Multi-Moded Acoustic Wave Oscillator

This invention relates to acoustic wave oscillators. Such oscillators can use surface acoustic waves (SAW) or bulk acoustic waves (BAW).

In a typical SAW oscillator a SAW delay line is used as the positive feedback element of an amplifier. A SAW delay line may comprise a quartz substrate carrying an input and an output transducer on one flat surface. Usually such SAW delay lines permit oscillation at a plurality of frequencies i.e. the various modes permitted by the substrate between the two transducers. A disadvantage of this oscillator is the control of the precise oscillation frequency. One arrangement for controlling the frequency of oscillation uses a high Q resonant circuit within the feedback loop but external to the delay line.

A quite different approach uses the frequency response of the transducers and intertransducer spacing to ensure oscillation at one frequency only; such an arrangement is described in U.K. patent application 7880/73, U.S. patent application Ser. No. 442,624 (now U.S. Pat. No. 3,950,713).

Neither of these prior art arrangements allow a quick and precise change of frequency of oscillation from one mode to another.

It is therefore an object of this invention to provide an oscillator capable of oscillating at a plurality of preset and controllable frequency modes.

According to this invention a multi-moded acoustic wave oscillator comprises an amplifier and a feedback loop incorporating an acoustic wave delay line, the delay line comprising a substrate able to support acoustic waves, an input transducer for launching acoustic waves along a track in the substrate, a plurality of output transducers each arranged on the track, phase changing means for supplying a phase difference between inputs to the amplifier from the different output transducers, and means interconnecting the output transducer for allowing each output transducer to be connected in turn to the amplifier, the arrangement being such that the frequency of oscillation is changed in preset steps as the output transducers are connected in turn to the amplifier.

The output transducers may be spaced different distances from the input transducers. Alternatively the output transducers may be equispaced from the input transducer and phase change circuits arranged between the output transducers and amplifiers.

The roles of the input and output transducers may be reversed, i.e. the delay line may use a plurality of input transducers and a single output transducer.

The substrate may be also piezo electric e.g. quartz or lithium niobate, or non-piezoelectric, e.g. glass, with pads of a piezoelectric material at the transducers.

The delay line may also take other forms e.g. a conventional bulk-wave delay line transducers at either end, and say two output transducers.

The transducers may have constant finger length and spacing, in which case the amplitude response is of the sin x/x type and the modes of oscillation are equally separated in frequency. However, it is possible to weight the finger strengths in various ways (c.f. design of SAW bandpass filters) to have, say a flatter top to the main lobe, thereby making all wanted modes more nearly equally favoured. This may be desirable in some circumstances.

It is also possible to make the finger spacings within the input and/or the output transducers vary continously along the length of the transducer (i.e. perpendicular to the fingers) in such a way that the delay line(s) becomes dispersive (as used in pulse compression radar). In these circumstances the effective path lengths are a function of frequency and the simple equations given later for the allowed frequencies of oscillation no longer obtain. The modes are no longer equally spaced in frequency but their values may be calculated readily from the theory of the pulse compression filters which is well-known in the art.

The invention will now be described by way of example only with reference to the accompanying drawings of which

FIG. 1 is a diagrammatic view of an oscillator having four output transducers;

FIG. 2 is a frequency response graph;

FIG. 3 shows a modification to FIG. 1;

FIG. 4 shows an alternative to the device of FIG. 1;

FIG. 5 shows a BAW delay line to replace the SAW delay line of FIG. 3.

FIG. 6 shows a modification of FIG. 4 using dispersive and weighted transducers.

The oscillator of FIG. 1 comprises a SAW delay line 1 and rotary switch 2 in the feedback loop 3 of an amplifier 4. The delay line 1 comprises a piezoelectric substrate 5, e.g. ST cut quartz, which carries an input transducer 6 and four output transducers 7a, b, c, d spaced .lambda..sub.0 /4 apart, measured along a SAW track 8 from the input transducer, .lambda..sub.o being wavelength at centre frequency f.sub.o. The output of each output transducer 7 is connected to the rotary switch 2 which has four equal resistors R.sub.1, R.sub.2, R.sub.3, R.sub.4 and a central rotatable wiper 9. Output from the wiper 9 is to the input of the amplifier 4 whose output is to the input transducer 6. The four output transducers 7 may have equal length finger electrodes, as shown, or unequal length transducers. For example, if it is intended to operate the device with transducer 7a (the others being used solely to change the frequency) then transducer 7a may have long electrodes, while smaller electrodes will suffice in the remaining three output transducers 7b, c, d.

The path length L between the input transducer 6 and the upper output transducer 7a is chosen to give the required device stability and/or the frequency separation of the modes.

When simple transducers of the type shown in FIG. 1 are employed it can be shown that the number of complete cycles, n, around the loop comprising the delay line and amplifier is satisfied by the equation

where

w = angular frequency of acoustic waves

L = path length

C = acoustic velocity on the substrate

.phi. = the small electrical phase shift associated with amplifier and transducers

n = an integer, often called the mode number.

Thus neglecting .phi. the allowed frequencies of oscillation are given by

The response of a transducer has the form sin x/x and is determined from standard formulae to provide a wide central lobe encompassing the desired number of modes, see FIG. 2. By weighting the transducers it is possible to flatten the top of this response to some extent which is a desirable improvement to the simple device.

FIG. 2 shows the frequency response 10 of the input transducer and the modes i.e., harmonic frequencies of oscillation 11a permitted by the path length L. Note that the modes 11a shown are those for the path length L between the input transducer and the upper output transducer 7a. Modes 11b, 11c, 11d also occur between those shown for the path lengths L + .lambda.o/4, L + .lambda.o/2, and L + 3.lambda.o/4 corresponding to the other output transducers 7b, 7c, 7d. Amplifier gain is set, as shown by line 12, so that side lobe responses cannot be excited.

In operation the device of FIG. 1 is switched on and, after a short stabilising period it will oscillate at one of the permitted modes. If the wiper 9 of the rotary switch 2 is gradually rotated anti-clockwise output into the amplifier 4 will be from the four output transducers 7a, b, c, d separately and in turn. Since the output transducers 7 are spaced .lambda..sub.o /4 along the SAW track the phase of signal into the amplifier 4 changes by steps of .pi./2. One complete revolution of the wiper 9 changes the frequency of oscillaton from one mode 11a (of path length L) to the next, decreasing the frequency for anti-clockwise rotation, or increasing the frequency for clockwise rotation.

If the wiper 9 is merely switched by 1/4 turn the device would settle down to oscillate at a frequency between the modes 11a shown in FIG. 2 i.e. a mode 7b or 7d corresponding to path length L + .lambda.o/4.

As shown in FIG. 1 all the output transducers 8 are on one side of the input transducer. In a modification transducer the four output transducers 7 are arranged two either side the input transducer.

More than four output transducers may be used, with correspondingly small steps in path length, arranged on one or both sides of the input transducer. In general such an arrangement gives smoother operation and permits operation over a larger bandwidth.

A typical device has a centre frequency of 60 MHz, with a path length L = 3.15 cms on ST-cut quartz the separation of modes is 0.1 MHz. The input transducer has 20 finger pairs and each output transducer has 20 finger pairs.

The device of FIG. 1 with its rotary switch 2 is a useful one for illustrating the principle of this invention. A more practical device would replace the rotary switch 2 with four PIN diodes suitably biassed in turn. Any other solid state electrically controlled switches such as voltage-controlled impedance e.g. varactor, or amplifiers can be used in place of the PIN diodes.

In operation it is not necessary to switch instantly from one output transducer to another. It is quite satisfactory (and often useful) to switch to a combined output at intermediate points, corresponding to intermediate points in the rotary potentiometer of FIG. 1, or to the use of make-before-break mechanical switches. This gives smoother operation, and also permits a further series of intermediate frequencies for example, midway between the frequencies which would obtain with transducers 7a or 7b in FIG. 1. FIG. 3 shows a modification of FIG. 1 having PIN diodes controlling the output from the delay line. The oscillator, as before, comprisies a SAW delay line 14 in the feedback loop 15 of an amplifier 16. This delay line 14 comprises a quartz substrate 17 carrying an interdigital input transducer 18 and two interdigital comb output transducers 19, 20 spaced .lambda..sub.o /4 apart measured along a SAW track 21 from the input transducer 18. Both the combs 19a, b, 20a, b forming each output transducer 19, 20 are connected to two PIN diodes 22a, 22b, 23a, 23b, 24a, 24b, 25a, 25b the common points of which are connected through 1 k ohm resistors to terminals of a two pole four way switch 27. One diode in each pair is connected to earth whilst the other is connected to a junction 28 and to the amplifier 16. Positive and negative voltages are supplied to the two poles of the switch 27 so that when a positive voltage is supplied to one comb of a transducer a negative voltage is supplied to the other comb thereby connecting one comb to the amplifier and the other to earth. In this manner the combs are connected in turn to the amplifier 16 as the switch 27 is rotated.

The two combs of an interdigital transducer are 180.degree. out of phase and therefore by reversing the earth and live terminals a 180.degree. change in phase can be affected. By this means the transducers 7c and 7d of FIG. 1 are effectively replaced using transducers 7a and 7d with their terminals reversed.

Thus the oscillator of FIG. 3 operates in the same manner as that of FIG. 1 to change modes, namely output is taken from:- the upper comb 19a of output transducer 19, (with the lower comb 19b earthed) followed by the upper comb 20a of output transducer 20 (with the lower comb 20b earthed), followed by output from the lower comb 19b of output transducer 19 (with the upper comb 19a earthed), followed by output from the lower comb 20b of output transducer 20 (with the upper comb 20a earthed).

As shown both output transducer 19, 20 are on one side of the input transducer 18; they may also be arranged one on either side of the input transducer 18 in which case all the transducers 18, 19, 20 may have equal finger length.

FIG. 4 shows another form of the invention which uses three output transducers. As before it comprises an amplifier 29 with a SAW delay line 30 in the feedback loop 31. The delay line 30 comprises a quartz substrate 32 carrying an input transducer 33 and three output transducers 34, 35, 36 spaced .lambda..sub.o /3 along a SAW track 37. Output from the output transducers 34, 35, 36 is through PIN diodes 38, 39, 40 to a junction 41 and then to the amplifier 29. These PIN diodes are biassed in turn by application of a suitable voltage through a make before break three way switch 42 and 1 k.ohm resistors to connect the output transducers 34, 35, 36 in turn to the amplifier 29. The 1 k.ohm resistors limit d.c. current to the PIN diodes and also act as high impedance blocks to RF signals.

To change the frequency of oscillation the output transducers 34, 35, 36 are connected in turn to the amplifier 29, in the same manner as for the oscillator of FIG. 1. One advantage of using three transducers (stepped by .lambda..sub.o /3) is that the amplitude of output from the output transducers is constant whilst output is being switched, even through the intermediate positions where two transducers are connected simultaneously.

For devices to operate in the manner described, it is necessary for each change in phase to be < 180.degree. (ideally < 120.degree.). Using such criteria, it can be shown that operating bandwidth of the device can be of the order of the centre frequency, fo. However for such large bandwidths the spacings of all the intermediate frequencies become unequal, e.g. in the example with four transducers, the spacing is exactly .lambda..sub.o /4 only at the centre frequency. This can be a problem if it is desired to use all the intermediate frequencies, e.g. in an f.m. tuner with equal channel separations.

One partial solution to this problem is to modify the positions of transducers 7c and 7d in FIG. 1, so that transducers 7a and 7c are equidistant from the input transducer 6 and 7b and 7d are equidistant from the input transducer 6. Additionally the outputs from 7c and 7d are taken from the lower combs instead of the upper combs shown in FIG. 1. By so reversing the connections to 7c and 7d a phase shift of 180.degree. is introduced which compensates for the displacement by .lambda..sub.o /2 at the centre frequency. By this means the frequencies from 7a and 7d have equal separation at al frequencies, and the same is true of the frequencies frm 7b and 7c.

Another solution to the above problem is to modify the arrangements shown in FIGS. 1 and 4 by replacing the non-dispersive transducers shown and use dispersive transducers. This is shown in FIG. 6 which is similar in operation to FIG. 4. The device of FIG. 6 comprises an amplifier 29 with a SAW delay line 65 in the feedback loop 31. The delay line 65 comprises a quartz substrate 66 carrying a dispersive input transducer 67 and three dispersive output transducers 68 69 70 spaced .lambda..sub.o /3 along a SAW track 71, each output transducer having varying finger length to provide weighting. As in FIG. 4 output from the transducers 68, 69, 70 is through PIN diodes 38, 39, 40 to a junction 41 and then to the amplifier 29. These PIN diodes are biased in turn by application of a suitable voltage through a make before break three way switch 42 and a 1 kohm resistor 72, 73, 74 to connect the output transducers 68, 69, 70, respectively in turn to the amplifier 29. The oscillator of FIG. 6 operates in the same way as the oscillator of FIG. 4.

With the dispersive transducers 67, and 68, 69, 70 shown in FIG. 6, different frequencies are effectively launched from different parts of the transducer; this is achieved by varying the interfinger spacing along the transducer. Thus, the low and high frequency ends of the input transducer 67 may be left and right ends respectively as seen in the drawing, while the low and high frequency ends of the output transducers 68, 69, 70 are also the left and right ends respectively. In these circumstances the dispersion characteristics of the input 67 and output 68, 69, 70 transducers tend to cancel each other out. It is arranged that the input transducer 67 and one output transducer, e.g., 68, have identical dispersion characteristics so that this delay line as a whole is non-dispersive and the mode spacing is 1/.gamma..sub.a at all frequencies (where .gamma..sub.a is the delay time which in these circumstances is independent of frequency). However, the dispersion characteristics of 69 and 70 are successively mismatched from the input transducer 67 in such a manner that for each delay line the intermode spacing is .DELTA.f.tbd.1/.gamma..sub.a at all frequencies.

An additional advantage of using dispersive transducers is in their relative ease of matching to a 50.OMEGA. source for minimum insertion loss over large bandwidths. For this reason there may be occasions where five transducers of identical dispersion characteristics are used in the arrangement of FIG. 1.

The arrangement shown in FIG. 3 may also be modified to use dispersive transducers in a manner similar to that described with reference to FIG. 1.

FIG. 5 shows a bulk acoustic wave (BAW) delay line 45 for use in the feedback loop of an amplifier to replace the delay line 15 of FIG. 3. It comprises an isopaustic glass block 46 with two polished flat and parallel ends 47, 48. One end 47 carries an input transducer 49 and the other end carries two output transducers 50, 51. Each transducer 49, 50, 51 comprises a quartz crystal 52, 53, 54 sandwiched between two gold sheet electrodes 55, 56, 57, 58, 59, 60. Signals from output transducers 50, 51 will be in phase; a 90.degree. phase shift between outputs may be achieved by a .lambda..sub.o /2 length of electromagnetic delay cable. Alternatively series/parallel tuning inductances may be added to one/both transducers so as to produce a phase difference between outputs.

The BAW delay line is used in the same manner as the SAW line of FIG. 3, namely to change the frequency of oscillation, output to the amplifier is changed in phase steps of 90.degree., by using the output transducers 50, 51 in turn and by reversing earth and live electrodes e.g. with biassed PIN diodes.

One problem with all the above oscillators is knowing the precise frequency mode the device is operating at out of the many permitted modes, e.g. after initial switch on. There are many possible solutions to this problem, four of these are as follows.

One solution is to ensure that one mode is more strongly favoured than all the others (e.g. the central mode in FIG. 2) and to switch the amplifier on slowly so that its gain gradually increases over a period of say 1 second. It is known that the build-up time of oscillation is about twenty times the acoustic delay time (typically 20 times 3.mu.sec = 60 .mu.sec) giving adequate time for the central lobe to build-up before the other modes are allowed.

A second solution could be as follows. Allow the device to come on at an arbitrary frequency, and then to wind the frequency up through at least the full range. By correct design it can be made to stay at the highest frequency when an attempt is made to wind it above the highest frequency i.e. rather than jump to an arbitrary frequency. Thus this procedure enables one to attain the highest allowed frequency, and all subsequent changes are referred to this (highest) frequency. Equally, the frequency could be wound down to the lowest allowed value.

A third solution is to use a SAW oscillator having a SAW delay line on the same substrate but with such oscillator having strong mode selection i.e. it operates at one frequency only. One such oscillator having strong mode selection is described in U.K. patent application 7880/73 U.S.A. patent applicaton Ser. No. 442,624 (now U.S. Pat. No. 3,950,713), and comprises (in a simple form) an input transducer whose effective length is equal to the centre to centre distance between input and output transducers. When using two delay lines on the same substrate the delay line having strong mode selection is switched on and allowed to stabilise its frequency. When the oscillator having many possible modes is then started it will oscillate at the frequency already present either because of the acoustic waves on the substrate travelling on both delay line tracks or because of electrical interconnections common to both oscillators.

A fourth solution is to incorporate a frequency discriminator into the oscillator, e.g. as described in U.K. patent application 52,237/73, U.S.A. Pat. No. 3,921,093. The voltage output from this discriminator (e.g. one or two SAW transducers) is a measure of the frequency of oscillation. A new frequency may be selected by push-buttons which generate a voltage equal to the discriminator output voltage at the desired frequency. Any difference between this push-button voltage and the discriminator output voltage causes the frequency changing mechanism to operate, until the difference is reduced to a negligible value.