摘要 |
In the digital filter with undersampling described, the rate 1/T of the input data is r times as large (r = 1, 2, 3...) as the rate of the output data. So that no unnecessary computations are carried out with such a filter and the number of required delay sections is as small as possible, the number N of filter coefficience is extended by such a number of coefficients with the value zero that it can be split up as product of integral factors of the form N = L*r. The number L determines the number of parallel arithmetic units contained in the filter and the total number of delay sections required. Each arithmetic unit (R0, R1, R2) contains a multiplier (M0, M1, M2), an accumulator (A0, A1, A2) and a chain of delay sections (K0, K1, K2) following the accumulator (A0, A1, A2). The coefficients of the filter are read out of a coefficient memory which is addressed by a counter (Z) cycling through r counts. Finally, the outputs of the L arithmetic units (R0, R1, R2) are connected to the inputs of an adder (AD) at the output of which the output data (Y) are present. <IMAGE>
|