That's spot on m_c. I'm not concerned with the accuracy, it will be plenty good enough and other issues such as backlash, drive belt tooth engagement, quill bearing precision etc will all be significantly greater.

There is very little, if any, torque reduction from 8 ustep to 16ustep it is just a bit smoother at low speeds. I could go down to 4 ustep, and it's still acceptable, if the operating frequency proves to be a problem. At high speeds the momentum of the rotor and load make the microstepping irrelevant and the driver reverts to full step to get a bit better torque.

The stepper drivers are current regulated so they deliver what ever voltage is needed to achieve the desired current up to the maximum the power supply can give. The higher the rpm needed from the motor the higher the frequency from the drive and the higher the reactance from the motor stators. Higher reactance then requires a higher voltage to reach the desired current.

The point about the step/direction/enable signals being opto isolated means I can wire them common anode or common cathode to suit the controller outputs, they will all be the same so no effect on timing.

USB manages to communicate totally reliably for many other devices, other than poor EMC design, why should a motion controller fail to talk properly on USB?

What motion controllers have you had least problems from?