NMR RECEIVER WITH SIGMA-DELTA A/D CONVERTER

TECHNICAL FIELD

The present invention relates in general to NMR ₁ 0 circuits and in particular to an NMR receiver that utilizes an analog-to-digital converter with a Sigma-Delta modulator.

BACKGROUND OF THE INVENTION

15

Nuclear magnetic resonance (NMR) spectroscopy is old and well known in the art. The signal receiver for processing the NMR signals generated during the NMR process operates in the RF frequency range. RF signals

20 are mixed with a local oscillator input to obtain an intermediate frequency which is filtered and amplified. That signal is mixed in two channels with a reference frequency, generally a 9 MHz signal, in one channel and the reference frequency shifted 90° in the other channel.

²⁵ In each of these channels, the signal is filtered, amplified and coupled to an analog-to-digital converter which includes a Nyquist or Delta modulator sampling circuit. Anti-aliasing filters must be utilized at the input to the A/D converters. The sample of the signal can '<* 30 be expressed in the frequency domain as the summation of the original signal component and signals that are <* ^■ frequency modulated by integer multiples of the sampling frequency; therefore, input signals above the Nyquist frequency, f , cannot be properly converted and create

new signals in the baseband which were not present in the original signal. This non-linear phenomenon is a signal distortion which is frequently referred to as aliasing. To compensate for this distortion, a lowpass filter, called the anti-aliasing filter, must be used and must have a flat response over the frequency band of interest (the baseband) and attenuate the frequencies above the Nyquist frequency enough to suppress them below the noise floor. Also, the non-linear phase distortion caused by the anti-aliasing filter may create harmonic distortion and degradation. Since the analog anti-aliasing filter is the limiting factor in controlling the bandwidth and phase distortion of the input signal, a high-performance anti-aliasing filter is required to obtain high resolution and minimum distortion. The output of the analog-to-digital converter in each channel is coupled to a digital signal processor which performs a complex Fast Fourier Transform on the signals and provides the zero order and first order phase corrections. At best, the prior art circuit has several disadvantages. First, the anti-aliasing analog filters are very complex. They must be able to achieve a narrow bandwidth while simultaneously being adjustable for a wide frequency range, usually 0-10 Kz. The prior art filters that have these stringent characteristics have a side effect of phase distortion within the operating frequency that must be corrected by using first order phase correction contained within the digital signal processing device. Another side effect is that the filters have difficulty in achieving equal amplification throughout the frequency range. Since there are two channels that have to be processed contemporaneously, any differences in amplification will cause inconsistent results while operating in different frequencies.

The present invention overcomes the disadvantages of the prior art. By utilizing a Sigma-Delta modulator in the analog-to-digital converter device, the problems with the anti-aliasing analog filter are resolved. In addition, the signal-to-noise ratio is improved.

Utilizing the Sigma-Delta modulation system, the input signal is oversampled and the system is able to use a much simpler analog anti-aliasing filter that has a wider bandwidth. Such a filter can be a simple RC filter. These characteristics allow for choosing identical analog filters in the two channels which do not induce phase distortion of the signal within the operating frequency. This eliminates the need for first order phase correction by the digital signal processor. Furthermore, there is no requirement to perform an adjustment in the analog filters since the correction is accomplished by the digital decimation filter contained within the analog-to-digital converter device and within the DSP by an adjustable digital decimation filter. This combination allows the achievement of an adjustable frequency sampling rate. This digital decimation filter gives the consistency of amplification that was lacking in conventional analog filters. Since the quantization noise generated by the sampling is assumed to be random, the differentiator in the Sigma-Delta modulator doubles the power of quantized noise. However, the error is pushed toward the high frequencies due to the differentiator factor. Therefore, provided that the analog input signal to the modulator, x(t) , is oversampled, the high frequency quantization noise can be removed by digital lowpass filters without affecting the input signal characteristics residing in the baseband. This lowpass digital filtering is part of the decimation process. After the digital

deci ation filtering process, the output signal has only the frequency components from 0 Hz to f(B) . In comparing the performance of Sigma-Delta modulators to conventional 1-bit Nyquist samplers and Delta modulation type oversamplers, the baseband (up to f(B)) noise of the Sigma-Delta converters is much smaller than Nyquist samplers or Delta modulators.

Thus it is an object of the present invention to provide an NMR receiver that utilizes a Sigma-Delta modulator in which the input signal from the anti-aliasing filter is sampled at a rate, f , where f is much greater than the input signal f , or f » f , to suppress quantization noise in the baseband and to generate quantization noise in the high frequency out-of-band frequency range.

It is also an object of the present invention to provide an NMR receiver which utilizes a simple RC anti-aliasing filter.

It is still another object of the present invention to provide an NMR receiver that utilizes a Sigma-Delta modulator for oversampling the input signal at a rate much greater than the Nyquist sampling rate.

It is a further object of the present invention to provide an NMR receiver having an analog-to-digital converter in which a digital decimation filter has a lowpass digital filter to remove the high frequency quantization noise from the oversampled signal without affecting the received signal characteristics in the baseband. It is another object of the present invention to provide an adjustable digital decimation filter in the digital signal processor to allow an adjustable frequency sampling rate.

SϋMMARY OF THE INVENTION

Thus the present invention relates to a circuit for receiving and processing an NMR analog signal, x(t) , comprising a resistance-capacitance anti-aliasing filter for receiving the NMR signal, x(t) , and passing a wide band frequency output signal, f ; a Sigma-Delta modulator for receiving the wide band frequency output signal, suppressing quantization noise in the baseband, filtering out high frequency quantization noise, and generating an output signal substantially free of first order quantization noise in the baseband; and a digital signal processor for receiving said output signal, providing an adjustable frequency sampling rate with an adjustable digital decimation filter, and performing a complex Fast Fourier Transform on said output signal with a zero order phase correction to provide a usable NMR signal.

BRIEF DESCRIPTION OF THE DRAWINGS

These and other objects of the present invention will be more fully understood in conjunction with the accompanying drawings in which like numbers indicate like components and in which:

FIG. 1 is a schematic block diagram of a conventional NMR receiver and analog-to-digital converter;

FIG. 2 is a schematic block diagram of a Sigma-Delta modulator;

FIG. 3 is a graph of the spectrum of the first order Sigma-Delta modulator;

FIG. 4 is a schematic block diagram of a modified NMR receiver and Sigma-Delta A/D converter; and

FIG. 5 is a schematic representation of a simple RC anti-aliasing filter.

DETAILED DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of the conventional prior art NMR receiver and analog-to-digital converter configuration. The RF signal containing the NMR data is coupled on line 12 through an RF preamplifier 14. The output is coupled to a mixer 15 where the signal is mixed with the local oscillator input on line 16. The IF frequency is filtered at 18, amplified at 20 and coupled on lines 22 and 24 to the first and second channels for processing. The two channels are identical except for the inputs. In the first channel, a reference signal, typically 9 MHz, is mixed with the IF signal on line 22 at mixer 24. In the second channel, the reference signal is shifted 90° at 30 and then mixed at 28 with the signal on line 24. The remainder of the two channels are identical and thus only one of them will be discussed. The output from mixer 24 in channel 1 is coupled to a filter 32 and amplified at 34. The amplified signal is coupled to a complex anti-aliasing adjustment filter 36 and then to a modulator which may be a Nyquist sampler and A/D converter 38. The output on line 40 is coupled to a digital signal processor 42 which does a complex Fast Fourier Transform on the signal and also provides zero order and first order phase correction. This conventional NMR receiver 10 with the A/D converters and the Nyquist sampler 38 have technical disadvantages. The complex anti-aliasing filters 36 are analog filters. These filters must be able

to achieve a narrow bandwidth while simultaneously being adjustable for a wide frequency range, usually 0-10 kHz. Nyquist rate A/D converters, such as that in block 38 in channel 1 and channel 2, sample analog signals which have maximum frequencies slightly less than the Nyquist frequency, f = f_/2, where f is the sampling frequency. The act of sampling is equivalent to modulating the input signal by carrier signals having frequencies at 0, f , 2f , . In other words, the sampled signal can be expressed in the frequency domain as the summation of the original signal component and signals frequency modulated by integer multiples of the sampling frequency. Thus, input signals above the Nyquist frequency, f.., cannot be properly converted and create new signals in the baseband which were not present in the original signal. This non-linear phenomenon is a signal distortion frequently referred to as "aliasing." The distortion can be prevented only by properly lowpass filtering the input signal up to the Nyquist frequency. This lowpass filter is known in the art as an anti-aliasing filter and it must have a flat response over the frequency band of interest (baseband) . It must also attenuate the frequencies above the Nyquist frequency sufficient to reduce them under the noise level. Also, the non-linear phase distortion caused by the anti-aliasing filter may create harmonic distortion in audible degradation of the signal. Since the analog anti-aliasing filter is the limiting factor in controlling the bandwidth and phase distortion of the input signal, a high-performance anti-aliasing filter is required to obtain high resolution and minimum distortion.

The prior art ilters that have these stringent characteristics have a side effect of phase distortion within the operating frequency that must be corrected by using irst order phase correction contained within the digital signal processing software. Another side effect is that the filters have difficulty in achieving equal amplification throughout the frequency range. These differences in amplification will cause inconsistent results when operating in different frequencies. By utilizing the Sigma-Delta modulation method implemented in the analog-to-digital converter device, the problems of the prior art can be resolved and the signal-to-noise ratio improved. It is well known that Sigma-Delta modulators oversample the signal and by oversampling the signal, a much simpler analog anti-aliasing filter can be used that has a wider bandwidth. These characteristics allow for choosing identical analog filters which do not induce phase distortion of the signal within the operating frequency. This eliminates the need for first order phase correction by the digital signal processor 42. Furthermore, it does not require any adjustment in the analog filters since the correction is done by the digital decimation filter contained within the analog-to-digital converter device. This digital decimation filter gives the consistency of amplification that was lacking in the conventional analog filters. Since the quantization noise is assumed to be random, the differentiator in the Sigma-Delta modulator doubles the power of quantized noise. However, the error is pushed toward high frequencies due to the differentiation factor. Therefore, provided that the analog input signal to the modulator, x(t) , is oversampled, the high frequency quantization noise can be

removed by digital lowpass filters without affecting the input signal characteristics residing in the baseband. This lowpass filtering is part of the decimation process. It is well known in the art that an oversampling converter generally uses a sequence of coarsely quantized data at the input oversampling rate, F , where F =

Nf followed by a digital-domain decimation process to compute a more precise estimate for the analog input at the lower output sampling rate, fs, which is the same as used by the Nyquist samplers. Oversampling has immediate benefits for the anti-aliasing filter. Consider a typical digital audio application using a Nyquist sampler as set forth on page 9 of the Motorola document entitled Principles of Delta-Sigma Modulation for A/D Converters and dated April 8, 1990. The data samples from the

Nyquist-rate converters are taken at a rate at least twice the highest signal frequency of interest. For example, a 48 kHz sampling rate allows signals up to 24 kHz to pass without aliasing but because of practical circuit limitation, the highest frequency that passes is actually about 22 kHz. Also, the anti-aliasing filter in the Nyquist A/D converters requires a flat response with no phase distortion over the frequency band of interest (such as 22 kHz in digital audio applications) . To prevent signal distortion due to aliasing, all signals above 24 kHz for a 48 kHz sampling rate must be attenuated by at least 96 dB for 16 bits of dynamic resolution. These requirements are difficult to meet with an analog lowpass filter. Now compare the Nyquist sampler with an oversampling approach twice that of the first case. In this case the same audio signal is sampled at 2f , or 96 kHz. The in-band quantization noise is moved out-of-band and the

anti-aliasing filter needs only to eliminate signals above 74 kHz. The filter has a flat response up to 22 kHz. Thus this is a simple filter that is easy to build because the transition band can be 52 kHz (22 kHz to 74 kHz) to reach the -96 dB point. Since the final sampling rate is 48 kHz, a sample rate reduction filter, commonly called a decimation filter, is required but it is implemented in the digital domain as opposed to anti-aliasing filters which are implemented with analog circuitry. Therefore, provided that the analog input signal to the modulator, x(t) , is oversampled, the high frequency quantization noise can be removed by digital lowpass filters without affecting the input signal characteristics residing in the baseband. This lowpass filtering is part of the decimation process.

Thus with the use of the Sigma-Delta modulator, an oversampled converter, considerably simpler anti-aliasing filters can be used than with Nyquist rate converters with similar performance because the complexity of the filter is a strong function of the ratio of the width of the transition band to the width of the pass band. For example, with N = 64, a simple RC lowpass filter such as shown in FIG. 5 can be used at the converter analog input and is often sufficient. A typical prior art Sigma-Delta modulator is shown in block diagram form in FIG. 2. It is called a "Sigma-Delta" modulator because the integrator (Sigma) is placed in front of the Delta modulator. The quantization noise characteristic (noise performance) on such a device is frequency-dependent in contrast to Delta modulation. The input signal on line 46 is mixed at 44 with an error feedback signal on line 48 and the difference coupled to integrator 50. The output of the integrator 50 is coupled

to a quantizer or comparator 52 whose output is coupled to sample and hold circuit 54 where the high sampling rate takes place. The output of the sample and hold circuit 54 on line 56 is coupled back on line 48 as the feedback signal and is also coupled to the digital decimation filter 57. It is well known that quantization noise is added to the signal by the quantizer or comparator 52. As the loop integrates the error between the sampled signal 48 and the input signal 46, it lowpass filters the signal and highpass filters the noise. In other words, the signal is left unchanged as long as its frequency content doesn't exceed the filter's cutoff frequency, but the Sigma-Delta loop pushes the noise into a higher frequency band. It is well known that grossly oversampling the input signal causes the quantization noise to spread over a wide bandwidth and the noise density in the bandwidth of interest, the baseband, to significantly decrease.

It is well known also that the output of the Sigma-Delta modulator illustrated in FIG. 2 is coupled to the digital decimator 57 which will be discussed hereafter in relation to the present invention as illustrated in FIG. 4. Filtering noise which could be aliased back into the baseband is the primary purpose of the digital filtering stage 57 or the digital decimation filtering. Its secondary purpose is to take the 1-bit data stream that has a high sample rate and transform it into a 16-bit data stream at a lower rate. This process is known as decimation and is both an averaging filter function and a rate reduction function performed simultaneously. The output of the modulator shown in FIG. 2 is a coarse quantization of the analog input. However, the modulator is oversampled by as much as 64 times higher than the Nyquist rate. High resolution is achieved by averaging

over 64 data points to interpolate between the coarse quantization levels of the modulator. The process of averaging is equivalent to lowpass filtering in the frequency domain. With the high frequency components of the quantization noise removed, the output sampling rate can be reduced to the Nyquist rate by decimation without aliasing noise into the baseband.

There are three basic tasks that are performed in the digital filter sections. First is the removal of the shaped quantization noise. The Sigma-Delta modulator is designed to suppress quantization noise in the baseband. Thus most of the quantization noise is at frequencies above the baseband. The main objective of the digital filter is to remove this out-of-band quantization noise. This leaves a small amount of baseband quantization noise and the band limited input signal component. Reducing the baseband quantization noise is equivalent to increasing the effect of resolution of the digital output.

Secondly, the output of the Sigma-Delta modulator is at a very high sampling rate. This is a fundamental characteristic of Sigma-Delta modulators because they use the high frequency portion of the spectrum in which to place the bulk of the quantization noise. After the high • frequency quantization noise is filtered out, it is possible to reduce the sampling rate. It is desirable to bring the sampling rate down to the Nyquist rate which minimizes the amount of information for subsequent transmission, storage or digital signal processing. Finally, in practice, the input signals are seldom completely band limited. Since the modulator is sampling much higher than the output Nyquist rate, the analog anti-aliasing filter before the modulator can roll off gradually. When the digital processor reduces the

sampling rate down to the Nyquist rate, it needs to provide the necessary additional aliasing rejection for the input signal as opposed to the internally generated quantization noise. Thus, after the digital decimation filtering processes, the output signal has only the frequency components from 0 Hz to f(B). The performance of the Sigma-Delta modulator can be compared to the conventional 1-bit Nyquist samplers and the Delta modulation type oversamplers. FIG. 3 shows the baseband (up to f(B)) noise of the Sigma-Delta converters to be much smaller than Nyquist samplers or Delta modulators. However, for the first order modulator discussed, the baseband noise cannot reach below the -96 dB signal-to-noise ratio needed for 16-bit A/D converters. The novel modified NMR receiver and Sigma-Delta A/D converter 60 is illustrated in FIG. 4. Again, the RF signal containing the pertinent NMR data on line 58 is amplified by RF preamplifier 61 and coupled to mixer 62. Local oscillator input on line 64 is mixed with the amplified NMR signal and coupled to filter 66 and IF amplifier 68. Again, the output of the IF amplifier 68 is coupled to two channels on lines 70 and 72 respectively. The signal on line 70 is coupled to mixer 74 where the 9 MHz reference signal on line 76 is also received and the output coupled to filter 82. In channel 2, the signal on line 72 is mixed in mixer 78 with the 9 MHz reference signal shifted 90° by phase shifter 80. The output of mixer 78 is coupled to filter 82. At this point, both channels continue to operate exactly the same and so only the first channel will be discussed hereafter. The filtered signal is amplified by amplifier 84 and coupled to a simple anti-aliasing filter 86 which may be of the type shown in FIG. 5. The output of the filter 86 is

coupled to the oversampled Sigma-Delta A/D converter 88 which includes the digital decimation filtering. The converter 88 is on an integrated circuit chip and includes the digital decimation filter. It is not adjustable on the chip. However, such filter could be made adjustable as is known in the prior art but it is a very complex filter to construct. The Sigma-Delta modulator used is similar to the one illustrated in FIG. 2 herein. As stated earlier, the Sigma-Delta modulator shown in FIG. 2 is designed to suppress quantization noise in the baseband. Thus, most of the quantization noise is at frequencies above the baseband. The main objective of the digital decimation filter 57 is to remove this out-of-band quantization noise. This leaves a small amount of baseband quantization noise and the band limited input signal component. Further, after the high frequency quantization noise is filtered out, the process of decimation reduces the sample reduction rate to the Nyquist rate which minimizes the amount of information for subsequent transmission, storage or digital signal processing.

The output of the A/D converter 88 on line 90 is typically a 16 bit data stream at a 24 kHz rate. When these signals from both channels are coupled on line 90 to the digital signal processor 92, the digital signal processor has only to provide the adjustable digital decimation filtering, perform the complex FFT transform and provide the zero order phase correction. It is no longer necessary to do the first order phase correction because of the use of the Sigma-Delta modulator, the digital decimation filter and the simple anti-aliasing filter. This circuit allows the proper operation for a fixed sampling rate unless the complex adjustable digital

decimation filter is used in the A/D converter 88 as stated previously. In the preferred embodiment herein, an adjustable digital decimation filter 94 is provided in the digital signal processor 92. It is preferably accomplished with the use of software and thus accommodates a variable frequency sampling rate.

The Sigma-Delta modulator is disclosed in an article by Mustansir H. Kheraluwala and Deepakraj M. Divan, "Delta Modulation Strategies For Resonant Link Inverters", IEEE Transactions On Power Electronics, Vol. 5, No. 2, pp.

220-228, April 1990 and in the Motorola document entitled Principles of Delta-Sigma Modulation for A/D Converter and dated April 8, 1990.

Thus there has been disclosed a novel NMR receiver that utilizes a Sigma-Delta A/D converter. By oversampling the signal with the Sigma-Delta modulator, a much simpler analog anti-aliasing filter can be used that has a wider bandwidth. It can be a simple RC filter. These characteristics allow for choosing identical analog filters which do not induce phase distortion of the signal within the operating frequency. It also eliminates the need for first order phase correction by the digital signal processor.

Furthermore, no adjustment is required in the analog filters since the adjustment is accomplished in the digital decimation filter contained within the analog-to-digital converter device or, preferably, in the digital signal processor. The digital decimation filter gives the consistency of amplification that was lacking in the conventional analog filters. Since the quantization noise is assumed to be random, the differentiator in the Sigma-Delta modulator doubles the power of quantized noise. However, the error is pushed towards high

frequencies due to the differentiator factor. Therefore, provided that the analog input signal to the modulator is oversampled, the high frequency quantization noise is removed by digital lowpass filters without affecting the input signal characteristics residing in the baseband. This digital lowpass filtering is part of the digital decimation process. After the digital decimation filtering process, the output signal has only the frequency components from 0 Hz to f(B) . The digital decimation filter can be made adjustable on either the A/D converter or the digital signal processor. It is preferable to provide the adjustment with software in the digital signal processor.

While the invention has been described in connection with a preferred embodiment, it is not intended to limit the scope of the invention to the particular form set forth, but, on the contrary, it is intended to cover such alternatives, modifications, and equivalents as may be included within the spirit and scope of the invention as defined by the appended claims.