You can try closely observing the behavior of the system. You must issue your "next" step pulse not only BEFORE the motion stops, but soon enough that the motor has time to respond with the next step before it begins to decelerate. This means, essentially, that each step is shorter than the previous one, until you reach the top speed at which you intend to operate.

If you accelerate too quickly, you'll miss steps, and if you do so too slowly, the system could "bounce" back and lose two steps. Knowing some things about the load dynamics will help considerably. The effect you want to avoid is "cogging" which is akin to the effect you feel when you manually rotate a stepper by hand when it is not powered. If you want to approach this analytically, you have to have a lot of informtation about your system, including rotational inertia of your pulleys, belts/cables, and also including friction and load inertia and lots of other things that tend to make one's eyes glaze over.

I've found that acoustics are very helpful. FIrst of all, if you can hear the stepping, it's too late, i.e. when accelerating, you have to step before the load begins to lose momentum. When decelerating, you have to know how much help you'll get from your system friction, and you can help yourself by cutting the width of the step pulse at its trailing end.

You can make empirical measurements. Have you tried to do that?

Keep in mind, if you can hear it or feel it, it's timed incorrectly. If it skips steps while accelerating or jumps ahead while decelerating, it's timed wrong, and not necessarily too fast or slow. Sometimes it's quite counterintuitive.

This kind of thing seems simple, but it's not.

RE