[Date Prev][Date Next][Thread Prev][Thread Next] - [Date Index][Thread Index][Author Index]

Patch Antenna Arrays

I apologize for joining the patch antenna discussion late, I just surfaced
(literally, I'm a U.S. Navy submariner) and have been trying to catch up on
the thread.

I have been working on an electrically rotated array of patch antennas for
the downlink of the 9600 bd Pacsats as time permits over the past two years.
I have built two trial systems and a total of 5 70 cm patch antennas.  I
don't yet have a full up system to report, but I believe the concept is
valid and probably fairly easy to achieve.

The concept is a total of 5 patch antennas, in a N/S/E/W/Up configuration.
Each patch has a MMIC preamp mounted directly at the feed point.  I've used
MAR-6 and ERA-3SMs, and I have a small quantity of a newer Motorola MMIC
with a better noise figure that I just need to etch some boards to go try
out.  The MMICs are placed on the same board with Motorola HSMP series PIN
diodes, microstripline style, and the preamp power also serves as bias power
for switching the diodes.  There are some new GaAsFet 3 and 5 watt TR
switches that could give such an array a small transmit capability.  The
idea was to get the preamps ahead of the switches to avoid the losses, yet
keep the overall cost and fabrication difficulty to a minimum.

Switching is controlled by software written by Al Lawler, WB1BQE, which
interfaces to Wisp, and toggles the bits on a parallel port any way the user
specifies.  The parallel port then controls switching of the power to the
preamps.  This has been tested satisfactorily.

I initially attempted to used 2 el vertical yagis, but the attempt failed
mostly due to my poor construction skills.  I evaluated both patch antennas
and two element full wave loops, feeding the driven element of the loop in
quadrature to go CP.  I settled on the patches due to not wanting to mess
with the baluns necessary for the quads.  Patch antennas are very low
profile, and might have some advantages in antenna restricted areas.

Rough dimensions for a 70 cm air dielectric patch (from memory, my notes are
back in home port) is about 33 cm square, spaced .5 cm off the ground plane.
A linear patch is fed about 9.5 cm in from the edge.  Ground plane sizes
less than one lambda start to distort the pattern, and I have settled on 24"
squares as they can be purchased precut.  One technique to go CP on such a
patch is to make it slightly rectangular and feed it on a diagonal.  My
reference here is "CAD for Microstrip Antennas" by Dr. Saintani.  This book
comes with software to calculate both nearly square CP patches fed on a
diagonal, and truncated corner patches fed along one axis.  Other sources of
CAD software include Dr. David Pozar's PCCAAD, a freeware version of which
is available on the RF GlobalNet site (http://www.rfglobalnet.com), and
software by Dr. Roger Cox, (WB0DGF, I think).  One way to visualize a patch
is a pair of half lambda slot radiators fed by a microstripline transmission
line.  They are difficult to model with method of moment techniques, such as
NEC and MiniNec type software, and the above software uses different
modeling techniques.  The CP feeding method is the "feed two elements at
right angles, one above resonance, one below, in parallel such that the
reactive components cancel with a 90 deg phase shift between elements"
technique.  Feed impedance varies from a high (several hundred ohm) value at
the edge to near 0 in the middle, and you just have to find the 50 ohm point
and feed it there.  The above programs do calculate feed points.

None of these programs will yield exact dimensions, but will get you within
a few percent, the remainder having to be done empirically.  A linear patch
is easier to trim in, and after a few failures on CP patches, I built a
linear 70 cm patch for testing.

When the bird is within the main beam of the patch, results were
outstanding.  Calculated gain of an air dielectric patch is just under 10
dbi.  However, for the 5 patch system, a beamwidth of 90 deg (not 3 db down,
but something greater 0 db, i.e. enough to copy the bird) is desired, and it
did not look like the air patch had that much.  I live in a heavily wooded
area, and that probably compromised my measurements.

I also built two FR4 PCB based patches.  You get about a 50 percent
reduction is size by going to FR4, but losses go up due to the dielectric.
Dimensional tolerance is pretty critical on all types of patches.  Probe fed
patches are narrow to begin with (2 to 3% 3 to 1 SWR bandwidth), and CP
patches can be expected to remain circular only in about a 1% range.  I ran
into too much variation in the relative permittivity value on FR4 to enable
me to accurately model them, and have put that design off for a while.

I just finished another air dielectric linear patch, which agreed pretty
closely to the model for the first time.  Jim, WB4GCS is testing that one
for me.  I think I now have pretty good modeling accuracy, and the next step
is to make a RHCP air dielectric patch and evaluate that, probably around

Hope this collection of notes is useful.  I'll try to get a drawing with
exact dimensions up on my web site when I get back to home port.  73 de
Chas, W4HFZ

Via the amsat-bb mailing list at AMSAT.ORG courtesy of AMSAT-NA.
To unsubscribe, send "unsubscribe amsat-bb" to Majordomo@amsat.org