A new normalised LMF (XE-NLMF) algorithm is proposed for channel equalisation to improve the convergence speed and bit error rate performance. The analysis of the LMF algorithm's initial convergence in the mean has facilitated the normalisation which is bounded by the error power and signal power. Hence, a combination of the signal power and error power, by means of mixing, is used to normalise the step size. The proposed normalisation improves the stability of the mean fourth cost function and achieves a faster convergence in both co-channel interference and white noise. The complexity of the normalised LMF algorithm is slightly higher than the NLMS algorithm but the convergence speed and bit error rate performance significantly outperform the NLMS algorithm.