Digital ladder networks

Abstract

for any M > 0 and wherefor any a and b for which the linear filter is asymptoti-(31) cally stable. Realizing this fact, the designer using the rms bound can be certain that the maximum value of the limit cycle d l not exceed this bound by a factor of more than two.4b for either b 5 0 or b > 0 and I a 1 2 -One should make the important observation that (32) is References identical to (27) over the region where a 2 -4 b 2 0 . Therefore, the rms bound (32) is also an absolute bound over the region a2-4b>0. Comparison of (29) and (33) over the region 6 > 0 and I a1 <4b/(l +b) shows that they differ by a factor of (1 +&. Note that (29) cannot exceed (33) by more than a factor of two over the specified region and for small values of b they are approximately the same. The following important result can be shown by careful comparison of the bounds:( 3 4 ) [l] I.

On one-multiplier implementations of FIR lattice structures

“…In this approach, the implementation technology has a great impact on the testing technique to be used which has necessitated developing special methods for each technology. Testing NORA (NO RAce) CMOS [l], [2] is the main scope of this paper.…”

“…NORA is a racefree dynamic CMOS for pipelined logic structures [2]. It operates using two clocks in racefree fashion regardless of their overlap time and clock skew.…”

“…It operates using two clocks in racefree fashion regardless of their overlap time and clock skew. The importance of NORA circuits stems from this property which overcome several problems of Domino CMOS circuits [2].…”

“…Since it resembles the LPC lattice which has one-multiplier realizations, it is natural to ask if we can reduce the number of multipliers per building block to one. In this correspondence we use the multiplier extraction approach [2] to show that it is not possible to get structures that have one real multiplier per building block. This is unlike the situation in conventional linear-prediction FIR lattice structures, where real one-multiplier realizations are possible.…”

New cascaded lattice structures for FIR filters having extremely low coefficent sensitivity

“…The vector GN_1(Z) [PN_1(Z) QN.1(z)] is said to be a lossless FIR vector, or simply an FIR ailpass function, in view of (5). We now show how to synthesize a structure of the form in Figure 1, to realize PN,..l(z) and QN_1(z).…”