[LazyH]

Hello, sorry if I posted this in the wrong area, but I'm working on making a cnc routing machine that has two linear and one rotational axis, similar to a polar 3d printer. The reason for this arrangement is that the rotating axis/table can be made incredibly rigid, while still allowing the majority of the machines footprint to be a usable work area, about the same efficiency of space as a moving gantry design but more rigid and stronger. The only issue I've had so far is finding the right control software. Ive been looking around at different 3d printing sites for some open source code or boards I could adapt or reprogram to suite subtractive rather than additive work but marlin 360 has been my only result and I've been having trouble getting that to work especially since there's little documentation that I've found and no discussions on its use anywhere I've found so far. Any advice with the marlin 360 software or ideas for alternatives would be really great. Once I get everything in order I'll be posting to the build forums next.

[LazyH]

Nobody? If I do find a way to make this controller work or any other info on this I'll post here so people can learn more about this.

Right now I'm searching through the marlin 360 and grbl libraries in an effort to see which ones might be responsible for converting xyz cartesian movement to r0z polar movement, if I can find the right one I'm hoping I can just copy from marlin 360 to grbl so I don't have to modify anything else in 360 to work with milling instead of printing.

[James Newton]

I tried a system like that in the past... it was difficult to account for all the sources of flex in the unit. In the end, cartesian works better. But... I haven't seen your mechanical design, so it's quite possible it's better than what I was able to do.

techref.massmind.org/techref/idea/lccam.htm

As to the firmware. The project referenced as the inspiration for my attempt in the link about has an archive, and there are some good notes in there:
https://sourceforge.net/p/freeandeas...lopers/?page=1

From an earlier posting for angles t(theta) and p(phi) for the two table
rotations, where R is the radius of the main circle (and through which
the centre of the other circle and the drill bit go), and some point
(x,y) which you want to drill:

t=2arcsin(r/2R)
p=t/2+arctan(y/x)
where r=sqr(x^2+y^2)

You can reverse these to:
r=2Rsin(t/2)
x=rcos(p-t/2)
y=rsin(p-t/2)

Of course that means doing floating point in real time, but these days, that's cake.

[109jb]

You would likely be better off using LinuxCNC opposed to grbl or Marlin. BTW Marlin has grbl at its core. Anyway, LinuxCNC has a kinematic engine that can handle robotic arms which would be more complex than what you are talking about. However I don't have any personal direct experience with this. Try over at the LinuxCNC.org site they may be able to help.

[websrvr]

I've seen grbl running on a STM32 with LCD and keypad (5-axis) that supports tool DIA/Len compensation so I wouldn't say grbl isn't suitable or appropriate.

The issue with grbl is that no one who has enhanced grbl's code is contributing their code enhancements publicly which means you need to make these changes yourself or finesse the enhanced code from the creator if you know one.

I am not aware of anyone using Marlin-360 to do anything other than 3D printing related stuff and it doesn't seem to be any better than the base grbl code if it's not a fork.

There are other gcode interpreters out there but finding one that is suited to your requirements is a crap-shoot so you might have to experiment with the various flavors before settling on something.

[LazyH]

I tried a system like that in the past... it was difficult to account for all the sources of flex in the unit. In the end, cartesian works better. But... I haven't seen your mechanical design, so it's quite possible it's better than what I was able to do.

techref.massmind.org/techref/idea/lccam.htm

As to the firmware. The project referenced as the inspiration for my attempt in the link about has an archive, and there are some good notes in there:
https://sourceforge.net/p/freeandeas...lopers/?page=1



Of course that means doing floating point in real time, but these days, that's cake.

Im working on a C++ lesson from Udemy, John from cave of programming has it and it seems pretty good. So I've been looking at the files in marlin 360 in case I can borrow any of the code and it looks like there are possibly 2-5 different files that all have an effect on just the polar motion, and my brain already feels like it's melted from trying to sort through lines of code I don't yet understand, but the more I look the more it starts to make sense. I've heard a bit about hal files and stuff like that in linux cnc for controlling motion and stuff like that so I'll have to take a closer look at how to modify that in case that'll be easier, it'll definitely be more functional if I did manage to get such a machine working.

I had a plan with heavy duty tapered bearings for the table, mine can support well over a ton and a half of force, then using the extra space by each side of the table and on top of the machine to use longer than "necessary" linear guideways for increased rigidity without impinging on the usable space of the machine. It'll be easier to draw a picture, but it uses a gantry design where the entire gantry/bridge travels in the z axis and the x axis only needs to travel half the distance of the table, since the table itself can rotate the other half towards the tool. I'm trying to model in fusion 360 so I'll try to post some pictures soon. Thanks for all the replies so far.

[James Newton]

. It'll be easier to draw a picture, but it uses a gantry design where the entire gantry/bridge travels in the z axis and the x axis only needs to travel half the distance of the table, since the table itself can rotate the other half towards the tool. I'm trying to model in fusion 360 so I'll try to post some pictures soon. Thanks for all the replies so far.

I'd love to see a picture.

[kanankeban]

Hello,
Any advances in your project?, I think I´ll go for the same way, thats a R2P robot...
Regrds,
Hector