Running large 3d code in linear interpolation mode (G1)

[chicoxiba]

Hi
Im interested in Auggie, specially since I saw it uses TEMPEST Trajectory Planner! I would love to use tempest with the kinks worked out. I did find a few problems while it ran with mach3 on the releases with tempest tp. From memory the inconsistencies were: Inability to maintain constant/satisfiable feedrate dependent of distance between points; Stuttering motion on some sections.


With Auggie so far I have only done the Demo runs as I don't have a pokeys board yet. I do however have several others.. ESS, UC100 and the mother of all... PP :)

For some reason Auggie throws exceptions after reading the G1 command, I would guess that linear interpolation is not implemented or allowed at the moment.

My code consists of 3 axis linear interpolation for the great majority.. 99%.. There are a few G0 moves only.

DeltaX is usually 4mm. Sometimes it is 1mm and sometimes it is 10mm.

Xaxis job dimesions are usually in the order of 6feet
Yaxis job dimensions are usually in the order of 20inches
Zaxis job dimensions are usually in the order of  6inches

Here is a complete job of my code, for reference.

I would like to testify here my interest in using Auggie as a machine controller and I am willing to give my best possible feedback, support and attention to make it a 100% preffered controller for the geeker machinist and community.

Regards,
Francisco

[ArtF]

Hi F:

  >>For some reason Auggie throws exception s after reading the G1 command

  How did this show up? Did it crash or only show an error? All linear interpolation shoudl work, but I amy have introduced
a bug or three. Ill run your code and see... Get back to you soon..

Art

[ArtF]

lol.. the second I loaded the file I knew you Francisco. Its amazing how many people I
remember from the olden days..  Your surfboard reminded me instantly of your operation,
and the troubles you had in days of yore. :)

  I see the trouble.. Modes were not sticking, Ive fixed it but wont release till I find another
error Im looking for. Ill do a release probably later today, early tomorrow that will run your
board files, I have them running here now, they seem fairly smooth, motor stats show
a pretty steady feedrate across the board, but we'll see how that machine of yours takes
to it.. lol


Thx
Art

[Ya-Nvr-No]

Id had that issue once too, found that if I used the single-MDI fold I could run just a G1 command and it reset it, the issue was finding the gcode line it had stopped on and starting the code back up from there. One of the many things I forgot to mention to Art.  :P

(Update: I just tested that don't work any longer,  :'()

I'm still amazed what has transpired in just 9 months. Will take time to work out all the kinks. But this will be one robust programmable machine controller, just needs imagineering applied.

Cool boards, but I'll still stay away from the sharks, thanks.

[chicoxiba]

wow thats nice to hear all!
yes Art that's me, still dreaming with the perfect motion:)

Art, I did download the latest rev. and ran it in sim only. I will look into putting a controller together with a pokeys board. Although I think the simulation gives a good glimpse of what's going on with the motion, pls correct me if i'm wrong.

Analyzing the simulation I see that the speed is not constant throughout the longitudinal axis. I can see that the motion hiccups in a smooth but unpredictable manner. In a way it reminds me of some ants I see at the house, as they do sprints while they walk.

I try to describe as true as possible what I see happening to me. As said I will look into getting motors turning. Meanwhile I will keep playing and will also demo Auggie with 2d code and compare. We do use a Syil mill everyday and mach3:)

One Idea I had a long time Art, for the screen, is if it was possible to create transparent buttons on top of the Gcode display window. I would make a giant gcode display and buttons on it. :)

cheers, F

Francisco

[ArtF]

Hi F:

  The planner is a work in progress, as you know Tempest sat for a long while used only by
those with heavy gantries primarily.  Im still working on it to see if I can smooth it out,
but its very specific to use, small line segments still can make it twitchy, but we'll see what happens
over time.

:)
Art

[chicoxiba]

Art

I imagine that it is a big ask, I remember you writing that the maths behind tempest is scary. I have kind of followed it since the beginning and have read most forum threads i think. Even downloaded docs on trajectory planning from google.. yes, right over the top lol. The main reason I started chasing these avenues is that I saw mach3 didn't have symmetrical accel/decel ramp during G1 (I do have a heavy load for not so powerfull motors). The accel respected the motor tuning curve but the deceleration was always quite abrupt and short, still is. On version 3.043.022 Brian did make the accel/decel symmetric for me, although there are 2 other imperfections he said he would fix it but in m4. I use it a lot and it works fine, I can get good job cut times out of it.

Since then I have tried many different controls. To my surprise the chinese pendants do a brilliant job maintaining CV versus cutting true to the path (bangbang tp). But I still like to have the computer and not have to be carrying files around on a flashdrive, nice toolpath display, customizable and way more fun.

In saying that I have never known an Scurve acceleration motion controller that works good apart from those million dollar machining centers. It would be SICK if you were able to get it there.

Thanks,
Francisco

F

[ArtF]

Hi F:

  I think I will do a write up on how tempest works so everyone has a chance at
understanding the why's as to its strengths and weaknesses..


Art

[chicoxiba]

Art,

Heres something I came across a few days ago and I remembered you. This guy improved the TP for linuxcnc recently. I have never seen it in action.

https://www.youtube.com/watch?v=412N5A-N8Fc

F

[ArtF]

Thanks for the link,

I havent followed EMC for a long time in terms of its code, but I have to say I could almost
copy that video as a primer on the Tempest planner. Not the same, exactly, but close in many ways
and I can hear, as the lecture continues, many of the math papers I read in developing tempest.
 
    It really shows in soem depth just how much there is to a planner, how they are a rocket
science of their own and a facinating puzzle to play with.. and I suspect each players puzzle
shows a somewhat different picture in the end.

Art