If you decide to solve the problem feeding A and B directly to the '51 beware of the old faulty "use B for direction and A for clock" it will be wrong at some transitions. I once could not figure out why the counts were occasionally wrong and wrote out all transitions an a graph and saw what was happening. You need to know both current state and previous state, then use a lookup table.

When you do this right your resolution will be 4 x the clock frequency.

Have fun,

Erik

PS the HP chip does it right.