Well, I can't understand why it would use 100.11...My suspicions are that the equation is Q=100.X / 2^32, and that an error is being introduced in the display. Possibly the user input being adjusted by the player is converted to DSP coefficients, and then reverse-calculated for display purposes. Given the lack of native floating point support within the ARM, the code probably uses fixed point math, which could introduce small errors, especially with a double conversion. Purely speculation...

I think that we can live with the accuracy of Q=100.X / 2^32 for anything we might want to do.