Chapter 4 FIR filter Design Todays engineer faces a challenge in selecting the proper mix of analog and digital

techniques to solve the signal processing task at hand. It is impossible to process realworld analog signals using purely digital techniques, since all sensors, including microphones, thermocouples, strain gages, piezoelectric crystals, and disk drive heads are analog sensors. Therefore, some sort of signal conditioning circuitry is required in order to prepare the sensor output for further signal processing, whether it be analog or digital. Signal conditioning circuits are, in reality, analog signal processors, performing such functions as multiplication (gain), isolation (instrumentation amplifiers and isolation amplifiers), detection in the presence of noise (high common-mode instrumentation amplifiers, line drivers, and line receivers), dynamic range compression (log amps, LOGDACs, and programmable gain amplifiers), and filtering (both passive and active).

Figure: Digital signal processing In general, as the ADC is moved closer to the actual sensor, more of the analog signal conditioning burden is now placed on the ADC. The added ADC complexity may take the form of increased sampling rate, wider dynamic range, higher resolution, input noise

rejection, input filtering, programmable gain amplifiers (PGAs), and on-chip voltage references, all of which add functionality and simplify the system. With todays high resolution/high sampling rate data converter technology, significant progress has been made in integrating more and more of the conditioning circuitry within the ADC/DAC itself. In the measurement area, for instance, 24-bit ADCs are available with built-in programmable gain amplifiers (PGAs) that allow full-scale bridge signals of 10 mV to be digitized directly with no further conditioning (e.g., AD773x series). At voice-band and audio frequencies, complete coder/decoders (codecs or analog front ends) are available with sufficient on-chip analog circuitry to minimize the requirements for external conditioning components (AD1819B and AD73322). At video speeds, analog front ends are also available for such applications as CCD image processing and others (e.g., AD9814, AD9816, and the AD984x series).

Finite Impulse Response (FIR) Filters There are two fundamental types of digital filters: finite impulse response (FIR) and infinite impulse response (IIR). As the terminology suggests, these classifications refer to the filters impulse response. By varying the weight of the coefficients and the number of filter taps, virtually any frequency response characteristic can be realized with a FIR filter. As has been shown, FIR filters can achieve performance levels that are not possible with analog filter techniques (such as perfect linear phase response). However, high performance FIR filters generally require a large number of multiplyaccumulates and therefore require fast and efficient DSPs. On the other hand, IIR filters tend to mimic the performance of traditional analog filters and make use of feedback, so their impulse response extends over an infinite period of time. Because of feedback, IIR filters can be implemented with fewer coefficients than for a FIR filter. Lattice filters are simply another way to implement either FIR or IIR filters and are often used in speech processing applications. Finally, digital filters lend themselves to adaptive filtering applications simply because of the speed and ease with which the filter characteristics can be changed by varying the filter coefficients. The most elementary form of a FIR filter is a moving average filter as shown in Figure shown below. Moving average filters are popular for smoothing data, such as in the

analysis of stock prices. The input samples, x(n) are passed through a series of buffer registers (labeled z1, corresponding to the z-transform representation of a delay element). In the example shown, there are four taps corresponding to a 4-point moving average. Each sample is multiplied by 0.25, and these results are added to yield the final moving average output y(n). The figure also shows the general equation of a moving average filter with N taps. Note again that N refers to the number of filter taps, and not the ADC or DAC resolution as in previous sections.

Figure 4-point moving average filter

The multiply-accumulate cycle is now required per tap for the generalized FIR filter. The essence of FIR filter design is the appropriate selection of the filter coefficients and the number of taps to realize the desired transfer function H(f). Various algorithms are

available to translate the frequency response H(f) into a set of FIR coefficients. Most of this software is commercially available and can be run on PCs. The key theorem of FIR filter design is that the coefficients h(n) of the FIR filter are simply the quantized values of the impulse response of the frequency transfer function H(f). Conversely, the impulse response is the discrete Fourier transform of H(f). The generalized form of an N-tap FIR filter is shown in Figure below. As has been discussed, an FIR filter must perform the following convolution equation:

where h(k) is the filter coefficient array and x(nk) is the input data array to the filter. The number N, in the equation, represents the number of taps of the filter and relates to the filter performance as has been discussed above. An N-tap FIR filter requires N multiply-accumulate cycles. FIR filter diagrams are often simplified as shown in Figure. The summations are represented by arrows pointing into the dots, and the multiplications are indicated by placing the h(k) coefficients next to the arrows on the lines. The z1 delay element is often shown by placing the label above or next to the appropriate line.

Figure: N-tap FIR flter