Linuxcnc - s-curve acceleration? (7176e)

That is the whole point of the planner... you can ask for impossible profiles and it makes them possible by clipping them to as close as the machine is capable.
The better it is at this, then the more aggressive you can be with speed, acceleration and jerk.
Whether your machine needs/can use this performance is a good question.

Chris M

Sure, just pointing out that many time the profile determine these values, not the machine,
though of course the machine determines the tangential acceleration when changing velocity.

The only time jerk is meaningful i find is in exact stop linear moves. Many arcs and constant velocity moves have what is effectively built in jerk control as mentioned above. Machining a square at high speed and high acceleration with G64 on is smooth as butter. With it off, the floor shakes

So... how hard is it to do jerk control ONLY for G0 and G1 exact stop moves as a starting point? This would cover most real life cases of bad jerk, and only leave a handful of special cases to be dealt with later. It sounds to me many of the previous tries got hung up on trying to cover all motion. I think we should be more incremental about it.

I have a 2000ipm machine with 0.5G acceleration, so jerk control is an important topic for me as the machine endures a lot of stress.

The cases where we will still have significant jerk ( the start and end of some g64 mode moves, shallow arcs in exact stop, etc) are a fair compromise to leave off.

You will still have to tune your aces for the worst case motion situations, but having smooth rapids and linear moves would be at least for me a vast improvement.

Yes, no, maybe?