Page 111 - Photodetection and Measurement - Maximizing Performance in Optical Systems
P. 111

System Noise and Synchronous Detection

            104   Chapter Five

                        tering. Once signal and reference are converted in analog-to-digital converters
                        (ADC), all multiplications and integrations in the Fourier analysis can be
                        carried out free from errors using digital hardware and software techniques. By
                        taking this approach, great flexibility and richness of functions can be provided.
                        It is, for instance, hardly more complex to compute the power in many har-
                        monics of the signal, each in a different bandwidth, or even the full Fourier
                        spectrum of the signal. With the relentless attack of digital solutions on areas
                        once considered solely analog, the increasing use of digital signal processing in
                        synchronous detection must be expected to continue. Low-frequency signals and
                        those recorded for off-line analysis have long been dealt with this way. Even the
                        humble radio receiver is being reduced to an aerial, ADC, and digital process-
                        ing chip.

            5.6 Walsh Demodulators

                        The ease of construction of digital (±1) multipliers hides a disadvantage of
                        increased noise. This is easily seen from the frequency-domain representations
                        of signal and reference. The passbands in the frequency domain formed by
                        modulation and synchronous demodulation are obtained by convolving the
                        frequency-domain representation of the reference signal with the double-
                        sided passband obtained from the postdemodulation low-pass filter. The Fourier
                        analysis of a sine-wave reference is just a single amplitude, so multiplication of
                        the input signal by a sine wave at f mod delivers a filter with a single passband,
                        width ±1/2pRC centered on f mod . The Fourier decomposition of a bipolar square-
                        wave reference clock is a set of odd harmonics of the fundamental frequency.
                        This produces passbands at f mod , 3f mod , 5f mod , etc., with relative amplitudes of 1,
                        1/3, 1/5, etc. (Fig. 5.11). Hence square-wave references open up not just a single
                        passband, but an infinite series of passbands. If the input has signal, noise, or
                        interference power at these harmonic frequencies, they will be detected and con-
                        tribute to the overall output. It is possible to roll off the receiver to give a
                        reduced power at three times the modulation frequency, but the suppression
                        will not be complete. When a single RC filter only achieves an ultimate cutoff
                        slope of -6dB/octave (-20dB/decade), it is difficult to get much suppression at
                        3f mod . Hence these harmonic responses can be a problem and must be kept in
                        mind if strange interference effects are seen.
                          A partial solution to the harmonic responses can be obtained by approxi-
                        mating the sine function needed as reference waveform for the synchronous
                        detector. If we can choose a small number of binary waveforms to approximate
                        the sine wave, these can then be applied to weighted binary multipliers and
                        summed together for an overall response similar to that of a sine wave refer-
                        ence. Just as Fourier synthesis can be used to approximate an arbitrary func-
                        tion using a small number of sine/cosine terms, we will use Walsh synthesis to
                        approximate the sine wave with a small number of Walsh functions.
                          Walsh functions are two-parameter binary (±1) functions that form an orthog-
                        onal series. They can be used just like the sine and cosine series of Fourier analy-


                   Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
                              Copyright © 2004 The McGraw-Hill Companies. All rights reserved.
                               Any use is subject to the Terms of Use as given at the website.
   106   107   108   109   110   111   112   113   114   115   116