There is always the synthesized method of generating the required pulses. This would likely only work if the absolute exact scale MPH wasn't important, only the general operation of the speed control.
My idea is to pick a place on the drivetrain to generate pulses based on the speed, something like Stan's slotted wheel.
Using an inexpensive microprocessor board, you'd time the spacing of the pulses from the slotted wheel. With that timing, you'd divide by your chosen ratio to simulate the DCS tach and output the tach pulses at the higher rate needed to simulate the motor tach. At each sample, you recompute the wheel pulse timing and adjust the frequency of the output tach pulses. With enough triggers per/rev from the slotted wheel, I think this might be a workable solution. The only fly in the ointment I can see with this approach is it's possible there'd have to be special handling for the initial start of motion and when stopping.
