GB2282879A - Optical gyroscopes - Google Patents

Abstract

In an open loop gyroscope clockwise (CW) and counterclockwise (CCW) beams from a common source 10 pass in opposite directions around a coil 20 of optical fibre and recombine on a detector 24. A phase modulator 22 applies a stepped phase modulation in which each step is of duration equal to the transit time of the coil 20 and in which alternate steps are zero. The intermediate steps increase linearly from zero to 2 pi . Without modulation, the CW and CCW beams combine on the detector with a combined intensity which varies as a cosine of rate. The above modulation causes two points to either side of the cosine fringe to be sampled which move progressively further as the amplitude of the stepped modulation increases. The sample data is processed to determine the rate applied to the gyroscope. <IMAGE>

Description

OPTICAL GYROSCOPES This invention relates to optical gyroscopes and in particular to open loop interferometric fibre optic gyroscope.

In an interferometric fibre optic gyroscope a clockwise (CW) beam and a counterclockwise (CCW) beam pass around a coil of optical fibre and are brought together on leaving the coil to interfere on a detector. Rotation applied to the gyroscope induces a rate-related non-reciprocal phase shift between the CW and CCW beams which causes the intensity on the detector to vary cosinusoidally with applied rate.

In general, two different methods of signal processing may be used. In closed loop systems, a compensatory nonreciprocal phase shift is applied to null the phase shift induced by rotation. In open loop systems, the intensity of the combined beams at the detector is used to determine the applied rate. In both methods, the measurand is linearly related to rate with the constant of proportionality being referred to as the Scale Factor.

Previously it has been thought necessary to use a closed loop architecture to achieve high accuracy of the scale factor. In this context, high accuracy means a scale factor performance which does not differ by more than 100ppm with respect to an ideal output over a rate range of +/-500 deg/sec or greater, which is the rate range needed for aircraft and missile applications. Such an accuracy is needed for inertial navigation in both civil and military applications, and for guidance of long range missiles.

For agile military aircraft or naval vessels, where typically a ring laser gyroscope system is currently used, an even higher scale factor accuracy of about 5ppm over the rate range is needed, and this represents one of the highest accuracy scale factor requirements.

To allow a fibre gyroscope to replace the more expensive ring laser gyroscope for these applications, scale factor performance at this level is required. Replacement of ring laser gyroscopes by fibre optic gyroscopes is likely to produce a much lower cost navigation system, as the cost of the gyroscope sensors is one of the largest elements in the cost of the overall system.

For a fibre optic gyroscope with closed loop architecture, optical modulation is applied so that at zero rate there is a null signal. At rate, after demodulation, a signal is derived which is proportional to rate. This is used to operate a feedback loop as the error signal, and a second signal is applied to a phase modulator to null out this error signal. In a typical example, the modulation comprises a square wave operating at the correct frequency for the fibre coil. This is the frequency that causes a phase shift of 1800 between the two directions when the phase modulator is placed at one end of the coil, after a circuit of the coil and recombination, and is given by 1/(2*loop transit time). The loop transit time is the time taken for light to propagate from one end of the coil to the other.The feedback signal is typically a serrodyne ramp applied to a phase modulator in the optical circuit. This comprises a linear phase modulation ramp going from 0 radians to 2 radians in a time t and then being reset to zero quickly, with the process then restarted. If the top of the ramp is exactly 2 radians, this corresponds to a frequency shift of 1/t Hz, and this frequency shift is that needed to null out the rate signal, and the frequency is then the rate output. This signal is normally applied at the other end of the sensing coil to the square wave modulation, to null out the applied rate. In this case, the output of the gyroscope is then the frequency of the serrodyne ramp which is proportional to rate.This gives very good scale factor performance over a broad rate range but suffers from the following disadvantages: (a) The light around the loop is at a different frequency in the CW and CCW directions, so that the gyroscope is fundamentally non reciprocal, and thus the gyroscope bias performance may be degraded.

(b) It is necessary to implement a very accurate servo loop to hold the serrodyne reset amplitude at exactly 2n radians. Methods of doing this have been devised, but there are problems in implementation at low rates when there is not a lot of information to drive the servo. In one implementation, the signal to drive the servo is obtained at the flybacks which occur seldom at very low rates.

(c) Any feedback loop which has the attributes of gain and feedback may suffer from lock-in behaviour at low rate. This causes extreme scale factor errors in a rate range around zero rate so that sensitivity may be severely affected or completely lost.

By contrast, the conventional open loop gyroscope schemes are believed to offer good gyroscope drift performance due to the high degree of reciprocity, but generally poor scale factor performance due to the difficulty of accurately following the cosine wave curve of light output against rate to very high precision.

In our earlier U.K. Patent Application No. 9304016.0 we describe a signal processing technique which involves the extraction of three signals at frequencies lf,2f and 4f (with a modulation frequency of lf) from the photodetector signal to account for three unknowns; the angular rate (i.e.

the desired output); the intensity of light on the photodetector, and the amplitude of the phase modulation.

This compensation is needed as the intensity of light and the amplitude of the phase modulator are both subject to change due to external perturbations such as temperature and time and these parameters enter into the equation determining the rate. In our earlier technique, two ratios are formed from the three signals to eliminate the intensity effects, and the inverse tangent is taken of one ratio (lf/2f) to extract the rate, and the other ratio (2f/4f) is used to determine the amplitude of phase modulation.

This is a fairly complicated scheme, and requires three signal channels at the three frequencies which need to be very closely matched so that the ratios are meaningful.

According to this invention there is provided an optical gyroscope, comprising: means defining a coil or ring around a sensing axis and around which light may propagate in a clockwise (CW) and a counterclockwise (CCW) direction; beam input means for introducing into said coil or ring a clockwise (CW) beam and a counterclockwise (CCW) beam to propagate in opposite directions around said coil or ring; means for combining said CW and CCW beams after passage around said coil or ring; detector means for detecting the intensity of said combined CW and CCW beams, phase modulator means for applying between said CW and CCW beams a plurality of different phase shifts, and processor means for monitoring the corresponding detected intensity for each of said phase shifts, thereby to sample the intensity across at least a major portion of a complete fringe.

Whilst the invention has been described above, it extends to any inventive combination of features set out above or in the following description.

The invention may be performed in various ways and two embodiments thereof will now be described by way of example, reference being made to the accompanying drawings, in which: - Figure 1 is a schematic figure illustrating a first embodiment of open loop fibre optic gyroscope; Figure 2 is a diagram illustrating the phase modulation applied to the CW and CCW beams; Figures 3 and 4 are diagrams illustrating the sum and difference signals for successive pairs of intensities detected, for a positive rotational rate and a negative rotational rate, respectively; Figures 5 and 6 are diagrams illustrating the Fourier Transforms of the sum data and the difference data of Figures 3 and 4 respectively; and Figure 7 is a schematic figure illustrating a second embodiment of an open loop fibre optic gyroscope which incorporates a control loop to set the maximum magnitude of the phase modulation.

Referring initially to Figure 1, light from a broad band source 10 such as an edge emitting light diode (ELED) or a superluminescent diode (SLD) is coupled into a single mode fibre 12 to a first 50:50 coupler 14. From there the light passes via a mode filter or polariser 16 to a second 50:50 coupler 18 which splits the beam to form CW and CCW beams which pass into the opposite ends of a coil 20 of optical fibre. At one end of the coil 20 is a broadband phase modulator 22 (typically of lithium niobate). The second coupler 18 and the phase modulator 22 may both be part of an integrated optics circuit.

Where an integrated optics circuit is used, there are two possibilities for connecting the ends of the coil 20 to the integrated optics circuit, depending on whether the coil fibre is of high birefringence fibre. When the coil is of high birefringence fibre, the integrated optics circuit is fibre-coupled to two fibres, typically of high birefringence fibre, which are fusion coupled to the high birefringence coil fibre.

Alternatively, where the coil is not of high birefringence fibre, depolarisers may be attached to the two high birefringence output fibres which are then attached to the relatively low birefringence coil. The depolarisers may each comprise a length of high birefringence fibre fusion spliced so that its fast and slow axes are at 45" to the corresponding axes in the high birefringent fibres from the integrated optics circuit.

After having passed around the coil, the CW and CCW beams are recombined at the second coupler 18, and then pass via the mode filter or polariser 16 to the first coupler 14 which passes a component of the combined beam to a photodetector 24.

The phase modulation applied by modulator 22 is shown in Figure 2. The upper trace shows the phase modulation seen by the CW beam, and the lower trace that seen by the CCW beam. Each phase modulation step has a duration of t, where t is the loop transit time, so that the CCW phase is equivalent to the CW phase delayed by one step, because the phase modulator is located at the CW end of the coil 20. The step height or amplitude is changed as a linear ramp, starting at zero, and increasing at a steady rate to 2 and then resetting to zero. Thus at time O < t < x, the step height is zero, at time t < t < 2t, the step height is A1, at time 3t < t < 4t the step height is A, and so on.

It will be seen that the modulation is made up of a series of pairs of positive and negative phase shifts A, the magnitude of each phase shift within the pair being equal, but the magnitude between pairs increasing stepwise from zero to 2n, and then resetting to zero. Thus the drive signal to the modulator 22 steps as follows o, vl,o,v2,o,v3.

In the absence of applied rate, and with no phase modulation, the intensity detected by the photodetector 24 should be at a maximum, corresponding to the peak of the cosine curve relating intensity to rate (otherwise referred to herein as the fringe). In general, with the series of progressively increasing phase shifts of alternate sign, the cosine curve is effectively sampled at a series of pairs of phase sample points symmetrically disposed relative to the zero phase modulation point. In the special case where there is no applied rotation and the intensity on the photodetector is at a maximum, the positive/negative pairs of samples will each be of the same magnitude as they move down either side of the curve.When rotation is applied, this will cause the fringe or cosine curve to shift to an extent and in a direction dependent upon the magnitude and sense of the applied rotation, so that the zero phase modulation point moves down one of the flanks of the cosine curve. The positive/negative pairs of samples will thus tend to be of different magnitude but they will still sample a complete 2 period of the cosine curve provided the phase modulation ramps up to 2W.

The reason for the shape of the phase steps can be understood from the following equations. The general expression for this intensity of the light I at the photodetector 22 may be written as: = = a0 + a1 (1 + cos(a2 + ##i)] ##i = = a3 aVi = a3 (Vj(t) - Vi (t-t)] (1) where a0 = electronic offsets a1 = intensity of fringe pattern a2 = S n where S is the scale factor of the gyroscope and n is the angular rate (radians/sec) a3 = scale factor of phase modulator (radians/V) Vi(t) is the input voltage to the phase modulator 22 and T is the loop transit time.

The phase modulation is applied at one end of the coil and suffers a delay of one loop transit time (t) between the two directions, so the difference appears in the equation (1). As set out above, for each positive value of Av it is possible to generate a negative value by reversing the sense of the voltage change.

A 16 bit digital to analogue convertor (DAC) 26 is used to apply voltages to the phase modulator 22 synchronised with the loop transit time. Thus a sequence O,V,0 from the DAC 26 causes a positive AV to be followed by an accurately matched negative AV at a time period t later. This causes two points either side of the cosine fringe to be sampled, which at zero rate will have the same intensity as explained above.For each magnitude Vi, the intensity Ii+ and Il-, of the two points sampled may be written: Ii+ = a0 + a1 [1 + cos (a2 + a3 #Vi)] Ij = a0 + a1 (1 + cos (a2 - a3 AVi)) (2) and for each magnitude Vi a difference AIi and a sum SIs of the two values may be obtained as follows: # Ii = Ii+ - Ii- = 2a1sina2 sina3 #Vi #Ii - 2(a0+a1) + 2a1cosa2 cosa3 #Vi (3) so that two signals are derived from the data, the sum (rIi) and the difference (#Ii).

An analogue to digital convertor (ADC) 28 with 16 bit resolution is used with a sample and hold amplifier (not shown) before it. The signal gathering is synchronised with the signal on the digital to analogue converter 26. The sample and hold amplifier has a broad bandwidth to ensure that the transitions between values are not spread into the data that is being collected.

The sum and difference signals have a cosine and sine wave dependence respectively on rate, and a cosine and sine dependence on the phase modulation amplitude ##i, which corresponds to the time axis.

In order to extract the rate from this data, various approaches are possible. One that is attractive is the use of fast Fourier transforms. A single complex Fourier transform can be used with the sum and difference as the real and imaginary parts on input, and on output the two transforms can be separated as follows: If h(t) is real, then H(f)=H(f)* where H(f) is the complex Fourier transform of h(t) If h(t) is imaginary then H(-f)=H(f)* thus H(-f)+H(f) = 2 H(f)reai and H(-f)-H(f) = H(f )imag so that from the single complex Fourier transform, the transforms of the real and imaginary parts can be separately extracted, corresponding to the sum and difference signals.

In order to use fast Fourier transforms (FFT) the data set needs to have 2n data points, where n is an integer. In a typical case n=7, corresponding to 128 points for both the sum and difference signals.

For a 600 metre fibre coil, the synchronous frequency is about 170kHz giving 340 kilosamples per second with each positive sample being followed by the corresponding negative sample. Analogue to digital converters and digital to analogue converters are available with a 16 bit resolution to give this frequency capability. To accumulate the 256 points will take 0.7 milliseconds, and typically 400 microseconds to carry out the FFT using a typical digital signal processor (DSP) 30 such as the AT & DSP32C which runs at 50MHZ clock with 25 million floating point operations per second (MFLOPS) processing power.

Figures 3 and 4 show the sum and difference signals for +10 deg/sec rate and -10 deg/sec rate with 128 data points. This clearly shows the cosine and sine dependence on time of the two sets of data for the sum and difference. Also shown is the phase inversion in the difference data between the positive and negative rates while the sum data has the same phase.

Figures 5 and 6 show the FFT, giving the amplitude in the channels for the sum and difference data respectively for -10 deg/sec rate. Each of the sum and difference signals has a DC offset which shows in the first channel of the FFT.

The second channel of the FFTs of the sum and the difference signals gives the rate signal. As there is no windowing in the FFTs, the information from the rate is spread over several channels, so that the correct amplitude can be written as:

Where Np is the noise power and S is the signal power for the Fourier transform of both the sum and difference signals. In the summation in equation (4) the (i)th element is the real part and the (i+l)th element is the imaginary part. This summation of the real and imaginary parts is needed in this particular technique, because the exact phase of the signals is not known. The summation for the signal goes from n=3, as n=l and n=2 correspond to the real and imaginary parts of the DC term, which does not correspond to the required signal. This summation applies for a FFT with 128 data points.The value of N in (3) defines the noise power per channel which is assumed to be white. The sum over the higher frequency components is used to give the noise power value, which is then subtracted from the signal information defining the signal power S. The rate is extracted as follows:

where S1 and S2 are the signals for the difference and sum terms respectively. Taking this ratio removes the effect of intensity variations. This ratio also needs to include the signs of the amplitude of the Fourier transforms, which are given by the sign of the imaginary part of the difference sign and the sign of the real part of the sum signal.

This inverse tangent gives the rate in terms of a phase angle, and this is converted into a rate in deg/sec by a knowledge of the scale factor of the gyroscope by previous calibration, as the phase angle (a2) rate (R) and scale factor (S) are related thus:- a2=SQ.

The benefits of the above approach using Fast Fourier Transforms, include the following: 1. There is only a single channel of information which is used to extract all the information, unlike the previous open loop gyroscope scheme described above, which has three channels.

2. The effect of intensity variation is eliminated by the taking of ratios to give the inverse tangent.

3. The effect of variation of the phase modulation amplitude is to change the frequency of the signal in the FFT. This will change by a small amount in relation to the width of the frequency channels of the FFT, and this small change will cause greater side band amplitudes. This however is not critical as it is known that there is only one signal to be considered in the FFT and the summation over the first few channels will take account of these signals. Therefore the method is relatively insensitive to the variation of the phase modulator scale factor.

4. The noise is eliminated by subtraction of the higher frequency components.

5. The analysis is done in the digital regime and so should be highly accurate after conversion to digits from the analogue signals. This includes the DAC and ADC, which are 16 bit devices in the practical implementation. It has been shown that 16 bits are sufficient to give the required level of accuracy to meet the needs of an inertial navigator.

6. All the data can be gathered and analysed in 1 millisecond, so that the real time extraction of rate can take place with a good bandwidth when a typical digital signal processor is used with typical DAC and ADC for a 600m coil, which is a typical length for a medium accuracy gyroscope.

7. The use of an open loop gyroscope gives better gyroscope bias stability as the reciprocity is better than an equivalent closed loop gyroscope which normally has a different frequency of light in the two directions.

An alternative approach to that set out above extracts the rate without requiring Fourier Transforms. Embodiments incorporating this alternative should be faster and offer greater accuracy and consistency.

In a second embodiment, in order to improve the operation of the digital signal processing, the peak voltage going into the phase modulator is controlled by a servo system in order to ensure that the scan is over exactly a complete fringe. In this embodiment, this is done by implementing a phase locked loop.

Referring now to the second embodiment illustrated in Figure 7, the optical arrangement is the same as in Figure 1, and the same form of phase modulation is applied as in Figure 2. The sum and difference signals are extracted as above to give a cosine and sine wave response respectively.

The frequency of these signals will be set by the voltage applied to the phase modulator. If a complete fringe is exceeded the frequency will be higher and vice versa. The sum and difference signals can be written as: AIi = 2alsina2 sina35Vi sIi = 2(aO + al) + 2a1cos a2cos a3 AVi (6) where a3 is the scale factor of the phase modulator (radians/V). This will vary over time and is not well determined, and so needs to be eliminated as a variable from the equation. This is the purpose of the phase locked loop. Four signals are extracted as follows:

where o is the desired frequency (determined by the extent of the phase shift and the time taken to sample the fringe), when a complete fringe is exactly scanned.When the frequency is incorrect, there will be a phase shift between the reference frequency o and the input signal. The sum signal is used to derive an error signal for the control loop as this signal is present even in the absence of rate, and only disappears at one particular rate which is fairly high in value. The phase at the end of the integral is given by:

down ((flfl))) (8) where n is the n th set of readings. In practice, the integral is performed by a summation of the data acquired on the ADC over the ramp interval, multiplying this data with a sine or cosine function. In appendix A there is the Fortran programme which simulates the signal processing, and this includes simulation of electronic noise on the photodetector.This noise does not affect the accuracy of the result.

The error signal for the servo is then the difference in phase between the n th and (n-l)th set of data, so that the peak voltage onto the phase modulator is given by; V(new)=V(old)-gain*(#n-#n-1) (9) where "gain" is the proportional gain of the servo, which is set by experiment to give a stable servo operation. After several cycles this will converge onto the correct value to give a complete fringe.

When the correct peak operating voltage is reached, which should happen after several cycles as in equation (9), then the four signals of (8) contain the information to extract the rate, as the integral over two u gives the correct Fourier components.

The angular rate is then given by:

as the sign of the S2 integral gives the sense of the rate.

This approach is believed to be quicker to implement than the use of the Fourier Transform in software, and the operation of the servo implies that there is always a scan across a complete fringe, so the results should be consistently accurate.

The schematic of the electronics is shown in Figure 7.

The DAC 26 that drives the phase modulator has the same step signals as shown in Figure 2 of the previous embodiment, with the data synchronised with the loop transit time. The additional element is a second DAC 32 which sets the reference voltage of the first DAC 26, by applying a voltage V(ref) as a reference to the first DAC. Changing the reference voltage V(ref) of the first DAC 26 changes the peak voltage.

The processor 30 adjusts the voltage V(ref) in accordance with the output of the servo. An integrator may additionally be included, so that when the correct condition is set, this is held. The reference voltage is then given by:

so that the phase modulation is controlled to sweep a complete 2 period of the intensity waveform or fringe.

Claims (15)

1. An optical gyroscope, comprising: means defining a coil or ring around a sensing axis and around which light may propagate in a clockwise (CW) and a counterclockwise (CCW) direction; beam input means for introducing into said coil or ring a clockwise (CW) beam and a counterclockwise (CCW) beam to propagate in opposite directions around said coil or ring; means for combining said CW and CCW beams after passage around said coil or ring; detector means for detecting the intensity of said combined CW and CCW beams, phase modulator means for applying between said CW and CCW beams a plurality of different phase shifts, and processor means for monitoring the corresponding detected intensity for each of said phase shifts, thereby to sample the intensity across at least a major portion of a complete fringe.

2. An optical gyroscope according to Claim 1, wherein the phase modulator applies between said CW and CCW beams a plurality of phase shifts sufficient to sample substantially a complete fringe.

3. An optical gyroscope according to Claim 1 or 2, wherein said phase modulator means applies a series of phase shifts comprising a plurality of positive phase shifts of predetermined magnitudes and a corresponding plurality of negative phase shifts of magnitudes equal to said positive magnitudes.

4. An optical gyroscope according to Claim 3, wherein said series comprises a plurality of pairs of positive and negative phase shifts of substantially the same magnitude, the magnitude of the pairs progressively varying through the series.

5. An optical gyroscope according to Claim 4, wherein the magnitude of the phase shift increases in stepped linear fashion from 0 to 2W.

6. An optical gyroscope according to Claim 5, including servo control means for controlling the maximum phase modulation to be 2X.

7. An optical gyroscope according to any of Claims 3 to 6, wherein said series is a repeating series.

8. An optical gyroscope according to any preceding claim, wherein said phase modulator means applies to both said CW and CCW beams, at one end of said coil or ring, a stepped modulation waveform wherein the duration of each step is substantially equal to the transit time of the coil or ring.

9. An optical gyroscope according to Claim 8, wherein the modulation waveform is of magnitude o,##1, o,##2,0,##3 .... 2, whereby the modulation sweeps across 2# of the intensity fringe.

10. An optical gyroscope according to any of Claims 3 to 9, wherein for each magnitude of phase shift, the processor means monitors the detected intensity for a positive shift and that for a negative shift.

11. An optical gyroscope according to Claim 10, wherein the processor means determines the sum and the difference of the two intensities monitored for each magnitude of phase shift, thereby to obtain a series of sum data and difference data respectively.

12. An optical gyroscope according to Claim 10, wherein said processor determines the rate experienced by the gyroscope is on the basis of at least one of said sum and difference data.

13. An optical gyroscope according to Claim 12, wherein said processor means applies a Fourier Transform to at least one of said sum and difference data, thereby to obtain the applied rate.

14. An optical gyroscope according to claim 12, wherein said processor means: (i) multiplies said sum data by a cosine function and integrates to give a first signal, (ii) multiplies said sum data by a sine function and integrates to give a second signal, (iii) multiplies said difference data by a cosine function and integrates to give a third signal,and (iv) multiplies said difference data by a sine function and integrates to give a fourth signal, and uses said first, second, third and fourth signals to obtain the rate applied.

15. Apparatus substantially as herein described with reference to and as illustrated in the accompanying drawings.