Search

Analog discrete-time finite-impulse-response (FIR) filters have been used as equalizers in digital communication receivers. For high speed applications, an FIR equalizer can be implemented using parallel sample-and-holds (S/Hs) and time-interleaved equalizer channels. Mismatches among the parallel S/Hs degrade the equalizer performance. This paper addresses mismatches of DC offsets, gain errors, sample-time errors, and bandwidths in the S/Hs. It is shown that having a different set of adapted coefficients in each equalizer channel can reduce the effects of mismatches. Simulation results are presented for different communication channels. [ABSTRACT FROM PUBLISHER]

In a conventional analog adaptive forward equalizer (FE), parasitic capacitance introduces an undesired RC pole at the output node of the equalizer. At high data rates, this pole can introduce intersymbol interference (ISI) that degrades performance. This paper considers the effect of the parasitic pole and presents two approaches to deal with the parasitic pole. The first approach uses an extra tap in the FE; the second uses chopping. Also, the filtered-x LMS (FX-LMS) algorithm is used to allow the equalizer to adapt and correct for the parasitic pole. Implementations of the FX-LMS algorithm and the chopper technique are presented. The conventional and new architectures and various adaptation algorithms are compared via simulation. [ABSTRACT FROM PUBLISHER]