@dc42 said in S-Curve/ sinusoidal , Jerk +acceleration:
I suspect that the move from jerk to JD is more to do with fashion than anything else.
always a possibility
I hate changes that are fashion only
Still, it doesn't make sense to treat X and Y jerk separately on delta printers and some other architectures.
IMO it should be separate for all axes, the gro of these el-cheapo PRC printers have a moving bed, you can't have same jerk settings for Y where you move 5kg bed and X where you move 100g head.. just like you cannot have same acceleration nor same max speed...
no clue about delta, corexy or some of those exotic architectures ... I assume corexy will have same X and Y settings but not sure what I base that assumption on
Another effect is that when using JD, if you reduce acceleration then the maximum cornering speeds will be reduced in line. This may not be desirable. For example, if you decide to reduce acceleration in an attempt to improve print quality, then curves may be printed jerkily instead of smoothly, because the
Yes, I know this (dunno much but this I experienced on my skin so I know), you have to change JD and acceleration at the same time, they are dependant on each other (I assumed it's the same with jerk?)
Perhaps a better approach would be to configure a "maximum speed to do a 90 degree corner" instead of a mythical deviation value, and then to calculate the maximum speed for a given angle using the same (complicated) speed vs. angle relationship that JD uses. Or perhaps there is a better relationship to use.
It's easier to explain for sure and easier to visualize. Maybe that's why klipper is doing it.
Even if using this approach or native JD, we'll still need to configure Z jerk (to handle instantaneous speed changes when doing bed compensation) and extruder jerk (to handle buggy slicers and pressure advance).
Z does introduce complication here for sure, but if implementation allow for separate JD setup (trough 90º or whatever settings) for each axis then the Z, E, A would just be another axis?
That "smoothing" klipper does, that looks like S-Curve, sounds great on paper, I assume it is unrelated to JD/JERK, is RRF doing that? Do you think that has merits? It looks to me that it would reduce shaking immensely and lot of slicers still generate those "shaky" codes filling in gaps.
Don't get me wrong I'm talking out of my %#$^^@$ here, trying to learn as much as I can from this conversation. I'm hardly an expert in these matters