If you're using the Henderson-Hasselbalch approximation then activity coefficients are unlikely to be relevant - using the approximation will already take you 0.1-0.4 pH points away from a true value.

Care to elaborate? HH equation is not an approximation, it is just a rearranged acid dissociation constant. Problems start when it is used blindly and without understanding limitations of the typical approach.

The HH equation as I know it is of the form:

[H+]=K

_{a}*(C

_{a}/C

_{s}) where C

_{s} is the initial concentration of your salt = concentration of the cation B+ present from salt BA, where you also have some HA present (initial concentration C

_{a}) in the mixture

This is heavily approximated. An exact solution is first obtaining

[H+]=K

_{a}*

**(**(C

_{a}-[H+]+(Kw/[H+]))/(C

_{s}+[H+]-(Kw/[H+]))

**)**After two approximations follows [H+]=K

_{a}*(C

_{a}/C

_{s}).

I have also seen the equation Ka=([H+]*[salt])/[weak acid], but am unclear what this means - presumably [salt]=Cs as all salts are assumed to dissociate 100% in solution, or for salt B

_{b}A

_{a}, [salt]=C

_{s}*b. Thus in calculating buffer composition you would work to find out what mass of salt would produce this concentration [salt] for your desired pH. But what does [weak acid] refer to and how could you calculate it exactly? My issue with that is the reason I don't see this so far as an exact method.