mirror (z=100)
26.05 (75GHz)-25.59 (400GHz)
Intermediate waist (z=580)
same as oblique
...same as oblique from here on ...same as oblique from here on
Mirror curvature radius
infinity in, 480 out (parabolic)
...same as oblique from here on
Polarization
The circular waveguide routing was computed in order to have the required
polarization at the input of the instrument. But inside the instrument, the geometrical
constraints described above make impossible a perfect polarization (and power, as a
consequence) balance between reference and moving arm of each interferometer. In
fact, correct operation of a rooftop reflector requires the input polarization to be at 45
deg from its tip, and the splitter grid was positioned to satify this requirement. As a
result, all channels have an intrinsic imbalance between the two arms: 39.75/60.25%
and 62.10/37.90% for the oblique ECE channels, and 69.11/30.89% for the
perpendicular ones. The consequent reduction in fringe amplitude is not significant
(97.9%, 97.0%, 92.4% respectively as compared to the ideal case).
REFERENCES
1.
2.
3.
4.
M. Zerbini et at, 15th Topical conf. on High Temperature Plasma Diagnostics, 2004
E. de la Luna et al, "Recent developments of ECE diagnostics at JET", EC-13, 2004
C.Sozzi et al, this Conference.
C. Sozzi et al, Fusion Engineering and Design 74, 691 (2005).