While emptying the carousel of the AVANCE 400 daily, I notice that some samples have a concentration gradient (i.e. the sample is more concentrated at the bottom of the sample column compared to the top). The samples are not mixed. Such samples will always give NMR spectra of a lower than necessary quality. Why? - because the gradient shimming routine used on the spectrometer uses a magnetic resonance image of the deuterium concentration across the tube. If the concentration of the solvent is not constant across the sample volume, then the deuterium gradient shimming routine will not do well and you will get a poorly resolved spectrum. In fact you will also get a poor spectrum when manual shimming is carried out based on the lock signal.
Mix your samples well - get better data.