[psorey]

I am familiar generating G-Code for an X-Y plasma table.

My new 3-D printing project has circular rails, 1) full 360-degree track in the X-Y plane for rotation around the Z-axis. 2) on that track, standing up vertically at 90 degrees to the circular track, rotates a half-circle arch gantry, and, 3) along that gantry travels a tool carriage holding a linear actuator with the printing nozzle that points towards the center of the hemisphere circumscribed by the arcs.

The machine is built. I'm pretty sure there is no out-of-the-box solution for generating the G-Code to place the nozzle at a particular point in cartesian coordinate space. G-Code assumes the tool travels along linear axes. We plan to use KFLOP for motion control.

My current plan is to convert back and forth between spherical and cartesian coordinate systems: We have a point xyz in cartesian coordinates, we translate that point into the machine's spherical coordinate system to find the distance that each actuator needs to travel to arrive at that point, generate the G-Code from those distances (xMachine,yMachine,zMachine) and I feed that point into the KFLOP which moves the nozzle to the desired xyz point!

Planned method for doing this is to build a 3-D model of the desired shape, query the model at regular intervals for xyz along the desired toolpath / translate xyz to machine spherical coordinates (latitude, longitude and 'elevation' or distance from the center point) / generate G-Code from that / send to KFLOP and we're there, right?

Thank you for any insights you may have.

[alan_3301]

can you post a pic of the machine?
Is this a known kinematics type or totally original?

[psorey]

Totally original. Let me get back to you on that.

[TomKerekes]

Hi psorey,

Very interesting. I think you have a choice of two fundamental approaches.

The one I believe you describe where you perform the Kinematic Conversions before creating the GCode where xyz coordinates would then map directly to your actuators. In this case I think you will need to also include special feedrate commands (probably using inverse time GCode mode) that you also calculate to produce the desired actual feedrate.

The other approach would be us use our Kinematic Layer in the KMotion Libraries to handle the Kinematics. The GCode could then be in standard xyz space. You should also be able to Jog in xyz space. In this case you would need to create a Kinematics module and compile it in. See the Kinematics3Rod.cpp example. Here is a video of a simple mechanism using that module (sorry for the poor quality).

https://youtu.be/MV42zmiT_4I

Regards

[psorey]

Thank you, I will check it out!

[psorey]

So, on this machine the X-axis is 'infinite'. How will I handle the transition from 359 degrees to zero degrees in the KFLOP without it going BACKWARDS to 0?

Also, I am looking for someone familiar with KFLOP so I can pick your brain, and you can help me set up a 'testing harness' with a power supply, AC servo, a stepper, an encoder, a limit switch, and ?? I can do this but am under a tight deadline and have other systems to work on. I am in Seattle. Will pay for a couple of days' time. My shop is in West Seattle. (is there another place i should post this?)

Thank you,

Paul

[alan_3301]

I will let Tom tell you for sure, but most controls have options for wrapped rotary axis, that would set whether it goes back to zero or not.

[psorey]

cool!

[TomKerekes]

Hi Paul,

So, on this machine the X-axis is 'infinite'. How will I handle the transition from 359 degrees to zero degrees in the KFLOP without it going BACKWARDS to 0?

You haven't explained your overall approach, but we don't currently have any special provisions for that. To move 1 degree forward from 359 degrees to 0 degrees would require you to command a move to 360 degrees

Also, I am looking for someone familiar with KFLOP so I can pick your brain, and you can help me set up a 'testing harness' with a power supply, AC servo, a stepper, an encoder, a limit switch, and ?? I can do this but am under a tight deadline and have other systems to work on. I am in Seattle. Will pay for a couple of days' time. My shop is in West Seattle. (is there another place i should post this?)

You might also post on our Yahoo Group
https://groups.yahoo.com/neo/groups/DynoMotion/info

Regards