I have seen some students collect 31P NMR data without proton decoupling. Their reason was that there were no directly bound protons to the phosphorus. What they have not taken into account are the remote protons which have small couplings to the phosphorus. There are usually many of these remote protons and the cumulative effect of the couplings is to broaden out the 31P resonance. This broadening causes a severe loss in the signal-to-noise ratio. The upper trace of the figure below is a single scan 31P spectrum of triphenylphosphate where the closest protons are 4 bonds away from the phosphorus atom. The lower trace was run under identical conditions with the proton decoupler turned on. In this case, the gain in signal-to-noise using proton decoupling is 17.