Study into phased arrays

Document Sample
Study into phased arrays Powered By Docstoc
					Study into Phased Arrays in

NGSO Earth Station Antennas
Summary
Currently, almost all satellites which provide broadcast, point-to-point and point-to-multipoint
networks to business and domestic customers are located in GeoStationary Orbit (GSO). Over the
next decade, it is anticipated that a number of new satellite systems will be launched, providing a
wide range of broadband services using Non-GeoStationary Orbits (NGSO). In the former case, the
satellite appears at a nominally fixed point in the sky and it is possible to maintain communications
with the satellite using a static ground terminal antenna; only a single satellite is used for the
communications link at all times. The key features which differentiate NGSO systems from GSO
networks are that the ground terminal must track relatively fast moving satellites as they traverse the
visible sky and communications must be handed from one satellite to another as one leaves this
visible region and another appears above the horizon.

The proposed systems, because they use Non-GeoStationary orbits, must employ a large number of
satellites if they are to provide continuous coverage over the entire globe. For such systems to be
commercially viable, it is essential that they attract a large customer base so that the substantial
investment in the satellite infrastructure can be recouped through access charges. Ultimately, the rate
of take-up of the systems and the total market volume will depend on the cost of ground terminal
equipment and access charges; based on the submissions of the key satellite providers, it is believed
that the cost of terminals must be in the region $1000-$2000 if customer acceptance is to be achieved.

Most potential satellite network operators and system suppliers claim that the targets set for the
terminal costs are achievable and many propose low profile phased array systems as the antenna
design solution. At this time, no such phased array systems have been designed which meet the cost
targets. This study provides a review of the state of the art in phased array technology and gives an
appreciation of the use of phased array antennas for the high volume ground terminal application.

Phased arrays can provide complete two dimensional scanning or single plane beam movement. For
a number of the NGSO scenarios, it appears that the latter is sufficient for a given location on the
earth; however, the antenna would need to be set to a given cross track angle to enable this to be
achieved. This has a major impact on the cost of phased array technology, since it offers the potential
to reduce the number of active modules used in the arrays significantly.

At present, there are no commercially available phased arrays available at any frequency. Such
antennas are, however, used up to Ku-Band for a number of military and earth observation
applications and Ka-Band units have been produced for use with US MILSTAR satellites. A number
of existing commercial applications, notably for TVRO satellite reception, could beneficially make
use of phased arrays to switch between satellites such as Astra and Eutelsat, for example; however,
but units have not been put on the market due to the continued high cost of such antennas.

The approximate cost of a T/R module for a military Ku-Band phased array is about $100. If this is
extrapolated to a Ka-Band communications application, where separation of the transmit and receive
functions is mandatory due to the frequency separation and need for high isolation between the
transmit and receive functions, it is reasonable to assume that each Transmit or Receive module
might also cost $100. On this basis, the active modules for a typical dual band phased array with 100
elements for both transmit and receive would cost $20,000. It is therefore concluded that, in order for
phased arrays to become commercially viable, a cost reduction for the active modules of at least two
orders of magnitude is required.
                                          Contents
                                                                 Page No.

1. Introduction                                                         9
      1.1 Background                                                    9
      1.2 Report Outline                                                9

2. Summary of the Requirements                                        10
      2.1 NGSO Fixed Satellite Services                                10
      2.2 Key Design Drivers                                           11
      2.3 Acquisition and Tracking Requirements                        11
      2.4 Cost Drivers                                                 12
      2.5 Basic Antenna Design Concepts                                12

3. Array Fundamentals                                                 13
      3.1 Array Design Constraints                                     13
              3.1.1 Required Aperture Size                             14
              3.1.2 Inter element spacing                              14
              3.1.3 Specific examples                                  15
              3.1.4 Impact of scanning on sidelobe levels              17
      3.2 Beamforming                                                  20
              3.2.1 Basic concepts                                     20
              3.2.2 Multi-beam operation                               22
      3.3 Quantisation and Tolerance Issues                            24
              3.3.1 Quantisation errors                                24
              3.3.2 Random errors                                      30

4. Antenna technologies                                               33
      4.1 Arrays 33
              4.1.1 Arrays for NGSO applications                       33
              4.1.2 Single plane scanning array                        34
              4.1.3 Two plane scanning array                           36
      4.2 Parabolic Reflector Antennas                                 38
              4.2.1 Parabolic reflectors for NGSO applications         38
              4.2.2 Basic design                                       40
              4.2.3 Production technology                              40

5. Solid State Components                                             41
      5.1 Device Technologies                                          41
               5.1.1 GaAs MESFET                                       42
               5.1.2 GaAs HEMT                                         42
               5.1.3 GaAs PHEMT                                        42
               5.1.4 InP HEMT                                          42
               5.1.5 HBT 43
      5.2 Components                                                   43
               5.2.1 High power amplifiers                             43
               5.2.2 Low noise amplifiers                              45
               5.2.3 Phase Shifters                                    46
               5.2.4 Integrated Transceivers                           47
      5.3 Solid State Device Summary                                   48
6. Antenna Cost Estimates                        49
     6.1 Basic cost components                   49
             6.1.1 MMICs                         49
             6.1.2 Array face production costs   52
             6.1.3 Reflector production costs    52
     6.2 Terminal antenna costs                  52

7. Summary and Conclusions                       55

8. References                                    56
     8.1 References for Section 3                56
     8.2 References for Section 4                56
     8.3 References for Section 5                56
     8.4 References For Section 6                58
                                            Tables List
Table 2.1: NGSO satellite systems                                                          10
Table 3.1: Sidelobe level variation with scan angle                                        20
Table 3.2: Effect of quantisation errors on antenna performance                            29
Table 4.1: Array requirements for NGSO scenarios                                           33
Table 4.2: Parabolic reflector aperture sizes                                              40
Table 5.1: Ferrite phase shifter characteristics                                           46
Table 6.1: MMIC dimensions and costs at $14 per mm2                                        50
Table 6.2: Transmit and Receive module costs                                               51
Table 6.3: Costing assumptions                                                             53
Table 6.4: Costing for example Ka-band system                                              53
Table 6.5: Costing for example Ku-band system                                              54



                                            Figures List
Figure 3.1 : Generation of grating lobes                                                   13
Figure 3.2: Basic antenna geometry                                                         14
Figure 3.3: Number of elements as a function of scan angle, 30 and 40 dBi antennas         16
Figure 3.4: Square array side length as a function of scan angle, 30 and 40 dBi antennas   17
Figure 3.5: Beam scanned to 0° and 30°                                                     18
Figure 3.6: Beam scanned to 50° and 70°                                                    19
Figure 3.7: Aperture distribution used                                                     19
Figure 3.8: Active and passive array configurations                                        21
Figure 3.9: Fixed beam array topologies                                                    22
Figure 3.10: Multi-beam fixed beam array                                                   23
Figure 3.11: Multibeam operation of an active array                                        23
Figure 3.12: Phase errors due to quantisation                                              25
Figure 3.13: Gain loss as a function of phase shifter bits                                 26
Figure 3.14: Degradation in RMS sidelobe levels with phase shifter bits                    27
Figure 3.15: Peak sidelobe level as a function of phase shifter bits                       28
Figure 3.16: Maximum pointing error as a function of phase shifter bits                    29
Figure 3.17: Relationship between theoretical, design and error sidelobe levels            31
Figure 3.18: Acceptable errors for a -25 dB linear array of 10 elements   32
Figure 4.1: Hybrid antenna solution                                       34
Figure 4.2: Hybrid antenna solution                                       35
Figure 4.3: Two beam system                                               35
Figure 4.4: Printed element solution for mm-Wave arrays                   37
Figure 4.5: Waveguide phased array module                                 38
Figure 4.6: Use of two reflectors with a NGSO system                      39
Figure 5.1: O/P power versus PAE for a range of amplifiers                44
Figure 5.2: Power limits of commercial solid state amplifiers             45
Figure 5.3: Commercial Solid State LNA Noise Performance Data             46
Figure 6.1: Reflector antenna cost model                                  52
                Abbreviations List
AlGaAs    Aluminium Gallium Arsenide
BER       Bit Error Rate
CDMA      Code Division Multiple Access
DBS       Direct Broadcast Satellite
Eb/No     Energy per bit / noise power density
GaAs      Gallium-arsenide
GEO       Geostationary Earth Orbit
GSO       GeoStationary Satellite Orbit
G/T       Gain / noise temperature ratio
HBT       Heterojunction Bipolar Transistor
HEMT      High Electron Mobility Transistor
HPA       High Power Amplifier
I-Band    Letter designation for approx. 8-10GHz
IF        Intermediate Frequency
InP       Indium Phosphide
IMPATT    Impact Avalanche Transit Time
InGaAs    Indium Gallium Arsenide
J-Band    Letter designation for approx. 10-20GHz
Ka-Band   Letter designation for approx. 20-36 GHz
LEO       Low Earth Orbit
LNA       Low Noise Amplifier
LO        Local Oscillator
Mbps      Million bits per second
MEO       Medium Earth Orbit
MESFET    Metal Semiconductor Field Effect Transistor
MMIC      Microwave Monolithic Integrated Circuit
n/a       Not available or not applicable
NGSO      Non-GeoStationary Orbit
PAE       Power Added Efficiency
PHEMT     Pseudomorphic High Electron Mobility Transistor
PS        Phase Shifter
QPSK      Quadrature Phase Shift Keying
RX        Receive
SLL       Sidelobe level
TBD       To be Determined
TDMA      Time Division Multiple Access
T/R       Transmit receive unit
TVRO      Television Receive Only (terminal)
TWTA      Travelling Wave Tube Amplifier
TX        Transmit
W-band    Letter designation for approx. 75 - 110 GHz
WWW       World Wide Web
X-band    Letter designation for approx. 8.2 - 12.4 GHz
         Wavelength
1.      Introduction
1.1     Background
Currently, almost all satellites which provide broadcast, point-to-point and point-to-multipoint
networks to business and domestic customers are located in GeoStationary Orbit (GSO). Over the
next decade, it is anticipated that a number of new satellite systems will be launched, providing a
wide range of broadband services using Non-GeoStationary Orbits (NGSO). In the former case, the
satellite appears at a nominally fixed point in the sky and it is possible to maintain communications
with the satellite using a static ground terminal antenna; only a single satellite is used for the
communications link at all times. The key features which differentiate NGSO systems from GSO
networks are that the ground terminal must track relatively fast moving satellites as they traverse the
visible sky and communications must be handed from one satellite to another as one leaves this
visible region and another appears above the horizon.

The proposed systems, because they use Non-GeoStationary orbits, must employ a large number of
satellites if they are to provide continuous coverage over the entire globe. For such systems to be
commercially viable, it is essential that they attract a large customer base so that the substantial
investment in the satellite infrastructure can be recouped through access charges. Ultimately, the rate
of take-up of the systems and the total market volume will depend on the cost of ground terminal
equipment and access charges; based on the submissions of the key satellite providers, it is believed
that the cost of terminals must be in the region $1000-$2000 if customer acceptance is to be achieved.

Most potential satellite network operators and system suppliers claim that the targets set for ground
terminal costs are achievable and many propose low profile phased array systems as the antenna
design solution. At this time, no such phased array systems have been designed which meet the cost
targets. This study provides a review of the state of the art in phased array technology and gives an
appreciation of the use of phased array antennas for the high volume ground terminal application.

1.2     Report Outline
This report is divided into six major sections. In Section 2, the generic requirements for NGSO
ground terminals are defined and potential antenna configurations identified. Section 3 provides a
summary of the fundamental limitations which define the performance of phased arrays and in
Section 4, a review of array technology is undertaken and compared to competing reflector antenna
solutions. Section 5 addresses the active device technology which would be needed to make phased
arrays a practical solution for NGSO ground terminals. Section 6 looks at the requirements for five
specific NGSO satellite constellations, including outline link budgets, system orbit characteristics and
antenna size and complexity, while in Section 7, preliminary cost estimates are put forward for each
system, based on current prices and technologies. Finally, Section 8 provides a summary of the
report and draws conclusions and Section 9 lists the references cited throughout the report.
2.      Summary of the Requirements
2.1     NGSO Fixed Satellite Services
It is expected that a number of different NGSO systems will be launched over the next decade.
Obtaining up-to-date information on these systems is often difficult, as the system specifications are
constantly being re-defined. Link budgets have been produced for many of the different proposed
systems, showing potential requirements for a very wide range of ground terminal antenna sizes. In
general, the size range is between about 40 cm diameter and 3m diameter, representing a gain range
of between 40 and 60 dBi at 30 GHz, though a number of lower data rate systems propose smaller
antennas. For most of these systems, the obvious antenna type is a reflector but the use of a phased
array offers potential benefits in terms of aesthetics and the absence of moving parts and it is phased
array antennas which are the principal focus of this study. Inevitably, given the very narrow
beamwidths associated with high gain antennas, any antennas used with NGSO systems will need to
incorporate a tracking facility.

Table 2-1 below provides a summary of the characteristics of a number of proposed NGSO systems.


                                 Table 0.1: NGSO satellite systems

  Constellation           Teledesic    Lugos          West              Celestri           Skybridge
  Orbit type              NGSO         NGSO           MEO / GSO         NGSO / GSO         NGSO
  Approx no of            288          161            <10 MEO, <2       64 (63 NGSO,       64
  satellites                                          GSO               1 GSO)
  Up-link frequency       Ka-band      Ka-band        Ka-band           Ka-band            Ku
  band                    (28.85)      (29.1)         (30)              (30)               (14)
  Down-link               Ka-band      Ka-band        Ka-band           Ka-band            Ku
  frequency band          (19.25)      (19.3)         (20.2)            (20)               (11)
  Up-link terminal        35.2         37.26          43.6              35.6               32.2
  antenna gain (dBi)
  Down-link terminal      34.1         33.69          40.2              35.6               30.6
  antenna gain (dBi)
  Maximum terminal        0.4          3              1                 4.7                0.7
  transmit power
  (dBW)



As previously stated, communicating with NGSO satellites is much more demanding than
communicating with GSO satellites, since the satellite moves rapidly over the sky. In order to
facilitate hand-over between satellites (as one leaves the visible sky and another comes into view)
two beam systems must be considered where the beam separation could be as high as 140°; in this
case, the use of electronically steered antennas becomes a highly attractive option. It is anticipated
that, in most cases, the terminals will need to incorporate two plane scanning to cater for satellites
which do not pass directly over the terminal.
Single beam systems can be used for NGSO satellites if the hand-over can be performed with
sufficient speed. A fully electronically steered phased array can accomplish this in less than 1s and
so this is likely to be a feasible option.

2.2     Key Design Drivers
There are design considerations which, when taken together, are unique to NGSO systems and these
have a major impact on the ground terminal antenna requirements. These are:

i)     dual frequency operation.
ii)    wide angular coverage.
iii)   high gain.
iv)    low sidelobes to minimise interference with other satellites and terrestrial services.
v)     rapid hand-over between satellites exiting the visible sky and a second satellite coming into
       view.
vi)    low antenna losses and, hence, good Gain to Noise Temperature (G/T) performance.
vii)   low recurring cost.

Taking a limited set of these criteria into account would be straightforward, for example a reflector
solution is ideal to satisfy (i), (iii), (iv), (vi) and (vii). However, rapid hand-over and wide angle
scanning can only be realised using two separate mechanically steered antennas.

A phased array, in principle, can readily satisfy (ii) and (v); however, using such an antenna, it will
be very difficult to satisfy the dual band requirement with a single antenna and the sidelobe
requirements over wide angular scanning using this approach.

2.3     Acquisition and Tracking Requirements
The ground terminal antennas are required to perform three distinct antenna functions, either with
manual intervention or automatically, depending on the application. These are:

i)     Acquisition. This relates to initial acquisition of the communications link prior to
       commencement of data transmission.
ii)    Tracking. This relates to maintaining accurate alignment of the antenna during data
       transmission.
iii)   Data transmission.

For a terminal tracking an NGSO satellite, knowledge of the location of the satellites will be
available in advance and this information could be used for alignment and tracking of the satellite.
For this application, the beam must move quite rapidly over a significant angular region. For
example, for a zenith pass, the rate of motion could reach many degrees per minute. In addition, the
rate of motion is variable, and complex scan control systems are required.

Unless the satellite passes directly overhead, it is not possible to define the track of the satellite with
respect to the terminal as a single plane. Tracking using single plane scanning may still be
considered, if the variation in the orthogonal plane is small compared to a beamwidth. This could be
achieved by mechanically aligning the antenna so that the satellite passes in a quasi-planar orbit with
respect to the antenna co-ordinate system. The antenna may require re-positioning mechanically for
the next satellite. The feasibility of this option depends on the satellite orbit geometry.

2.4     Cost Drivers
A critical parameter in any phased array design suitable for the ground terminal market must be
production cost. In principle, the array face itself can be manufactured at a low cost, either by using
printed technology (ie. microstrip patch elements and feed networks) or by adopting a waveguide
array approach in which the waveguides are formed from metallised plastics. The key cost drivers
will inevitably be the active components which will be required to steer the beam. MMIC technology
for Ka-Band applications is in its infancy, and developments in this area have been relatively slow
due to the lack of high volume applications. This may change in the near future as automotive radar
and point-to-multi-point communications applications become more widely deployed.

The current cost of a single Transmit/Receive (T/R) module for a phased array operating in I/J-band
is in the region of $100. Given the need for a higher frequency of operation, and typically 100 such
modules per antenna, the cost will need to be reduced by at least two orders of magnitude before such
a technology could be applied to the consumer market.

2.5     Basic Antenna Design Concepts
A total of four principle design concepts can be considered for the NGSO systems, divided into
arrays and reflector solutions:

i)     Active flat plate array using two plane electronic scan.
ii)    Active flat plate array using single plane electronic scan (together with mechanical scan in the
       orthogonal plane, if needed).
iii)   Passive flat plate array, with two plane mechanical scanning.
iv)    Parabolic reflector antenna, with two plane mechanical scanning.

This report focuses on the first two of these and uses the latter for comparison purposes only.
3.       Array Fundamentals
3.1      Array Design Constraints
Array solutions can be used to replace the reflector based configurations which are currently used in
the majority of ground terminals in several ways. Firstly, fixed arrays can be used in conjunction
with the same mechanical positioning equipment to offer a lower profile solution having the same
functionality. Alternatively, phased arrays can be proposed, making it possible to remove all
mechanical movement and provide a static low profile solution. A third solution may also be
considered which uses single plane electronic scanning in conjunction with orthogonal plane
mechanical scan.

For fixed beam arrays (which could only be used for NGSO in conjunction with a mechanical
positioner), the use of a single aperture for both the up- and down-link frequencies is possible. When
elements are placed too far apart, a single high gain beam is no longer produced. A number of beams
of similar gain are generated, which are referred to as a main beam and "grating lobes". This is
illustrated in Figure 0.1.


                                                                              Main lobe
                                            Grating                                Grating lobe
      Main lobe                             lobe




                  Antenna pattern for closely             Antenna pattern for widely
                    spaced (<1) elements                 spaced (>1 elements




                                   Figure 0.1 : Generation of grating lobes

In order to ensure no grating lobes in the radiation pattern, the element size must be <1 at the
highest frequency of operation (ie. 30 GHz), which is some 0.67 at 20 GHz. For any wide angle
scanning array solutions, two separate arrays will be needed, since the element size must now be
substantially reduced to ensure that grating lobes do not appear at maximum scan. As an example,
for a 40º scan away from mechanical boresight, the inter-element spacing must be less than 0.6,
which is only 0.4 at the lowest frequency; waveguide elements cannot be used for this separation
unless they are dielectrically loaded. This adds significant complexity to the antenna design due to
the need to match the elements to a beamforming network and to free space and, potentially, reduces
the antenna efficiency due to dielectric losses. Printed elements could be considered but these are
very difficult to design covering such a wide bandwidth and would have an unacceptably high loss at
the frequencies in question. The use of two separate arrays, one for the up-link and another for the
down-link, is therefore the only solution; however, the total size of the antenna sub-system will be
increased.

3.1.1 Required Aperture Size

The approximate antenna aperture size required in order to realise any given gain specification is
                     G2
given by A                    , where G is the gain requirement  is the efficiency  is the
                4 cos cos 
wavelength, and  ,  are the scan angles in azimuth and elevation away from the normal to the
aperture surface, see Figure 0.2Figure 0.2. This concept is used extensively throughout the report as
the basis for estimating the antenna size requirements.




                                                                        
                                                                                    Array face
            Look direction




              Mechanical
              boresight
                                               




                                     Figure 0.2: Basic antenna geometry



3.1.2 Inter element spacing

When an array is used to generate a steerable beam, the inter-element spacing is constrained by the
need to ensure that no grating lobes are generated. These reduce gain and increase the potential for
interference. The criterion which must be satisfied is that the inter-element spacing (d) in a given
                     d           1
plane must satisfy                    , where  is the scan angle in that plane.
                            1  sin 
3.1.3 Specific examples

Figure 0.3 illustrates the relationship between the number of antenna elements required to achieve 30
and 40 dBi gains as a function of scan angle, taking into account the increased aperture size
requirements discussed above. It can be seen that the number of elements required increases very
rapidly for scan ranges between no scan and 70º scan in two planes.

Similarly, Figure 0.4 shows graphically how the side length of a square array increases as a function
of scan angle requirements for 30 dBi and 40 dBi antennas operating at 30 GHz, assuming 100%
aperture efficiency. It is clear from the figure that the scan angle requirement has a major impact on
the size of the array.
                                         Number of elements as a function of scan angle, 30 dBi gain at 30 GHz




                     3000



                     2500
Number of elements




                      2000



                      1500
                                                                                                                       70
                                                                                                                  60
                      1000                                                                                   50
                                                                                                        40

                       500                                                                         30    Elevation scan angle (deg)

                                                                                              20
                             0
                                                                                         10
                                 70 65 60
                                          55 50 45
                                                    40 35 30                         0
                                                               25 20 15
                                                                        10   5
                                      Azimuth scan angle (deg)                   0




                                         Number of elements as a function of scan angle, 40 dBi gain at 30 GHz



 Number of elements



                     30000


                     25000


                     20000


                     15000
                                                                                                                       70
                                                                                                                  60
                      10000                                                                                  50
                                                                                                        40
                       5000                                                                        30    Elevation scan angle (deg)
                                                                                              20
                             0
                                                                                         10
                                 70 65 60
                                          55 50 45
                                                    40 35 30                         0
                                                               25 20 15
                                                                        10 5     0
                                      Azimuth scan angle (deg)




Figure 0.3: Number of elements as a function of scan angle, 30 and 40 dBi antennas
                                          Square array side length as a function of scan angle, 30 dBi gain at 30 GHz




                          0.9

                          0.8

                          0.7

                          0.6
        Side length (m)




                          0.5

                                                                                                                          70
                           0.4
                                                                                                                     60
                           0.3                                                                                  50
                                                                                                           40
                           0.2
                                                                                                      30    Elevation scan angle (deg)
                           0.1
                                                                                                 20
                                0
                                                                                            10
                                    70 65 60
                                             55 50 45
                                                       40 35 30                         0
                                                                  25 20 15
                                                                           10   5
                                         Azimuth scan angle (deg)                   0




                                          Square array side length as a function of scan angle, 40 dBi gain at 30 GHz




                          0.9

                          0.8

                          0.7

                          0.6
        Side length (m)




                          0.5

                                                                                                                          70
                           0.4
                                                                                                                     60
                           0.3                                                                                  50
                                                                                                           40
                           0.2
                                                                                                      30    Elevation scan angle (deg)
                           0.1
                                                                                                 20
                                0
                                                                                            10
                                    70 65 60
                                             55 50 45
                                                       40 35 30                         0
                                                                  25 20 15
                                                                           10   5
                                         Azimuth scan angle (deg)                   0




    Figure 0.4: Square array side length as a function of scan angle, 30 and 40 dBi antennas

3.1.4 Impact of scanning on sidelobe levels

It is inevitable that ground terminal antennas for Ka-Band systems will have to satisfy stringent
sidelobe specifications similar to those imposed on current Ku-Band systems. Arrays can be
designed to satisfy low sidelobe specifications; however, the sidelobe performance degrades as the
scan angle is increased and maintenance of low sidelobes at wide scan angles will, at best, be very
difficult and may be impossible.

Figure 0.5 and Figure 0.6 below show a set of ideal radiation plots based on a low sidelobe array
distribution for a 10 element array with a 0.5 inter-element spacing as the beam is scanned to 70º.

The distribution used to generate these patterns gives a sidelobe distribution which falls off with
increasing angle in accordance with the behaviour required to fit most regulatory templates, and is
illustrated in Figure 0.7.

In all the plots shown, the aperture amplitude distribution above has been used. The beam has been
scanned by applying an appropriate phase taper to the elements and each pattern has been normalised
to its peak value (ie. scan losses are ignored). This allows a direct comparison of sidelobe levels to
be made.




                                                 0
     -90           -70   -50     -30      -10         10     30        50       70        90


                                                 -5



                                                -10



                                                -15
                                                                                                   0°
                                                                                                   30°
                                                -20
  Amplitude (dB)


                                                -25



                                                -30



                                                -35
                                            Angle (deg)




                               Figure 0.5: Beam scanned to 0° and 30°
                                                                        0
                        -90     -70      -50        -30          -10             10         30       50         70       90


                                                                        -5



                                                                       -10



                                                                       -15
                                                                                                                               50°
                                                                                                                               70°
                                                                       -20
  Amplitude (dB)


                                                                       -25



                                                                       -30



                                                                       -35
                                                                   Angle (deg)




                                               Figure 0.6: Beam scanned to 50° and 70°

                                                                         Excitation

                                                                       0.4



                                                                   0.35



                                                                       0.3



                                                                   0.25
  Relative Excitation




                                                                       0.2



                                                                   0.15



                                                                       0.1



                                                                   0.05



                                                                         0
                        -2.5   -2     -1.5     -1         -0.5               0        0.5        1        1.5        2   2.5
                                                             Distance (Wavelengths)




                                               Figure 0.7: Aperture distribution used
Table 0.1 below shows the sidelobe level relative to the peak of each beam.
                        Table 0.1: Sidelobe level variation with scan angle

                                 Scan angle (°)        Relative sidelobe
                                                             level
                                       0                     -19.1
                                       30                    -17.7
                                       50                    -16.1
                                       70                     -6.0



3.2     Beamforming

3.2.1 Basic concepts

A beamforming network can be designed to generate a fixed beam, for single plane scanning or for
two plane scanning and it can be passive or active. Transmit and receive networks will be separated
at some point if the antenna is to be used for both functions. This can be done most cheaply using an
external diplexer. Where two separate antennas are used, this component is not required, although
filtering may still be necessary.

Where no scanning is required, there is no need for control components (phase shifters). In single
plane scanning, elements can be grouped in rows and a phase shifter is required for each row. Using
two plane scanning, a phase shifter is required for each element. In a passive network, the only
components are power splitters and phase shifters and the antenna is connected to a single HPA or
LNA; in an active network there are many HPAs and/or LNAs. Depending on how the elements are
inter-connected, there could be one HPA and/or LNA for every element or there could be one for a
group of elements. The latter case would typically arise where a row of elements is fed together and
only a single plane of scan is required.

Figure 0.8 shows generic active and passive configurations for an typical linear array. In the passive
case, the phase shifters must be low loss whereas in the active case there is no such requirement.
Receive arrays are similar except that the amplifiers are reversed.
                                Antenna              Active configuration.
                                elements             Separate amplifiers for each
                                                     element.
                                 A


                                 A


                                 A                                 Input


                                 A
                                                  Phase shifters




                                Antenna              Passive configuration.
                                elements             Common amplifier for
                                                     all elements




                                                       A
                                                                   Input




                                                   Phase shifters
                                                   (must be low loss)




                       Figure 0.8: Active and passive array configurations



A basic fixed beam array consists of an array of radiators interconnected using a low loss power
dividing or combining network, see Figure 0.9. In the figure, two different topologies are shown.
The first is easily configured for single plane scan by inserting a phase shifter for each column. The
second topology could only be used for two plane scanning. As shown, both have equal path lengths
to every element.
                    Row / column approach                       Subarray approach




                             Figure 0.9: Fixed beam array topologies
The beamformers themselves can take many forms. At lower frequencies, printed circuit power
dividers etched on to low loss dielectric sheets are used while, at Ka-Band frequencies, the use of
waveguide power dividing networks offers a much lower loss solution. The latter solution is
mandatory for a passive array; however, printed technology can be considered even at Ka-band for
active arrays.

3.2.2 Multi-beam operation

Multi-beam operation may be necessary for NGSO operation if there is a requirement to
communicate with two satellites simultaneously (for example to provide a redundant data pathway)
or if it is felt that hand-over cannot be achieved with sufficient speed using one beam. This latter
concern may not be significant for a phased array since hand-over time can be below 1s. If an array
can scan in one plane only then a second beam would also be in the same plane.

Multi-beam operation is achieved by effectively overlaying two different beamformers, and exciting
the elements with one or other of the networks. If the two beams required are in the same plane, there
are a number of low loss passive networks which can be used to provide multiple beams (ie. Butler
and Blass matrices), provided that the beams generated are "orthogonal", ie. they do not significantly
overlap in space. Only the first of the topologies shown in Figure 0.9 can be used with these multi-
beam networks; Figure 0.10 below shows how the beamformer is re-configured for multi-beam
operation, in this case, showing two beams. These networks are inevitably narrowband; in a dual
frequency application, there would need to be a diplexer on each column and a separate beamformer
for each frequency, as illustrated in the figure.
                 Multi-coupler network


    Beam inputs


                           Figure 0.10: Multi-beam fixed beam array
With the number of columns required for the proposed antennas, it is not considered viable to
propose a multi-beam passive array due to the complexity and consequent size and losses associated
with the beamforming function. In an active array environment, the complex multi-coupler networks
can be replaced by simple multi-way power dividers and a conventional corporate power dividing
network. In this case, phase shifters are required on each column, one for every beam; however,
beam locations in a single plane are now completely un-constrained. This active concept is shown in
Figure 0.11, in this case for a two beam TX array.




                                                                              HPA
                                                            Power combiner
                                                           TX phase shifter

                                                 TX divider 1
                                            TX divider 2


                                                            TX


                      Figure 0.11: Multibeam operation of an active array
Complete freedom to place beams anywhere within the coverage sector served by the array is only
achieved by individually addressing the phase at each element. To achieve simultaneous multi-beam
operation, it is necessary to have M sets of phase shifters with M beamformers, where M is the
number of beams required. It is clear that the complexity of this approach rapidly becomes
unmanageable; an alternative is to use digital beamforming techniques where multiple beams can be
generated digitally using a single set of RF signals. This approach is not considered commercially
viable for the current application at this time.

It is not considered that any of the NGSO FSS systems considered in this report require simultaneous
multi-beam operation.

3.3     Quantisation and Tolerance Issues
It is likely that ground terminal antennas operating in Ka-Band will have to satisfy very stringent
sidelobe templates, as is common at Ku band. In order for this to be possible, high tolerance designs
with multi-bit phase shifting elements will be required. Analogue devices can be manufactured, but
are difficult to realise in a low cost way, and they will in any case, almost certainly be controlled
digitally. This section discusses the effect of tolerance and phase shifter quantisation with respect to
sidelobes, gain and beam location.

3.3.1 Quantisation errors

In this section, the effect of phase quantisation errors is considered. It is assumed that the required
amplitude excitation can be realised without errors. This will only not be true if it is necessary to
vary the gain of distributed amplifiers for different scan angles. Phase quantisation is clearly
illustrated in Figure 0.12 below, in which, for simplicity, it is assumed that one bit phase shifters are
available. In the figure, the ideal linear phase taper required for a particular scan angle is
approximated by a step function, where the step size is defined by the number of phase states
available in each phase shifter.

Phase errors affect the gain, the sidelobe performance and the pointing accuracy of an array. [Ref 1]
gives a good treatment of the effects.
                                  8

                                  7

                                  6
                                                   Realisable phase
                                  5                taper

  Phase shift                     4
  (radians / 2
                                  3

                                                                                 Ideal
                                  2                                              phase
                                                                                 taper
                                  1

                                  0

                                         1           Element number                        n




                                  Figure 0.12: Phase errors due to quantisation
The analysis is based on determining an RMS phase error, which can be derived from the peak error
                          
as follows:  rms        n
                                  , where n is the number of bits in each phase shifter.
                      2       3

3.3.1.1 Gain degradation

Loss in gain is simply related to RMS phase error as follows:

           rms
             2

G  1           , where G is the gain loss
            2

This function is plotted in Figure 0.13 below. For n>4, the degradation is negligible.
                                           Gain loss vs phase shifter bits

                     -5.00
                     -4.50
                     -4.00
                     -3.50
    Gain loss (dB)




                     -3.00
                     -2.50
                     -2.00
                     -1.50
                     -1.00
                     -0.50
                     0.00
                             1       2           3           4           5          6         7   8
                                                      Phase shifter bits




                                 Figure 0.13: Gain loss as a function of phase shifter bits



3.3.1.2 RMS sidelobe degradation

Sidelobe degradation takes two forms, degradation in the mean sidelobe level relative to the peak of
the beam and an increase in the peak sidelobe level. An estimate of the level of RMS sidelobes
(relative to the peak of the beam) resulting from phase quantisation effects is given by:

                 rms
                   2

Rrms       
             N (1   rms )
                      2




where N is the total number of elements. This function is illustrated in Figure 0.14 for phase shifters
with 1-8 bits and for arrays with between 10 and 60 elements.
                                                       RMS sidelobe level vs number of phase shifter bits

                             -70
                                                                                                                     No of
                                                                                                                     elements

                             -60



                             -50
   RMS sidelobe level (dB)




                                                                                                                            10
                             -40                                                                                            20
                                                                                                                            30
                                                                                                                            40
                             -30                                                                                            50
                                                                                                                            60


                             -20



                             -10



                              0
                                   1         2          3              4                   5         6      7    8
                                                                      No of phase shifter bits




                                       Figure 0.14: Degradation in RMS sidelobe levels with phase shifter bits
3.3.1.3 Peak sidelobe degradation

The degradation of peak sidelobe level is related to both the maximum error and the rate of change of
phase error over the aperture, which is clearly scan angle dependant. Therefore, as expected, the
maximum increase in peak sidelobe level resulting from quantisation effects will occur at maximum
scan, where the deviation from an ideal uniform phase taper is greatest. It should be noted that this
effect is entirely different to and independent of the effects due simply to array geometry which are
described in Section 0.

                              1  m       s
V                                     2 n sin  , where n is the number of phase shifter bits,  is the scan angle, m is
                             2  m        
in the range (-N/2  N/2, N is the number of elements) and s is the inter-element spacing. The
aperture distribution can be divided into two separate superimposed distributions, one of which is the
"ideal" distribution with no phase errors and the second of which is described as an "error"
distribution. Using this definition, it is possible to calculate the peak sidelobe level to be:

             1 v ,m 1                   
Rpk  20 log
             2m v ,m 1 2n Vm sinVmm , where Rpk is the peak sidelobe level in dB. Solving
                                           
this gives a very simple expression for Rpk,,

             1
Rpk  20 log n  . This function is shown in Figure 0.15.
            2 
                               -50

                               -45

                               -40
   Peak sidelob e level (dB)




                               -35

                               -30

                               -25

                               -20

                               -15

                               -10

                                -5

                                 0
                                     1           2           3           4             5        6          7       8
                                                                        Phase shifter bits




                                            Figure 0.15: Peak sidelobe level as a function of phase shifter bits



3.3.1.4 Pointing accuracy

Pointing errors are created by error lobes appearing in and distorting the main beam region. Pointing
errors vary with scan angle, and are maximum when the peak phase errors are maximum. At these
points the beam shift, , in beamwidths, is given by

                                 
                                       . This function is shown in Figure 0.16.
                               4  2n

3.3.1.5 Summary

A summary of the results for arrays with 100 and 300 elements is shown in Table 0.2 below.
                                  0.40

                                  0.35
    Pointing error (beamwidths)




                                  0.30

                                  0.25

                                  0.20

                                  0.15

                                  0.10

                                  0.05

                                  0.00
                                         1           2             3       4             5         6            7          8
                                                                          Phase shifter bits




                                             Figure 0.16: Maximum pointing error as a function of phase shifter bits




                                                Table 0.2: Effect of quantisation errors on antenna performance

Phase                                        RMS           Drop in     RMS Sidelobe level (dB)         Peak         Beam squint
shifter                                      phase error   gain        100              300            Sidelobe     (beamwidths)
bits                                         (º)           (dB)        elements         elements       level (dB)
1                                            0.91          -4.60       -13.25           -18.03         -6.02        0.39
2                                            0.45          -0.94       -25.78           -30.55         -12.04       0.20
3                                            0.23          -0.23       -32.57           -37.34         -18.06       0.10
4                                            0.11          -0.06       -38.77           -43.54         -24.08       0.05
5                                            0.06          -0.01       -44.83           -49.60         -30.10       0.02
6                                            0.03          0.00        -50.86           -55.63         -36.12       0.01
7                                            0.01          0.00        -56.88           -61.66         -42.14       0.01
8                                            0.01          0.00        -62.91           -67.68         -48.16       0.00



This table suggests that phase shifters with 5 bit (ie. 36/2 5 = 11.25°) or 6 bit (360/26 = 5.625°)
accuracy will be required for NGSO antennas, irrespective of the antenna size and configuration
since for those values, gain and sidelobe levels are hardly degraded. Using a fewer number of bits,
the peak sidelobe levels increase rapidly and are likely to exceed sidelobe templates which will be
imposed on NGSO ground terminal systems.

3.3.2
          Random errors

Random errors are caused by imperfect manufacturing and/or by non-ideal components such as
amplifiers and phase shifters. By determining RMS levels for amplitude and phase errors, it is
possible to calculate an overall error power level and to relate this to the design sidelobe level of an
array and the realised sidelobe level.

Using a similar approach to that discussed in section 0, the aperture distribution can be divided into
an "ideal" distribution (ie. one that gives the "design" sidelobe level) and an "error" distribution.

The power contained in the error distribution, 2, is calculated as follows:

    2
 
  2
       , where M is the number of elements and  is the aperture efficiency of the distribution
    M
[Ref 2]. The total mean square error  2 is given by  2  2   2 , where 2                                          is the mean

square amplitude error expressed as a ratio and  2                        is the mean square phase error in radians
squared.

The mean amplitude error is defined as


       A                                                                     P                                 
      M                                                                         M

             i ( act )    Ai ( ideal )                                                i ( act )    Pi ( ideal )
                                              and the mean phase error as  
      i 1                                                                      i 1

                    M                                                                         M

where Ai(act) is the actual amplitude of the ith element
      Ai(ideal) is the ideal amplitude of the ith element
      Pi(act) is the actual phase of the ith element
      Pi(ideal) is the ideal phase of the ith element

This relationship is illustrated in Figure 0.17 below.
                                     Sidelobe level as a function of design and error levels




                            -45
                                                                                                                           -45--40
                             -40
                                                                                                                           -40--35
                             -35
      Achieved level (dB)




                                                                                                                           -35--30
                             -30
                                                                                                                           -30--25
                             -25
                                                                                                                           -25--20
                             -20                                                                           -50
                                                                                                                           -20--15
                                                                                                        -42
                             -15                                                                                           -15--10
                                                                                                  -34
                             -10                                                                                           -10--5
                                                                                            -26         Error level (dB)
                               -5                                                                                          -5-0
                               0                                                      -18
                                      -46

                                            -40

                                                  -34




                                                                                -10
                                                        -28

                                                              -22

                                                                    -16

                                                                          -10




                                    Design level (dB)



                            Figure 0.17: Relationship between theoretical, design and error sidelobe levels


In the graph, "achieved level" is the worst case sidelobe level (relative to the beam peak) realised in
the presence of errors. The "design level" is the sidelobe level (relative to the beam peak) of the ideal
distribution with no errors and the "error level" is the power (relative to the ideal distribution)
contained in the error distribution.

As an example, the levels required to realise a -25 dB sidelobe level with an array of 10 elements is
shown for various design sidelobe levels (SLL) in Figure 0.18 below. It is particularly interesting to
note that, for large arrays, the tolerances required are, in general, much less severe than for smaller
arrays.
                    7.00

                                                                                                                   Design
                                                                                                                   SLL
                    6.00

                                                                                                                        -25.50
                                                                                                                        -26.50
                    5.00                                                                                                -27.50
                                                                                                                        -28.50
                                                                                                                        -29.50
Phase error (deg)




                    4.00                                                                                                -30.50
                                                                                                                        -31.50
                                                                                                                        -32.50
                    3.00                                                                                                -33.50
                                                                                                                        -34.50
                                                                                                                        -35.50

                    2.00




                    1.00




                    0.00
                        0.00   0.10   0.20    0.30    0.40         0.50         0.60   0.70   0.80   0.90   1.00
                                                         Amplitude error (dB)




                               Figure 0.18: Acceptable errors for a -25 dB linear array of 10 elements
4.      Antenna technologies
In this section, a comparison of the production technologies associated with the passive array antenna
components and the competing reflector technology is given.

4.1     Arrays

4.1.1 Arrays for NGSO applications

Table 0.1 below summarises the array requirements for an NGSO application with an example scan
angle of 70°. All three array options can be considered.


                       Table 0.1: Array requirements for NGSO scenarios

                                          Single plane       Two plane          Fixed beam,
                                          electronic scan    electronic scan    mechanically
                                          antenna                               scanned
  Electronic scan       Azimuth           None               ±70
  requirement (º)       Elevation         ±70                ±70
  Array size            30 dBi            152 x 152          260 x 260          89 x 89
  required, 30 GHz
  (mm x mm)             40 dBi            482 x 482          824 x 824          280 x 280
  No of elements        30 dBi            451                2559               152
  required              40 dBi            4513               25594              795
  No of phase           30 dBi            21                 2559
  controls required     40 dBi            67                 25594



Using single plane scanning arrays, two antennas would be required to ensure that hand-over was
accomplished without loss of signal unless the hand-over was to another satellite in the same plane,
since insufficient independent control of the two beams is available. A single phased array could be
used, generating one rapid switching beam that could accomplish hand-over without a dual beam
system. Finally, fixed beam arrays can be used as a direct replacement for reflectors. In this
application, where the antennas are nominally pointing towards the zenith, this could offer substantial
benefits in terms of profile and hence wind loading.

The total cost of a phased array solution is dominated by the number of phase controls needed. If
both transmit and receive frequencies are close together and relatively low, a single Transmit /
Receive (T/R) module can be built; however, in most of the systems under consideration here, the
frequencies are well separated and relatively high (above 18 GHz). This means that, in general,
separate transmit and receive modules will be needed. In addition, where the transmit and receive
frequencies are well separated and scanning to large angles is required, it will be necessary to use
separate transmit and receive antennas.
4.1.2 Single plane scanning array

4.1.2.1 Basic design concepts

A single plane scanning array solution achieves a two plane scanning regime using electronic
scanning in one plane and mechanical scanning, if required, in the other. This is illustrated in Figure
0.1 below.


                                                                  Single plane
                                                                  electronic
                                                                  scan




                                              Single plane mechanical scan




                                Figure 0.1: Hybrid antenna solution
This approach allows a substantial simplification to the mechanical gimballing required to achieve
full two plane scanning with only a minimal increase in antenna complexity; each row of the array
must be phase scanned instead of every element in a full phased array.

Figure 0.2 shows a typical array configuration for a single beam, two frequency application.
Unfortunately, the solution is not appropriate for tracking two satellites unless they are in the same
plane. For some satellite constellations this may be possible, in which case there would be a very
short switching time of typically 1 s. If fully independent simultaneous tracking of two satellites in
one plane was required, a system such as that illustrated in Figure 0.3 could be used. The figures
show separate beamformers for TX and RX functions.
                                                   Diplexer
       LNA                                                                                        HPA

                  RX phase shifter                                            TX phase shifter

                                                                       TX divider
                  RX combiner



                                              Azimuth rotating joint




                                RX                                                    TX




                                     Figure 0.2: Hybrid antenna solution




                                                  Diplexer
     LNA                                                                                          HPA

                 RX phase shifter                                              TX phase shifter
                                                                       TX divider 1
                                                              TX divider 2
                RX combiner 1
           RX combiner 2

                                             Azimuth rotating joint



                   Beam 1       Beam 2                                     Beam 1      Beam 2
                             RX                                                       TX




                                        Figure 0.3: Two beam system

4.1.2.2 Production issues

An active single plane scanning array has amplification at the input to each array column and so the
columns themselves need to have low loss.
It is considered that the only viable fabrication technique for high volume applications is plastic
injection moulding. ERA has collaborated with a specialist moulding company to develop the
process and, jointly, the two companies have developed considerable expertise in this manufacturing
technique. ERA has designed a number of array antennas suitable for this fabrication technology,
including a recent development of a flat plate DBS antenna operating at Ku-Band.

All the antenna components are manufactured from injection moulded, metallised ABS. The antenna
is manufactured as a set of plastic components which are individually moulded and then metallised.
The selection of the type of plastic is critical to the viability of the process, and significant effort has
gone into determining the most appropriate material. The ABS used is designed to be very
temperature stable (exceeding the performance of aluminium) and also highly resistant to Ultra-
Violet radiation; it is therefore ideally suited to an outdoor environment.

Copper metallisation is carried out using an electro-less (dipping) process which is both very rapid
and highly robust. Both temperature and extended life cycle tests have been carried out on samples
of metallised plastics with no degradation in bond strength. The components are doweled together
and bonded using a conducting adhesive. Once more, extensive environmental testing has been
carried out on the adhesive, which has been shown to be stronger than the base plastic itself.

For this approach to be commercially viable, a production volume generally in excess of 10,000 is
required so that the tooling costs involved can be amortised. Typically, prototype tooling suitable for
low volume production runs can be designed and manufactured for about £100,000. This tooling is
adequate for many tens of thousands of units. Unit costs, again highly dependant on volume and
component size, would typically be about $30-60 for an antenna of the lower gain requirement to
satisfy the current specification for a production run of 10,000 units. For higher volumes this can be
substantially reduced to about $100 per square metre. For larger gain requirements, the technique
becomes more difficult and expensive to implement and it is unlikely to represent a viable solution
for arrays above about 450mm square at the current state of development. In the further term larger
antennas may be viable.

Beamforming could be carried out using power dividers fabricated using low profile dielectric
substrates.

4.1.3 Two plane scanning array

The use of a fully phased array solutions may have benefits for the NGSO application, where the only
alternative is to use two fixed beam arrays with independent mechanical scanning.

In this scenario, a static, low profile array with a zenith directed mechanical boresight could be
designed to cover a wide conical sector using phase scanning and, with suitable design of the
beamforming networks, two independent beams could be incorporated, if required. In practice, it is
unlikely that this would be necessary, since the beam location of the phase steered beam could be
moved anywhere within the sector very rapidly (< 1s), allowing near seamless hand-over using a
single beam.
For a two plane phased array, an active architecture is mandatory in order to maintain low losses and
a good terminal G/T ratio. Based on this premise, both waveguide and printed design solutions can
and have been adopted. Two Ka-Band phased arrays have been designed for the US MILSTAR1
programme which illustrate the two technologies. These are described below.

4.1.3.1 Printed radiating elements

For the printed approach to be successful, it is necessary to consider the choice of materials very
carefully. The approach allows integration of radiating elements and active components; however,
the two components have different requirements. The radiating element must use low dielectric
constant materials. If this is not done, surface waves can be excited within the radiating structure,
resulting in poor radiation characteristics and low efficiency. The active circuitry, however,
generally makes use of high dielectric constant substrates to reduce size. These conflicting
requirements are generally overcome by using a two layer structure, as shown in Figure 0.4 below.
This approach has been adopted by Texas Instruments [Ref 1].



                                                              Element on suspended,
                                                              low dielectric constant
               Cavity                                         substrate


                                                                                Ground plane
                                                                                with excitation
                                                                                slot
                                                 Active circuitry on
                                                 high dielectric
                                                 constant substrate




                        Figure 0.4: Printed element solution for mm-Wave arrays
This solution is really only applicable to an active array where individual LNAs or HPAs are placed
very close to the elements, since losses in the high dielectric constant substrate are very high at mm-
Wave frequencies; where high power applications are being addressed, the power handling of the
substrate must be given careful consideration. A further factor which must be included in any
assessment of this technique is that a single printed patch element pattern has a relatively narrow
(±40°) 3 dB beamwidth. If the element pattern is narrow, an array will have a significant scan loss at
wide angles of scan because the element does not radiate strongly in the scan direction. For example,
an element with a 3 dB beamwidth of ±35° will have a scan loss of 3 dB at a scan angle of 35°.




1
    The US MILSTAR satellite constellation is a network of military satellites with an EHF (20/44 GHz) payload.
4.1.3.2 Waveguide technology

The second approach which has been adopted is to use a waveguide radiating element. This solution
has the advantage that it is comparatively low loss and can handle high power levels; however, the
structures tend to be much larger and the interface with active electronics is much more difficult to
achieve. In order to get wide angle scanning without grating lobe radiation in two planes, it is
necessary to load the waveguides to reduce the spacing below/2, since unloaded waveguides
become cut off when their dimensions are less than this value. Figure 0.5 below shows one
implementation of this concept which has been used by Boeing Defence [Ref 2].


                    Probe excitation of
                    waveguides at input                      Active components
                    and output                               on alumina
                                                             substrate              Input power
                                                                                    divider
      Wide angle
      impedance
      matching
      dielectric sheet                           Dielectrically
                                                 loaded circular
                                                 waveguide




                                      Arraying of circular waveguides using
                                                triangular lattice



                               Figure 0.5: Waveguide phased array module



This approach has been used very successfully to create a very wide angle scanning phased array for
use on an airborne platform with the US MILSTAR satellite system.

4.2       Parabolic Reflector Antennas
As a comparison, this section looks briefly at reflector antenna technology.

4.2.1 Parabolic reflectors for NGSO applications

At the gain levels proposed, no single parabolic reflector and mechanical positioner is able to provide
the very wide angle coverage and seamless hand-over needed for the NGSO scenario and hence two
separate antennas would be required to fulfil this requirement. For this application, the tracking of
two moving satellites simultaneously would be accomplished by placing each reflector on a separate
positioner. On hand-over, one reflector would act as the principle communications path. The other
would move mechanically back to the location where the next satellite would appear and begin to
track as soon as the satellite became available. Hand-over between the dishes would take place and,
once this had been accomplished, the first reflector would return to lock on to the next satellite as it
appeared. Thus, seamless communications could be achieved using a simple mechanical system. In
the systems considered, the locations of the satellites are known very accurately, and therefore the
possibility to track without a beacon using a programme tracking technique is available. The
principle is illustrated in Figure 0.6 below.




                      Figure 0.6: Use of two reflectors with a NGSO system

4.2.2
        Basic design

A mechanically steered parabolic reflector represents by far the most straightforward antenna design
solution and, if its characteristics can be matched to all the key specification requirements, is likely to
be the lowest cost and technical risk. It is therefore the antenna type with which a phased array must
compete in the marketplace. A front fed parabolic reflector will have an aperture efficiency of some
60%; an offset design can achieve higher efficiency, typically approaching 70%, and better sidelobe
performance and is currently preferred for most applications. There is, of course, no antenna scan
loss to be considered in this case, though pointing losses may need to be included. Based on typical
gain requirements, the anticipated sizes of the projected apertures for the two solutions are given in
Table 0.2 below.


                            Table 0.2: Parabolic reflector aperture sizes

      Gain required at             Front fed reflector                      Offset reflector
        30 GHz (dBi)              Size         Approximate             Size           Approximate
                                                beamwidth                              beamwidth
              30              140 mm dia            7.00°          120 mm dia              8.00°

              40              440 mm dia            2.20°          380 mm dia              2.60°

              50              1390 mm dia           0.70°          1200 mm dia             0.80°

              60              4400 mm dia           0.22°          3800 mm dia             0.26°



4.2.3 Production technology

The cost of reflector antennas can be very low for small reflector sizes (ie up to 1m) and has been
driven down by the satellite TVRO and cellular back haul markets. Production techniques based on
the use of pressed aluminium reflectors offer very low costs in high volumes. For larger reflector
sizes, different manufacturing techniques have to be used. Up to about 2.4m, spun or moulded
reflectors can be manufactured at a reasonable cost. Above about 2.4m, reflectors must be assembled
from a number of panels; at Ka-Band frequencies, the alignment tolerance becomes critical and hence
costs are relatively high. In addition to the cost of the reflectors themselves, support structures must
become substantially more robust as the size increases, since wind loading forces increase according
to the square of the reflector area.
5.      Solid State Components
Ka-band phased array antennas will require one or more of the following solid state devices:

i)     High Power Amplifier (HPA)
ii)    Low Noise Amplifier (LNA)
iii)   Phase Shifter (PS)

Amplifiers are readily available at Ku-band, in particular, LNAs, as a result of the wide application of
satellite TVRO systems. Phase shifters are only found in the military arena. The use of these
components at Ka-Band frequencies has been investigated by various laboratories around the world
(primarily in the USA) and manufactured in small batches (typically hundreds). The requirement for
state of the art electrical performance, small size, low intrinsic cost and low assembly cost demands
that these components are produced using the most advanced MMIC processes available. This
section concentrates on Ka-band technology.

5.1     Device Technologies
Only three terminal devices are considered here, loosely referred to as transistors. Two terminal
devices are generally only used in an application when transistors are simply not available at elevated
frequencies. Historically, as technology has advanced, devices such as Gunn Diodes, IMPATTs and
Parametric Amplifiers have been displaced to the higher frequencies where transistor technology
cannot be used. Today, the advantages of high performance, large bandwidth and higher degree of
integration associated with transistor circuits are available up to at least 94 GHz (which is well above
the Ka-Band under consideration here).

A number of transistor technologies are described below, along with their relevance to mm-wave
antenna activities.

MESFETs, HEMTs and PHEMTs are all field effect transistors with horizontal current flow through
an active region which is lithographically defined. Present state of the art (P)HEMTs use gate
lengths of typically 0.25 m but may reduce to 0.1 m when high power is not the main concern.
Gate lengths of 0.1 m are almost essential at W band. This is pushing lithography to the limit.

HBTs are bipolar junction devices with vertical current flow across the boundaries of epitaxially
defined layers. With present technology, epitaxial growth and doping can be better controlled than
lithography, allowing the thin active layers necessary for mm-wave operation to be readily realised.

It is difficult to arrive at a single figure of merit which indicates how good a technology may be for
mm-wave applications. One speed indicator commonly quoted in publications is the unity current
gain frequency (fT). This represents the frequency at which the input current equals the output
current. This does not necessarily equate to unity power gain because the device may still provide
significant voltage gain. Unity power gain frequency (f MAX) is sometimes quoted. Both fT and
fMAX are usually inferred from lower frequency measurements. Values for f T in excess of 300 GHz
have been reported, making direct measurement very difficult.
A second key performance parameter is Power Added Efficiency (PAE). This relates the total RF
output power to the sum of the RF and DC input powers. Clearly, in an ideal device, the total output
power would equal the total input power (PAE = 100%); in practice, values of between 10 and 30%
are typical at Ka-band.

5.1.1 GaAs MESFET

The MESFET is based on a doped GaAs channel. The presence of dopant in the critical conducting
channel restricts the electron transport properties by causing scattering. Modest electron transport
properties preclude standard MESFET technology from being used in Ka-Band. Typical maximum
values for fT are 100 GHz. Available gain (and therefore efficiency) is low at upper Ka-Band.

5.1.2 GaAs HEMT

The HEMT is based on the heterojunction of a GaAs channel and an AlGaAs donor layer. Separation
of the donor material from the channel allows the device to achieve electron transport properties
which are not restricted by scattering. Electron transport properties are determined by the perfect
GaAs lattice in the channel. All layers of the device are lattice matched to the GaAs substrate.

Performance is improved compared to MESFETs with fT in excess of 110 GHz available.
Performance in Ka-Band is modest but usable devices can be fabricated.

5.1.3 GaAs PHEMT

The PHEMT is based on the heterojunction of an InGaAs channel and an AlGaAs donor layer. This
is an extension of the GaAs HEMT concept, modifying the composition of the channel to improve the
electron transport properties. Some of the Gallium atoms are substituted by Indium atoms. The
structure is still a regular lattice with no scattering sites. The resulting layer structure is not lattice
matched to the GaAs substrate. The proportion of In that can be introduced into the channel to
improve performance is limited by the amount of strain that can be elastically absorbed at the
interface of the mismatched layers without lattice faults developing.

Performance is excellent with reported values of f T up to 150 GHz. High gain in Ka-Band and good
electron transport properties enable high PAE. Excellent noise figures are also possible.

After MESFETs, the PHEMT is the most widely produced of the “high performance” technologies
due to the similarity of processes and materials to those used in the well established GaAs MESFET
industry. Low noise PHEMTs are produced in enormous volumes for direct broadcast satellite
receivers. The greater maturity of the GaAs based industry has allowed the PHEMT to effectively
compete with InP HEMTs despite the theoretically superior performance allowed by the enhanced
electron transport properties of InP based devices.

5.1.4 InP HEMT

The InP HEMT is based on the heterojunction of an InGaAs channel and an AlInAs donor layer.
It is desirable to improve on the PHEMT transport properties by increasing the amount of Indium in
the InGaAs channel. This cannot be done on a GaAs substrate due to dislocations occurring as a
result of too great a lattice mismatch. This can be overcome by reverting to a basic HEMT
configuration and employing a substrate with a different lattice constant. Using an InP substrate
allows a significant increase in the Indium channel content and results in superior transport
properties. Performance is excellent with reported values of f T up to 340GHz and fMAX of
600GHz. High gain in Ka-Band and good electron transport properties enable high Power Added
Efficiency (PAE).

InP HEMTs are capable of outstanding noise figures, particularly at higher mm-wave frequencies.
The improved electron transport properties possible with InP based devices theoretically allows better
high frequency performance than GaAs based PHEMTs can provide. In practice the greater maturity
of the GaAs based industry and the amount of ongoing development work has generally allowed
PHEMT devices to keep pace despite some outstanding results for InP devices. As the InP HEMT
industry matures, it can be expected that the full benefits of enhanced material properties relative to
GaAs based products will be realised.

5.1.5 HBT

Performance is excellent with reported values of f T up to 150 GHz. High gain in Ka-Band, high
power density and good electron transport properties enable high Power Added Efficiency (PAE).

HBTs typically demonstrate higher breakdown voltages than (P)HEMTs and therefore greater power
density. By enabling a small device with reduced parasitic losses to produce relatively high power
outputs an efficient design can be realised.

The HBT possesses a number of other advantages that the FET based (P)HEMT technologies do not
have. Low 1/f noise makes the HBT very suitable for oscillators where close to carrier phase noise is
important, for example Doppler radar or communication links (depending on the type of modulation
scheme in use). The HBT also has significantly better control of its DC parameters. Biasing,
precision low frequency and DC circuits are more easily realised than with HEMTs. These
comparisons are entirely analogous to the mature Silicon industry BJTs and FETs.

5.2     Components
In this section, a guide to the current performance levels of published devices is provided. This is
made difficult by the broad spread of power levels, frequencies, gains and other parameters for the
different devices. Each designer has chosen a different set of compromises to demonstrate their
technology or application. The examples given below are not an exhaustive list but are chosen to
demonstrate the level of performance which has been reported at Ka-Band.

5.2.1 High power amplifiers

At the upper end of the power scale TRW have reported a two stage 35 GHz PHEMT MMIC with
1W output, 25% efficiency and 10 dB gain.
General Electric have reported a three stage 35 GHz PHEMT MMIC with 200 mW output, 18%
efficiency and 20 dB gain.

At more modest power levels General Electric have reported a single stage 35 GHz PHEMT MMIC
with 95 mW output, 50% efficiency and 8 dB gain.

Efficiency clearly decreases as power levels are increased. If more power output is required, the use
of physically larger devices with greater parasitic problems is needed or power combining of a
number of smaller unit cells. Both approaches result in loss and therefore degraded efficiency. The
attainable efficiency of a power amplifier is largely determined by the characteristics of the final
stage(s) where the greater DC power consumption dominates the lesser contribution from earlier
stages. Using this assumption, the efficiency of an optimised cascade of devices should approach that
of the output device.

TRW have recently reported two high efficiency Q-Band MMICs. The first achieves 1100 mW,
9.4 dB gain and 32% PAE at 43.5 GHz using a PHEMT process [Ref 6]. The second achieves
850 mW, 11.3 dB gain and 34% PAE at 45.5 GHz using a PHEMT process [Ref 7]. This latter
device is particularly attractive for a phased array antenna, its narrow width of 1.6 mm easing
packaging problems in transmit modules.

Figure 0.1 shows the power versus PAE attainable with devices developed by the industry. These
devices are not necessarily production devices but are mostly laboratory units.

Figure 0.2 indicates the power range limits of some commercial solid state amplifiers over the
microwave range.

                      1200


                      1100


                      1000                          Ref 16


                      900


                      800                             Ref 15
  Output Power (mW)




                      700


                      600


                      500                                              Ref 11


                      400


                      300


                      200         Ref 12

                                           Ref 14                                        Ref 13
                      100                                                                                   Ref 13
                                                                                Ref 13                        Ref 13
                        0
                             10   15           20            25   30             35               40   45            50   55   60
                                                                  Power Added Efficiency (%)




                                       Figure 0.1: O/P power versus PAE for a range of amplifiers
                                                                 Commercial Solid state power amplifiers


                  1000
                                                             Philips silicon bipolar
                                                             (1% duty cycle)


                                           Silicon bipolar




                   100         Motorola silicon bipolar (CW)
                                                                            Matcom inc
  Power (watts)




                                                                                                SSPA inc


                                                                                            GaAs FET (MESFET)


                                                                     Thompson
                    10              Semelab silicon                  MMIC
                                    bipolar (CW)
                                                                             Matcom
                                                                             MIC                           Impatt
                                                                                            Miteq
                                                                                            MMIC

                                                                  H-P / Avantek module

                                                                 MA-com & H-P MMICs             Aydin
                     1
                         0.1                          1                                10                       100   1000
                                                                             Frequency (GHz)




                                        Figure 0.2: Power limits of commercial solid state amplifiers

5.2.2 Low noise amplifiers

InP devices generally offer the lowest noise figures for mm-wave applications. A survey of the best
reported InP devices up to 142 GHz [Ref 1] describes the state-of-the-art. HEMT noise figures of
0.7 dB have been reported for devices operating to 60 GHz and 0.3 dB at 18 GHz. Two MMIC LNAs
are reported covering the 43-46 GHz band. Ref. 2 and Ref. 3 claim noise figures of 2.3 and 2.0 dB
with associated gains of 25 and 22 dB respectively. Hughes Space and Communications Company
have claimed an average noise figure of 1.8 dB and 31 dB of gain from 43.3 to 45.7 GHz [Ref 5].
Lower frequency MMICs have reported minimum noise figures of 1.1 dB with 38 dB of gain over the
frequency range 19-22 GHz [Ref 4].

Hybrid MIC construction can provide improved noise figures (closer to bare device levels) at the
expense of other factors. MMIC construction is lower cost, physically smaller and more repeatable.
Although little work has been published for Ka-band MMICs, a noise figure estimate of 1 dB at
20 GHz and 1.5 dB at 30 GHz is reasonable based on the published 22 and 46 GHz figures.
                      10




                                                                                                       Miteq amplifiers


                                                                               CTT (amplifier)


                                              Thomson
                                              (GaAs FET)
  Noise figure (dB)




                                                                      Thomson amplifiers
                       1

                                                GMMT
                                                (HEMT)




                                                                                        Berkshire Technologies' HEMT
                                                                                        (cooled to 15°K, gain 25 dB)




                      0.1
                            0.1                      1                                   10                               100
                                                             Frequency (GHz)




                                  Figure 0.3: Commercial Solid State LNA Noise Performance Data

5.2.3 Phase Shifters

5.2.3.1 MMIC devices

TRW have published results for a Q-band (44 GHz) 3 bit phase shifter [Ref 9]. This device is a
PHEMT MMIC based on a switched line design. The mid-band loss is typically 7.5 dB and the phase
error better than 7º. Overall chip size is 2.8 x 2.0 mm.

Honeywell have produced a Ka-band 4-bit switched line phase shifter using FET technology
[Ref 10]. A measurement sample of thirty MMICs produce a mean insertion loss of 10.5 dB and
standard deviation for phase accuracy of 19%.

At this time, MMIC phase shifter technology is limited to military spheres of activity and there are no
reported devices at Ka-band which have the required resolution for a very low sidelobe application.

5.2.3.2 Ferrite devices

Ferrite phase shifters are essentially digitally controlled analogue devices which have the
characteristics of very low loss together with high cost. The leading world supplier of Ferrite phase
shifters, EMS Technologies (Atlanta, Georgia), has been contacted and has provided an outline
specification for a 5 bit device which is reproduced in Table 0.1 below.


                                           Table 0.1: Ferrite phase shifter characteristics
                          Parameter                   Specification

                          Bandwidth                   10%

                          Power handling              0.4W CW

                          Insertion loss              0.7 dB

                          VSWR                        1.2:1

                          Switching time              5 ms

                          Size (mm)                   30 x 20 x 4

                          Driver size (mm)            54 x 23 x 5

                          Cost (100 off quantity)     $4,000 each



It is clear from this specification that the device would be completely unsuited to a full phased array
antenna due to its size; it could, however, be used in a single plane scanning array. It is clear that the
cost of ferrite technology is very high; there is little prospect that these costs will be brought down in
the foreseeable future.

5.2.4 Integrated Transceivers

The complexity and yield problems associated with mm-wave MMIC devices have largely kept the
degree of integration to a low level. Amplifiers and phase shifters are typically up to 4 or 5 stages.
TRW have recently published details of a Ka-band transceiver in which they claim to have integrated
the equivalent of seven separate chips and a mm-wave filter on a single MMIC [Ref 8].

The PHEMT MMIC integrates the following functions:

i)     LNA
ii)    Image reject filter
iii)   Receive mixer
iv)    IF amplifier
v)     LO frequency multiplier
vi)    Transmitter driver amplifier.

The claimed performance from 38.0 to 38.5 GHz is as follows:

i)     Conversion gain   28 dB
ii)    NF                6.5 dB
iii)   Output Power      56 mW
iv)    DC consumption    2.5 W
The performance, when compared to specialised MMIC components, is modest and needs to be
supplemented by an external LNA and power amplifier. This circuit represents a significant increase
in the level of integration.

No yield figures are given, so it is not possible to judge whether it is cost effective to use this
increased level of integration.

As yields increase and high integration becomes feasible for commercial production, designers are
expected to move towards solutions such as this TRW design. The Ka-Band earth station designer
should benefit from lower costs with improved repeatability and tracking between units by using such
an approach.

5.3     Solid State Device Summary
All of the major components required to implement the RF functions of a Ka-band ground station
have been investigated by MMIC foundries world-wide. The greatest part of this research has been
carried out using GaAs PHEMT processes. InP HEMT offer advantages for LNAs while HBTs offer
improved close to carrier phase noise capabilities.

The relatively greater maturity of GaAs PHEMT processing and its good all round capability suggests
that it is most likely to be the technology chosen for cost sensitive applications such as Ka-band
satellite user terminals. However, at this time, the costs are at least at the level of a T/R module in X-
band ($100), potential process yield is low and the world-wide capacity to produce high volumes is
lagging behind what would be required to launch a major consumer product at a realistic cost.
6.      Antenna Cost Estimates
In this section, cost estimates are put forward for phased array antennas which are compared to
parabolic reflector systems providing a similar gain and power output. For each scenario, two plane
scanning phased arrays, single plane electronic/single plane mechanical scanning phased arrays and a
two reflector mechanically scanning system are considered. The cost estimates which follow assume
separate transmit and receive arrays will be required, and these are compared with dual band reflector
antennas.

6.1     Basic cost components

6.1.1 MMICs

6.1.1.1 Chip costs

The likely costs of Ka-Band MMIC in large volumes are extremely difficult to assess. During the
late 1980s and early 1990s a numbers of foundries capable of MMIC production have been developed
but not without very high investment costs. Currently, a 50% yield is considered good for low
complexity MMIC circuits. It is considered unlikely that yields will be much higher than 60% for the
type of device required for the Ka-Band terminals in the near future.

In the late 1980s GaAs wafers were being offered at approximately $10 per mm2 ( figures supplied
by Plessey to ERA for Ku-band terminal studies work). In late 1996, the Director of Engineering of
TRW indicated that their 3" wafer line costs approximately $20 per mm2 which could be reduced to
$5 in the future. He also stated that when their 6" wafer fabrication facility was fully optimised the
costs could go as low as $1 per mm2 [Ref 1]. Other foundries have indicated costs of the order of $4
per mm2 [Ref 2]. For the purposes of this evaluation, a value of $10 mm2 has been used as a
benchmark for cost estimating. Although not detailed by the suppliers of these figures it is assumed
that the cost will need to be inflated to allow for a yield of 60% which would increase the cost to
$14 mm2.

It is worthy of note that Alpha Industries have indicated costs for a 38 GHz voltage controlled
oscillator at $300 in small quantities going to $15 each for quantities exceeding 5000 [Ref 3].

The parameters given in Table 0.1 have been used to assess the MMIC costs for ground terminal
antennas based on the dimensions of prototype units detailed in the public domain and assuming
separate phase shifters for up-link and down-link. The cost of a diplexer at each module is not
included here since it is anticipated that in most cases the antennas will be separated.
                     Table 0.1: MMIC dimensions and costs at $14 per mm2

                     Component        Length     Width     Area      cost     Comments
                                      (mm)       (mm)      (mm2)     ($)
     Transmit        HPA (1W)         3.0        1.5       4.5       63.0     4 stage, ref 4
     components      Driver amp       1.5        1.5       2.3       31.5     2 stages, half HPA
                     Phase shifter    2.8        2.0       5.6       78.4     ref 9, section 0
     Receive         LNA              1.6        1.2       1.9       26.9     ref 4 & ref 5
     components      Phase shifter    2.8        2.0       5.6       78.4     ref 9, section 0



6.1.1.2 Volume considerations

It can be seen that up to about 18 mm2 of MMIC will be required for each set of devices for a
terminal. It is instructive to analyse the number of wafers required to produce a large volume of earth
stations. 4" diameter wafers are relatively common now although development of 6" wafers is well
advanced. A 4" wafer has a usable area of approximately 80% of its total area. Thus some 360 sets
can be produced from one wafer. To generate a volume of 1 million sets, as might be required for a
domestic subscriber application world-wide, a total of 2775 wafers would be required. With
reasonably advanced techniques approximately one 4" wafer is produced per hour by a foundry
[Ref 2]. Assuming 300 days a year and 24 hour operation it will take approximately five months to
produce enough wafers. According to [Ref 9], Alpha Industries shipped 2,000 chip sets of monolithic
38 GHz equipment in 1994 and were shipping at a rate of 700 per month in 1995. At the same rate
for Ka-Band satellite terminals it would take 119 years to ship one million sets! Given the world-
wide capacity and potential for accelerating production it is still considered unlikely that a million
units could be shipped in less than five years.

In 1995, the global GaAs MMIC market was of the order of $227m [Ref 7 and 8]. Given the above
assumptions which would result in a cost of the order of $240m for one million chip sets then the
capacity required is greater than the total world-wide demand in 1995, a significant demand!

6.1.1.3 RF housing assembly costs

Because of the small size and delicate nature of MMIC devices, it is necessary to use some form of
carrier to support the device and provide a strength member for the interconnections. Furthermore,
the devices can generate significant heat energy that must be conducted away from the small MMIC
device and dissipated. The carriers are consequently some form of precision material such as ceramic
which is firmly bonded to a precision machined metal housing. The MMIC is connected to
conductors on the carrier by means of precision wire bonds. Carrier production, cementing down of
the devices and bonding can be automated but requires dedicated machines using precision alignment
and supervision. It is envisaged that a cast metal housing of the form used for TVRO low-noise block
converters (LNB) would be used, however accurate surface machining for mounting the carriers
would be required for Ka-Band operation.
The material costs for the housing of the MMIC devices is insignificant in quantities but the
processing time and machine preparation and maintenance is significant. Based on ERA experience
of assembling structures containing MMIC chips an estimated cost of $30 per module is anticipated
for large volume machine production. It is worthy of note that this is not significantly different from
figures quoted for the future mass production of automotive radar units.

6.1.1.4 Component and unit testing

It would be vital to conduct on-wafer RF probe testing of the MMIC units to ensure that failed units
are detected accurately. Other testing and alignment would be automated as much as possible. Based
on testing of TVROs and similar devices costs of the order of 1% for the outdoor unit are envisaged.
Thus typical testing processes would cost of the order of $9 per unit in volume numbers (> 5000).

6.1.1.5 Total costs

Collating the results above gives the costs in Table 0.2 for transmit and receive module costs
excluding any diplexer or other filtering.


                           Table 0.2: Transmit and Receive module costs

                     Costs in $             Transmit               Receive
                     MMIC                                   174                    105
                     Housing                                 30                     30
                     Testing                                  9                      9
                     TOTAL                                  213                    144



The current most optimistic estimates for X-band T/R modules in volume is about $100. Given the
increased frequency of operation, the above figures seem to be realistic. For the rest of this report, a
value of $100 per unit is assumed on the basis that continued process development will reduce these
costs. However, it is important to bear in mind that $100 is optimistic at the present time.

6.1.1.6 Cost reduction potential

The technology to make transmit and receive modules exists but it is at present close to the leading
edge of the state of the art and therefore expensive. The costs outlined in Section 0 mean that a
phased array is unlikely to be commercially viable at present but it may be useful to consider an
analogy with silicon processing. Moore’s law, first suggested by Gordon Moore, the founder of Intel
in 1965, states that the density of transistors on a silicon wafer doubles every 18 months. Originally
only a suggestion, it has stood the test of time until today and it is now anticipated (by Intel at least)
that it will remain valid to 2017 when atomic limits are reached. Doubling the density of the
transistors means that the power of a microprocessor doubles or the cost of the same power halves. It
is the latter situation which is of interest here. If Moore’s law holds for MMIC technology, then a 10-
fold reduction in cost would take 5¾ years while a 100 fold reduction would take 11½ years. This
sort of cost reduction is what is required to move phased arrays from interesting technical devices to
mainstream commercial reality.
6.1.2 Array face production costs

For the purposes of this assessment, we assume that the passive array face can be manufactured, for
example using plastic injection moulding techniques, for $100/m2.

6.1.3 Reflector production costs

An outline cost model for reflector antennas has been developed for this application, for a range of
antenna sizes, as shown in Figure 0.1 below.



              4500
              4000
              3500
              3000
   Cost ($)




              2500
              2000
              1500
              1000
               500
                 0
                     0    0.5            1          1.5           2            2.5           3
                                             Antenna size (m)



                                Figure 0.1: Reflector antenna cost model

6.2           Terminal antenna costs
The costs in this section are for the basic antenna and do not include costs for up-conversion, down-
conversion or other costs associated with the electronics of the antenna. In all the scenarios above,
two antennas are required. It is assumed that, for arrays, a transmit and a receive antenna are required
while for a reflector two dual band antennas are required.

Based on the cost structure identified above, the production cost for each of the terminal
requirements discussed in Section 6 is given in below.
                                 Table 0.3: Costing assumptions

        Item                 Standard           Exceptions /                  Comments
                            assumption           deviations
Module cost               $100                $50 at Ku band        Optimistic value
Array face cost           $100/m2                                   Assume 0.3 m  0.3 m except
                                                                    for significantly larger systems
Control circuitry         $200 for arrays     $160 for reflectors
Reflector cost            $58 up to 0.65m     $100 above this       maximum of 1 metre
Mount cost                $58                                       one for each antenna, reflector
                                                                    only
Mechanical positioner     $150                                      one for each axis


                        Table 0.4: Costing for example Ka-band system

                                                         TX             RX
                     Gain                                      36            33
                     Along track scan ()                   75           75
                    Phased array
                     Side (m)                              0.3 m         0.3 m
                    Two plane phased array
                     No of elements                        2424           1301
                     No of modules                         2424           1301
                     Module cost                       $242,400      $130,100
                     Array face cost                           $9            $9
                     Control circuitry                     $200           $200
                     TOTAL                             $242,609      $130,309
                    Single plane phased array
                     No of elements                        2144           1151
                     No of modules                             47            34
                     Module cost                         $4,700         $3,400
                     Array face cost                           $9            $9
                     Control circuitry cost                $200           $200
                     Mechanical positioner cost            $150           $150
                     TOTAL                               $5,059         $3,759


                    Reflector                         Antenna 1     Antenna 2
                     Reflector cost                           $58          $58
                     Mount cost                               $58          $58
                     Control circuitry                     $160           $160
                     Mechanical positioner                 $300           $300
                     TOTAL                                 $576           $576
  Table 0.5: Costing for example Ku-band system

                                TX           RX
 Gain                                 31          32
 Along track scan ()             70          70
Phased array
 Side (m)                          0.42         0.44
Two plane phased array
 No of elements                      817        552
 No of modules                       817        552
 Module cost                   $40,850      $27,600
 Array face cost                     $18        $19
 Control circuitry                $200         $200
 TOTAL                         $41,068      $28,923
Single plane phased array
 No of elements                      748        506
 No of modules                        28          23
 Module cost                     $1,400       $1,150
 Array face cost                     $18        $19
 Control circuitry cost           $200         $200
 Mechanical positioner cost       $150         $150
 TOTAL                           $1,768       $1,518


 Reflector                    Antenna 1    Antenna 2
 Reflector cost                      $58        $58
 Mount cost                          $58        $58
 Control circuitry                $160         $160
 Mechanical positioner            $300         $300
 TOTAL                            $576         $576
7.      Summary and Conclusions
Any use of phased arrays for Ka-band or Ku-band NGSO applications will be based on active arrays
with distributed amplifiers and MMIC phase shifters. Hence, their cost-effectiveness depends almost
entirely on the number of Transmit and Receive modules required, since the cost of these components
completely dominates the overall cost of a phased array. For some of the systems investigated, single
plane scanning may suffice and for those, the cost of modules must be reduced by five to ten times.
For those situations where two plane scanning is required, the commercial viability of phased arrays
will only be realised when at least a 100-fold reduction in costs is achieved. A reflector antenna,
even where two are required for hand-over, is a much lower cost option at present and is likely to be
the antenna of choice, at least at the time of market launch.

Using available knowledge of the technologies and anticipated costs of implementing such Ka-Band
terminals the following summary conclusions can be drawn.

1. The technology for building viable Ka-Band terminals exists today but will require significant
   engineering effort to be expended in creating a really low cost approach.

2. Phased array systems, in terms of single plane and fully steerable arrays, are very unlikely to
   become economically attractive in the near term. This fact is primarily due to the cost of the
   transmit or receive modules.

3. After examining a wide range of technologies for the vital antenna component of the terminal it is
   evident that simple reflector based antennas still provide the lowest cost approach.

4. The impact of adding mechanical steering to such an antenna is significant even when quantities
   are high. This impact may be reduced by innovative development of integral motor/encoders and
   high volume low-cost tracking processors (intelligent step-track or programme track) but it will
   still remain significant. However, in comparison with phased arrays, the cost remains very low.

5. Production of large quantities of MMIC devices may constrain the production roll-out of the
   terminals. There are significant potential demands for high volumes of MMIC in the same
   timeframe as the requirements for the Ka-Band terminals, including but not limited, to automotive
   radar at 77 GHz, new generation mobile telephones and terminals for the emerging generations of
   mobile satellite systems such as Iridium, Globalstar, ICO etc.

6. The achievement of low costs will only be possible by very significant investment in advanced
   processes. This applies to MMIC production, terminal assembly and test as well as antenna
   production. Thus a strong commitment will be required to derive sufficient cost effectiveness. It
   is estimated that volumes of small terminals must be of the order of 20,000 per order to be worth
   the tooling/production costs and follow-on orders of 10-50 times this number will be required to
   assure an adequate base over which to spread the high up-front costs.
8.    References
8.1   References for Section 3
      1.   Miller C.J.
           Minimizing the effects of phase quantisation errors in an electronically scanned array
           Westinghouse Electrical Corporation, reference unknown

      2.   Carlier P.
           Random excitation errors in array antennas and their influence on the choice of array
           distribution
           Marconi Review, Second Quarter, 1979,pp119-134

8.2   References for Section 4
      1.   Sanzgiri S., et al.
           Active Subarray Module Development for Ka Band Satellite Communication
           SystemProceedings of the 1994 Antennas and Propagation Symposium, AP-S,
           Seattle, WA, USA, Jun 1994, pp860-863

      2.   Fitzsimmons G.W., et al.
           A connector-less module for an EHF Phased-Array Antenna
           Microwave Journal, Vol 37, Jan 1994, pp 114-126

8.3   References for Section 5
      1.   Smith P.M.
           Status of InP HEMT Technology for Microwave Receiver Applications (Invited)
           IEEE MTT-S Digest, pp. 5-8, 1996

      2.   Lo D.C.W, et al.
           A High-Performance Monolithic Q-Band InP Based HEMT Low-Noise Amplifier
           IEEE Microwave and Guided Wave Lett., vol3, no. 9, pp. 299-301, Sept. 1993.

      3.   Isobe R., et al.
           Q- and V-Band MMIC Chip Set Using 0.1 µm Millimeter-Wave Low Noise InP
           HEMTs IEEE MTT-S Digest, pp. 1133-1136, 1995

      4.   Duh K.H.G. Private Communication Presented at IEEE MILCOM Conf., San Diego,
           CA, Nov. 1995

      5.   Tran L., et al. High Performance, High Yield Millimeter-Wave MMIC LNAs Using
           InP HEMTs IEEE MTT-S Digest, pp. 9-12, 1996

      6.   Hwang Y., et al. Fully Matched, High Efficiency Q-Band 1 Watt MMIC Solid State
           Power Amplifier IEEE MTT-S Digest, pp. 149-152, 1996

      7.   J.A. Lester J.A., et al.Highly Efficient Compact Q-Band MMIC Power Amplifier
           Using 2-Mil Substrate and Partially Matched Output IEEE MTT-S Digest, pp. 153-
           155, 1996

      8.   Lin E.W., et al. An Advanced Single Chip Ka-band Transceiver IEEE MTT-S Digest,
           pp. 513-515, 1996
      9.    Wang H., et al. Monolithic Q Band Active Array Module and Antenna Applied
            Microwave, pp. 88-100, Winter 1993

      10.   Riley A.L., et al. A Ka-Band MMIC Array Feed Transmitter for Deep Space
            Applications

      11.   IEEE Microwave and Millimeter-Wave Monolithic Circuits Symposium, pp. 11-14,
            1991

      12.   Huang J.C., et.al. A double-recessed Al0.24GaAs/In0.16GaAs pseudomorphic
            HEMT for Ka- and Q-band power applications IEEE Electron Device Letters, Vol
            14, No 9, Sep 1994, pp 456-458

      13.   Ferguson D.W., et.al.35GHz Pseudomorphic HEMT MMIC Power amplifier IEEE
            MTT-S Digest, 1991, Boston MA, USA, 10-14 Jun 1991, pp 335-338

      14.   Streit D.C., et. al. High-gain W-band pseudomorphic InGaAs power HEMT's IEEE
            Electronic Device Letters, Vol 12, No 4, Apr 1991, pp 149-150

      15.   Lan G.L., et. al. Millimeter-wave pseudomorphic HEMT MMIC phased array
            components for space components SPIE Vol.1475 Monolithic Microwave Integrated
            Circuits for Sensors, Radar, and Communications Systems, Orlando FL, USA, 2-4
            Apr 1991, pp 184-192

      16.   Schellenberg J.M., et. al. A 0.8-watt, Ka-band power amplifier IEEE MTT-S Digest,
            Albuquerque NM, USA, Jun 1992, pp 529-532

      17.   Dow G.S., et. al. Ka-band high efficiency 1 watt power amplifier. IEEE MTT-S
            Digest, Albuquerque, NM, USA, Jun 1992, pp 579-582

      18.   Yarborough R. et.al.Performance comparison of 1 Watt Ka-Band MMIC Amplifiers
            using Pseudomorphic HEMTs and Ion-Implanted MESFETs IEEE MTT-S Digest,
            San Francisco, CA, USA, June 1996, pp 21-24

      19.   Simon K. et.al.K through Ka-Band Driver and Power Amplifiers IEEE MTT-S
            Digest, San Francisco, CA, USA, June 1996, pp 29-32

      20.   Freundorfer A.P. et.al.A Ka-Band GaInP/GaAs HBT Four-stage LNA IEEE MTT-S
            Digest, San Francisco, CA, USA, June 1996, pp 141-144

8.4
References For Section 6
1.    Presentation by Raoul Dixit, the Director of Engineering Automotive Technology
      Automotive Electronics - Challenges & Opportunities
      IEEE meeting, September 1996

2.    Raffaelli L., et. al.
      Technology Comparisons for Millimeter-wave Automotive Radars
      M+RF 96 Conference Digest, Wembley, England, 8-10 October 1996, pp 118-123

3.    Private communications between Alpha Industries & M Philippakis of ERA, 1994.

4.    Byzery G.D. et. al.
      A K-Band Chip Set for Radio Links Applications
      M+RF 96 Conference Digest, Wembley, England, 8-10 October 1996, pp 438 - 443

5.    Nam S. Robertson I.D.
      Design and Performance of a Two-Stage Ka-Band Monolithic pHEMT Amplifier
      M+RF 96 Conference Digest, Wembley, England, 8-10 October 1996, pp 432 - 437

6.    Leckey J.G, et. al.
      Design of a MMIC 35 GHz Buffered Dielectric Resonator Oscillator
      M+RF 96 Conference Digest, Wembley, England, 8-10 October 1996

7.    Turner J.
      The World-Wide Competitiveness of European Made GaAs Components
      M+RF 96 Conference Digest, Wembley, England, 8-10 October 1996, pp 288 - 290

8.    Hess R.
      MMICs for Commercial Applications; The Low-cost High Volume Production
      Techniques
      MMT-S Symposium Digest, San Francisco, CA, USA, June 1996, pp 3- 6

9.    Meinel H.H.
      The Market for Short-Haul Line-of-Sight Millimeterwave Transmission Links
      IEEE MTT-s Digest, June 1996, pp 487 - 489

10.   Hiltek Ltd.
      The European Microwave Report Microwave Engineering Europe Market Research
      Report, 1993

11.   Cohen E.D.
      Trends in the Development of MMICs and Packages for Active Electronically
      Scanned Arrays (AESAs)
      IEEE Symposium on Phased Array Systems & Technology, Boston USA, 15-18
      October 1996, pp 1-4

                              _______________

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:27
posted:4/7/2010
language:English
pages:58