| OCR Text |
Show 25 som saturation of the modulator can be tolerated in x hang for th in d precision and decreased quantization noise that results from a low r valu of 6. 3.4 Decimation Sigma-delta modulation yields a signal with one-bit words at a rate much higher than the Nyquist Rate. It is desirable to reduce the sampling rate and incr as the word length of this signal before subsequent processing. This process is called decirnation. As rnentioned in the previous paragraph, the quantization noise produced by a sigma-delta modulator is concentrated in frequencies above baseband as long as the sampling rate is high. Using decimation techniques, the sigma-delta modulated signal can be downsampled without aliasing this noise back into the baseband. The result is a digital signal that is at a lower rate and has longer word lengths. Different types of decimators can be used depending on the type of filtering that is required, and the effects the sigma-delta modulator has on the input signal. The decimator that we are using for the first stage of decimation is a simple accumulate and dump filter. Since the signal is oversampled at 20 times the base frequency, the decimator accumulates 20 single bit values and dumps out a 5-bit accumulated number at a frequency that is 20 times less than the input sampling frequency. This is achieved by having a divide by 20 counter, that divides the input clock by 20, and sends out a clear signal to the accumulator every 20 cycles. The accumulator meanwhile adds the last 20 values and dumps the final value out to the next stage, each time the counter issues a clear signal. It then clears its registers and accumulates the next 20 values. Thus we have a 5-bit output from the sigma-delta modulator which is at Nyquist rate and which is sufficient to recover the input signal completely. 3.5 Advantages of Sigma-Delta Modulation A/D converters utilizing sigma-delta modulation have many desirable properties when compared to other methods of A/D conversion(7]. One advantage of this approach over other high-speed A/D conversion methods is that a smaller number |