Abstract: An adaptive network includes a summation node for receiving an input signal and delivering an output signal. A decision unit receives the output signal of the summation node as an input signal and has an output for delivering an output signal being a differential signal between the input and output signals of the decision unit. An adaptive feedback branch receives the output signal of the decision unit and has status variables, coefficients and an output. A changeover switch unit couples the output of the feedback branch to the summation node. A monitoring unit monitors an error magnitude of the output signal of the decision unit. The monitoring unit triggers the changeover switch unit in the next clock cycle if a first signal value having an amplitude exceeding a certain threshold value appears, for causing the summation node to be acted upon by a digital zero signal, for setting the status variables of the feedback branch to zero and for freezing the coefficients of the feedback branch.

Abstract: A method and a circuit configuration for filtering a digital value with improved noise behavior include adding a digital input value and a feedback value to form a digital output value. The feedback value is equal to a first variable from which a second variable is subtracted, from which the second variable shifted m places to the right is subtracted, and to which the first variable is added. The first variable is equal to a bottom m bits of a previous digital output value and always represents a positive value. The second variable is equal to a previous first variable from which a previous second variable shifted arithmetically by m bits to the right is subtracted. The feedback value is calculated with at least one additional least significant bit until at least a last addition with the first variable. Subsequently, an expansion is no longer taken into account.