GB2154803A - Circular antenna array, and phase comparator for use therewith - Google Patents

Abstract

A circular antenna array (1-4) of directional elements as described in GSA-1507674 feeds a Butler matrix (10). Two phase mode outputs of the matrix are selectively applied to a phase comparator (30) which measures the phase difference between them and shifts the phase of one of the outputs to bring it into phase with the other. It does this by adding (38) an oscillatory phase perturbation at a frequency f<1> to one of the signals and combining it (40) with the other for application to a receiver 42 with a.m. detector 44. The component at frequency f<1> is selected (46) and applied to a phase sensitive detector (48) to determine the sense of the phase error. The phase introduced by the phase shifter (36) is then adjusted to tend to reduce the error. The system can be used for direction finding and simultaneously for communications reception. <IMAGE>

Description

SPECIFICATION Circular antenna array, and phase comparator for use therewith This invention concern circular antenna (aerial) arrays, such as may be used for communications and direction finding purposes. The invention also relates to a phase comparator which may be used with such an antenna array.

The basic principles of operation of circular antenna arrays are well known and reference is made to "The Handbook of Antenna Design" 1983, Vol. 2, published by Peter Peregrinus in the IEE Electromagnetic Wave .Series, and edited by A.W. Rudge, K. Milne, A.D. Oliver and P. Knight for details; see particularly Chapter 1 2 "Circular Arrays" by D.E.N. Davies. Also reference should be made to a paper by J.R.F. Guy and D.E.N. Davies entitled "UHF Circular arrays incroporating open-loop null steering for communications" published in IEE PROC., Vol. 130, Pts.F and H, No. 1, February 1983, pages 67-77, which describes how a circular array can be made to exhibit a steerable directional response by appropriate combination of the constituent phase modes which are separated by the now well-known Butler matrix (see Electronic Design, April 1961 pages 170-173).

The symmetry of circular arrays and the cyclic nature of the phase mode excitations make them particularly suitable for use in direct phase-measuring direction finders. If a single signal is received by an idealised circular array fed by a Butler matrix, the phase difference between the zero-oder and firstorder phase modes will be directly related to the angle of arrival relative to a reference defined by the direction in which the two modes are in phase. Thus, if this reference is 0" and a signal at bearing ao is received, the zero-order and first-order mode outputs would have a phase difference of (E . Conversely therefore measurement of the phase difference gives directly the bearing from which the signal is received.

It is known that for large arrays (with more than four elements) the theoretical sensitivity of such a circular array can be improved by comparing the outputs from higher-order modes. An example of this is given by S.

Rehnmark in a paper entitled "Improved angular discrimination for digital ESM systems" published in the Proceedings of the First Military Microwaves Conference, London, 22nd-24th October 1980, pages 157-162.

However, there are problems in designing a 2-channel or other multi-channel receiver and phase detector system which will enable the inherent measurement accuracy of the circular array to be realised in practice. This largely arises because of the need to equalise the two receiver channels in relation to the phase of the received signals.

The invention in its various aspects is defined in the appended claims, to which reference should now be made.

The invention will be described in more detail, by way of example, with reference to the drawings, in which: Figure 1 is a block circuit diagram of a direction finding system embodying the invention; Figure 2 illustrates combinations of phase vectors at the out-of-phase and in-phase positions; Figure 3 is a block circuit diagram of a microprocess-controlled direction finding system based on that of Fig. 1 but with a number of improvements and modifications; Figures 4 and 5 are flow charts illustrating the operation of the microcomputer 94 of Fig.

3; and Figure 6 illustrates a further modification which can be applied to the direction finding systems to remove interference from an unwanted signal at a different bearing to the wanted signal.

The circular array illustrated in Fig. 1 has four antenna elements 1,2,3 and 4 arranged in a circle. The interelement spacing around the circle should in theory be not greater than half a wavelength at the highest operating frequency envisaged. The array illustrated is designed to operate over the H.F. band in the broad frequency range of 2 to 30 MHz, and the diameter of the circle formed by the four antenna elements is in this case 6.4m.

Each of the antenna elements is constituted by a loop which is of small dimensions compared to the wavelength and in this example may have a diameter of 0.7m. The loops used are however preferably fed so as to produce a cardioid radiation pattern by the method described in British Patent No. 1 507 674, namely an unbalanced amplifier is used at the feed point and a discontinuity is introduced at the point on the loop diametrically opposed to the feed point, the two resultant halves of the loop then being connected by a suitable load impedance. The use of such directional aerials is particularly desirable where a wide operating bandwidth is required.

When N elements are used a maximum of N distinguishable phase modes can be obtained. Thus the network required to feed the array must have N input ports and N output ports, with the m th input port producing N outputs of equal amplitude but with an incremental adjacent phase difference of 2nm/N.

This network is constituted by a Butler matrix 1 0. The design of the Butler matrix is complicated by the requirement for the array to operate over a ten-to-one frequency range.

The matrix employs three 180 > wideband hybrid junctions all supplied by Adams-Russel, ANZAC Division, 80 Cambridge Street, Burlingotn, Massachusetts 01803, U.S.A.

The 180 hybrids may be ANZAC type HH-106 and the 90 hybrid ANZAC type JH-6-4.

In the Butler matrix illustrated, a first 180 hybrid 1 2 receives the outputs of antenna elements 1 and 3 at its A and B inputs and provides their sum at its C output and their difference at its D output. Similarly, hybrid 14 generates the sum and difference of the outputs of antenna elements 2 and 4. The third 180 hybrid 1 6 takes the two C outputs of hybrids 1 2 and 14 and generates their sum at its C output. This then comprises the sum of the four antenna element outputs and is the monopole or zero-order mode output. The D output of hybrid 1 6 provides the second order mode output but this is not used in Fig. 1 and is appropriately terminated.Finally the quadrature phase hybrid 18 introduces a 90 phase shift so as to generate the first order modes, namely the + 1 and - 1 modes, at its two outputs.

In a four element antenna array the secondorder phase mode suffers from severe "ripple" in that it reduces to zero at certain points and is not therefore used.

As previously explained, it is now necessary to compare the zero-order mode output with one of the first order mode outputs to ascertain the bearing of a received signal by determining the phase error. In actual fact the hybrid junctions 12, 1 4, 16, 1 8 in the Butler matrix introduce transfer phase delays which vary with the operating frequency. The transfer phase delays introduced in the + 1 and - 1 modes are similar, and therefore it is preferred to compare these two modes with each other rather than comparing the zeroorder mode with one of them. Comparing the + 1 and - 1 modes however introduces a 1804 ambiguity into the resultant, and comparison with the zero-order mode is still required to resolve this ambiguity.

The phase delays could be removed by the addition of further dummy hybrids arranged to equalise the delays, but this is expensive and causes a loss of half the power output of the matrix 10. However we have appreciated that the major phase error is caused in the quadrature hybrid 1 6 and that its effects can be sufficiently minimised by using an extra length of cable in the zero-order mode path to the phase detector circuit, diagrammatically illustrated at 20. A 3.3m length of cable was found to increase the accuracy sufficiently to enable the zero-order/first-order mode comparison to be sufficiently accurate to resolve the ambiguity generated in the + 1 / - 1 mode comparison.

In addition to the transfer phase delays introduced by the hybrids, they also have phase imbalances, namely deviations from the stated phase differences of their outputs, which can not be so readily removed. Instead, compensation can if necessary be introduced on an empirical basis by measuring the phase imbalances at various frequencies and producing correction graphs or tables for the phase detector. Similar small errors are introduced by mis-positioning of the antenna elements.

The - 1 phase mode is found to be the most consistent in phase and this is therefore compared with the zero-order and + 1 modes in succession, as selected by a switch 22.

As will be appreciated, care has thus been taken in providing two signals to be compared, the relative phasing of which has not been distorted. Conventionally each of these signals would now be applied to a respective receiver channel and the receiver outputs compared in phase. In practice it is not possible to maintain the phase accuracy of two such receiver channels over the wise operating band of the system, at least at reasonable cost.

The phase comparator 30 is therefore of different construction. It has two inputs 32, 34 for receiving the two signals the phases of which are to be compared. The phases are compared by a null system, which involves introducing a controlled phase shift into one of the signals until the signals are in phase, and then seeing how much phase shift had to be introduced in this way. To that end, a controlled phase shift circuit 36 is connected to the input 34 and a phase perturbation circuit 38, the operation of which will be described in more detail below, is connected to the output of phase shifter 36. A summing circuit 40 combines the output of the phase perturbation circuit 38 with the signal at input 32 and applies the resultant to a receiver 42, which may be a RACAL type RA 1 7 receiver for example.An amplitude-modulation envelope detector 44 is connected to (or may form part of) the receiver 42, and the output of this is applied to a band-pass or low-pass filter 46.

The filter output is applied to one input of a phase sensitive detector 48. An oscillatory signal is applied from an input 50 both to the phase perturbation circuit 38 and to the phase sensitive detector 48. The error signal output of the phase sensitive detector is applied as the control input to the phase shifter 36.

The operation of the phase comparator 30 is essentially as follows. The aim is to adjust the phase shifter 36 so that its output is in phase with the signal at input 32. To tell when it is in phase, a + 10 oscillatory phase perturbation at frequency fl is applied to the signal by circuit 38. The two signals are then combined vectorically in summer 40.

Fig. 2 illustrates the combination of two vector signals A and B. At (a) they are seen to be out of phase. If signal B is subject to a + 10 perturbation it moves between the extreme positions shown in dashed lines about the mean position shown by the full line. The resultant vector C Ikewise moves between the dashed line positions and it will be seen that it does not only vary in phase, but also varies in amplitude above and below its mean value at frequency fl.

However, if the signals A and B are in phase then the vectors are as shown at (b) in Fig. 2. Vector B varies to either side of its mean position, and vector C again varies in phase. It will also still vary in amplitude by a small amount, but in this case the length of the vector reduces in the same way either side of the mean position; thus this is a variation based on frequency 2fl and there is no component at frequency fl. Thus if the phase introduced by phase shifter 36 is adjusted until the fl component of the amplitude of the combined signal is zero (or at least a minimum), the in-phase position will have been found and the phase of phase shifter 36 will represent the phase difference of the input signals.

Referring to Fig. 1, the output of summer 40 will, when the signals applied to it are out of phase, be amplitude modulated at the frequency fl of the phase perturbations. The signal is demodulated by the single receiver 42 and detected in detector 44. The fl frequency component is selected by filter 46 and applied to the phase sensitive detector 48.

Detector 48 produces an error signal output the sense of which will depend on the sense of the phase error between the inputs to summer 40 and which will adjust the phase shifter 36 so as to tend to reduce the phase error. Eventually the signals will be brought into phase.

The frequency fl will normally be chosen to be outside the usual communication speech band of say 300Hz to 3kHz. A frequency of 70 or 80Hz is preferred, though a value of 3.5kHz could be used.

The amount of phase perturbation can vary upwards and downards from the value of 10 degrees given as an example. Another convenient value would be 20 degrees. However any value up to 90 degrees can be used with advantage and indeed the system of Fig. 3, described below, uses a phase perturbation equivalent to + 90 .

One feature of the system of Fig. 1 is that it can be used simultaneously for direction finding and for communications reception. The output for the communications receiver can be taken at 45 from the output of detector 44, or alternatively the average of the signal at input 32 and the phase shifted output of shifter 36 can be used, as supplied by a circuit 52. In either case, the effecive radiation pattern for the aerial is such as to have a maximum in the direction of the wanted signal, as the effect of the phase shifter 36 is to steer the beam to the specified bearing at the same time as it does the direction finding. This furthermore increases the sensitivity of the receiver system.

More generally, the four outputs of the Butler matrix could be separately combined, if desired, for application to a separate communications receiver.

While the system of Fig. 1 has been shown as using only 4 individual antenna elements in the array, a greater number of elements can be employed with advantage, for example 8 or 1 6 elements.

Fig. 3 illustrates a microprocessor-based implementation of the circuit of Fig. 1 with one or two further modifications included.

First, referring to Fig. 1, it will be appreciated that the phase perturbation could be introduced into either channel to the summer 40. In Fig. 3, it is preferred to have the phase perturbation circuit 38 in one channel and the phase shifter circuit 36 in the other. In this way some of the phase errors introduced in these circuits will tend to cancel.

Secondly, the phase shifter circuit 36 works by dividing the - 1 mode signal into two equal-amplitude quadrature components by using a 90 hybrid device 60. The in-phase and quadrature-phase signals are applied to respective attenuators or multipliers 62, 64 and their outputs combined in a combiner 66, the output of which forms one input to the combiner 40.

In the other channel, the 0 and + 1 phase mode outputs from the Butler matrix 10 are applied respectively to multipliers 68 and 70.

The multipliers act as gates and receive multiplying inputs through a switch unit 72 either from the 80Hz square wave source 50 or from fixed potentials giving unity and zero multiplying factors respectively.

The action of the multipliers 68 and 70 is as follows: Control Input Output + 5V Unity gain-zero phase shift O Zero - 5V Unity gain-180" phase shift Thus at any given time the control input to one of these multipliers is zero and that multiplier acts as a closed gate. If the other multiplier is required to provide perturbation, its control input is switched by source 50 between the + 5V and - 5V input values to cause a phase perturbation of + 90 about a mean value of 90 . Fixed compensation has to be made in the phase sensitive detector for this mean offset of 90 .

The outputs of gates 68 and 70 are applied to a 90 hybrid 74 where they are combined, and also suffer the same phase delays as are introduced to the - 1 phase mode channel by hybrid 60. For similar reasons a combiner 76 is connected between the hybrid 74 and combiner 40.

The output of combiner 40 is as before applied to receiver 42, and the base-band detected signal is applied to an 80Hz bandpass filter 46. As the subsequent circuitry is now digital, the filter output is applied to a limiter 78 to produce square-wave pulses.

Phase sensitive detector 48 compares these pulses with the output of frequency source 50 to produce a binary output indicating the sense of the relative phases of the inputs to the combiner 40. These pulses are applied to the up/down count input of an up/down counter 80. The count held by the counter is representative of the number of degress of phase shift to be introduced by the phase shifter circuit 36. The value in degrees is applied to two EPROM (erasable programable read only memory) circuits 82,84 which contain sine and cosine functions in the form of a look-up table and provide the values of the multipliers required by multiplier circuits 62,64. The look-up table can also be adjusted to take account of any non-linearities in the circuits forming the feedback loop.The outputs of the EPROM circuits 82,84 are applied through digital to analogue converters 86,88 to the multipliers 62,64.

When the two signals at combiner 40 are very close in phase the detected AM modulation is very small, and circuit 78 will not produce a pulse train but rather a constant level output. Then the error signal from detecotr 48 must remain as it was before and the feed-back loop overshoots until an AM modulation of opposite sense (phase) is detected again. This will then cause a reversal in the direction of count of counter 80 which will again overshoot in the other direction. Thus the counter will hunt or cycle around the null value. The microcomputer 90 is used to take the average of these readings so as to eliminate the effect of the oscillations. Averaging many readings in this way will also reduce errors due to noise. A further important advantage of this system is that it removes an inherent ambiguity in the output of the phase sensitive detector 48.Reference to Fig. 2 will reveal that the fl component of the combined signal from combiner 40 will be zero not only at the in-phase position but also at the antiphase position. However, the system will tend to overshoot from the antiphase position until it can be pulled in by the feedback loop towards the in-phase position.

Means are provided to disable the counter 80 in the event that no signal is detected by the receiver 42. For this purpose the IF output of the receiver is applied to a level detector circuit 92 through an R.M.S. to D.C. converter circuit 94. In the absence of a received signal of sufficient amplitude the IF signal will be insufficient to cause the ciruit 92 to enable the counter 80.

The microcomputer 90 provides general control sequencing of the operation of the system to compare the - 1 phase mode signal with the + 1 and zero modes, and also provides certain corrections to the phase angle hald in counter 80. These corrections comprise calibration corrections, discussed above, held in a record 94, and ripple error corrections held in a table 96. Ripple errors arise because a finite number of antenna elements are used in the array. The corrected bearing can then be applied to a display 98.

Fig. 4 shows a flow chart for the operation of the microcomputer 90 and will be briefly described, though much of the operation will be clear from the preceding description. In Fig. 4 after startup 100 the phase perturbation is switched on 102 and the frequency of the RF signal of which the direction is to be found read in 104. The correction data and phase shifter calibration for this frequency are loaded from disc into record 94 in step 106.

The operator is required to specify the number N of successive readings which are to be taken before an output is obtained in step 108.

In step 110 the operator indicates whether a new zero/first order mode comparison is required to resolve the 180 ambiguity in the - 1 / + 1 phase mode comparison, and if so in step 113 sets switch 72 in Fig. 3 to pass the zero order mode signal from the Butler matrix 10 to the phase perturbation circuit 38. A data capture subroutine is initiated at step 114, this being described in more detail below. After the reading is taken, step 11 6 resets switch 72 to select the + 1 mode output of matrix 10.

In step 11 8 a variable I or cycle count is set to zero and in step 1 20 is incremented by one. The data capture subroutine is again called in at step 1 22 to calculate the bearing from the + 1 / - 1 phase mode comparison.

The necessary corrections from record 94 are applied in step 1 24 and the resultant value divided by two in 1 26. This is necessary because the comparison between the two firstorder phase modes produces an output twice as large as is required and as is obtained by the zero/first-order mode comparison. The latter comparison is used in step 1 28 to resolve the 180 ambiguity, and then the ripple errors stored in table 96 are subtracted. The resultant value is stored in step 1 32 in a store 5(l).

If the required number N of determinations requested in step 108 have not been completed, the cycle beings again at step 120, but if in step 1 34 it is found that the required number of readings have been taken then in step 136 the mean of the N stored values S(l) is taken. To provide an indication of the accuracy of the resultant, the standard deviation can also be calculated. These results are displayed on display 98 in step 1 38.

The data capture subroutine called in in steps 114 and 1 22 utilizes the output of the up/down counter 80 as previously described.

In digital terms, the phase sensitive detector 48 will provide an output for every system clock pulse which will take the same value until the feedback loop overshoots, whereu pon it will reverse in value and overshoot again. The phase shifter 36 will change by one unit of phase for each clock pulse.

The microcomputer detects transitions in the output of the detector 48, and when the feedback loop is hunting around the correct value these transitions will occur fairly regularly. Transitions in one sense are used as high priority interrupt pulses, and the clock pulses occurring between these interrupt pulses indicate the range of values through which the system hunts. These values are averaged to give a mean value. If the range of values exceeds a preset maximum, the reading is regarded as unreliable and is discarded.

Making the adaptive loop hunt in this way improves the noise immunity and also eliminates any hysterisis effects in the loop arising from servo delays.

Fig. 5 illustrates how the system of Fig. 3 can subsequently be used to steer the radiation pattern of the array so that the null of the pattern or the maximum response is set at a selected direction.

After completion of the steps of Fig. 4, the operator selects in step 140 whether to continue direction finding with the same chosen options and if so is returned to step 11 8. If not, the operator can then select in step 142 to continue direction finding with different options and will be returned to step 1 04.

Failing that the operator can select in step 144 "null or beam steering", failing which the program ends 146.

If null or beam steering is selected in step 148 the phase perturbation circuit 38 is switched off by appropriate operation of switch 72. The phase pertubation is of course only required for direction finding. In step 150 the operator enters the bearing of the required null or beam and indicates in step 152 whether setting should be on the basis of the zero/first order mode phase comparison or the + 1 / - 1 mode phase comparison.

Switch 72 is set appropriately in step 154 or 156. In step 158 the operator indicates whether a null or beam is being specified and in step 160 or 162 the appropriate phase shift setting is calculated, taking into account the device calibration errors and theoretical ripple errors, in order for the two chosen phase modes to be added in phase (in the case of a beam setting) or in antiphase (in the case of null setting) at the specified bearing.

In step 164 the phase shifter 36 is set to the calculated value. Subsequently in step 166 the operator can select whether to steer to another beam or null and if so is returned to step 1 50. If not the operator can in step 168 elect to continue with direction finding, and will be returned to step 102 on Fig. 4, or stop 170.

Fig. 6 illustrates a modification to the system which can be used to locate a wanted signal in the present of an unwanted interfering signal coming from a different direction. It is assumed that the wanted signal of amplitude Dw arrives from a bearing a and that the interfering unwanted signal Dun arrives from a bearing 80. The circuit of Fig. 6 operates to steer a null at the bearing 0o so as to cancel out the interference and then carry out a direction finding exercise on the wanted signal at bearing 0.

To this end the circuit makes use of the zero, + 1 and - 1 order modes from the Butler matrix 10. The signal A from the zero mode output is divided by a two-way power divider 180 to give two signals A, and A2.

The gains of the signals A, and A2 and of the signals B and C from the + 1 and - 1 mode outputs of the Butler matrix are all adjusted to be the same by a suitable form of amplitude equalization. The A, and A2 signals are then applied to respective phase shifter circuits 182 and 184 which provide phase shifts of II + On and II - On respectively.The zero mode signals A, and A2 are given by: A, = A2 = Dwei + Dunelo After phase shifting the resultant signals A', and A'2 are: A'1 = Dwej(II + #0) + Dunej(II + #0) = - De' - D00e'0 A'2 = Dwej(II - #0) + Dunej(II-#0)= - D,,e-J0' - D0 e - Also the + 1 and - 1 mode outputs will be B = Dwej# + Dunej#0 C = Dwe-ls + D,,e-j A combining circuit 186 receives the signals A', and B to provide an output R, and a combining circuit 188 receives the signals A'2 and C to provide an output S.The signals R and S are given by: R = A', + B = Dw(ej# - ) + Dun( & - e'0') S = A'2 + C = D(e-'0 - e-'0') + Q 0(e-IU - e - In each case the second term in the expression is zero, so that: R = D,(e' - S = D,,(e-'0- e-'0) Hence the signals R and S are free from the interfering signal Dun. It will be seen that what has happened is that the + 1 and 0 modes have been combined so as to steer a null in the direction 80, and the - 1 and 0 modes combined to steer a separate null in the same direction. The two resultant signals R and S can be applied to the inputs 32, 34 of the phase comparator 30 to extract the bearing information of the wanted signal.

It will be appreciated from the above expressions for R and S, that the phase of R is (O+ On) - IT and the phase of S is - 0- On) - II. The phase difference between Rand S is thus simply 0+00, so that if On is determined 8 is known.

The system can be used to locate relatively low level wanted signals in the presence of quite high level interfering signals.

If more phase modes are available, due to there being more antenna elements in the array, more nulls can be steered simultaneously while direction finding can still be carried out on the wanted signal.

Claims (14)

1. A circular antenna array, comprising a plurality of antenna elements connected to a Butler matrix to produce at least two phase mode outputs, and a phase comparator connected to receive the said phase mode outputs and to measure the phase difference therebetween and to shift the phase of one of the outputs so as to bring it substantially into phase with the other output.

2. An array according to claim 1, in which the phase comparator comprises inputs for receiving signals the phases of which are to be compared, a combining circuit, automatically-adjustable phase shift means and oscillatory phase perturbation means connected between the inputs and the combining circuit, and means connected to the combining circuit to determine the amplitude of the signal component from the combining circuit at the frequency of the phase perturbation oscillation, and to alter the phase of the phase shift means so as to tend to reduce the said amplitude of the said signal component.

3. An array according to claim 1 or 2, in which the individual antenna elements each have a directional response.

4. An array according to claim 1, 2 or 3, including null steering means connected between the Butler matrix and the phase comparator and comprising power dividing means connected to the zero phase mode output of the Butler matrix, phase shift means con nected,to the two outputs of the power dividing means to provide respective phase shifts the mean value of which is 1 80', and combining means for combining the outputs of the phase shift means respectively with the + 1 and - 1 phase mode outputs of the Butler matrix to provide two signals for application to the phase comparator.

5. A phase comparator having two inputs for receiving signals the phases of which are to be compared, a combining circuit, automatically-adjustable phase shift means and oscillatory phase perturbation means connected between the inputs and the combining circuit, and means connected to the combining circuit to determine the amplitude of the signal component from the combining circuit at the frequency of the phase perturbation oscillation, and to alter the phase of the phase shift means so as to tend to reduce the said amplitude of the said signal component.

6. Apparatus according to claim 2 or claim 5, in which the phase shift means is connected to one input of the phase comparator and the phase perturbation means to the other input.

7. Apparatus according to claim 2, 5 or 6, in which the phase comparator is provided with means for combining the output of the phase shift means with the other input of the phase comparator to provide a signal output.

8. Apparatus according to any of claims 2, 5 6 or 7, in which the said means connected to the combining circuit comprises a receiver with an amplitude modulation envelope detector.

9. Apparatus according to claim 8, including a phase sensitive detector connected to compare the receiver/detector output with the phase of the phase perturbation means to indicate the sense of the phase error, the output of the phase sensitive detector being applied to the phase shift means such as to tend to reduce the phase error.

10. Apparatus according to claim 9, in which correction of the phase shift means continues until the phase sensitive detector detects an error in the opposite sense whereby the phase shift means hunts cyclically about the required mean phase shift.

11. Apparatus according to claim 8, 9 or 10, in which the phase sensitive detector and control of the phase shift means operates using digital processing techniques.

1 2. Apparatus according to any of claims 8 to 11, including means for determining when the output of the receiver/detector falls below a threshold and for disabling adjustment of the phase shift means in response thereto.

1 3. A circular antenna array comprising a plurality of like antenna elements connected to a Butler matrix to produce at leasts two phase mode outputs, in which the individual antenna elements each have a directional response.

14. An array according to claim 3 or 13, in which each antenna element comprises a loop having a termination at one point on the loop, the impedance around the loop being increased relative to the rest of the loop in a region on the opposite side of the loop from the termination by an at least partially resistive impedance, the loop being provided with means forming an unbalanced feed from the loop.

1 5. A circular antenna array provided with a phase comparator, substantially as herein described with reference to Fig. 1 of the drawings, with or without the modification of Fig. 6.

1 6. A circular antenna array provided with a phase comparator, substantially as herein described with reference to Fig. 3 of the drawings, with or without the modification of Fig. 6.