Note: Descriptions are shown in the official language in which they were submitted.
3~;78
-1- 35-EL-1583
MIXER FOR USE IN A MICROWAVE SYSTEM
. .
Back~round~ of the Invention
._
1. Field o;f the Invention;:
The present invention reIates to microwave
systems, typically communications systems, such as find
use in the communications band of 21. 8 to 23.2 GHz
carrier frequency. The invention further relates to a
novel mixer for effecting a stable, low cost, single
conversion of a microwave signal to an intermediate
frequency suitably low (50 MHz) for convenient ampli-
fication. In a mixer operating at these disparate
freguencies, efficient transmission paths and efficient
filtering must be provided. The disclosed transmission
paths include waveguides, microstrip circuits, trans-
itions between waveguide and microstrip circuits, andsuitable filters connected before and after the mixer.
2. Des_ription 'of' the Pr'ior Art:
A need has arisen for broadband, short range,
low cost and generally directional point to point
communications channels. Typically, such channels require
adequate bandwidths for television channel capability or
for wideband data transmission suited for computer
applications. Higher microwave frequencies, i.e., "K" band,
are now accessible with the advent of relatively low
cost Gunn diode oscillators. (These currently operate
up to about 30 GHz). The Gunn diode oscillator
provides both a low cost source of modulatable RF energy
or low power signal transmission, and a low cost local
oscillator for use in the signal conversion process.
31 2~357~3
35-EL-1583
--2--
The Gunn diode oscillator is of a highly stable design,
and permits use of reIatively low IF frequencies,
usually without the need for a closed loop frequency
control. Thus, a simple, single conversion system is
practical. Low cost dish antennas and efficient antenna
waveguide feeds are well known (see Sichak article infra).
In addition, low cost, high frequency diodes are
available and packaged suitably for use as balanced
mixers at these frequencies. In the mixer design, however,
the question arises as to the most efficient and most
eaonomical method of providing the essential filtering
and of providing the essential signal transmission paths.
It is at this point that the use of microstrip circuits
is suggested. Known microstrip circuits provide low cost
construction of transmission paths, hybrid couplers, and
filter circuits. These developments find application
to the present invention.
SUMMARY OF THE INVENTION
Accordingly, it is an object of the invention
to provide an improved mixer for high frequency operation.
It is another object of the invention to provide
an improved mixer for microwave operation.
It is a further object of the invention to
provide an improved single conversion mixer for microwave
operation.
It is an additional object of the invention to
provide a single conversion mixer for microwave operation
having improved input and output transmission paths~
It is another object of the invention to provide
a single conversion mixer for microwave operatior. having
improved filtering in the input and output transmission
paths.
It is still another object of the invention to
provide an improved single conversion mixer employing
microstrip transmission paths efficientl~ fed from
sources coupled by waveguides.
3~
35-EL-1583
These and other objects of the invention are
achieved in a novel mixer comprising waveguide means,
a local oscillator and signal coupling means, both
coupled to the waveguide means, a microstrip circuit,
in which balanced detection occurs, and a pair of
impedance transformers for deriving local oscilla~or
and signal energy from the waveguide means and coupling
it efficiently to the microstrip circuit.
More particularly, the waveguide means propagates
waves in a TE 10 mode and contains an electrically
shorted, transverse wall, which divides the waveguide
means into two electrically isolated waveguide sections.
The local oscillator couples high frequency energy into
the first waveguide section and the signal (usually
from an antenna) is coupled into the second waveguide
section.
The microstrip circuit, in which detection occurs,
comprises a ground plane, a thin dielectric layer,
and surface conductors formed on the dielectric layer.
The microstrip circuit further comprises a four part
hybrid coupler formed of surface conductors; a balanced
detector comprising a pair of diodes, each having one
electrode connected via a surface conductor to one hybrid
coupler output port, the other diode electrodes being
joined and connected to a surface conductor at which
the detector output~ containing heterodynes appears;
and a low pass filter formed of surface conductors, and
connected to the detector output for selecting the
desired low frequency heterodyne and rejecting the
3a undesired higher frequency heterodynes.
The microstrip circuit is disposed adjacent a
wide face of the waveguide means in proximity to the
transverse wall for convenient interconnection. More
particularly, the filter impedance transformer is formed
from a first resonant stub of approximately 1/4
f ~
35-EL-1583
--4--
eIectrical waveIength projecting through an aperture
in the wide face of the waveguide into the interior
thereof for deriving local oscillator energy present
in the first ~aveguide section and for supplying it
to one input port of the hybrid coupler.
The second impedance transformer is formed from
a second resonant stub also of approximately 1/4
electrical waveIength projecting through an aperture
in the wide face of the waveguide into the interior
thereof for deriving the signal energy present in the
second waveguide section and for applying it to the
other input port of the hybrid coupler. The two stubs
form effective impedance transformers for matching the
impedance of the waveguide section to the impedance of
the hybrid coupler input ports.
Preferably, the frequency of the local oscillator
and the signal are selected to be sufficiently close
such that the transverse dimensions of the two waveguide
sections may be the same without causing a non-TE l-0 mode
of the propagation; the frequency selection should also
permit the hybrid to accommodate both frequencies without
a broadbanded design; and the frequency difference
should produce an IF frequency sufficiently low for
convenient solid state amplification. Typical frequency
differences in the 23 GHz communication band permit an
intermediate frequency of 50 MHz.
In accordance with another facet of the invention,
the axes of the resonant stubs are oriented perpendicular
to the wide faces of ~he waveguide and are spaced from the
transverse wall by amounts adequate to provide efficient
transfer of microwave energy from the waveguide section
into the stub with the inner end of the stub being spaced
~rom the lower inner surface of the waveguide to avoid
shorting the electrical field. An insulator having a
substantial dielectric constant is provided encircling
each stub for increasing the electrical length of the stub
s~
35-EL-1583
to approximately 1/4 wavelength to provide an approximate
match of the impedance of the waveguide section (at the
position of the probe) to the characteristic impedance
of the associated hybrid input port on the microstrip
circuit.
For maximum seLectivity, which conveniently
embraces a band having approximately 6% relative band-
width (e.g. 2I.8 GH2 to 23.2 GHz), the axes of the stub
are placed approximately l/8 electrical wavelength from
the transverse wall in the waveguide and on the center
line of the waveguide section. This placement also
allows for eficient energy transfer from each waveguide
section into the microstrip circuit.
For optimum tuning, the electrical lengths of
the stubs are set slightly in excess of l/4 electrical
wavelength to produce a net inductive effect, and an
additional capacitance is provided at each of the hybrid
coupler input ports to optimize the energy transfer.
In addition, the two surface conductors, each of
which is connected between a hybrid coupler output port
and a detector diode, are made of equal electrical length
for effecting cancellation of local oscillator noise
at the detector output. For impedance matching the
hybrid coupler output ports to the detector diodes,
which are of lower impedance, each detector connected
surface conductor is thinned to present an effective
serial inductance. To complete the matching network
a shunt capacitance is provided at the detector output.
The low pass detector output filter is a pi network
consisting of an input shunt capacitance, a serial
inductance r and an output capacitance. The low pass
filter is connected to the diode output by a l/2 wave-
length line segment, the disposition causing the filter
input capacitance to be reflected back to the diode output
for the dual use thereof.
:~;2!~3~
35-EL-1583
Preferably, the microstrip circuit uses conductors
whi'ch form transmission paths having a characteristic
impedance in the vicinity of 50'ohms. For instance, the
serial branches of the hybrid coupler have characteristic
impedances of approximateIy 35 ohms, the shunt branches
of the hybrid coupler have characteristic impedance of
approximately 50'ohms and the hybrid coupler ports have
characteristic impedances of 50 ohms. In addition the
1/2 waveIength line segment from the detector output to
the low pass filter, both filter input and filter output
double stubs and the signal path from the low pass filter
have characteristic impedances of approximately 50 ohms.
This impedance choice provides for efficient signal
transfer at microwave frequencies.
The mixer is typically constructed of three rectangular
waveguide sections interconnected by mating flanges.
BRIEF DESCRIPTION OF THE DRAWINGS
The novel and distinctive features of the invention
are set forth'in the claims of the present application.
The' invention itself, however, together with further
objec'ts and advantages thereof, may best be understood
by reference to the following description and accompanying
drawings in which:
Figure 1 is a block diagram of a 23 GHz microwave
communication system employing a novel mixer constructed
in accordance with the present invention;
Figure 2 is a cut-away perspective view of a
receiving installation of the Figure 1 communication system.
Figure 3 is a more detailed perspective view of
the receiving installation, illustrating the mechanical
detai.ls of a novel mixer employing microstrip trans-
mission paths and including the waveguide microstrip
circuit interconnections;
Figure 4 is a detailed plan view of the microstrip
circuit in which mixing occurs; and
Figure 5 is an electrical equivalent circuit
representation of the microstrip circuit of Figure 4.
35-EL-1583
DESCRIPTION OF THE P~REFERRED EMBODIMENT
.__
Referring now to Figure 1, there is shown a block
diagram of a 23 GH~ microwave communication system
employing a noveI mixer, which is the subject of the
present invention. The communication system comprises
a transmitter portion driving a transmitting antenna and
a receiver portion coupled to the receiving antenna, the
latter arranged to receive transmissions from the
transmitting antenna.
More particularly, the transmitter comprises a
high frequency oscillator 11, typically a Gunn oscillator,
which is subject to amplitude modulation in the AM
modulator 12, the modulations being applied from a bias
control. unit 13. The output of the AM modulator 12 is
coupled to the transmitting antenna 14. Typically, the
Gunn oscillator 11 operates at a power level of 100
milliwatts and the antenna takes the form of a small
parabolic, dish antenna , slighting under 12" in diameter.
The modulator 12 is a pin diode modulator installed in
a waveguide, not illustrated, interconnecting the Gunn
oscillator to the antenna 14. The blocks 11, 12 and
13 are of conventional design.
The antenna 14 is similar to that illustrated in
Figure 2 and is driven by the waveguide end remote from
the Gunn oscillator. The waveguide feed is split by a
flat conductive member lying in a plane parallel to the
upper and lower surfaces of the waveguide, the conductor
being used to support an inner active dipole, which is
fed by the waveguide, and an outer reflective dipole. The
second dipole i5 used as a reflector at the termination
to reduce the back lobe of the feed and thus increase
the efficiency of the antenna. The back lobe can be
further reduced by tapering the thickness of the wave-
guide. This refinement was not required in the present
design. The double dipole antenna (with the tapering
features is described in the Rad Lab Series Vol. 12,
~L2~3S~3
35-EL-1583
"Microwave Antenna Theory and Design", Section 8.9,
Double-Dipole Feeds, b Waveguide System. The waveguide
double-dipole feed is shown in Figure 8.14. A footnote
to that figure cites RL Report No. 54-25, June 26, 1943,
W. Sichak, "Double Dipole Rectangular Waveguide Antennas".
In Fig. 2, which shows the receiving installation, the
waveguide is shown at 22, the dipole mounting strip is
shown at 23, the active dipole is shown at 24 and the
reflective dipole is shown at 25.
Recapitulating, the energy developed from the
Gunn oscillator 11 and then modulated by the ~M modulator
12 is coupled via the waveguide to the antenna feed,
which by means of the paraboIic antenna 14 develops a
focused beam. In the present practical embodiment, the
input for the AM modulator may take the form of video
information or digital information. As will be seen,
the useful bandwidth of the communication system is
approximately 15 MHz.
The receiving system includes a receiving
antenna 15, similar in design to that of the transmitting
antenna and is the antenna specifically illustrated
in Figure 2. Similar to the transmitting antenna, the
receiving antenna 15 consists of a parabolic dish 21
feeding the waveguide 22 by means of the short reflective
dipole 25 arranged outwardly of the active dipole 24.
In reception, waves are collected from the surface of
the dish 21, and focused upon the reflective dipole 25.
The latter reflects the signal to the active dipole 24,
which feeds vertically polarized signal energy into the
waveguide 22 directed inwardly toward the novel mixer 26.
The receiver portion further includes the local
oscillator 16, which in cooperation with the mixer 16
converts the received signal at a 50 MHz IF frequency.
As will be described, the converter has both input and
output bandpass filtering. The signal at IF frequency
is then amplified in the preamplifier 18, optimally
additionally filtered at the IF amplifier 19, and
3S~7~
35-EL-1583
_..9_
finally amplified to a level suitable for detection in
the video data detector 20.
If the information is video information, for
example in the form of a television picturel the signal
is applied to a suitable monitor, not shown. In the
event that the information supplied is digital information
for computation, the output of the detector 20 is coupled
to the input of a processor and then a computer.
The communication system is designed for a low
cost market. The dish antenna and associated electronics
being housed in a conventional large signal light housing,
with a thin plastic enclosure protecting the antenna from
exposure to the weather. The antenna is linearly
polarized and has a radiating angle at 23 GHz of about
2.5 and a gain of 34 db. At the indicated 100 mw power
levels, the microwave communication system provides "all
weather" communication in a line of sight for up to about
5 miles.
In the interests of cost reduction, the mixer
achieves single conversion from 23 GHz to a 50MHz IF
frequency. Single conversion is feasible since FCC
channel allocations in this band minimize the image
problem, (alternate channel usage avoiding the image).
Single conversion intrinsically leads to reduced cost
of conversion. In addition, the mixer would be costly
if constructed using both a waveguide input and waveguide
output. As will be seen, the present approach of using
a waveguide for the high 23 GHz input frequency and
mic~ostrip circuitry at the mixer, provides a lower
cost solution by a factor of lO than an all waveguide
solution. The novel balanced mixer design achieves
reduced local oscillator radiation at the antenna and
produces a reduced noise figure for the mixer.
The mixer 16 and its components are shown in
Figures 2, 3, 4 and the equivalent circuit diagram is
shown in Figure 5. Figure 2 is a mechanical drawing
~L2~3~'7~
35-EL-1583
--10--
showing the principal external parts of the mixer, which
include the waveguide 22, the local oscillator 17 and the
microstrip circuit 26 in which the actual frequency
conversion or mixing takes place. The same elements are
shown in greater detail in Figure 3, the view being
broken open and exploded for improved clarity.
Figure 4 is a detailed plan view of the microstrip
circuit 26. Figure 5, the equivalent circuit diagram of
the mixer, commences at the point where the signal at 23
GHz is derived from the waveguide and introduced to the
microstrip circuit and continues to the point where the
low frequency signal at IF frequency (50 MHz) is
transferred from the microstrip circuit to the video
preamplifier 18. The explanation will now proceed with
reference to Figures 2-5.
As best seen ln Figure 3, the waveguide structure
is a three part assembly (22, 27, 35) fastened together
by mounting flanges. The antenna elements 23, 24, 25
are supported on the right end of the waveguide 22
(using the orientations of Figure 3), the right flange 39
of which fastens to the left flange 37 of a central
waveguide member 27. The antenna is vertically
polarized, with the received wave (TE 10) being directed
to the left into the waveguide 22 toward the waveguide
member 27. The local oscillator 17, a Gunn oscillator,
also includes a flanged waveguide section 351 the flange
36 of which is mounted on the left flange 38 of the
waveguide member 27. As seen in Figure 3, a small orifice
28 is provided in the flange 26 to couple local oscillations
into the left opening of the waveguide 27. The Gunn
oscillator has screw adjustments, i.e., 29, 30, which
permit it to be precisely tuned to a selected channel
in the 23 GHz band. The approximate frequency setting is
determined by the interior dimensions of the "cavity"
formed by the waveguide section 35. The length of the
waveguide section 35 is approximately 1 wavelength long.
5~
35-EL-1533
--11--
The Gunn diode, which acts as a dc energized, negative
resistance diode oscillator, is electrically connected
between the top and bottom surface of the waveguide section
at approximately 1/4 wavelength from the back ~leftmost)
interior surface. The tuning screw 29 is approximately
3/4 wavelength from the back interior surface of the
cavity. The arrangement generates a vertically polarized
wave (TE 10), which is coupled through the orifice 28
into the left opening of the waveguide member 27. The
orifice 28 allows sufficient decoupling (typically, the
coupling coefficient is 10-15%) between the oscillator
and the central waveguide load for oscillator frequency
stability.
The central waveguide 27, which receives signals
from the antenna entering the right opening and local
oscillations from the Gunn oscillator entering the left
opening, and thus operating in the TE 10 mode, is the
means for coupling both waves into the microstrip circuit
26. As seen in Figure 3, the waveguide 27 has a trans-
verse shorting partition 43 dividing the waveyuide intotwo isolated sections, and causing backward reflections
of any energy directed into the waveguide toward two
conductive probes or stubs 31 and 32 arranged respectively,
to the right and to the left of the central partition.
The stubs 31, 32 are the means for coupling energy from
the two interior regions of the waveguide 27 into the
microstrip circuit 26.
The stubs are inserted into the waveguide 27
through circular op~nings in the top, wide-dimensioned,
surface of the waveguide. The openings are sufficiently
large to admit the smaller radius central conductors
(31, 32) of the stubs and the two larger radius cylind-
rical insulators 33, 34 providing for their support.
The cylinders 33, 34 extend downardly through the
waveguide openings into contact with the upper surface
of the bottom of the waveguide, while the conductive
probes are arranged to be spaced typically 35 mils above
?3~7~
35-EL-1583
that upper surface of the bbttom of the waveguide.
Continuing with a description of the design and
placement of the probes, the waveguide 22 is a WR42
brass wave~uide having an interior dimension of
0.17" x 0.42" and wall thickness of 0.04". The probes
3I~ 32 are approximately 0.048" in diameter, and the
supporting insulating cylinders are 0.160" in diameter.
The assembly is installed perpendicular to the upper
surface of the waveguide and substantially in contact
with the central partition 43 on the center line of
the waveguide.
The microstrip circuit 26 has a central insulating
layer 40l 0.010" in thickness, spaced individual con-
ductors on the upper surface, and a continuous conductive
underlayer 41. The individual conductors on the upper
surface of the microstrip circuit make electrical contact
with the probes 31, 32 and having line width designed for
a proper impedance match to the probes. The conductive
underlayer 41 is mounted in electrical contact with a
bracket 42 resting on the upper surfaces of the waveguide.
Adjacent the top surface of the supporting cylinders 33,
34, a circular region of the bracket 42 and of the lower
conductive underlayer 41 is removed.
Mechanically, the supporting cylinders 33, 34
stabilize the probes from vibration, etc., and together
with the probes 31, 32 and holes for reception of the
cylinders and probes, align the waveguide 27, the support
bracket 42, and the microstrip circuit 26.
The probe design influences the coupling factors
re~uired to couple energy from the waveguide sections into
each microstrip input connection; the bandwidth of the
coupling action; and the correct impedance transformation
ratio to match the waveguide section impedance to the
microstrip impedance.
The probes are positioned on the center line of
the waveguide at a distance selected to minimize back
35'~1~
35-EL-1583
-13-
reflections out of the waveguide over the pass band and
in order to optimize the transfer of energy from the
probe into the microstrip circuit 26 within the same
pass band. The placement of the probes 31, 32 has a
very significant filtering effect upon the converter.
Measurements indicate that optimum placement of the
probe (for greater selectivity) produces a bandpass
characteristic into the microstrip over the 21.8 to
23.2 GHz communication band (approximately 6~ bandwidth)
generally in excess of 15 db (return loss) and in regions
of maximum rejection exceeding 30 db. The return loss
(into the antenna) corresponds to a VSWR of approximately
1.4 over the pass band. Outside of the pass band, the
return loss is near zero. Thus, a very low loss
communication band, bandpass filter with a good coupling
Eactor into the mixer input is provided.
The axial placement of the probe in relation to
the partition 43 also varies the bandwidth. The actual
distance of the center line of the probe to the partition
20 is 0.080", and is slightly in excess of 1/8 electrical
wavelength along the waveguide. (This provides a
minimum passband design. A wider passband would be
achieved by increasing the distance from the partition 43).
In principle/ each probe is a 1/4 wavelength
impedance transformer for matching each waveguide section
to the associated microstrip input. The physcial
configuration prevents the actual lengths of the probes
from having the desired electrical length, but this
problem is overcome by the insulating cylinders 33, 34.
The lower end of the probes must be spaced from the lower
waveguide wall to prevent the vertical E field established
in the waveguide from being shorted. The upper ends of
the probes 31, 32 extend slightly above the upper surfaces
of the waveguide and slightly above the upper surface of
the supporting cylinders 33, 34 for an amount adequate to
allow electrical contact between the upper ends of probes
and conductive members on the upper surface of the microstrip
: . :
35~8
35-EL-1583
-14-
circuit 26. The upper end of the probe passes through
the upper surface of the waveguide 27, the mounting
bracket ~2, and the microstrip circuit 26. Where the
probe passes through apertured walls, a virtual, short
coaxial line is created having a characteristic
impedance near 50 ohms, which is approximately that of
the hybrid coupler input port. The inner surface of the
upper wall of the waveguide r which is the point where
the probe enters the virtual coaxial line, defines the
upper physical limit of the probe, and due to the gap
required at the lower limit, forces the probe length to
be less than ~/4 of this application. The characteristic
impedance of the WR42 waveguide at this frequency is
approximately 480 ohms. The (vertical) E field dimension
of the wavegulde is ~/4 of the probe bounded as described
above cannot be quite this long. The use of the dielectric
supporting cylinder thus permits the electrical length
of the probe to be lengthened from about 5/6 oE A/4
to somewhat over ~/4. In free space, a quarter-wave
pole would exhibit an input transfer impedance of 337
ohms. This addition of the dielectric thus brings the
transfer impedance somewhat past that corresponding
to 1/4 length (and thus slightly inductive) and to a
value somewhat less than 480 ohms. The microstrip
25 impedance is designed to be 50-100 ohms. The slightly
inductive impedance permits a small capacitance stub
with a calculated Z ~97~ (45) to provide a good
impedance match into the input ports of the hybrid
coupler 53.
The mixing operation takes place on the micro-
strip circuit 26. The plan view in Figure 4 illustrates
the conductive runs on the upper surface of the microstrip
circuit. An equivalent circuit representation of the
electrical effect of these conductive runs is shown in
Figure 5. In summary, the microstrip circuit is seen
to include matched input connections, having capacitive
reactances 51, 52 for matching the probes 31, 32 to
33~
35-EL-1583
-15-
the input ports of a microwave hybrid coupler 53
represented in Figure 3 by two summation networks (with
the symbol "~ " superimposed) and two phase shift net-
works (with the symbol 90 superimposed). The signals
S~jL and L+jS
from the output ports of the hybrid coupler are coupled
via inductive impedance matching arms to the balanced
detector diodes 57~ 58~ The hybrid coupler 53 is
commonly referred to as a branch line hybrid coupler.
It has four ports interconnected by four 1/4 wavelength
paths~ The design permits 3 db power division at each
of the four ports. The design is capable of being
carried out in waveguide, stripline circuitry and, as
herein described, in microstrip circuitry.
An equivalent capacitance 59 between the diode
interconnection and ground completes the input circuit
for the balanced diode detector. The detector output
at 50 MHz is coupled via the filter 60, 61, 62 which
selects the desired heterodyne frequency (50 MHz) and
couples it to the output terminal 63. The necessary
r filtered, dc return path for setting the balanced
diodes in the correct operating range is shown at 74,
751 76.
The mechanical and electrical details of the
microstrip circuit are best seen with reference to Fig. 4.
The microstrip circuit 26 has a lowermost conductive layer
~1, typically of 1 oz. copper, a 0.010" thick insulating
layer 40, with a dielectric constant of 2.17, and
suitable for operation at the indicated frequencies, a
plurality of apertures for acceptance of the input and
output signal connections and operation potentials, and
a plurality of conductive runs on the upper surface of
the microstrip circuit 26.
3~ 35'7~
35-EI,-1583
-16-
These conductive runs define the microstrip paths
through the mixer and provide the necessary operating
energization for the detector diodes. The line widths
of the conductive runs together with the dielectric
constant and dimensions of the material forming the
dielectric layer, define the impedances of these lines.
In order to reduce far field radiation and propagation
losses, the line widths on the microstrip are selected
to be approximately 50 ohms. This reduces the losses
very significantly in respect to that of a 200-300 ohm
microstrip line.
The signal paths on the microstrip will now be
described with greater particularity. The antenna
signal is coupled from the probe 31 by the widened
conduction run 51 to the input port 63 of the hybrid
coupler 53. The run 51 is of minimum length to
reduce signal loss. The widening is dimensioned to
provide a capacitance to tune the inductive driving point
impedance to the nominal 50 ohms impedance of the hybrid
coupler 53. The local oscillator signal from the probe
32 is similarly coupled by a widened matching strip 52 to
the hybrid coupler input port 65. The exact length of
the local oscillator run is not critical and is connected
by a somewhat longer connection.
The hybrid coupler 53 has input ports 64, 65, and
output ports 66, 67, each connected by a l/4 ~branch.
The series ~ranches between ports 64, 67 and 65, 66~
respectively, have 35 ohm characteristic impedances (they
are wider)~ The shunt branches between ports 64, 65 and
66, 67 respectively, have 50 ohm characterisitc impedances.
(They are relatively narrower than the series arms).
Each corner of the hybrid coupler is notched to equalize
the path length around the hybrid and to direct the waves
efficiently in the widened sections. The electrical
design produces the complex quantities indicated in
Figure 5 at the hybrid coupler output ports 66, 67.
35~7~3
35-EL-1583
The next task is to take the complex quantities
available at the hybrid coupler output ports 66, 67 and
couple them efficiently to the balanced detector (57, 58).
The output port 66 is coupled via a first line segment
66 - 54, and a second line segment 54 - 68 to the anode
of diode 57 of the balanced diode pair (57, 58). The
output port 67 is similarly coupled via a first line
segment 67 - 55 and a second line segment 55 - 69 to
the cathode of the diode 58 of the pair. The cathode of
diode 57 and the anode of diode 58 are joined and are
the point from which the detected output appears. The
line segments 66 - 54 and 67 - 55 have an electrical
impedance of 50 ohms and are of equal electrical length.
The diodes 57~ 58, at 23 GHz and at a given local
oscillator power level, have an impedance of about 25 ohms.
(The selected local oscillator power level is 5-10
milliwatts). The diode impedance must be matched, first
regarding the diodes as the load for the hybrid coupler,
and secondly regarding the diodes as the generator of the
heterodyne siynal supplied to the preamplifier.
The line segments 54-68 and 55-69 (which have
been thinned) are a part of the matching networks requlred
to match the detector diodes as loads to the hybrid
coupler output ports 66, 67. The line segments are
locally increased in impedance (both inductive and
resistive) at locations just past the points 54 and 55
and which extend between 3/8~ and 7/16~ to obtain the
desired series impedances.
The diode matching networks are completed by the
application of an effective shunt capacitance (Fig. 5,
capacitor 59) between the detector output and ground.
This capacitance is obtained by use of a half-wave
].ine segment 70 serially connected between the junction
output point and the dual input matching stubs (60) of
the output pi network (60, 61, 62). This choice of
length for segment 70 reflects the capacitive impedance
~Z~3S~
35-EL-1583
-18-
of the .input stubs 60 back to the detector output.
The capacitance 59 thus completes the network for matching
the diodes 57, 58 to the hybrid coupler 53. The
inductive impedance of each short line segment (54-68;
55-69), in series with the dynamic diode load; and
the capacitance 59 and the self capacitances of the
diodes 57, 58, in shunt with the diode load, form an
impedance matching "L" network in which the diode loads
presented to the hybrid coupler becomes essentially
resistive and properly terminated.
A second effect of the capacitance 59 is to
present a low impedance to ground and thereby provide
further isolation between the antenna signal and the
local oscillator.
Assuming now that all of the measures indicated
above to effect a proper impedance transformation into
the diodes 57, 58 has taken place, the branch line
hybrid coupler causes the local oscillator signal to
be shifted in phaseso that any noise produced by the
local oscillator is cancelled at the balanced diode
detector. To maximize this cancellation, the two lines
coming to the diodes 57, 58 must be of identical
electrical length. Reduction of the noise is critical to
the noise figure of the converter and provides an
approximately 3 db improvement over the figure that one
would obtain using a single-ended mixing circuit.
Finally, the source impedance of the balanced
diode detector is in the vicinity of 25-50 ohms at the
heterodyne frequency. This source impedance provides an
approximate impedance match into the output pi filter,
whose input impedance is approximately 50 ohms.
~ s earlier noted, the diodes 57, 58 provide the
detected IF signal at 50 MHz which drives the output
pi network 60, 61 and 62. The diode output is coupled
via the 5Q ohm line segment 70 of one-half wavelength to
a short double tuning stud bearing the reference numeral
60. The use of the 1/2 wavelength line segment 70 allows
3~
35-EL-1583
--19--
the capacitance obtained from the double tuning stub 60
to be reflected back to the diodes where it appears as
the capacitance 59 in the Figure 5 diode load circuit.
It also appears as the first capacity 60 of the Figure 4
pi network. The first double tuning stub has a line
width corresponding to 50 ohms and extends above and
below the center of the line segment 70 by approximately
1/8 of a wavelength. This permits a first high shunt
capacitance. The serial inductive member 61 is of
approximately 200 ohm line width and is approximately
1/4 wavelength (at 23 GHz). This presents a large
serial inductance. The second dual stub of the pi
network has a line width of 50 ohms and each stub extends
1/4 wavelength above and below the center line. This
presents a very high shunt capacitance and an equivalent
a.c. ground at the microwave frequencies. The 50 ohm
output line segment 71 couples filtered output to the
output pad 63 for connection to the preamplifier.
The output pi network is designed to filter
out all microwave frequencies from the detected output.
As discussed, it is a pi, low pass filter of the
electrical configuration shown in Figure 5. The design
eliminates the microwave components in the region of 23 GHz
and above. The output impedance of the filter at 50 MHz is
about 500 ohms at the IF frequency and fits the impedance
of the mixer output to the input impedance of the
preampli~ier.
The diodes 57, 58 must be operated at a desired
dynamic impedance level. For this reason~ a carefully
filtered dc return path for the diode is provided which
is symbolically illustrated in Figure 5 by the inductance
74~ inductance 75 and the capacitance 76. As seen in
Figure 4, a high impedance 1/4 wavelength line 74 is
provided at the local oscillator input. It is connected
to a line segment 76, which is also of 1/4 wavelength
(50 ohm line width). This stub configuration creates
3~
35-EL-1583
-20-
an ac short at 23 GHz, which reflected to the take-off
point, becomes an infinite impedance and prevents
significant ac coupling of local oscillator output into
the line. The dc return continues as a high impedance
5 mil line entering a serpentine configuration which
increases the self-inductance and series resistance of
the line until it reaches the pad 77 to which the
preamplifier is also grounded and which is the ground
reference of the diode output. When the dc paths are
traced, it may be seen that the diodes 57, 58 are
operating at a zero dc bias. Thecorrect dynamic impedance
level is achieved by using the correct level of the local
oscillator. The 5 to 10 milliwatt level achieves this
objective. The balanced configuration permits both diodes
(àssuming they are closely matched) to be balanced. The
diodes 57l 58 are beam load supported with an 0.1 pico-
farad package to ground capacity. The diodes are of
Schottky design (5600 Alpha Industries, Inc., Woburn,
MA, 01801).
The practical embodiments of the invention so
far described has been optimized for use at the 21.8
to 33.2 GHz communications band. The WR~2 waveguide
selected is designed to propagate waves over the
frequency range of from 18 to 26.5 GHz, in a TE 10 mode.
A larger waveguide~ as for instance oneof double size,
would pose the problem of supporting spurious modes, as
for instance a TE 11 mode at the same frequency. The
wa~eguide selectionl if the operating band is changed,
should be scaled for restriction to the TE 10 mode of
propagation.
Similarly, the microstrip circuit design must be
selected in relation to the operating frequency (21.8-23.2
GHz). In generall 25 ohm lines are too wide for this
f~equency, and exhibit moding problems, and excessive
propagation path problems. Thus, the hybrid coupler
cannot ordinarily be designed to the exact dynamic
impedance (25 ohms) of the detector diodes, but must
~L~ 35'713
35-EL-1583
be matched through a transformer. The selection of 50
ohms and 35 ohms for the hybrid coupler branches and
other microstrip signal paths is practical~ and in the
case of the hybrid coupler~ the use of corner notches
permits low impedance operation while avoiding multi-
path effects. Had the operating frequency been lower
(higher), wider tnarrower) lines would have been
tolerated.
The microstripcircuit selected for the present
application is of 10 mil 2.17 dielectric constant
material manufactured by Minnesota Mining and
Manufacturing Company, Inc. The selection permits
flexibility without breakage, and good electrical
efficiency. In general, at the 21.8 - 23.2 GHz
communications band7 this thickness is optimum, although
slightly thinner materials could be used. At lower
microwave frequencies, the dielectric material may be
thicker (perhaps .125 at 3 GHz) since the hybrid coupler
may have longer lines for a given wavelength and wider
conductors (also a function of the wavelength) without
multipath problems.
The local oscillator frequency is selected to
be close to the signal frequency: a 50 MHz difference
at 22 GHz. This close frequency relation permits the
heterodyne to be of a low enough frequency for efficient
low cost amplification of the heterodyne. The frequency
selection also permits the hybrid coupler to work
efficiently without broad banding for both local
oscillator and signal frequency isolation. The frequency
selection also permits use of a common waveguide (27),
both sections of which support the loca] oscillator and
signal in the TE 10 mode.
While the practical embodiment so far described
is optimized for the 21.8 to 23.2 GHz communications
band, the general design, in which wave~uide and micro-
strip circuits form the principal microwave conduits,
can be efficiently scaled up or scaled down in frequency.
135'78
35-EL-1583
-22-
The upper band limit currently available is R-band
(26.5 - 40 GHz) and this limit is set by available
low cost microwave sources. The lowermost band likely
to be of value is S-band (2.60 - 3.95 GHz). Frequencies
lower than this suggest the use of something other than
waveguide for the high frequency portions of the circuit
to avoid excessive costs. In other words, the invention
is generally applicable to microwave frequencies by which
is meant frequencies of about 1 gigahertz and higher.
The range of optimum application of the invention appears
to be the centimeter wave region (3-30 gigahertz) and
slightly above that (30-40 gigahertz).
The noise figure of the design is 7 db of which
1-1/2 db is assignable to the preamplifier coupled to the
output 63 of the detector. The preamplifier is conven-
iently formed of two stages of low noise amplification
such as are provided bv two Plessey SL 560C low noise
RF amplifiers~ the two amplifiers being cascaded and
followed by bandpass filters. The arrangement provides
30 db of gain and available filters permit 80 db skirts
with a 50 MHz center frequency and a signal bandwidth
~f ~ 25 MHz.
-
The probes 31, 32 are of standard 0.050" (nominal)
~ diameter (0.0048" actual), and were selected because
Of their availabiLity. The insulators 51, 52 are alsoof standard diameter 0.90" and available at low cost.
The probe and insulator diameters affect the tuning of
the 1/4 resonant stub, and the matching between waveguide
and hybrid coupler. The total design permits use of these
arbitrary sizes, and in particular the positioning of the
probe and the size of the capacitances 51 t 52 give that
flexibility. The capacitance in particular can be
dimensioned in the etching of the microstrip surface
conductor to have the desired exact value.
