Question : 5axis simultaneous setup in LCNC

Hello world,
A new subscriber here to ask some questions.

I am playing with LCNC inside a virtual box, 2.8 dev version and to grab some idea on multi-axis configuration.
I can successfully install and launch the program to see how kinematics works and here I am facing a challenge.



(ps. okay, I got how to insert link... the above is when I input g0A45, and g0A-45 only.. nothing else)

I just captured and uploaded this just to ask a question.

Why Y/Z axis move (seems like linked) together when I put G0A45 or G0A-45 here??
Weird thing is that while rotating A, I am clearly seeing Y/Z move together but DRO shows no change.
It does seem like kinematics/graphics engine desperately tries to point its original spot on the surface while keeping same distance but I don't get this when cam software does all that kind of complex coordinate transformation internally and makes me gcodes.
Is it just a matter of graphical showing issue?
I am trying to modify and cleanup a postprocessor and I want my linuxcnc running inside my workstation for immediate test and proving.
Can someone tell me if it's from vismach gui script or kinematic engine itself?

Regards,
KevKim

If you are doing all the kinematics work in CAM then you don't want to do it again in Linuxcnc. Just set up a config with trivkins for this.

Use a 5 axis kinematics configuration if your CAM is outputting simple Cartesian code and you want Linuxcnc to do the tool tip calculations. (What it looks like you have in your simulation.)

I always change the youtube link to https: //www.youtube.com /watch?v=id instead of the youtu.be/id syntax. Maybe someone can add this feature?

Exactly, it is the kinematics that does it. You surely have a kinematic where x,y,z,a,c means the tool's position and orientation. It is up to linuxcnc to position the axis' so this happens. You get the effect that several axis can/will move when you think you ask it to make a simple translation.

Hi,
many tnx for the quick answer!

So then what I will be needing is trivkins instead of xyzackins.
I picked up a sim-9axis, which already has trivkins, and inserted vismach gui within .ini file
and here comes a new challenge.

# vismach xyzac-trt-gui items
HALCMD = loadusr -W ./xyzac-trt-gui
HALCMD = net :table-x joint.0.pos-fb xyzac-trt-gui.table-x
HALCMD = net :saddle-y joint.1.pos-fb xyzac-trt-gui.saddle-y
HALCMD = net :spindle-z joint.2.pos-fb xyzac-trt-gui.spindle-z
HALCMD = net :tilt-a joint.3.pos-fb xyzac-trt-gui.tilt-a
HALCMD = net :rotate-c joint.5.pos-fb xyzac-trt-gui.rotate-c
HALCMD = net :tool-offset motion.tooloffset.z
# HALCMD = net :tool-offset trivkins.tool-offset xyzac-trt-gui.tool-offset
# HALCMD = net :y-offset trivkins.y-offset xyzac-trt-gui.y-offset
# HALCMD = net :z-offset trivkins.z-offset xyzac-trt-gui.z-offset
# HALCMD = sets :y-offset 20
# HALCMD = sets :z-offset 10

Remark-treated lines are making error due to absence of pins(that's what error message saying).
Without these lines, gui model works as I wanted it to behave.

Should I add pins to .hal file? Is that all I need to work it or any further process needed?
If so, what material to read to learn these stuffs?
I am no software expert, no linuxcnc expert.
I just want it to be setup for my work.

environment : current wheezy version(stock iso) + LCNC 2.8 recent dev version(stock built one) inside vmware

regards,
Kevin

One thing that just hit my head is that....


Do I really need those offsets for trivkins?
It looks like they are needed because they have to move in harmony, but with trivkins, I don't think I have to consider that.
Am I in right direction?

As far as I understand you don't need the offsets in linuxcnc.
With trivkins you need that info in the CAM post processor. The machine description is now
moved to the CAM system.

With the little experimentation I have done on this subject, I think you are best off with a tcp kinematics
(that is, NOT trivkins) for a 5 axis machine. The movements are interpolated in world coordinates, not in
axis (or rather joint) coordinates.
The challenge comes when jogging. Fortunately it is possible to hit "$" or F12 in many guis to enter joint jogging mode..
Jogging then takes place in joint coordinates and one doesn't need to twist the brain.

Tnx all,
I am pretty much noob at this subject, just used whatever provided so far.
To me, I think giving CAM same geometry and see it working is more straightforward to understand.
Hence I today worked a lot to resolve this issue.

I guess its more than 50% working as intended.
- modified vismach gui for the size/work envelope
- trivkins
was never an easy job for me to complete with little accumulated understanding....



Anything weird?

KevKim

IN Most cases the Machine builder will try to get the Rotation Point in the AXIS mid point
so all is zero lined up and the Standard postprocessors of all multiaxis CAM will get yoiu a perfect Gcode

so it is as a Machine joints Cam not just build a machine and hope the cam will fit your need

Quite some hard learning while customizing this software for my needs.
As mentioned, whole environment is from most recent distribution of linuxcnc iso file,
then deleted 2.7 and added 2.8, inside VMware.

This is the part of my .ini file, which is a sample sim and I modified just a little bit

---

[DISPLAY]
OPEN_FILE = ./examples/impeller-7bl-xyzac.ngc
JOG_AXES = XYZC
DISPLAY = gmoccapy
POSITION_OFFSET = RELATIVE
POSITION_FEEDBACK = ACTUAL
MAX_FEED_OVERRIDE = 2
PROGRAM_PREFIX = ../../nc_files
INTRO_GRAPHIC = emc2.gif
INTRO_TIME = 1
EDITOR = gedit
TOOL_EDITOR = tooledit z diam
GEOMETRY = -A-B-CXYZUVW
TKPKG = Ngcgui 1.0
NGCGUI_FONT = Helvetica -12 normal
NGCGUI_SUBFILE = test-xyzac.ngc

---

I tried some other GEOMETRY order but nothing is satisfactory.
Vismach is working correctly, I think it happens only in the toolpath window within linuxcnc.
Tried normal axisgui too but result is same - which I think is natural, they use same graphics module....

Here is my progress, LCNC screen is starting at about 2:10.
Can someone bring me some light on this toolpath-drawing issue?

regards,
Kevin