[BobKinsman]

I'm starting a cnc conversion on an Optimum BF46 mill and have just purchased a Kflop board to get things rolling.

Running the board off the machine, I see that the setup time between a change in the Dir signal and the falling edge of the next Step pulse is about 0.5 microseconds, whereas the stepper driver I was hoping to use requires 20 microseconds. The manual and discussion somewhere on cnczone suggests that there is not enough adjustment in the board setup to satisfy this requirement (didn't spot that little problem before ordering!)

As far as I can see, there is always at least 500 microseconds between the last Step pulse in one direction (due to deceleration) and the Dir signal changing state, so am proposing to simply delay the step signals by the required 20 microseconds. The Dir signal is constant while running fast, so top speed is not affected.

The question is-
Are there any circumstances under which an axis can reverse direction without the controller ramping the speed down to zero? It would seem to be a recipe for lost steps or a broken machine, but there might be something obscure. I know that all the other axes could be going at full speed while this happens.

Thanks in anticipation
Bob

[TomKerekes]

Hi Bob,

If you invert the Step Pulse to that trailing edge of the pulse is what is used by the drive then the Dir will have 4us of setup before the step. But this is still insufficient for your 20us setup requirement.

I believe delaying (or stretching if high step rates aren't used) the step pulse should work. The actual worst case situation is a path that goes right along the boundary between two steps dithering back and forth. If it was exactly on the boundary then only a microscopic motion would be required to generate +1, -1 steps. KFLOP introduces 1/8th uStep of hysterisis to avoid this problem. Also KFLOP only updates the trajectory path every 90us.

HTH
Regards

[BobKinsman]

Thanks for the reply Tom. Your answer is very reassuring.

Would I be correct in thinking that the dithering situation would only occur in a closed loop configuration? I can't imagine why a pre-calculated, open loop trajectory would ever do this.

Also, if the Kflop recalculates the trajectory at most every 90microseconds, does that mean that the minimum gap between two Step pulses that span a change of Dir signal is 90microseconds?

Thanks, Bob

[TomKerekes]

Hi Bob,

Yes a CNC path would have to be pathological to dither at such a high rate,

It is fairly complicated, but I believe you will be ok.

The Direction can not change more often than every 90us. However step pulses and direction change may occur anywhere within the 90us period.

Additionally, there is +/- 1/8th (1/4 total) steps of hysteresis built in. Which basically means that on a trajectory reversal there will be no steps generated (and the direction will not switch) until the trajectory has reversed the minimum distance of 1/4 step.

Regards