3.4028235*10^38 is the largest possible number, however the second largest is 3.4028233*10^38, which is quite a significant step.
Calculating you would have 2^exp*(2-eps), where eps is the amount by which the fraction is smaller than two (since it can never actually be 2). now all you need to calculate is how large can you make the exponent, so that subtracting 2^(-22) from the maximum of the fraction makes such an insignificant difference that the physics don't start badly malfunctioning.
Taking 0.001m as a minimum needed distance your float should be somewhere around 2^14*(max_fraction). That is equivalent to 32km, which is not actually very much, correct me if I'm wrong.
Here you can play with floats and their limits: http://www.h-schmidt.net/FloatConverter/IEEE754.html