Do I understand you correctly that you're trying to find a single buffer system that spans a wide pH range? To do that, you'll definitely need to consider buffers with polyprotic acid components.
However, why limit yourself to a single buffer composition. It's very common to screen the spectra in your protein using a wide variety of buffers to see which yield high quality spectra.
As a starting point, take a look at this web page from the University of Buffalo:
http://www.nmr2.buffalo.edu/nesg.wiki/Buffer_optimization
There they give series of common buffers that span a large pH range. It's easy enough to buffer-exchange your protein and then record a quick spectrum. Since you're doing 31P NMR, simply exclude those buffers that contain phosphorous if you don't want to worry about extra signals appearing in your spectrum. good luck.