The present invention relates to transversal equalisers and more particularly d.c. compensation in adaptive equalisers.
The presence of a d.c. offset in an input signal to an adaptive equaliser causes confusion in the coefficient update algorithm because the error signal does not correlate properly with the data signal samples at each tap of the equaliser. The d.c. offset in the input signal has therefore to be reduced to zero before the signal enters the equaliser. This is normally accomplished by using adjust-on-test resistors or potentiometers providing in-life adjustments to correct for ageing drift in analogue components such as operational amplifiers, sample-and-hold circuits and the analogue-to-digital converter itself.
According to the present invention there is provided an adaptive equaliser including a plurality of tap stages, the first stage receives an input signal and subsequent stages receive time spaced samples of the input signal and each stage includes a multiplier and a correlator both of which are fed by an input signal, the output signal from the multiplier are summed by an accumulator which applies a signal to a decision circuit providing an output signal and an error signal, the latter is applied to each correlator to correct residual error, said adaptive equaliser includes an additional tap stage having its respective multiplier and correlator connected to a fixed voltage source, the output signal from the multiplier is summed with the other multiplier output signals by said accumulator and the error signal produced by said decision circuit is correlated with said fixed voltage by the additional correlator to control the gain of the additional tap stage multiplier thereby compensating for an original d.c. offset in the input signal.
According to a feature of the invention the adaptive equaliser finds application in modulator/demodulator equipment commonly known as modems.
The invention will be readily understood from the following exemplary embodiment thereof and should be read in conjunction with the following drawing wherein:
FIG. 1 shows a known form of adaptive equaliser; and,
FIG. 2 shows an adaptive equaliser in accordance with the present invention .
Referring to FIG. 1, the time-spaced samples of the input signal are provided by delay lines T. The input signal xi, and the time-spaced samples x(i-T), x(i-2T) are multiplied by the four-quadrant multipliers M1, M2, and M3 respectively and then summed in an accumulator A. The output of accumulator A is fed to decision circuit D and compared with a set of reference levels or with a locally generated reference signal to produce an output signal Y.sub.K and error signal E.sub.K. The error signal E.sub.K is correlated with the signal samples in the correlators C1, C2 and C3 respectively, the outputs of which control the gains of associated multipliers M.sub.1, M.sub.2 and M.sub.3 to minimise the correlation of each tap stage.
If there is a d.c. offset in the input signal the correlators C.sub.1, C.sub.2 and C.sub.3 will attempt to correct the residual error, but since there is no linear relationship between the d.c. offset and the signal samples the effect of this attempt to remove the d.c. offset will result in non-optimum settings for the multipliers M.sub.1, M.sub.2 and M.sub.3 and distortion in the output signal Y.sub.K will occur.
Referring to FIG. 2 an equaliser according to the invention is shown and functions as described with reference to FIG. 1, but includes a fourth tap stage connected to a fixed voltage source V. The voltage V is multiplied by a multiplier M.sub.4 and summed with the outputs of multipliers M.sub.1, M.sub.2 and M.sub.3 by accumulator A. The error signal E.sub.K is correlated with the fixed voltage V in the correlator C.sub.4. The output of correlator C.sub.4 is used to control the gain of multiplier M.sub.4 until the d.c. offset of the output of accumulator A is reduced to zero, allowing multipliers M.sub.1, M.sub.2 M.sub.3 to assume their optimum gain values.
The above description has been of one embodiment only and is not intended to limit the scope of the invention.
For example, the invention could be applied to adaptive equalisers having complex-valued tap gain coefficients by having two additional taps as described, one to remove the d.c. in the inphase channel and one to remove the d.c. in the so-called Quadradure channel.