GB2412737A

GB2412737A - Enhanced sonde recognition

Info

Publication number: GB2412737A
Application number: GB0407355A
Authority: GB
Inventors: John Mark Royle
Original assignee: RADIODETECTION Ltd; Radiodetection Ltd
Current assignee: RADIODETECTION Ltd; Radiodetection Ltd
Priority date: 2004-03-31
Filing date: 2004-03-31
Publication date: 2005-10-05
Also published as: GB0407355D0

Abstract

A detector for detecting an oscillating magnetic signal produced by a sonde comprises a plurality of antennas for detecting electromagnetic signals with a plurality of frequencies comprising the magnetic signal, an analogue to digital convener (11) to produce a digital output, and digital processing means for processing the output to isolate the magnetic signal produced by the sonde. The digital processing means comprises a phase feedback loop (21) to allow the processing means to follow variations in the frequency of oscillation of the magnetic signal produced by the sonde. The sonde is mounted in a drill head used for drilling underground trenches and enables locating of the drill head underground.

Description

24 1 2737 Enhanced sonde recognition

Field of the invention

The present invention relates to apparatus and a computer program for and a method of digital signal processing, and in particular to apparatus and a computer program for and a method of sonde recognition.

Background of the invention

When laying a new utility cable or pipe underground, an underground drill can be used to avoid the need to dig and fill-in a trench. When performing such trenchless drilling, the drill operator must be aware of the exact location of the drill head, in order to steer the drill in the correct direction and to avoid buried utilities or underground structures.

To locate the drill head, a sonde can be inserted therein. The sonde generates an alternating dipole magnetic field with a frequency of oscillation in the range 200Hz to 83KHz. At the lower of these frequencies, attenuation due to the drill bit enclosure and distortion due to ground effects are minimized. The magnetic signal can be modulated to transmit useful information, such as data relating to the tilt and roll angle of the sonde, a temperature reading and/or a battery life reading.

The magnetic field produced by the sonde is detectable by a portable detector, which can be hand-held or mounted on a tripod. The detector incorporates an array of highly sensitive antennas which detect electromagnetic signals from 50Hz up to around 250kHz. The signal generated in each antenna is processed to identify and isolate the sonde signal. The field magnitude, field gradient and relative phase of the sonde signal detected at each antenna are evaluated and these measurements are processed to determine the location and orientation of the sonde and associated drill head. The detector incorporates a digital signal processor (DSP) to process the received electromagnetic signals. The degree of complexity of the algorithms used in the DSP depends on a number of factors, including the power available from the detector's batteries, the requirement to provide a reasonable drilling speed and the desired degree of accuracy when pinpointing the sonde.

Prior art apparatus for sonde recognition and location, comprises an analogue or digital receiver of either super-heterodyne or homodyne architecture, a phase sensitive signal processing system and use of a complex discrete Fourier transform (CDFT) to arbitrate the sonde present criterion. Such processes were limited to measuring the signal to noise ratio (SNR) of the carrier frequency with respect to the broadband noise floor.

The effects of use of Blackman or Hamming window filters to suppress the spectral leakage which is inherent to the CDFT algorithm for frequency components which are not an integral number of periods within the transform length are shown in Figure 6 which shows a sonde spectrum processed through a fast Fourier transform (FFT) using a Hamming window. Comparison with Figure 3, which shows a typical frequency spectrum of a magnetic signal produced by a sonde, illustrates the advantage of the window function which has pushed the overall noise floor down by6dB. Although useful, these window functions introduce distortion and nonlinearity into the process which is undesirable.

It is an aim of the present invention to provide apparatus and a computer program for and a method of improving the sensitivity of a detector for a sonde, to provide more accurate and faster location of a sonde and to provide detection of a sonde at a greater separation between a sonde and a detector, compared with that achievable with currently available detectors. This enhanced sonde recognition is achievable without a significant increase in power consumption by the detector, so that new algorithm does not adversely affect the battery life of the device.

Whereas the maximum distance at which currently available sondes can be detected is around 20m, by using a new algorithm the present invention can achieve a maximum working separation between a sonde and a detector of around 28m, with a drill speed comparable to the drill speed achievable with currently available sondes at a 20m separation. As the strength of the magnetic field produced by a sonde at a distance along the axis of the magnetic field is governed by an inverse cube relationship, such an increase in the maximum working separation is a substantial improvement.

Summarv of the invention A first aspect of the present invention provides a detector for detecting an oscillating magnetic signal produced by a sonde, the detector comprising a plurality of antennas for detecting electromagnetic signals with a plurality of frequencies comprising the magnetic signal produced by the sonde, each antenna having an analogue output; an analogue to digital converter for sampling the analogue output of the antennas to produce a digital output; and digital processing means, for processing the digital output from the analogue to digital converter to isolate the magnetic signal produced by the sonde, wherein the digital processing means comprises a phase feedback loop to allow the digital processing means to follow variations in the frequency of oscillation of the magnetic signal produced by the sonde.

One of the key properties of sonde location is the quality and integrity of the signal processing. A poor SNR leads to unstable vector geometry and a display read-out which is prone to jitter. The invention can improve the clarity of the received signal at intermediate separations, making the readings more reliable and the display read-out less prone to jitter, thereby increasing the drilling speed and providing more efficient hardware deployment. Commercial sondes typically use a clock source with an accuracy of20ppm. Therefore to maintain the maximum SNR the receiver should phase lock to the sonde carrier frequency to follow any drift in frequency of the sonde.

A key variable which governs SNR is the locate bandwidth, as the amount of noise present in a signal is proportional to the square root of bandwidth. A high integrity receiver should have a noise floor determined by the Johnson noise in the antenna windings only. Other noise sources, such as op-amp voltage noise, digital quantisation and spurious magnetic feedback (a consequence of noise currents in the receiver) are unacceptable if they dominate the intrinsic antenna noise.

In an embodiment of the invention, the digital processing means comprises frequency shifting means for shifting the digital output from the analogue to digital converter in the frequency domain so that the frequency spectrum of the magnetic signal produced by the sonde is shifted to DC.

In an embodiment of the invention, the frequency shifting means comprises mixing means, preferably a homodyne mixer.

In an embodiment of the invention, the digital processing means comprises a filter having a frequency response of a since filter, where m 2 2, to produce a since filtered output. Preferably the filter is a finite impulse response (FIR) filter. Preferably the FIR filter comprises N.m taps, having coefficients am,n, where: n+N am n = am-l,k; ken aotn= 0; aO, N m=1; and N = the number of samples in the aperture In an embodiment of the invention, the since filtered output is passed to a downsampler which downsamples every N samples to produce one sample.

In an embodiment of the invention, the digital processing means comprises a low-pass filter to isolate the magnetic signal produced by the sonde. Preferably the bandwidth of the low-pass filter is less than 2.0Hz at -1 30dB, further preferably the bandwidth of the low-pass filter is less than l.OHz at -135dB the bandwidth of the low-pass filter is less than 0. 5Hz at-140dB.

In an embodiment of the invention, the analogue to digital converter comprises a delta- sigma analogue to digital converter.

In an embodiment of the invention, the detector comprises a frequency spectrum analyser having frequency bins and an output, wherein the frequency bins of the frequency spectrum analyser are substantially aligned with the fundamental frequency and a plurality of the harmonic frequencies of the magnetic signal produced by the sonde.

Figure 3 shows a typical frequency spectrum of a magnetic signal produced by a sonde, the signal having a carrier frequency of 8192Hz. The magnetic signal is modulated with a "Pseudo Manchester Bi-phase" 100% amplitude modulation scheme, at 37.5 bits per second, as described below. To recognise the presence of the sonde, a detector should identify this spectrum pattern. Therefore, a detector for detecting the presence of a sonde requires analysis of the signal across a bandwidth which is large (1 OOHz) compared to the bandwidth used to identify the position of the sonde (<lHz), to ensure that energy from data modulation is visible in the side bands around the carrier frequency.

The detector should arbitrate that a signal originates from a sonde and not from a source of interference, e.g. a high order harmonic frequency from a mains electricity supply grid. Furthermore, such interference could appear to be phase correlated, which could erroneously cause the detector to track phase with respect to the supply grid and not the sonde.

The detector analyses the frequency spectrum with a hyperfine resolution of between 0.01Hz and 0. lHz around the carrier and dominant side band frequencies. By aligning the frequency bins of the frequency spectrum analyser with the fundamental frequency and a plurality of the harmonic frequencies of the magnetic signal produced by the sonde, the SNR reading for the fundamental frequency and harmonic frequencies can be maximised for a given signal.

In an embodiment of the invention, the frequency bins follow frequency variation of the magnetic signal produced by the sonde. As stated above the digital processing means comprises a phase feedback loop to allow the digital processing means to follow variations in the frequency of oscillation of the magnetic signal produced by the sonde.

The frequency spectrum analyser is similarly phase locked to the sonde carrier frequency so that the entire bandwidth of the frequency spectrum analyser can 'slide' in the frequency domain. This ensures that the frequency bins of the frequency spectrum analyser are aligned with the fundamental carrier frequency and the harmonic frequencies, leading to enhanced SNR.

In an embodiment of the invention, the output from the spectrum analyser is passed to a matched filter which produces a rank correlation coefficient. The rank correlation coefficient is a measure of the certainty of the sonde being present. Hence the recognition process can be regarded as a matched frequency spectrum correlator.

In an embodiment of the invention, the output from the matched filter is passed to an arbitrator which decides whether the sonde is detectable by the detector or not.

The physical parameters governing the ability to successfully arbitrate are the SNR of the carrier and side band peaks, the phase noise and the frequency resolution.

In an embodiment of the invention, the frequency spectrum analyser comprises a CDFT.

In an embodiment of the invention, one of the frequency bins of the frequency spectrum analyser is aligned within 20mHz of the fundamental frequency of the magnetic signal produced by the sonde, preferably one of the frequency bins of the frequency spectrum analyser is aligned within lOmHz of the fundamental frequency of the magnetic signal produced by the sonde, further preferably one of the frequency bins of the frequency spectrum analyser is aligned within 5mHz of the fundamental frequency of the magnetic signal produced by the sonde.

The output of the frequency spectrum analyser is comprised of measurements of the fundamental frequency and harmonic frequencies of the magnetic signal produced by the sonde and measurements of noise, allowing SNR measurements to be calculated for the fundamental frequency and some of the harmonic frequencies. Inspection of Figure 3 shows the SNR at a nominal 70dB. The modulation scheme, data rate and protocol result in the first side band artefact appearing at 6Hz from the carrier frequency. A figure of 70dB SNR as shown in Figure 3 represents the situation with a sonde and a receiver in close proximity (5m separation). A more challenging scenario arises at a 30m separation, when achieving a 6dB SNR is more difficult.

In an embodiment of the invention, the SNR for at least 4 of the harmonic frequencies is greater than 6dB, preferably the SNR for at least 8 of the harmonic frequencies is greater than 6dB, further preferably the SNR for at least 16 of the harmonic frequencies is greater than 6dB.

According to a further aspect of the invention there is provided a method for detecting an oscillating magnetic signal produced by a sonde, the method comprising detecting electromagnetic signals with a plurality of frequencies comprising the magnetic signal produced by the sonde; sampling the detected signal to produce a digital output; processing the digital output with a digital signal processor to isolate the magnetic signal produced by the sonde; converting the isolated signal to its magnitude and phase measurements; and feeding back the phase of the isolated signal to allow the processor to follow variations in the frequency of oscillation of the magnetic signal produced by the sonde.

Preferably the processing comprises shifting the frequency of the digital output in the frequency domain so that the frequency spectrum of the magnetic signal produced by the sonde is shifted to DC.

Preferably the frequency shifting comprises mixing, preferably homodyne mixing.

Preferably the processing comprises filtering the frequency shifted signal with a filter having a frequency response of a since filter, where m > 2, to produce a since filtered output. Preferably the filter is a finite impulse response filter comprised of N.m taps, having coefficients am n, where: n+N amn = am-lk; k=n ao,n= 0; aO,Nm=1; and N = the number of samples in the aperture Preferably the since filtered output is passed to a downsampler which downsamples every N samples to produce one sample. Preferably the downsampled signal is passed to a low-pass filter to isolate the magnetic signal produced by the sonde.

The present invention can be implemented either in hardware or on software in a general purpose computer. Further the present invention can be implemented in a combination of hardware and software. The present invention can also be implemented by a single processing apparatus or a distributed network of processing apparatuses.

Since the present invention can be implemented by software, the present invention encompasses computer code provided to a general purpose computer or processor on any suitable carrier medium. The carrier medium can comprise any storage medium such as a floppy disk, a CD ROM, a magnetic device or a programmable memory device, or any transient medium such as any signal e.g. an electrical, optical or microwave signal.

Brief description of the drawings

Embodiments of the invention will now be described, by way of example, with reference to the accompanying drawings, in which: Figure 1 is an overview block diagram of a signal isolator module of a detector embodying the invention; Figures 2a, 2b, 2c and 2d are detailed block diagrams of the signal isolator module of Figure 1; Figure 3 is a frequency spectrum of a known modulated magnetic signal produced by a sonde; Figure 4 shows frequency responses of a sinc3 and a sinc5 filter; Figure 5 is a frequency spectrum of a signal which is output from a complex discrete Fourier transform block; and Figure 6 is a frequency spectrum of a signal which is processed through a fast Fourier transform (FFT) using a Hamming window.

Description of preferred embodiment (R74)

Figures 2a to 2d are parts of a Matlab model of the digital signal isolator module embodying the invention which has been simplified in Figure 1. A signal isolator module 1 is provided for each receive antenna. To provide accurate location of a sonde, four receive antennas are deployed. More than four receive antennas can be deployed to provide more accurate geometrical resolution, but this would be more expensive, more power hungry and less responsive due to the additional processing required. The signal isolator modules 1 of the detector simultaneously perform the signal processing of the received electromagnetic signals from the analogue outputs 9 of the associated antennas to produce a signal magnitude measurement 3 and a phase angle measurement 5 for the electromagnetic signal induced by the sonde in each antenna. The signal bandwidths are processed down to one of four selectable outputs (2.0, 1.0, 0.5 and 0.25Hz) as is explained below. The selectivity requirement is 120dB rejection in the stop bands of these filters. The signal magnitude lul 3 and phase angle flu 5 measurements of each of the signal isolator modules 1 are then presented to a geometrical evaluation module (not shown) which determines the relative location of the sonde.

A secondary application for the signal isolator module 1 is to compute a CDFT in a bandwidth of 100Hz around a defined sonde frequency. The requirement is to provide a reliability measurement 7 for the received signal to determine the integrity of the signal magnitude measurement lul 3 and phase angle measurement flu 5.

The processor chosen for the detector is a Texas Instruments TMS320VC33150. The device is a 32-bit floating point DSP and in this embodiment the processor is clocked at its maximum frequency of 75MHz. The DSP serves as a slave processor under the full control of a Texas Instruments MSP430 mixed signal processor. The MSP430 is also responsible for boot-loading the DSP. A half duplex synchronous serial bus is used as the communication link between the two processors.

The signal isolator module 1 functions as a homodyne receiver in the band 50Hz to 40kHz, whereby the entire bandwidth is sampled by a 24-bit stereo coder- decoder (CODEC) 11 with a maximum sampling frequency of 96 kHz. The CODEC should have a Signal to Noise + Distortion Ratio (S / (N+D)) which exceeds 90dB referred to a 96kHz bandwidth. The minimum dynamic range of the converter at this bandwidth should be 100dB. The receiver is phase sensitive. The CODEC data is memory mapped to the external data bus of the TMS320VC33-150 processor.

As stated above, the frequency of oscillation of the magnetic field produced by a sonde is typically between 200Hz and 83KHz. In this embodiment, the magnetic field produced by the sonde oscillates sinusoidally at 8192Hz. This frequency is produced by an oscillator with a stability of better than 20ppm, giving a frequency accuracy of +0.16Hz. The modulation scheme used in this embodiment is 100% 'pseudo Manchester bi-phase", as devised by Fling et al and as described in WO/03012483-A1, with data encrypted at 37.5 bps. This 100% modulation scheme transmits either a pure sine wave of 8192 Hz or nothing, and is known as 'bang-bang' modulation. The frequency spectrum of such a modulated signal is shown in Figure 3. The first side band artefacts appear at +6Hz from the carrier frequency.

Each antenna has a passband between 50Hz and 250kHz, hence the output from the antenna is a broadband signal containing the sonde signal at 8192Hz and a large amount of noise. Referring to Figure 1, an amplified analogue output from the antenna (not shown) is fed into the CODEC 11. The CODEC 11 is based on an oversampling Delta- Sigma ADC which produces a 24-bit output at a sampling rate of 96kHz. A feature of this CODEC 11 is that it incorporates an anti-alias filter for all signal components above a frequency near to the Nyquist sampling frequency, in this case for all frequency components above 48kHz.

The output of the CODEC 11 is split into two paths which are multiplied by a sine and a cosine wave, each of a nominal 8192Hz, in a homodyne mixer 13. The sine and cosine waveforms are generated in a digital phasequadrature oscillator 15, as described below, and the oscillating frequency tracks the carrier frequency of the sonde. Both sine and cosine components are used so that the homodyning process preserves the signal phase, when the signals are recombined downstream.

The effect of this mixing on the frequency spectrum of the signals is to produce a frequency sum component and a frequency difference component. The frequency difference component will move the sonde signal from 8192Hz to DC, so that a narrowband low-pass filter can be used downstream in the device to isolate the sonde signal. The frequency sum component is an unwanted frequency component and when the processed signal is later downsampled, this frequency component aliases in the base-band of the down-sampled bandwidth (200Hz). It is essential to find a decimation parameter which ensures that the aliased frequency sum component falls outside the bandwidth of the selective filter (2.0, 1.0, 0.5 or 0.25Hz) which is applied at a later stage to isolate the signal transmitted by the sonde. Without this constraint the sum term only experiences rejection due to the decimating filter, which is a relatively poor low- pass filter. This in turn would lead to an unstable signal magnitude and phase.

A narrowband filter computed at the ADC sampling frequency, Fs' (96kHz) would require a massive number of filter taps and is not realistic. To produce a realizable filter with a frequency response which is highly selective in a narrow bandwidth, it is necessary to decimate the data rate down to 200Hz. This is achieved by using a since decimator and down sampler 17.

The frequency shifted signals are passed to the since decimator and down sampler 17.

A since decimator is a cascade of m sine filters and is said to have order m. The since decimator is simply a moving average filter with N samples in its buffers at any instant in time.

To achieve the required selectivity either a sinc3 or a sinc5 decimator is chosen, being a cascade of 3 or 5 moving sum averaging filters respectively, as described in our US application no. 10/406597. The choice of which decimator is used will depend on the distance of the detector from the sonde, and this decision is made by the processor. The frequency response of these two filters is shown in Figure 4. The since decimator is chosen as it is a function that, when down-sampled, gives an aliased spectrum which does not interfere with the required frequency response. Figure 4 shows that at certain frequencies (in this case for multiples of 200Hz) the corresponding component of the signal is attenuated. The degree of attenuation increases as the order of the filter increases. When the incoming signal is filtered and then decimated, aliasing causes the side lobes of the filtered signal to fall onto the main lobe. If the decimation parameter (down-sampling ratio), is chosen correctly, the nulls of the frequency response of the since filter are aliased onto the DC component which represents the sonde signal, leaving the sonde signal unaffected. It is an essential requirement that the decimation parameter is equal to the aperture width, N. of the since filter. If this requirement is not met, erroneous aliasing and poor selectivity will occur. In this embodiment the sampling rate, fs, is 96kHz and the aperture width, N. is 480 which gives a decimation parameter of 200Hz to correspond to the null response region of the since filter, as shown in Figure 4.

Since filters may be implemented by the simple recursive difference equation: y(n) = y(n-l) + x(n) - x(n-N) where: y(n) is the current filter output x(n) is the current filter input y(n-l) is the previous value of the filters output x(n-N) is filter input delayed by 'N' samples Evaluation of this difference equation reveals a truncation problem caused by the integrating feedback term yn-l). This eventually leads to integrator wind-up in the sumTning junction. Over an extended period of iterations, this becomes a significant problem and leads to gain instability in the DSP.

To solve this wind-up problem the since decimator is computed as a finite impulse response (FIR) filter, which has a transfer function identical to that of a cascade of m moving average filters, but without the recursive term in the computation. The coefficients of this filter are computed at run-time. This approach is justified by the fact that the TMS320VC33 can perform a floating point multiply-accumulate (MAC) in one instruction cycle, which includes incrementing the pointers for the next filter tap. As a decimating filter the FIR only needs to be calculated at the downsampled data rate (200Hz) although the maintenance of its history buffers has to be performed at the full sample rate Fs.

In this embodiment, the since decimator is computed as a FIR by convolving a rectangle aperture of width N samples on itself m times. This simple algorithm is well suited to computation by the DSP at run- time. The FIR filter comprises N.m taps, and the coefficients are calculated as: n+N am = am-,k; where a0,n= 0 and aO,N.m=1. ken

The output of the since FIR filter is downsampled by factor N=480 to give a signal with a spectrum which aliases into the first lobe of the frequency response of Figure 4. This decimation and downsampling is performed simultaneously for the two homodyned signals of each signal isolator module 1 and the resulant signals are output from the decimator and down sampler 17 to a low-pass filter block 19.

The low-pass filter block 19 is comprised of multi-stage cascaded pairs of filters and down-samplers, as shown on the bottom portion of Figure 2. The number of FIR low- pass filter stages used governs the selectivity of the system and also the response time or inherent lag as seen at the output. The first low- pass filter pair have a 2.0Hz bandwidth.

If greater selectivity is required the two filtered signals are then passed to a second down-sampler which down-samples by a factor of 5 and the down-sampled signals are then filtered by a low-pass filter of l.OHz bandwidth. If further selectivity is required there are two further downsampling and filter stages which each down-sample by a factor of 2 and low-pass filter to produce an output of 0.5Hz and 0.25Hz bandwidth respectively. (Rrr)

The filter coefficients are calculated in Mathcad. using the Remez exchange algorithm and are stored in the DSP 'constant' memory segment. The Remez function implements the Parks-McClellan algorithm, which uses the Remez exchange algorithm and Chebyshev approximation theory to design filters with optimal fits between the desired and actual frequency responses. The filters are optimal in the sense that they minimize the maximum error between the desired frequency response and the actual frequency response. Filters designed in this way exhibit an equiripple behaviour in their frequency response. The Parks-McClellan FIR filter design algorithm accepts filter specifications in terms of passband and stopband frequencies, passband ripple, and stopband attenuation. The MSP430 microcontroller controls the degree of filtering.

The two outputs from the low-pass filter block 19, I and Q. are combined in the complex signal combiner 35 to produce magnitude 1ul and phase flu information in accordance with the following equations: IUI = [Q] Inverse tangent is computed in the first octant and then by applying some simple transformations the remaining angles can be computed to 360 . The transformations are dependent on the relative magnitudes and signs of I and Q. The inverse tangent is itself computed using a truncated polynomial approximation to the exact function. The approximation gives a function which is considerably more efficient than the standard C library function whilst being accurate to +0.1 . The phase information flu for each channel is cast to a 16-bit unsigned integer such that 2n - 65536 and is passed to a phase control feedback loop 21 and the geometrical evaluation module (not shown).

A call to the floating point SQRT function in the standard C library 'math.h' is adequate for magnitude calculation. The magnitude information lul remains in 32-bit floating point format (EKE standard 754) and ispassed to the geometrical evaluation module.

The magnitude lul 5 and phase information flu 3 for each channel is transmitted to the micro-controller at 200Hz (the down-sampled data rate).

The phase control feedback loop 21 serves to match the DSP local oscillator 15 (phase- quadrature synthesiser) to the exact sonde frequency. The requirement is to trim the frequency error between the signal isolator 1 and the sonde frequency within a 250ppm tolerance band.

The frequency error is computed in the differentiating filter 23 as the time derivative of phase (d Veldt) computed in an 8 Hz bandwidth. It is necessary to cast the phase to a 16-bit integer in order to map the natural wrap-around at 2n to a continuous function.

The phase error is processed through a simple proportional and integral (PI) controller and the output is used to adjust the seed values in the phase quadrature (PQ) oscillator 15. The PI coefficients are set for dominant integral control law and the loop dynamics are tuned for critical damping. The loop dynamics are dominated by the step response functions of both the decimator and down-sampler 17 and the FIR low-pass filter 19.

Stable control of the feedback loop 21 is achieved by asserting an integral gain of 2/ and a proportional gain of 1/ in the error feedback. This control law leads to a servo mechanism which can 'lock-in' to a negligible error with a response time of 800ms.

The absolute accuracy of the frequency tracking is governed by the resolution of the 32- bit floating point representation - this is typically between 5 and 6 decimal places.

The phase control servo mechanism operates under the control of the MSP430 micro- controller which controls the DSP to either track, hold-track or return to the nominal sonde frequency.

The DSP reports signal magnitude to the MSP430 at a data rate of 6Hz. The choice of the applicable bandwidth is the domain of the MSP430. The MSP430 is also responsible for arbitrating the phase reference channel, this decision being based on the largest signal magnitude. A 6dB hysteresis band is used to avoid rapid channel hopping.

The signals which are output from the decimator and down-sampler 17 are passed to the low-pass filter 19, as described above, and they are also passed to a lOOHz low-pass filter 27 before being passed to a complex discrete Fourier transform (CDFT) block 29.

The idea of the CDFT 29 is to provide a frequency spectrum analyser in a bandwidth of lOOHz around the nominal sonde frequency in order to verify that the signal which has been isolated by the low-pass filter block 19 originates from the sonde.

As the DSP has already resolved the signal into its in-phase and quadrature components these act as the real and imaginary inputs to the CDFT. The CDFT can be computed at a variable sample rate of between 50Hz and lKHz, allowing a frequency resolution down to 0.01Hz, but typically a frequency resolution (frequency bin spacing) of O. l Hz is used. The frequency bin width is determined by: fain N.DEC where Fs is the ADC sample rate (96kHz) DEC is the decimation ratio (480) N is the number of samples in the CDFT Tuning the frequency bin width to be an exact multiple of the Sonde data rate ensures the sidebands also align with the CDFT. The sample rate and corresponding frequency resolution are determined by the processor on a case-by-case basis based on pre- programmed scenarios.

As the PQ oscillator is servoed onto the sonde carrier frequency, the CDFT is similarly phase locked to the sonde carrier frequency so that the entire bandwidth of the CDFT can 'slide' in the frequency domain. This ensures that the frequency bins of the CDFT are aligned with the fundamental carrier frequency and the harmonic frequencies, leading to enhanced SNR. Put another way the zeros due to the CDFT modulation are coincident with the sonde harmonics. This gives superior performance to an approach using standard window functions. The degree of alignment between the fundamental carrier frequency and the corresponding frequency bin of the frequency spectrum analyser is less than 20mHz, preferably less than lOmHz and further preferably less than 5mHz.

Figure 5 demonstrates a 20dB improvement with respect to the prior art method, where the signal is processed through an FFT using a Hamming window, as shown in Figure 6.

The output of the CDFT block 29 is passed to a matched filter 31 which determines whether the known sonde spectrum is present in the received electromagnetic signal by looking at the energy in the fundamental carrier frequency and in each harmonic frequency which is identifiable by an SNR of at least 6dB. The sonde spectrum is correlated against the matched filter 31 as a function of the modulation scheme, data rate and encryption method. The 'sonde present' decision is arbitrated in the arbitrator 33 from a rank correlation coefficient. As a minimum, 5 or 6 harmonic frequencies should have an SNR of at least 6dB to allow the arbitrator to recognise the sonde. Due to this enhanced SNR of the fundamental and harmonic frequencies, a positive 'sonde present' decision can result at increased separations.

Various modifications will be apparent to those in the art and it is desired to include all such modifications as fall within the scope of the accompanying claims.

Claims

CLAIMS: 1. A detector for detecting an oscillating magnetic signal

produced by a sonde, the detector comprising: a plurality of antennas for detecting electromagnetic signals with a plurality of frequencies comprising the magnetic signal produced by the sonde, each antenna having an analogue output; an analogue to digital converter for sampling the analogue output of the antennas to produce a digital output; and digital processing means, for processing the digital output from the analogue to digital converter to isolate the magnetic signal produced by the sonde, wherein the digital processing means comprises a phase feedback loop to allow the digital processing means to follow variations in the frequency of oscillation of the magnetic signal produced by the sonde.
2. A detector as claimed in claim 1, wherein the digital processing means comprises frequency shifting means for shifting the digital output from the analogue to digital converter in the frequency domain so that the frequency spectrum of the magnetic signal produced by the sonde is shifted to DC.
3. A detector as claimed in claim 2, wherein the frequency shifting means comprises mlxmg means.
4. A detector as claimed in claim 3, wherein the mixing means comprises a homodyne mixer.
5. A detector as claimed in any one of claims 1 to 4, wherein the digital processing means comprises a filter having a frequency response of a since filter, where m 2 2, to produce a since filtered output.
6. A detector as claimed in claim 5, wherein the filter is a finite impulse response filter.
7. A detector as claimed in claim 6, wherein the finite impulse response filter comprises N.m taps, having coefficients amp, where: n+N amn = am-l, k; k=n ao,n= 0; ae,Nm=1; and N = the number of samples in the aperture
8. A detector as claimed in any one of claims 5, 6 or 7, wherein the since filtered output is passed to a downsampler which downsamples every N samples to produce one sample.
9. A detector as claimed in any one of the preceding claims, wherein the digital processing means comprises a low-pass filter to isolate the magnetic signal produced by the sonde.
10. A detector as claimed in claim 9, wherein the bandwidth of the lowpass filter is less than 2.0Hz at -130dB.
11. A detector as claimed in any one of claims 9 or 10, wherein the bandwidth of the low-pass filter is less than l.OHz at -135dB.
12. A detector as claimed in any one of claims 9, 10 or 11, wherein the bandwidth of the low-pass filter is less than 0.5Hz at -140dB.
13. A detector as claimed in any one of the preceding claims, wherein the analogue to digital converter comprises a delta-sigma analogue to digital converter.
14. A detector as claimed in any one of the preceding claims, the detector comprising a frequency spectrum analyser having frequency bins and an output, wherein the frequency bins of the frequency spectrum analyser are substantially aligned with the fundamental frequency and a plurality of the harmonic frequencies of the magnetic signal produced by the sonde.
15. A detector as claimed in claim 14, wherein the frequency bins are adapted to follow frequency variation of the magnetic signal produced by the sonde.
16. A detector as claimed in any one of claims 14 or 15, wherein the output from the spectrum analyser is passed to a matched filter which produces a rank correlation coefficient.
17. A detector as claimed in claim 16, wherein the output from the matched filter is passed to an arbitrator which decides whether the sonde is detectable by the detector or not.
18. A detector as claimed in any one of claims 14 to 17, wherein the frequency spectrum analyser comprises a complex discrete Fourier transform.
19. A detector as claimed in any one of claims 14 to 18, wherein one of the frequency bins of the frequency spectrum analyser is aligned within 20mHz of the fundamental frequency of the magnetic signal produced by the sonde.
20. A detector as claimed in any one of claims 14 to 19, wherein one of the frequency bins of the frequency spectrum analyser is aligned within 1 OmHz of the fundamental frequency of the magnetic signal produced by the sonde.
21. A detector as claimed in any one of claims 14 to 20, wherein one of the frequency bins of the frequency spectrum analyser is aligned within 5mHz of the fundamental frequency of the magnetic signal produced by the sonde
22. A detector as claimed in any one of claims 14 to 21, the output of the frequency spectrum analyser being comprised of measurements of the fundamental frequency and harmonic frequencies of the magnetic signal produced by the sonde and measurements of noise, allowing a signal to noise ratio to be calculated for the fundamental frequency and harmonic frequencies, wherein the signal to noise ratio for at least 4 of the harmonic frequencies is greater than 6dB.
23. A detector as claimed in any one of claims 14 to 22, wherein the signal to noise ratio for at least 8 of the harmonic frequencies is greater than 6dB.
24. A detector as claimed in any one of claims 14 to 23, wherein the signal to noise ratio for at least 16 of the harmonic frequencies is greater than 6dB.
25. A method for detecting an oscillating magnetic signal produced by a sonde, the method comprising: detecting electromagnetic signals with a plurality of frequencies comprising the magnetic signal produced by the sonde; sampling the detected signal to produce a digital output; processing the digital output with a digital signal processor to isolate the magnetic signal produced by the sonde; converting the isolated signal to its magnitude and phase measurements; and feeding back the phase of the isolated signal to allow the processor to follow variations in the frequency of oscillation of the magnetic signal produced by the sonde.
26. A method as claimed in claim 25, wherein the processing comprises shifting the frequency of the digital output in the frequency domain so that the frequency spectrum of the magnetic signal produced by the sonde is shifted to DC.
27. A method as claimed in claim 26, wherein the frequency shifting comprises mixing.
28. A method as claimed in claim 27, wherein the mixing comprises homodyne mlxmg.
29. A method as claimed in any one of claims 25 to 28, wherein the processing comprises filtering the frequency shifted signal with a filter having a frequency response of a since filter, where m 2 2, to produce a since filtered output.
30. A method as claimed in claim 29, wherein the filter is a finite impulse response filter.
31. A method as claimed in claim 30, wherein the finite impulse response filter comprises N.m taps, having coefficients arn,n, where: n+N amn = aml,k; k=n a0,n = 0; aO,Nm=l; and N = the number of samples in the aperture
32. A method as claimed in any one of claims 29, 30 or 31, wherein the since filtered output is passed to a downsampler which downsamples every N samples to produce one sample.
33. A method as claimed in claim 32, wherein the downsampled signal is passed to a low-pass filter to isolate the magnetic signal produced by the sonde.
34. A method as claimed in claim 33, wherein the bandwidth of the lowpass filter is less than 2.0Hz at-130dB.
35. A method as claimed in any one of claims 33 or 34, wherein the bandwidth of the low-pass filter is less than l.OHz at -135dB.
36. A method as claimed in any one of claims 33, 34 or 35, wherein the bandwidth of the low-pass filter is less than 0.5Hz at -140dB.
37. A method as claimed in any one of claims 25 to 36, wherein the sampling is performed by a delta-sigma analogue to digital converter.
38. A method as claimed in any one of claims 25 to 37, the magnetic signal having a fundamental frequency and harmonic frequencies, the method comprising analysing the frequency spectrum of the magnetic signal with a frequency spectrum analyser having frequency bins and an output, wherein the frequency bins of the frequency spectrum analyser are substantially aligned with the fundamental frequency and a plurality of the harmonic frequencies of the magnetic signal produced by the sonde.
39. A method as claimed in claim 38, wherein the frequency bins follow frequency variation of the magnetic signal produced by the sonde.
40. A method as claimed in any one of claims 38 or 39, wherein the output from the spectrum analyser is passed to a matched filter which produces a rank correlation coefficient.
41. A method as claimed in claim 40, wherein the output from the matched filter is passed to an arbitrator which decides whether the sonde is detectable by the detector or not.
42. A method as claimed in any one of claims 38 to 41, wherein the frequency spectrum analyser comprises a complex discrete Fourier transform.
43. A method as claimed in any one of claims 38 to 42, wherein one of the frequency bins of the frequency spectrum analyser is aligned within 20mHz of the fundamental frequency of the magnetic signal produced by the sonde.
44. A method as claimed in any one of claims 38 to 43, wherein one of the frequency bins of the frequency spectrum analyser is aligned within lOmHz of the fundamental frequency of the magnetic signal produced by the sonde.
45. A method as claimed in any one of claims 38 to 44, wherein one of the frequency bins of the frequency spectrum analyser is aligned within 5mHz of the fundamental frequency of the magnetic signal produced by the sonde.
46. A method as claimed in any one of claims 38 to 45, the output of the frequency spectrum analyser being comprised of measurements of the fundamental frequency and hannonic frequencies of the magnetic signal produced by the sonde and measurements of noise, the method comprising measuring the signal to noise ratio of the fundamental frequency and harmonic frequencies, wherein the signal to noise ratio for at least 4 of the harmonic frequencies is greater than 6dB.
47. A method as claimed in any one of claims 38 to 46, wherein the signal to noise ratio for at least 8 of the harmonic frequencies is greater than 6dB.
48. A method as claimed in any one of claims 38 to 47, wherein the signal to noise ratio for at least 16 of the harmonic frequencies is greater than 6dB.
49. A computer program for detecting an oscillating magnetic signal produced by a sonde in an electromagnetic signal with a plurality of frequencies comprising the magnetic signal produced by the sonde, the computer program comprising: sampling the electromagnetic signal to produce a digital output; processing the digital output to isolate the magnetic signal produced by the sonde; converting the isolated signal to its magnitude and phase measurements; and feeding back the phase of the isolated signal to allow the processing to follow variations in the frequency of oscillation of the magnetic signal produced by the sonde.
50. A computer program as claimed in claim 49, wherein the processing comprises shifting the frequency of the digital output in the frequency domain so that the frequency spectrum of the magnetic signal produced by the sonde is shifted to DC.
51. A computer program as claimed in claim 50, wherein the frequency shifting comprises mixing.
52. A computer program as claimed in claim 51, wherein the mixing comprises homodyne mixing.
53. A computer program as claimed in any one of claims 49 to 52, wherein the processing comprises filtering the frequency shifted signal with a filter having a frequency response of a since filter, where m 2 2, to produce a since filtered output.
54. A computer program as claimed in claim 53, wherein the filter is a finite impulse response filter.
55. A computer program as claimed in claim 54, wherein the finite impulse response filter comprises N.m taps, having coefficients a,n n, where: n+N amn = am-lk; k=n ao,n= 0; aO N.m=1; and N = the number of samples in the aperture
56. A computer program as claimed in any one of claims 53, 54 or 55, wherein the since filtered output is passed to a downsampler which downsamples every N samples to produce one sample.
57. A computer program as claimed in claim 56, wherein the downsampled signal is passed to a low-pass filter to isolate the magnetic signal produced by the sonde.
58. A computer program as claimed in claim 57, wherein the bandwidth of the low- pass filter is less than 2.0Hz at -130dB.
59. A computer program as claimed in any one of claims 57 or 58, wherein the bandwidth of the low-pass filter is less than l.OHz at -135dB.
60. A computer program as claimed in any one of claims 57, 58 or 59, wherein the bandwidth of the low-pass filter is less than 0.5Hz at -140dB.
61. A computer program as claimed in any one of claims 49 to 60, wherein the sampling is performed by a delta-sigma analogue to digital converter.
62. A computer program as claimed in any one of claims 49 to 61, the magnetic signal having a fundamental frequency and harmonic frequencies, the computer program comprising analysing the frequency spectrum of the magnetic signal with a frequency spectrum analyser having frequency bins and an output, wherein the frequency bins of the frequency spectrum analyser are substantially aligned with the fundaTnenta1 frequency and a plurality of the harmonic frequencies of the magnetic signal produced by the sonde.
63. A computer program as claimed in claim 62, wherein the frequency bins follow frequency variation of the magnetic signal produced by the sonde.
64. A computer program as claimed in any one of claims 62 or 63, wherein the output from the spectrum analyser is passed to a matched filter which produces a rank correlation coefficient.
65. A computer program as claimed in claim 64, wherein the output from the matched filter is passed to an arbitrator which decides whether the sonde is detectable by the detector or not.
66. A computer program as claimed in any one of claims 62 to 65, wherein the frequency spectrum analyser comprises a complex discrete Fourier transform.
67. A computer program as claimed in any one of claims 62 to 66, wherein one of the frequency bins of the frequency spectrum analyser is aligned within 20mHz of the fundamental frequency of the magnetic signal produced by the sonde.
68. A computer program as claimed in any one of claims 62 to 67, wherein one of the frequency bins of the frequency spectrum analyser is aligned within lOmHz of the fundamental frequency of the magnetic signal produced by the sonde.
69. A computer program as claimed in any one of claims 62 to 68, wherein one of the frequency bins of the frequency spectrum analyser is aligned within 5mHz of the fundamental frequency of the magnetic signal produced by the sonde.
70. A computer program as claimed in any one of claims 62 to 69, the output of the frequency spectrum analyser being comprised of measurements of the fundamental frequency and harmonic frequencies of the magnetic signal produced by the sonde and measurements of noise, the method comprising measuring the signal to noise ratio of the fundamental frequency and harmonic frequencies, wherein the signal to noise ratio for at least 4 of the harmonic frequencies is greater than 6dB.
71. A computer program as claimed in any one of claims 62 to 70, wherein the signal to noise ratio for at least 8 of the harmonic frequencies is greater than 6dB.
72. A computer program as claimed in any one of claims 62 to 71, wherein the signal to noise ratio for at least 16 of the harmonic frequencies is greater than 6dB.
73. A detector for detecting an oscillating magnetic signal produced by a sonde, substantially as hereinbefore described with reference to Figures 1 to 6 of the accompanying drawings.
74. A method of detecting an oscillating magnetic signal produced by a sonde, substantially as hereinbefore described with reference to Figures 1 to 6 of the accompanying drawings.