TCP 5-axis kinematics

More
04 Apr 2018 16:03 #108387 by jsskangas
Replied by jsskangas on topic TCP 5-axis kinematics
Finally Its alive!!!

I will take some video tonight.
Thanks for all of you that helped me.

Im still having a encoder count rate problem I need to solve first.

Please Log in or Create an account to join the conversation.

More
04 Apr 2018 17:46 #108391 by jsskangas
Replied by jsskangas on topic TCP 5-axis kinematics
Here is video of transferring between two different kinematics.


I will need some time to finish this setup, but I will upload it to be used and further developed.
The following user(s) said Thank You: PCW

Please Log in or Create an account to join the conversation.

More
08 Apr 2018 20:56 #108625 by jsskangas
Replied by jsskangas on topic TCP 5-axis kinematics
I did first run on Full 5-axis.

Please Log in or Create an account to join the conversation.

More
08 Apr 2018 21:45 #108629 by andypugh
Replied by andypugh on topic TCP 5-axis kinematics
Nice!
Where do you get the CAM from?

Please Log in or Create an account to join the conversation.

More
09 Apr 2018 02:21 #108638 by jsskangas
Replied by jsskangas on topic TCP 5-axis kinematics
I work as a CAM expert in company that provides Siemens PLM Products.
My work is to make (mainly coding) post processors and machine tool simulation models. (NC code simulation )

Please Log in or Create an account to join the conversation.

More
09 Apr 2018 09:11 #108647 by andypugh
Replied by andypugh on topic TCP 5-axis kinematics

jsskangas wrote: I work as a CAM expert in company that provides Siemens PLM Products.
My work is to make (mainly coding) post processors and machine tool simulation models. (NC code simulation )


Ah! That's cheating :-)

It does mean that you are probably in a good position to say if LinuxCNC works "properly" in multi-axis mode. I understand that the blending of non-linear is a difficult problem and one not fully addressed even in the new trajectory planner.

Please Log in or Create an account to join the conversation.

More
09 Apr 2018 10:17 #108648 by skunkworks
Replied by skunkworks on topic TCP 5-axis kinematics
Yes. Any motion involving rotary falls back to one segment look a head.

Please Log in or Create an account to join the conversation.

More
09 Apr 2018 17:14 - 09 Apr 2018 17:22 #108666 by jsskangas
Replied by jsskangas on topic TCP 5-axis kinematics
I just got full 5-axis working.
That video was a second attempt on full 5-axis milling.
first attempt was that same roughing in half way.
At the moment I cant comment yet. I need to run whole impeller and finishing path to see results.

One thing I noticed, there was considerable amount of Jerk in movements.
I did set acceleration down to compensate that.
I'm running 1800 mm/s2, on my axis, for 5-axis work 700-900 max.
Only limiting factor is jerk. Whit S-curve acceleration I could go to "close" 1G.


What I can say is that Linuxcnc is maybe the most user friendly control system I have used.
That is my personal opinion.

But it is far from commercial systems, regarding trajectory planner.

What are done better in commercial control systems are kinematic support, 5-axis trajectory planner.
S curve accelerations ect..

5-axis trajectory planner or even 3 axis S-curve acceleration is not easy task to make.

S-curve acceleration / path smoothing would be my no. 1 to be developed in linuxcnc.
So that amount of smoothing and size of S-curve could controlled from G-code.
This way one could run smooth roughing paths and tight finishing paths.

Please do not take this as a fend for Linuxcnc, I like it and I will support it as best as i can,
Last edit: 09 Apr 2018 17:22 by jsskangas.

Please Log in or Create an account to join the conversation.

More
09 Apr 2018 22:04 #108678 by robh
Replied by robh on topic TCP 5-axis kinematics
nice work, will you be sharing some of the setup , as would be intresting to see how got this to work and any info on setting it up etc..

Please Log in or Create an account to join the conversation.

More
10 Apr 2018 05:16 - 10 Apr 2018 05:28 #108704 by jsskangas
Replied by jsskangas on topic TCP 5-axis kinematics
Hello

I included nc files and setup that hold kinematics source file as well.

Nc files has one example, but please be careful running it.
My post is not complete yet and at operation endings it does not lift Z home between operations.
So when it crosses over impeller it will hit a bolt that holds impeller down.
So if someone wants to run that nc file fix each operation ending so that it will lit tool higher.
It is only includes rough cut operations, no finishing paths.

Another limitation in Linuxcnc is that you can only change kinematics when rotary axes are at 0.00 position.
This limits things in some degree. You can not pre-index part with rotary axes, then move close and turn TCP on.

I made my own post so that it evades most of collisions.
It seems to work best this way:
G0 Xxx.xx Yyy.yy
G43 Hxx
M128 (TCP ON)
Ccc.cc
Zzz.zz Bbb.bb
.
.
.
.
.
.
G0 Zzz.zz B0.
G0 C0.
M129 (TCP OFF)
G91 G28 Z0.

Kinematic definitions are located in HAL file:
#############################################3
# Kinematics

setp XYZBCsmkins.X-offset 0.0
setp XYZBCsmkins.Z-offset 0.0

setp XYZBCsmkins.X-rotpoint 103.0820
setp XYZBCsmkins.Y-rotpoint -0.861
setp XYZBCsmkins.Z-rotpoint -455.1930
net tool-offset motion.tooloffset.z XYZBCsmkins.Tool_Z_Offset

X-offset and Z-offset are misalignment between rotaary axes, only need to set X.
I have not yet measured my own offset value.

X-rotpoint, Y-rotpoint, are center of C-axis rotation in regarding your machine zero.
Z-rotpoint is B axis coordinate from machine zero.
Attachments:
Last edit: 10 Apr 2018 05:28 by jsskangas.
The following user(s) said Thank You: Shu

Please Log in or Create an account to join the conversation.

Time to create page: 0.221 seconds
Powered by Kunena Forum