TCP 5-axis kinematics

Hello

I got my core2 duo dumpster find working and linuxcnc installed.

I took look into vismach sim xyzbc kinematics and there are some problems.
B and C axis rotations are incorrect.

CAM software's output toolpath as tool tip location + IJK vectors.
This is called CLD (cutter location data).
Post-processor take this data in and translates it to NC-code and calculates rotary axis position from tool orientation.

In general all CAM outputted nc-code is tooltip data and tool orientation.
It is a machine position where tool is located and oriented as it is in CAM.

All machines I have seen until now has followed this same rule.
To layman's therms, you don't move machine axis you move tool, this applies to rotary axes as well.


If you have machine like xyzbc vismach sim example, X axis is table and moves part.
To move tool X+ direction , part and table needs to move X- direction.
Like it does, and it is correct.

But same rule applies to rotary axes.
When you look trough axis to its positive direction, CW rotation is positive rotation.
When you rotate tool positive direction it would be CW.
But in this case positive rotation of B and C axis would be CCW, cause we are rotating part.
Then tool moves CW relative to part.

Current vismach xyzbc sim example kinematics does not behave this way.
Both B and C axis rotation is incorrect.
I assume that xyzac kinematics have same issue.

Do we change these?

andypugh wrote:

plopes9000 wrote:

Finally, see attached test gcode file and video.

I thought that your video was very instructive about what TCP is, and the Linuxcnc home page needed a new Showcase.

So your video is now on the front page of LinuxCNC

Andy, could you please update the TCP feature video on the home page with this new video link?



Had to create a new version - thank you !

Hello All, This is a really neat feature!
I am still reading through this thread and I read a lot on AC and BC configurations is it possible to use this "switchkins branch" on an AB configured machine. I have a rotary table for the A axis and the spindle rotates in the B axis. I have attached a picture for clarity.
I guess I am confused as to how I would go about handling the changing offsets between A and B. I hope I am over thinking it.

How do i compile this switchkinetics.c file to get this action

is the Joint numbering open from 0 to max joints e.g 0=x 1=Y 2=z 3=A 4=C ?
or is it fixed as in AXIS_0=X 1=Y 2=Z 3=A 4=B 5=C

Hi there
thanks for this great forum and for your support to newbies like me, I come from Mach3 experience and
I have installed the linuxcnc mint 19.2 RT 2.8 pre, I want to test the xyzac-trt-kins on my 5 axis mill table/table, I have read the topic but I have not understood where I have to put the the trt values of x.rot.point, y.rot.point, z.rot.point (vs the machine0 home), maybe in the machine.ini file together with the Y and Z offset of the A/C axis?

next I will test your great project regarding the kinematics switch in order to activate RTCP function during gcode execution , but I need to adapt the xyzac kinematics to a xyzab one cause my mill is an horizontal machine, I can simulate an A/C configuration homing the tilt table at 90°, but normally my turning table is along Y axis (in my case the vertical axis) as a normal trunnion table,
someone has done a kinematics mod for an horizontal 5 axis mill?

thanks in advance
hello from Italy!

where I have to put the the trt values of
x.rot.point, y.rot.point, z.rot.point (vs the machine0
home), maybe in the machine.ini file together with the
Y and Z offset of the A/C axis?

Setting hal pins ('setp' commands) can be included in a
file specified as [HAL]HALFILE=filename or within the
ini file using the [HAL]HALCMD=command

Ref: linuxcnc.org/docs/master/html/config/ini...ig.html#_hal_section

The latter method is used in some of the s_kins branch sim
configs:

github.com/LinuxCNC/linuxcnc/blob/dgarr/...ng/xyzbc-trt.ini#L44

Notes:

1) In the s_kins branch, both the xyzac-trt-kins and
xyzbc-trt-kins kins modules include
xyz[ab]c-trt-kins.[xyz]-rot-point hal pins but are not
implemented within the vismach guis and are lightly
tested.

2) The s_kins branch is usually force-pushed, notes
are updated at:
www.panix.com/~dgarrett/stuff/s_kins_guide.txt

3) If someone writes *and* tests kinematics functions
for an xyzab configuration (following the design
pattern implemented in trtfuncs.c), they could likely
be added to the branch

Ref: github.com/LinuxCNC/linuxcnc/blob/dgarr/...inematics/trtfuncs.c

I don't understand because pin motion.tooloffset.z in xyzbc-trt.ini eample, not is refreshed whe use a tool with a positive tool lenght value.
How kinematic can calculate if not have this value .
In hal file existe connetion, but the pin motion.tooloffset.z is alway zero value.

Thanks

the pin motion.tooloffset.z is alway zero value

maybe try using g43/g49 to activate/cancel tool length compensation

Ref: www.linuxcnc.org/docs/html/gcode/g-code.html#gcode:g43
Ref: www.linuxcnc.org/docs/html/gcode/g-code.html#gcode:g49

more thanks.

Hi.
I would like some clarification about the kinematics parameters.
The parameter xyzbc-trt-kins.x-rot-point, xyzbc-trt-kins.y-rot-point, xyzbc-trt-kins.z-rot-point.
On my machine, I have linear, incremental encoders. The machine performs the zeros and axes are distant from the center of the table X = 830mm, Y700mm and Z 750mm.
After finishing the references the machine position for X, Y, Z has a value of zero.

Should I write the values of x = 800, Y700, Z750 in the file .ini [JOINT_0] HOME_OFFSET = X .ecc. or in xyzbc-trt-kins.x-rot-point?

Thanks