Slope Overload Noise in Linear Delta Modulators With Gaussian Inputs

l.l Objective This paper presents some new theoretical results on slope overload noise in linear delta modulators. In particular, we derive the mean power of the (unfiltered) slope overload noise at the demodulator output when the modulator input is a stationary Gaussian random process and the modulator feedback path contains an ideal integrator, Fig. 1. This same problem has already been treated by other investigators, notably Zetterberg,1 Rice (with O'Neal), 2 Protonotarios,3 and Abate, 4 but several factors have prompted a reexamination. One is 387 388 THE BELL SYSTEM TECHNICAL JOURNAL, MARCH 1973 RATE, 1/T QUANTIZER 1 y (t) LOW PASS FILTER OUTPUT, x(t> I DEMODULATOR IDEAL INTEGRATOR I (a) x (t),y (t) x(t) y (t) x (t). y (t) x'(t) = + x' 0 (b) TIME TIME (C) Fig. 1--Linear AM codec with perfect feedback integrator, (a) Equivalent block diagram, (b) Input and feedback signals, (c) Model used.