A good 2H lock signal with a high signal-to-noise ratio is a real advantage for maintaining a stable magnetic field for long data acquisitions and also for shimming the magnet using the lock signal. Sometimes, however it is desirable to run NMR spectra for samples with only a very small quantity of deuterated solvent and therefore a very weak lock signal. Such may be the case when one is monitoring a chemical reaction by removing aliquots and adding a drop or two of a deuterated solvent to help with magnet shimming using the 2H lock signal. Although one may be able to shim a magnet using a very weak lock signal (with difficulty), running the spectrum locked may not be a good idea. Running a spectrum while locked on a very weak lock signal can lead to distortions in the spectrum. It is often better to use the weak lock signal to shim the magnet as best you can and then run the spectrum unlocked. This is demonstrated in the figure below.
The figure shows two single scan 1H NMR spectra of a sample of acetone (one drop) in CCl4 with a drop of CDCl3. The spectrum on the left was acquired using the 2H lock and the one on the right was acquired unlocked. One can clearly see the distortion in the 1H spectrum caused by locking on a very weak 2H signal.