is there 5 axis

[sliptonic]

... sponsors willing to make 5 axis machines available to developers...

.
Isn't this problem really a matter of seeing with your own eyes how the tip of the milling tool moves in space?

Not really otherwise simulation would be enough. Can you spot where the toolpath goes wrong in this model? Besides, a lot of what happens with toolpaths are to compensate for real-world effects. For example, in theory it shouldn't be necessary to mill outside the boundary of a pocketed area. But anyone who as actually run the CNC will know how important the edge extension feature is. There's just no substitute for seeing how the machine actually runs the path.

mess.png (80.2 KiB) Viewed 2601 times

If so, I'd say we design a small and simple 5-axis machine made for 3D printing. Then it would be "just" a matter of getting your hands on a printer and some stepper motors.
I mean, just to make the code working, you shouldn't need a perfect real Mega $ machine, right?
5-axis milling for the people!

I'm not suggesting that a developer needs to own a 'perfect real Mega $ machine'. Sometimes a hobby-class machine is fine. But access to real hardware is essential and if screw-around factor is too high then it just isn't worth it. We have the drag-knife dressup because Donek donated a drag knife that I could experiment with.

Also, it was a lot of fun! Chrisb is exactly right on that point.

[Thorhian]

It's too bad I don't plan on building a 5-axis machine anytime soon, although I do want to lol. All I will be able to do is test 3-axis code.

[sliptonic]

Hi Pauvres_honteux ,

... 5-axis machine made for 3D printing. .

Is there a 3D printing system which needs/uses 5 axis ? I don't know one...

Hmm. Interesting idea. Slice a model with a series of nested spherical 'shells' starting at the inside and progressing out. Instead of building the 3D print from the bottom to the top, print it from the inside out on a 5 axis machine. I dunno. Would that work? It would be fun to see it print!

i think one 5 axis machine isn't enough, because there are different types of machines out and i think they need differnt code to.

Most use some flavor of gcode so we can account for differences in the postprocessor. But when a machine has a special capability, then it's a different story. For example we have the drag-knife dressup which is great. But there's also a thing called a "tangential oscillating knife" It cuts kind of like a dragknife for leather, cardboard, cloth, etc but actively turns the knife head. It needs special gcode to control the head position. I would love to be able to support TOK but I don't have one to play with.

Since about 2 years now i'm thinking about building a little 5 axis router, but i'm not sure how to build the B axis...
A turns the bridge of the router, but for B i'm changing between turning the spindle or just turning a part of the table. Turning a part of the table would be more easy, turning the spindle allows bigger parts.
But for buth machine types the code must be different to in my mind...

Or you could mount a trunnion to the table and mount the part to the trunnion.

[routalot]

I'm having a hard time imagining a 5 axis printer working,wouldn't the filament just sag with the head tilted?I am forever encountering descriptions of the machines people would like to build with the ability to machine a tall work envelope and no understanding of the potential for parts of the machine fouling the job.I didn't know there was much development of the 3+2 capabilities under way as that could be very useful.Even so,it would be nice to have the ability to add boundaries to the machining area to prevent the travel taking corners of the machine into parts of the workpiece.I look forward to seeing how this aspect of a great piece of software develops.

[Pauvres_honteux]

Is there a 3D printing system which needs/uses 5 axis ? I don't know one..

Try searching youtube for "additive manufacturing robot arm". Not all 3D additive manufacturing systems on this planet consists of molten thermoplastic.
.
In addition to that I had the fortunate to experience, in first hand, an offer from a spanish manufacturer who showed us how to manufacture an instrument panel structure with a robot arm.
The end of the arm consisted of an epoxy extruder + a fiber glass conveyer/feeder + a laser-like beam of (for that particular epoxy) a specific wave length.
With that setup they were able to create almost any 3D shape with non shopped fiber glass!

[Pauvres_honteux]

Can you spot where the toolpath goes wrong in this model?

To overcome the "stripieness" I would try to break down the model into its surfaces and process them individually.
E.g. an inner radius will look good if you mill it paralell to its centerline.
Perhaps this will even be possible to run in paralell threads on the computer?
How about this processing order:
.
- Divide the model into a square mesh/net and number each square uniqly.
- Within each square extract the surfaces and number them, say clockvise.
- Determine for each surface its properties, planar, radius, single curved or double curved.
- Planar surfaces can be milled in any direction, go to next surface.
- Radius surfaces shall be milled paralell to its center axis, i.e. extract its center axis. This axis will probably determine the main milling direction for this square. Store the axis unit vector in conjunction with the square number.
- Single curved surfaces shall also be milled along its curvature axis. To figure that out we need to come up with an algorithm that can identify that axis. One way forward could be to check for extrema points in surface normal direction. Then connect these dots/points with a curve and check that curve for smoothness. If okey, then that curve will dictate the milling direction for that surface.
- Double curved aurfaces may not gain anything in appearence milling direction vice?
.

... access to real hardware is essential and if screw-around factor is too high then it just isn't worth it.

To me that sounds as if it's not worth start programming at all if real world hardware is not in the hands of the programmer?
Wouldn't simulation be good enough to start the programming adventure?

[sliptonic]

To me that sounds as if it's not worth start programming at all if real world hardware is not in the hands of the programmer?
Wouldn't simulation be good enough to start the programming adventure?

Sure. And people have messed around with it. But this thread (and a LOT of other threads and tweets) are asking why we don't have it now and why progress isn't faster.

I think almost everyone who's working on Path has a CNC machine in their garage/basement/workshop. Every one of these machines that I'm aware of is a 3 axis mill or router. Some have a 4th axis rotary and there might be a CNC lathe in the mix but that's it. I don't know of one developer that has a 5 axis machine or even has access to one.

Eric Raymond said, "Every good work of software starts by scratching a developer's personal itch." If people want to see more progress in this area, the only thing they can do, besides coding it themselves, is cause the developers to get itchier.

[routalot]

Can you spot where the toolpath goes wrong in this model?

To overcome the "stripieness" I would try to break down the model into its surfaces and process them individually.
E.g. an inner radius will look good if you mill it paralell to its centerline.
Perhaps this will even be possible to run in paralell threads on the computer?
How about this processing order:
.
- Divide the model into a square mesh/net and number each square uniqly.
- Within each square extract the surfaces and number them, say clockvise.
- Determine for each surface its properties, planar, radius, single curved or double curved.
- Planar surfaces can be milled in any direction, go to next surface.
- Radius surfaces shall be milled paralell to its center axis, i.e. extract its center axis. This axis will probably determine the main milling direction for this square. Store the axis unit vector in conjunction with the square number.
- Single curved surfaces shall also be milled along its curvature axis. To figure that out we need to come up with an algorithm that can identify that axis. One way forward could be to check for extrema points in surface normal direction. Then connect these dots/points with a curve and check that curve for smoothness. If okey, then that curve will dictate the milling direction for that surface.
- Double curved aurfaces may not gain anything in appearence milling direction vice?
.

... access to real hardware is essential and if screw-around factor is too high then it just isn't worth it.

To me that sounds as if it's not worth start programming at all if real world hardware is not in the hands of the programmer?
Wouldn't simulation be good enough to start the programming adventure?

The quoted post adds some interesting thoughts.I have used other software that allowed you to draw boundaries in any construction plane that could be used to limit the area for which toolpaths were to be calculated.This is really useful if you have a model that has a region with a particularly small fillet radius as you can use the boundary to limit a small tool to the small fillet radius and to use a much larger tool,with a correspondingly larger stepover,to machine the rest of the surface in a quicker and more efficient manner prior to using the finishing tool.As things are,I have enough difficulty with post processing individual toolpaths as I don't have a toolchanger and have yet to find a way to post process individual toolpaths.

[Thorhian]

Eric Raymond said, "Every good work of software starts by scratching a developer's personal itch." If people want to see more progress in this area, the only thing they can do, besides coding it themselves, is cause the developers to get itchier.

So what you are saying is that as a community we should fund one or more devs to get a 5-Axis machine to make them "itchier"? *Evil Laughter*

[sliptonic]

Eric Raymond said, "Every good work of software starts by scratching a developer's personal itch." If people want to see more progress in this area, the only thing they can do, besides coding it themselves, is cause the developers to get itchier.

So what you are saying is that as a community we should fund one or more devs to get a 5-Axis machine to make them "itchier"? *Evil Laughter*

Well, yes. But we can think more creatively than that. Owning the stuff isn't what's important. it's access. There's plenty of room to share some equipment. I certainly don't want to end up with a lot of hardware that I don't use. I have the Donek drag-knife sitting on a shelf that I used when developing the dressup. Cool piece of hardware but I only use it rarely. If someone has an idea to improve the dressup or implement a new feature, contact me. I'll happily drop it in the mail.

Second, there are other ways to make a developer itchy. How about a community effort to design a low-cost plasma cutter? I have in mind a design that would work well in maker spaces and could be built for a couple hundred bucks, not counting the plasma torch. I started a design but haven't had time to finish it.

The more people get involved in projects like this, the more inspiring and engaging they become.

plasma.png (27.14 KiB) Viewed 2435 times