A circularly polarised phased array for land-mobile satcoms.
`
`P.V. Brennan
`
`1 Introduction
`
`Satellites have been used for many years to provide conununications between fixed land
`stations, ships and local telephone networks. However, the matter of land-mobile satcoms has
`only recently begun to be addressed. lnmarsat standard-C is probably the first service of this
`kind and uses terminals with low gain omni-directional antennas to provide low speed data
`communications. Inmarsat are now developing standard-M, a land-mobile service capable of
`carrying digital voice conununications. Also, there is the U.S. MSA T program, which has
`similar objectives. Both these programs rely on the use of high-technology steerable antenna
`systems mounted on the land-mobile. The fundamental antenna specifications [1,2] are, in
`fact, very similar. In both cases an antenna gain of around 10-12 dBi is required, with circular
`polarisation and operating at L-band from 1.537 GHz to 1.644 GHz for Inmarsat and 1.545
`GHz to 1.6605 GHz forM-SAT.
`
`In this paper, a number of cost-saving techniques are described aimed at producing an
`economical phased array design capable of good performance in these applications.
`
`2 The prototype array
`
`The basic layout of the array design arrived at in this work is shown in figure 1. It consists of
`18 half-wavelength spaced elements arranged in the standard triangular lattice with the absence
`of a central element The array comprises 6 identical sub-arrays each containing 3 phase(cid:173)
`controlled elements. The contributions of each sub-array are combined in a central 6-way
`power combiner. A symmetrical phasing technique is used in which the array is arranged in
`pairs of elements about a conunon phase centre and equal but opposite phase shifts are applied
`to the elements of each pair.
`
`An important benefit of the use of such a sub-array structure is that accurate path length
`matching between the centre of the array and each element is easy to achieve. This is in
`contrast to many other approaches in which a single 19-way power combiner is used to
`distribute signals to each of the elements. However, a penalty of the proposed geometry is that
`it is more difficult to include the central element for both mechanical reasons, because of the
`central power combiner, and electrical reasons, since its contribution would need to be
`appropriately weighted in both amplitude and phase. This gives rise to a small amount of
`pattern degradation and also, assuming equal element weightings, a gain penalty of 0.23 dB.
`The use of symmetrical phasing halves the number of bits that are required to control the array.
`This allows a simplification in the control circuitry and also an improvement in the agility of the
`array since fewer phase shift combinations are available. The prototype design uses 2-bit
`
`P.V. Brennan is with the Department of Electronic & Electrical Engineering, University
`College London, Torrington Place, London WCIE 7JE.
`
`<./1
`
`Petitioners' Ex. 1038 - Page 1
`
`

`
`phase shifters so that just 18 bits are required to control the array. Because circularly-polarised
`elements are used and each pair consists of 2 elements orientated at 180° to each other, the
`phase shifts are in fact related by <Pn • = 180°- <Pn·
`
`The circuit diagram of each sub-array is shown in figure 2. The array uses printed cavity(cid:173)
`backed crossed-dipole elements which are fed with the appropriate phase shift via 0°, 90°,
`180° or 270° line lengths. This alleviates the need for any baluns in the design which
`improves its mechanical simplicity. A 2-bit switched-line phase shifter is included in the path
`to each element and the 3 elements of each sub-array are combined in a reactive combiner at the
`centre of the sub-array. The design is implemented in a printed form and is capable of accurate
`and repeatable phase matching. Because phase shifts are achieved by means of line lengths,
`the design is only suitable for narrow-band applications such as those considered here where
`the entire transmit-receive fractional bandwidth is less than 7%.
`
`The proposed beamforming strategy consists of two stages. On start-up, a relatively low
`number of preset beams (perhaps 50) are each formed in turn and the received signal strength is
`monitored. With a data rate of 9600 bits/s, 1 ms per beam should be sufficient, so this process
`takes some 50 ms to complete. Once the approximate signal direction has been established the
`system switches to adaptive mode and then each pair of phase shifters in turn is altered to its
`adjacent value, on either side of the current value, to determine the optimum setting. This
`process takes around 27 ms to complete. but results in a small loss of gain which may be
`reduced by spending a greater proportion of time idling at the expense of reduced array agility.
`
`3 Results obtained with the prototype array
`
`The completed prototype array consists of 6 identical sub-arrays each containing three
`elements. Each sub-array is fonned from a sandwich of two printed circuit boards, one
`containing the printed crossed-dipole elements and the other containing the power splitters,
`phase shifters and feeds to the four individual arms of each crossed-dipole. The complete
`phased array system is shown diagramatically in figure 3.
`
`The variation of gain and return loss of an individual sub-array is plotted in figure 4. This
`measurement was performed with the phase shifters all set to 0°, producing a broadside beam.
`The gain is flat to within 2 dB over the lnmarsat/MSAT band (with some of this variation
`accounted for by the increasing free-space loss with frequency) and the return loss is at least 20
`dB.
`
`The performance of the beamforming board itself has also been measured. The return loss is
`around 20 dB for all phase shift positions and the insertion loss averages around 2. 7 dB over
`the lnmarsat/MSAT band. This insertion loss is somewhat on the high side and is partly due to
`the low cost epoxy glass printed circuit board used in the design. It is estimated that the
`microstrip line losses are around 1.5 dB and that a 1 dB reduction in insertion loss would be
`obtained by using a higher quality substrate.
`
`The array geometry used in the design has 60° rotational symmetry. It is convenient,
`therefore, to predict the radiation pattern as a function of elevation angle at two extreme azimuth
`angles differing by 30°. The results of a very simple analysis of the array, neglecting mutual
`coupling and the element radiation patterns, is shown in figure 5. In both cases there are deep
`
`2/2
`
`Petitioners' Ex. 1038 - Page 2
`
`

`
`nulls at ±30° and. within this range of angles, the plots are virtually identical. Beyond ±30°,
`however, the patterns are quite different with maximum sidelobe levels of -12 dB in the 0°
`azimuth position and -15 dB in the 30° azimuth position. Also, there is a -34 dB null in the
`30° azimuth plot at ±70° from zenith which does not appear in the 0° azimuth plot. Sidelobe
`levels are quite high for two principal reasons. Firstly, the analysis assumes that all elements
`are equally weighted where, in practice, an edge taper could easily be introduced to reduce
`sidelobe levels. Secondly, the absence of a central element slightly reduces the geometrical
`amplitude taper, again increasing sidelobe levels.
`
`Radiation pattern measurements were performed on the array with phase shifters adaptively set
`for beam pointing directions over a range of ±75° from zenith. The pattern obtained with an
`intended beam pointing direction of 0° elevation and 10° azimuth orientation is plotted in figure
`6. The result of figure 7 was obtained with phase shifters set for a beam deflection of 30° in
`elevation. A rotating source antenna was used so that the axial ratio could also be measured.
`Compared with the 0° plot, the main beam level has decreased by 0.83 dB and the sidelobe
`levels have increased slightly, as would be expected. The axial ratio is quite good in the
`intended pointing direction - around 1.5 dB - and, in fact, a good axial ratio was measured in
`the main beam direction over a wide range of angles.
`
`4 Conclusions
`
`A number of techniques have been investigated, aimed at producing a low cost steerable phased
`array for use with land-mobile satcoms. A sub-array geometry has been developed which
`allows convenient and accurate path length matching from each element to the centre of the
`array. Symmetrical phasing has been used in conjunction with 2-bit phase shifters to enable
`the array to be controlled by just 18 bits, thus reducing control system complexity and
`improving array agility. An entirely adaptive control system has been proposed to achieve
`beamforming and tracking without the aid of any costly open-loop technique.
`
`Predicted and experimental results have confirmed the feasibility of these techniques and have
`shown that there is little sacrifice in array performance. A prototype planar array has
`demonstrated the beamforming capability of the system and has shown that good performance
`is achievable over a ±45° range of beam deflections from zenith.
`
`5 References
`
`1. "Inmarsat Standard-M system defmition manual", draft 2.1, Inmarsat, May 1989.
`
`2. Huang, J., "L-Band phased array antennas for mobile satellite communications", 37th IEEE
`Vehicular Technology Conference, pp.ll3-117, June 1987.
`
`;:;3
`
`Petitioners' Ex. 1038 - Page 3
`
`

`
`Elemeot pooitioo
`
`Sigoallevel
`
`Cootrol
`
`Figure 1. Layout of the 18 element symmetrically phased array.
`
`Figure 3. The complete phased array system.
`
`to priotecl X-dipole
`
`100
`
`to other oectioo
`(via 120' pbaoe obit\)
`I
`
`100
`
`to other eectioo
`(via 240' pbaoe obift)
`
`loH
`L.-.1---IH
`
`100
`
`180" cootrol bit
`
`loceramic
`90" cootrol bit
`
`Figure 2. Sub-array circuit diagram.
`
`2/4 ~·
`
`Petitioners' Ex. 1038 - Page 4
`
`

`
`lnmaraat/MSAT Band
`
`Frequency (GHz)
`
`Figure 4. Measured gain and return loss of a sub-array.
`
`s
`~
`c
`"ii a
`
`·10
`
`-20
`
`-30
`
`-40
`-90
`
`r -.- T -~-,-
`
`0° azimuth
`
`30° azimuth
`
`-60
`
`·30
`
`30
`
`60
`
`90
`
`Elevabon angle from zenith ( c )
`
`Figure 5. Predicted radiation patterns of the prototype array.
`
`·!lO
`
`-60
`
`-30
`
`30
`
`60
`
`90
`
`-90
`
`-60
`
`-30
`
`JO
`
`60
`
`90
`
`Elevation angie from zenith ( o )
`
`Elevation angie from zenith ( 0
`
`)
`
`Figure 6. Measured pattern, oo elevation, 10° azimuth cut.
`
`Figure 7. Measured pattern, 30° elevation, 10° azimuth cut.
`
`------~---
`
`2/5
`
`Petitioners' Ex. 1038 - Page 5