Septum Polarizers and Feeds

Paul Wade, W1GHZ ©2003

The septum feed1,2 was first described by Zdenek, OK1DFC, at the 10th International
EME Conference 2002 in Prague. On-the-air results were promising, but, like any new
antenna, there were questions as to how well it really works. To find out, I ran some
computer simulations and published the results3 suggesting that this feed should work
well, and also suggesting some variations to allow use over a range of dish f/D.
However, I later realized that I was not properly analyzing the circular polarization, so
this paper includes revised analysis of the previous feeds as well as additional results.

OK1DFC septum feed

The septum feed as described by OK1DFC is an
unflared square horn, or simply a square
waveguide, with an internal stepped septum to
generate circular polarization. Figure 1 is the
view looking into the horn, and Figure 2 is a
photo of a partially assembled feed with the
septum in place, and Figure 3 is a cartoon of a
septum feed. The horn is excited by inputs on
either side of the septum, with the two sides

exciting opposite senses of circular

polarization. For EME, this provides
separate transmit and receive ports of
opposite sense of polarization –
reflecting off the moon reverses the
sense of the polarization. The
excitation may come from two
rectangular waveguides, each
matching the dimensions of one-half of
the square horn, or from a perpendicular
probe on each side of the septum acting as
an integral transition from coax to the
waveguide. The two methods should
provide identical results provided that the
waveguide section before the septum is
long enough to suppress any spurious
modes.
The radiating element, at the aperture, is simply a square horn. Rotated 45 degrees, it is
identical to a diagonal horn4; if the diagonal horn is excited with circular polarization,
then the radiated pattern should be identical. N7ART has shown5 the diagonal horn to be
a good feed, so we might expect the septum feed to be also. The version described by
N7ART used phased crossed dipoles to generate circular polarization, an arrangement
that seems awkward at higher frequencies. The septum could be a better way to generate
circular polarization.

Circular Polarization
Most antennas radiate linear polarization; most communication antennas use either
horizontal or vertical linear polarization. Only a few types, like the helical antenna, have
inherent circular polarization. Polarization is defined as the plane in which the electric
field, the E-field, varies. For example, a vertically-polarized antenna, like a vertical
dipole, has an electric field which at one instant might be positive at the top and negative
at the bottom; half a cycle later, it would reverse direction, to be positive at the bottom.
In between, a quarter-cycle from the peaks, it would instantaneously pass through zero.

To generate circular polarization with linearly-polarized antennas, we must add a second

radiator polarized at right angles to the first, and excite it a quarter cycle (90°) later than
the first, so that the electric field of the second radiator reaches a peak as the first passes
through zero, and vice-versa. Thus, the positive end of the electric field travels in a circle
rather than just reversing along a line. Since the field is also radiating from the antenna at
the speed of light as it travels in a circle, we might visualize the positive end as travelling
along a corkscrew. Circular polarization is characterized by the direction of travel –
right-hand (RHCP) or left-hand (LHCP), like the threads on a machine screw.

One way to excite the second radiator a quarter-cycle later is to add an extra quarter-
wavelength of transmission line; choosing which linear polarization is delayed controls
the direction of circular rotation. Another common method for producing the delay is the
use of a 90° hybrid – a directional coupler with two outputs of equal amplitude but 90°
phase difference. In waveguide, a thin dielectric sheet or card will delay energy polarized
parallel to the plane of the sheet, but not perpendicularly polarized energy; the length of
the sheet may be chosen to provide a quarter-wavelength of delay. A circular waveguide
linearly excited at a 45° angle to a card with ¼λ delay will generate circular polarization.
The 45° excitation is mathematically equivalent to two orthogonal components, but only
the component parallel to the dielectric is delayed. The dielectric may be a material, like
Teflon, or an artificial dielectric, for instance, a row of screws6 in the waveguide,
adjusted to provide the desired delay for circular polarization.

The septum is a bit more complicated. A circularly polarized wave entering the aperture
may be considered to have two polarization components with a 90º phase difference, one
parallel to the septum and one perpendicular. The parallel component is divided equally
by the septum and passes to the two rectangular input waveguides. The cutoff frequency
for the perpendicular component is changed by the septum, so that the wavelength for the
perpendicular component is shorter. Thus, the electrical length of the septum is longer
for the perpendicular component than for the parallel component; if the difference in
length is ¼ λ, or 90°, then the horizontal and vertical components arrive in phase at the
input. The components add together on one side and cancel on the other, depending on
the sense of circular polarization, so that the two ports are isolated from each other. In
order to achieve the difference in electrical lengths in a reasonable physical distance, the
septum polarizer operates near the cutoff wavelength of the waveguides.

Figure 4 shows the simulated E-field distribution in a

transparent septum feed, with the polarization
component perpendicular to the septum visible through
the left wall and the parallel component visible through
the top wall. The red areas of high field intensity are
separated by ½ λg along each wall, so we can see that
the top and side are ¼ λg apart at the aperture end, but
go through a more complex difference around the steps
in the septum. The cancellation at the input probe on
the far side is also clear.

The first septum polarizers7 used a sloping septum,

with a linear taper around 30°, like the cartoon cutaway
in Figure 5. This version apparently8 offered limited
bandwidth and isolation, so the stepped septum9 was
developed to improve isolation and bandwidth. These
references state that the field solution around the
septum is very difficult so no analytic procedure is
available. However, Chen & Tsandoulas9 show an
example with dimensions. OK1DFC used their
dimensions to build a spreadsheet that scales the design
for other frequencies. Since the example is for a 0.635λ square horn, and we don’t know
how to calculate septum dimensions for other horn sizes, we are limited to this size horn
for the septum section. It should work equally well at any frequency as long as all the
dimensions are scaled.
 When I originally analyzed the
septum feeds, I saw that the
polarization ratio, the ratio of desired
to undesired polarizations, was high.
Thus, I felt it would be reasonably
accurate to calculate efficiency using
only the desired circular polarization.
Later, I noticed that some of the
septum variations had large rear
lobes containing undesired
polarizations, even though the
polarization ratio in the main lobe was large. As a result, I modified the efficiency
calculation to more correctly calculate efficiency as the energy illuminating the dish with
the desired polarization divided by the total power in all polarizations, integrated over the
reflector as described by Cutler10 and W7PUA11.

Simulations
A septum feed for 1296 MHz with dimensions specified by OK1DFC was simulated
using Ansoft HFSS software12. The calculated radiation patterns in Figure 6 show the
broad illumination expected of a small horn. Like other open waveguide feeds, the rear
lobes are relatively large, only about 10 dB down, reducing the calculated efficiency.
The efficiency is further reduced by cross-polarization (XPOL) losses to about 61%, with
best f/D around 0.35 to 0.4. Cross-polarization losses are due to the wasted energy in
polarizations other than the desired sense of circular polarization, since the undesired
polarizations will not be received by a circularly polarized antenna. In Figure 6, the total
radiated power is included in the polar plot in addition to the RHCP (right-hand circular
polarization) radiation pattern. The main lobe, which would illuminate a dish, consists
almost entirely of RHCP energy, so the radiation in the beam reflected from the dish
would have good circular polarization. However, the side and back lobes contain
significant energy in unwanted polarizations, spillover energy which is lost and reduces
dish efficiency.

The patterns for right and left-hand circular polarization, when excited by the appropriate
input port, are pretty much identical. Patterns were calculated for both probe excitation
and rectangular waveguide excitation; they were very similar, showing that the distance
from the probe to the septum is adequate.

Previous simulations of diagonal feeds13 with linear polarization showed good radiation
patterns, but with efficiency reduced by the large rear sidelobes typical of open
waveguide feeds. Square horns14, with linear polarization parallel to the sides, show
large additional sidelobes in the E-plane. Since the circular polarization vector is
constantly rotating between these two conditions, we might expect the radiation pattern to
OK1DFC Septum feed with step polarizer, RHCP at 1296 MHz
Figure 6
90
E-plane
Feed Radiation Pattern

E&H-planes

Feed Phase Angle

67.5
H-plane
45

TOTAL POWER
22.5
0
0 dB -10 -20 -30
-22.5
-45
-67.5
-90
45˚-planes 0 10 20 30 40 50 60 70 80 90
Rotation Angle around specified
Dish diameter = 10 λ Feed diameter = 0.9 λ Phase Center = 0 λ beyond aperture
MAX Possible Efficiency without XPOL or Phase error
MAX Possible Efficiency with XPOL loss & Phase error
90
MAX Efficiency with phase error ONLY AFTER LOSSES:
Parabolic Dish Efficiency %

REAL WORLD at least 15% lower Illumination

80 1 dB
Spillover
Feed Blockage
70

2 dB
60

50 3 dB

40 4 dB

5 dB
30
6 dB
20 7 dB
8 dB

10

0.25 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Parabolic Dish f/D

W1GHZ 1998, 2003
be a composite of a diagonal horn and a
square horn. The circularly polarized
pattern of the septum feed, shown in
3D in Figure 7, looks like we might
imagine this composite, showing
sidelobes on the four corners like the
diagonal horn, generated as the
polarization vector passes through
horizontal and vertical polarization in
the square horn. The sidelobes on the
corners reduce the calculated efficiency
compared to a calculation using only
the traditional horizontal and vertical
pattern cuts.

The circular polarization is quite good,

with cross polarization about 21 dB
down in the main lobe, and the pattern
circularity is good. Isolation between
the two ports is about 24 dB at 1296
MHz, with reasonable bandwidth, showing good isolation from at least 1.2 to 1.4 GHz.
Note that reflection from the parabolic reflector reverses the sense of circular
polarization, so that the reflection coming back into the horn will appear at the other port
and reduce the isolation.

Simulation of the version with a simple 30° tapered septum like Figure 5 showed
performance at 1296 MHz very similar to the performance of the stepped version shown
in Figure 6, but the isolation between ports was high only over a smaller bandwidth,
roughly 100 MHz. This is quite adequate for amateur use, and the sloping septum might
be easier to fabricate at higher frequencies.

The calculated efficiency of this feed is not as high as some. High efficiency feeds often
have a larger blockage shadow, so the septum feed may be the best performer on a small
dish where circular polarization is required.

Variations
The OK1DFC septum feed consists of three sections: the input excitation, the septum
polarizer, and the radiating aperture. The input excitation can be provided by a
waveguide transition or by a probe which provides an integral coax-to-waveguide
transition. The radiation pattern is controlled only by the aperture dimensions, so we may
change the aperture to adjust the pattern and provide better illumination to dishes of
various f/D.
Flared horns
The radiating aperture is a simple square waveguide, equivalent for circular polarization
to a diagonal horn. A diagonal horn may be tailored to illuminate a desired f/D by
varying the dimensions of the diagonal section, or by adding a flared section for larger
f/D. Since we only know the correct dimensions a septum to generate circular
polarization for one waveguide dimension, 0.635λ, the square cross-section is fixed at
0.635λ for a given operating frequency. However, a flare section may be added to
increase the aperture size to optimize the horn for any larger f/D, so that the septum feed
may be used for any dish with f/D > 0.3. Since there are no good feeds for very deep
dishes, the septum feed is probably as good as any for deeper dishes.

The flare section is similar to a rectangular

waveguide horn, except that it should
maintain a square cross-section. Rectangular
horns need different aperture dimensions in
the E- and H-planes to achieve the same
beamwidth, but the circular polarization of the
septum feed has no fixed planes – they are
constantly rotating in a circle – so the square
cross-section should be maintained. The flare
should have a gentle taper, like the cartoon in
Figure 8, with one wall cut away to reveal the septum.

I first tried a adding a flare section with an

aperture 1.4λ square and a flare angle of 30°
(15° halfangle on each side of the septum),
since this size diagonal horn with linear
polarization is a good feed for an offset dish
with an equivalent f/D around 0.7. With the
septum feed generating circular polarization,
the calculated efficiency in Figure 9 is best
for f/D is around 0.7 to 0.85, suitable for
many offset dishes. This horn also had high
rear sidelobes on the corners, so that the 3D
pattern for RHCP in Figure 10 looks like a
rocket with fins. These sidelobes are even
larger when undesired polarizations are
included, reducing the calculated efficiency
to about 66% when feeding an offset dish.

An intermediate size flare, with an aperture

1.1λ square, produces the radiation patterns
in Figure 11 with about 66% calculated
efficiency at intermediate f/D, best around
Septum feed with horn flare to 1.4λ sq aperture, RHCP at 1296 Septum feed with horn flare 1.1λ square, RHCP at 1296 MHz
Figure 9 Figure 11
90 90
E&H-planes
E-plane E&H-planes
E-plane
67.5 67.5
H-plane H-plane
45 45
22.5 22.5
0 0
0 dB -10 -20 -30 0 dB -10 -20 -30
-22.5 -22.5
-45 -45
-67.5 -67.5

Feed Phase Angle

Feed Phase Angle

-90 -90
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90

Feed Radiation Pattern

Feed Radiation Pattern
45˚-planes 45˚-planes
Rotation Angle around specified Rotation Angle around specified
Dish diameter = 10 λ Feed diameter = 0.5 λ Phase Center = 0.2 λ inside aperture Dish diameter = 10 λ Feed diameter = 1.5 λ Phase Center = 0.04 λ inside aperture
MAX Possible Efficiency without XPOL or Phase error MAX Possible Efficiency without XPOL or Phase error
MAX Possible Efficiency with XPOL loss & Phase error MAX Possible Efficiency with XPOL loss & Phase error
90 90
MAX Efficiency with phase error ONLY AFTER LOSSES: MAX Efficiency with phase error ONLY AFTER LOSSES:
REAL WORLD at least 15% lower Illumination REAL WORLD at least 15% lower Illumination
80 1 dB 80 1 dB
Spillover Spillover
Feed Blockage Feed Blockage
70 70

2 dB 2 dB
60 60

50 3 dB 50 3 dB

40 4 dB 40 4 dB

5 dB 5 dB
30 30
6 dB 6 dB
20 7 dB 20 7 dB

Parabolic Dish Efficiency %

Parabolic Dish Efficiency %

8 dB 8 dB

10 10

0.25 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.25 0.3 1.00.4 0.5 0.6 0.7 0.8 0.9

Parabolic Dish f/D Parabolic Dish f/D

W1GHZ 1998, 2003 W1GHZ 1998, 2003
0.5 to 0.6. The corner lobes for this horn are
less pronounced in the 3D pattern, Figure 12,
and have less effect on efficiency.

Since the corners produce undesired sidelobes,

I tried cutting them off and making the flare
sections octagonal. The results, shown in
Figure 13 for a 1.4λ aperture and Figure 14 for
a 1.1λ aperture, have significantly lower XPOL
(cross-polarization) loss. Calculated efficiency
for both sizes increases to about 70%.

The flared septum horns show good isolation

and cross-polarization in the main lobe. Since
horn beamwidth is inversely related to aperture
size, we can choose an appropriate aperture for
the flare for any f/D by interpolating between
the results for the three sizes above, 0.63λ square, 1.1λ square, and 1.4λ square. For
smaller apertures, the flare angle should be small so that the flare length is reasonably
long.

Chokes
The VE4MA feed15 adds a choke ring around
a circular waveguide feed to reduce side and
back lobes, thus increasing feed efficiency by
putting more of the energy on the reflector.
Since the unflared septum feed has rather
large rear lobes, perhaps a choke would

improve the septum feed also; the cartoon in

Figure 15 shows one with a circular choke and
one wall cut away to reveal the septum.
Adding a circular choke with the same
dimensions as the VE4MA feed, 419 mm in
diameter and 121 mm deep at 1296 MHz,
resulted in a significant reduction in the back
lobes, as shown in Figure16. Calculated
efficiency improved to 63% with best f/D
around 0.4 with the choke rim flush with the
horn aperture. Sliding the choke back 10 to
20 mm favors deeper dishes, with best f/D
around 0.35, but sliding the choke further
back brings out the corner lobes and reduces
Septum feed flared to 1.4λ octagonal horn, RHCP at 1296 MHz Septum feed flared to 1.1λ octagonal horn, RHCP at 1296 MHz
Figure 13 Figure 14
90 90
E&H-planes
E-plane E&H-planes
E-plane
67.5 67.5
H-plane H-plane
45 45
22.5 22.5
0 0
0 dB -10 -20 -30 0 dB -10 -20 -30
-22.5 -22.5
-45 -45
-67.5 -67.5

Feed Phase Angle

Feed Phase Angle

-90 -90
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90

Feed Radiation Pattern

Feed Radiation Pattern
45˚-planes 45˚-planes
Rotation Angle around specified Rotation Angle around specified
Dish diameter = 10 λ Feed diameter = 0.5 λ Phase Center = 0 λ beyond aperture Dish diameter = 10 λ Feed diameter = 1.1 λ Phase Center = 0.05 λ inside aperture
MAX Possible Efficiency without XPOL or Phase error MAX Possible Efficiency without XPOL or Phase error
MAX Possible Efficiency with XPOL loss & Phase error MAX Possible Efficiency with XPOL loss & Phase error
90 90
MAX Efficiency with phase error ONLY AFTER LOSSES: MAX Efficiency with phase error ONLY AFTER LOSSES:
REAL WORLD at least 15% lower Illumination REAL WORLD at least 15% lower Illumination
80 1 dB 80 1 dB
Spillover Spillover
Feed Blockage Feed Blockage
70 70

2 dB 2 dB
60 60

50 3 dB 50 3 dB

40 4 dB 40 4 dB

5 dB 5 dB
30 30
6 dB 6 dB
20 7 dB 20 7 dB

Parabolic Dish Efficiency %

Parabolic Dish Efficiency %

8 dB 8 dB

10 10

0.25 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.25 0.3 1.00.4 0.5 0.6 0.7 0.8 0.9

Parabolic Dish f/D Parabolic Dish f/D

W1GHZ 1998, 2003 W1GHZ 1998, 2003
efficiency. The 3D pattern with the choke flush, Figure 17, shows that the round choke
eliminates the corner lobes. Isolation between the ports is still good, but the cross-
polarization ratio is slightly lower at about 19 dB. However, the increased efficiency
comes at a price: the blockage shadow has increased from 0.63λ square to 1.8λ diameter,
a significant difference for a small dish. The improvement would be greater on a larger
dish with smaller blockage loss.

Since the septum feed may be fabricated by

cutting and bending sheet metal, it seemed that
a square choke like the cartoon in Figure 18
might be easier to fabricate than a round one.
From the round choke, I estimated choke
dimensions of 1.7λ square and the same 121
mm deep. As shown in Figure 19, the square choke does not work nearly as well as the
round one, with very low efficiency and a 3D pattern, in Figure 20, resembling a pepper.
I don’t know whether it is the shape or the dimensions that reduce the performance, but
I’d recommend sticking with the round choke.

Circular waveguide with septum polarizer

The septum polarizer seems to do a good job of producing circular polarization in a
square cross-section, with no adjustment required. Would a septum polarizer work in
circular waveguide? One reference16 described a stepped septum in circular waveguide,
talked about using a computer program to solve the wave equation, and plotted the
relative cutoff wavelengths of the parallel and perpendicular components vs. step height.
Unfortunately, the authors found that the septum dimensions had to adjusted
experimentally to achieve the 90° phase shift necessary for circular polarization.
 Septum feed with round choke flush, RHCP at 1296 MHz Septum feed with 1.7λ square choke, RHCP at 1296 MHz
Figure 16 Figure 19
90 90
E&H-planes
E-plane E&H-planes
E-plane
67.5 67.5
H-plane H-plane
45 45
22.5 22.5
0 0
0 dB -10 -20 -30 0 dB -10 -20 -30
-22.5 -22.5
-45 -45
-67.5 -67.5

Feed Phase Angle

Feed Phase Angle

-90 -90
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90

Feed Radiation Pattern

Feed Radiation Pattern
45˚-planes 45˚-planes
Rotation Angle around specified Rotation Angle around specified
Dish diameter = 10 λ Feed diameter = 1.8 λ Phase Center = 0.3 λ inside aperture Dish diameter = 10 λ Feed diameter = 1.8 λ Phase Center = 0 λ beyond aperture
MAX Possible Efficiency without XPOL or Phase error MAX Possible Efficiency without XPOL or Phase error
MAX Possible Efficiency with XPOL loss & Phase error MAX Possible Efficiency with XPOL loss & Phase error
90 90
MAX Efficiency with phase error ONLY AFTER LOSSES: MAX Efficiency with phase error ONLY AFTER LOSSES:
REAL WORLD at least 15% lower Illumination REAL WORLD at least 15% lower Illumination
80 1 dB 80 1 dB
Spillover Spillover
Feed Blockage Feed Blockage
70 70

2 dB 2 dB
60 60

50 3 dB 50 3 dB

40 4 dB 40 4 dB

5 dB 5 dB
30 30
6 dB 6 dB
20 7 dB 20 7 dB

Parabolic Dish Efficiency %

Parabolic Dish Efficiency %

8 dB 8 dB

10 10

0.25 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.25 0.3 1.00.4 0.5 0.6 0.7 0.8 0.9

Parabolic Dish f/D Parabolic Dish f/D

W1GHZ 1998, 2003 W1GHZ 1998, 2003
VE4MA feed
The simplest septum polarizer is just a linear taper, and it seems to work well in the
square feed. Davis, et. al., suggest7 30° as an appropriate taper for a septum polarizer, so
I added a 30° sloping septum to a VE4MA feed15 with relatively small diameter circular
waveguide, so that it would be close to cutoff. The resulting radiation patterns were OK,
but the polarization ratio (the ratio of desired to undesired polarization) was not: the
undesired sense was only about 10 dB down, so that efficiency would be reduced since
about 1/10th of the energy is in the wrong polarization.

The polarization ratio was also sensitive to frequency, so I fiddled with the waveguide
diameter. The best I could find with a 30° septum taper was about a 13 dB polarization
ratio, at a waveguide diameter of about 0.7λ. Since this is far worse than the stepped
taper in a square guide, perhaps the stepped taper might be needed in a circular guide
also.

Lacking a spreadsheet for circular guide,

I decided to try a quick approximation. I
adjusted the frequency in Zdenek’s
spreadsheet1 for square guide until the
septum height matched the 0.7λ
waveguide diameter, then used the
calculated step dimensions. A cartoon
of this feed is shown in Figure 21. I
don’t know whether these are the best
dimensions, but they work pretty well: a

VE4MA feed with these dimensions has a

polarization ratio around 22 dB at 1296
MHz, and good isolation between ports.
The calculated radiation patterns and
efficiency are shown in Figure 22; best
calculated efficiency is about 61% at an f/D
around 0.35 to 0.45. This efficiency is
similar to the original square septum feed –
the improved illumination is negated by a
larger blockage shadow. As might be
expected, the 3D pattern in Figure 23 has no
corner lobes, since all the cross-sections are
round – the aperture is axisymmetric.
VE4MA feed 0.71λ dia with step septum, RHCP at 1296 MHz
Figure 22
90
E-plane E-plane
Feed Radiation Pattern

Feed Phase Angle

67.5
H-plane
45
22.5
0
0 dB -10 -20 -30
-22.5
-45
-67.5
-90
H-plane 0 10 20 30 40 50 60 70 80 90
Rotation Angle around specified
Dish diameter = 10 λ Feed diameter = 1.8 λ Phase Center = 0.13 λ inside aperture
MAX Possible Efficiency without XPOL or Phase error
MAX Possible Efficiency with XPOL loss & Phase error
90
MAX Efficiency with phase error ONLY AFTER LOSSES:
Parabolic Dish Efficiency %

REAL WORLD at least 15% lower Illumination

80 1 dB
Spillover
Feed Blockage
70

2 dB
60

50 3 dB

40 4 dB

5 dB
30
6 dB
20 7 dB
8 dB

10

0.25 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Parabolic Dish f/D

W1GHZ 1998, 2003
Coffee-can feed
To verify that the septum polarizer works in circular waveguide, I calculated radiation
patterns for a simple open-ended circular waveguide, the popular coffee-can feed.
Figures 24 and 25 show the radiation patterns and feed performance for a 0.71λ diameter
open waveguide with slope and step septums, respectively. The two patterns, with total
power in all polarizations shown also, are noticeably different, so the two types of
septums are not equivalent in circular waveguide. With the step polarizer, calculated
efficiency was good, about 65%, while the efficiency with the slope polarizer was lower
due to larger XPOL loss. For either, f/D for best efficiency is about 0.35. Like the
VE4MA feed, slope septums and other diameters did not work as well, since we do not
know how to compensate the septum dimensions. This looks like a good feed for small
dishes where feed blockage is an important consideration.

W2IMU dual-mode feed

A number of good feeds are
based on circular waveguide.
Once we have established that
the septum polarizer works in
circular waveguide, all these
feeds are possible candidates. A
popular one with good
performance is the W2IMU17 dual-
mode feedhorn, known to provide
good circular polarization using
conventional techniques. With a
septum polarizer, like the cartoon
in Figure 26, the W2IMU feed
provides very good performance.
As shown in Figure 27, the
calculated efficiency is about 73%
at an f/D around 0.66. The dual-
mode feed is designed to have low
sidelobes, which helps to
minimize XPOL loss. The 3d
view of the radiation pattern
shown in Figure 28 illustrates the
low sidelobes.
 Coffee-can feed 0.71 dia with slope septum RHCP at 1296 Coffee-can feed 0.71λ dia with step septum RHCP at 1296 MHz
Figure 24 Figure 25
90 90
E&H-planes
E-plane E&H-plane E-plane
67.5 67.5
H-plane H-plane
45 45

TOTAL POWER
22.5 TOTAL POWER
22.5
0 0
0 dB -10 -20 -30 0 dB -10 -20 -30
-22.5 -22.5
-45 -45
-67.5 -67.5

Feed Phase Angle

Feed Phase Angle

-90 -90
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90

Feed Radiation Pattern

Feed Radiation Pattern
45˚-planes 45˚-planes
Rotation Angle around specified Rotation Angle around specified
Dish diameter = 10 λ Feed diameter = 0.71 λ Phase Center = 0.03 λ beyond aperture Dish diameter = 10 λ Feed diameter = 0.71 λ Phase Center = 0 λ beyond aperture
MAX Possible Efficiency without XPOL or Phase error MAX Possible Efficiency without XPOL or Phase error
MAX Possible Efficiency with XPOL loss & Phase error MAX Possible Efficiency with XPOL loss & Phase error
90 90
MAX Efficiency with phase error ONLY AFTER LOSSES: MAX Efficiency with phase error ONLY AFTER LOSSES:
REAL WORLD at least 15% lower Illumination REAL WORLD at least 15% lower Illumination
80 1 dB 80 1 dB
Spillover Spillover
Feed Blockage Feed Blockage
70 70

2 dB 2 dB
60 60

50 3 dB 50 3 dB

40 4 dB 40 4 dB

5 dB 5 dB
30 30
6 dB 6 dB
20 7 dB 20 7 dB

Parabolic Dish Efficiency %

Parabolic Dish Efficiency %

8 dB 8 dB

10 10

0.25 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.25 0.3 1.00.4 0.5 0.6 0.7 0.8 0.9

Parabolic Dish f/D Parabolic Dish f/D

W1GHZ 1998, 2003 W1GHZ 1998, 2003
Input probes
Since the intent was to examine feedhorn performance, these simulations were all done
with single-mode waveguide excitation. The only exception was one with input probes to
the OK1DFC dimensions, to make sure the probes do not affect the basic radiation
pattern. The calculated patterns and efficiency were identical, but the probe version had
higher isolation, shown in Figure 29. Achieving such high isolation in an actual feed
would require careful construction, and then reflections from the dish surface would still
reduce the isolation as described above.

Because the septum cuts the guide in half in the input area, the probe length is limited.
The optimum probe length might be very close to the septum, particularly in circular
guide, close enough to arc over with high power. For the square guide, Zdenek has
shortened the probe length and compensated with a tuning screw to add capacitance, a
reasonable solution. A better solution would be to increase the probe diameter – at 1296
MHz, ¼” diameter probes should provide a good match. Anyone wishing to add a
septum polarizer to a VE4MA feed with probe excitation will have to empirically find
probe and tuning screw dimensions for the circular guide. Since the polarization and
isolation are taken care of by the septum, only the VSWR of each probe must be adjusted.
W2IMU dual-mode feed with step septum, RHCP at 1296 MHz
Figure 27
90
E-plane
Feed Radiation Pattern

E&H-planes

Feed Phase Angle

67.5
H-plane
45
22.5
0
0 dB -10 -20 -30
-22.5
-45
-67.5
-90
45˚-planes 0 10 20 30 40 50 60 70 80 90
Rotation Angle around specified
Dish diameter = 10 λ Feed diameter = 1.3 λ Phase Center = 0.02 λ beyond aperture
MAX Possible Efficiency without XPOL or Phase error
MAX Possible Efficiency with XPOL loss & Phase error
90
MAX Efficiency with phase error ONLY AFTER LOSSES:
Parabolic Dish Efficiency %

REAL WORLD at least 15% lower Illumination

80 1 dB
Spillover
Feed Blockage
70

2 dB
60

50 3 dB

40 4 dB

5 dB
30
6 dB
20 7 dB
8 dB

10

0.25 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Parabolic Dish f/D

W1GHZ 1998, 2003
Summary
The septum feeds are impressive – the septum polarizer provides good circular
polarization performance with no adjustments. The other common ways to achieve
circular polarization use two orthogonal probes phased by 90°; the phasing is achieved
with an external 3-dB hybrid or a phasing section in the guide, often a series of screws, to
provide a slow-wave structure for one polarization. The screws require careful
adjustment to achieve good circularity and good VSWR, while the external hybrid and
cabling adds some losses. The septum polarizer offers the possibility of good, low-loss,
circular polarization with no adjustments.

The septum polarizer is applicable to a variety of feedhorns The simple square cross-
section described by OK1DFC is ideal for low blockage on small deep dishes, while a
choke may be added for better performance on larger dishes. A flare section to increase
the aperture will better illuminate shallow and offset dishes. In square waveguide, the
step and slope septums both appear to work, but only the step septum polarizer works
well in cylindrical horns like the VE4MA feed and the W2IMU dual-mode feed. Best
performance is found on feeds with low sidelobes, since sidelobes contain unwanted
polarizations and increase cross-polarization loss.
References
1. Zdenek Samek, OK1DFC, “Feed for Parabolic Dish with Circular Polarization,”
10th International EME Conference 2002, Prague, 2002. www.qsl.net/ok1dfc
2. Zdenek Samek, OK1DFC , “Information and practical hints for the construction
of a septum feed,” DUBUS, 1/2003, pp. 39-47.
3. P. Wade, W1GHZ, “Analysis of the OK1DFC Septum Feed,” DUBUS, 1/2003,
pp. 22-38.
4. A.W. Love, “The Diagonal Horn Antenna,” Microwave Journal, March 1962,
pp. 117-122. (reprinted in A.W. Love, Electromagnetic Horn Antennas, IEEE,
1976, pp. 189-194.)
5. R. Miller, N7ART, “A 23cm Diagonal Waveguide Feed,” DUBUS, 2/1997,
pp. 5-14.
6. D. Turrin, W2IMU, “A Circularly Polarized Feed Antenna for 1296 mc/s,”
Crawford Hill Technical Report #9, December 1971. (reprinted in Proceedings of
Microwave Update ’99, ARRL, 1999, pp. 487-491.)
7. D. Davis, O. J. Digiondomenico, and J. A. Kempic, "A new type of circularly
polarized antenna element," 1967 IEEE Group on Antennas and Propagation
International Symposium Digest, vol. 5, pp. 26-33, October 1967.
8. H. Schrank; Antenna designer's notebook, IEEE Antennas and Propagation
Society Newsletter, vol. 25, pp. 23-24, October 1983.
9. Ming Hui Chen, G. N. Tsandoulas; A wide-band square-waveguide array
polarizer, IEEE Transactions on Antennas and Propagation, vol. 21, pp. 389-391,
May 1973
10. C.C. Cutler, “Parabolic-Antenna Design for Microwaves, Proceedings of the IRE,
Nov. 1947, pp. 1284-1294. (reprinted in A.W. Love, Reflector Antennas, IEEE,
1978, pp. 16-26.)
11. B. Larkin, W7PUA, “Dipole-Reflector Parabolic Dish Feeds for f/D of 0.2-0.4,”
QEX, February 1996, pp. 3-11.
12. www.ansoft.com
13. P. Wade, W1GHZ, The W1GHZ Microwave Antenna Book – Online, Section
6.5.3, www.w1ghz.org
14. P. Wade, W1GHZ, The W1GHZ Microwave Antenna Book – Online, Section
6.4.2, www.w1ghz.org
15. B.W. Malowanchuk,VE4MA, “Selection of an Optimum Dish Feed,”
Proceedings of the 23rd Conference of the Central States VHF Society, ARRL,
1989, pp. 35-43.
16. R. Behe and P. Brachat, "Compact duplexer-polarizer with semicircular
waveguide," IEEE Transactions on Antennas and Propagation, vol. 39, pp. 1222-
1224, August 1991.
17. R.H. Turrin, (W2IMU), “Dual Mode Small-Aperture Antennas,” IEEE
Transactions on Antennas and Propagation, AP-15, March 1967, pp. 307-308.
(reprinted in A.W. Love, Reflector Antennas, IEEE, 1978, pp. 214-215.)