Am Freitag, dem 06.12.2024 um 23:43 +0200 schrieb Viesturs Lācis:
> Hello!
> 
> One of the Biesse retrofits I have been working on lately is finally
> receiving some of its intended final shape:
> https://photos.app.goo.gl/cKi8pysM3Tnb9Egh6
> 
> That wood box on top of it contains 600W laser CO2 tube (some pics of
> it also included). Since I have no idea how to lift it without
> braking
> I decided to mount it with all the box and encapsulate the tube in
> foam.
> 

It's really none of my business but that laser-tube "installation"
really looks scary. Are you going to run those tubes in that enclosure?
How do you maintain alignment with this kind of non-rigidity?

> So the question for 10K is: what is the current status of jerk
> limitation? Searching mailing list and forum shows quite a few
> discussions but I did not see anything about it being implemented (in
> separate branch or whatever). I do recall discussion that it would be
> hard to make it work with spindle synchronized moves. I do not need
> that here, motion only with G0, G1 and G2/G3.

You probably found my hack I did some years ago with replacing
trapezoidal acceleration profile with something "smooth" bezier-based.
Unfortunately there are some corner-cases where accel-spikes happen
and/or the thing just stops and even then it would not be a complete
solution as it only looks at "tangential" movement and won't help a bit
with line->line transitions (that have infinite acceleration, current
TP should replace that already with circular arcs where possible) or
line->circle transitions (that have infinite jerk).

Some swiss school made a linuxcnc fork that has a jerk limited TP, but
it's not easy to integrate that with stock linuxcnc, the code structure
is dissolved in acid, digested, and then reconstituted in a new
directory structure/build system/everything. And the TP mainly consists
of some MATLAB-generated C-Code 

If you want to look yourself, I recently found prunt3d.com which uses
"pythagorean hodographs" to round corners and generate jerk-limited
trajectories. But it is written in (of all languages) ada!, and the
preprint i found of the paper it is based on is somewhat hard to read.
The published paper is locked behind a paywall.


regards
Robert



_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to