Timing and amplitude error estimation for time-interleaved

Abstract

Methods for calibrating a plurality of N time-interleaved analog-to-digital converters (ADCs) for any integer N.gtoreq.2 by detecting and/or correcting timing and amplitude errors therein, where each of the ADCs, numbered j=1, 2, . . . , N has an identical clock frequency f.sub.c and period T.sub.c =(f.sub.c).sup.-1. Ideally, the N ADCs are clocked uniformly in time so that two consecutively clocked ADCs number j1 and j1+1 have effective sampling times that differ by time intervals of uniform length .DELTA.t=T.sub.c /N. Ideally, the N ADCs also have identical gains from their analog input signals to their digital output signals. In a calibration procedure, all ADCs receive a common analog input signal S.sub.in (t)=Asin[2 .pi.(1+s)f.sub.c t], where s is a non-zero fractional number satisfying -0.5<s<0.5. The output signal values from ADC number j provide a set of ordered pairs of values (t.sub.i, D.sub.ij) for sampling times t.sub.i, and the set of values {D.sub.ij }.sub.i with j fixed is approximated by another set of values { V.sub.ij =A.sub.j sin[2 .pi.sf.sub.c t.sub.i +.phi..sub.j ]}.sub.i. Here, the amplitudes A.sub.j and phase angles .phi..sub.j are chosen to provide a best numerical fit to the set {D.sub.ij }.sub.i for each j=1, 2, . . . , N, according to some criterion. The phase differences .DELTA..phi..sub.j2,j1 =.phi..sub.j2 -.phi..sub.j1 are then determined and compared with the ideal phase difference value .DELTA..PHI..sub.j2,j1 =2 .pi.(j2-j1)/N to determine if a relative timing error exists between ADCs number j1 and j2. In a similar manner, the amplitude differences A.sub.j2 -A.sub.j1 are compared to determine if a relative gain error exists between ADCs number j1 and j2.