Frage Manual Toolpath Generation using G02/G03 Arcs

Hi.

I have experienced problems with the G Code toolpath for my Red Special pickguard. Actually, I think the problem lies with TurboCAD V19.2 Pro which has a bug/glitch that sometimes causes arcs to degenerate into line segments. Unfortunately this is not a simple display issue because it translates through to the geometry, STL and MeshCAM faithfully generates G Code for it.

No matter how I much I try, I couldn't work around it or design it out. Very frustrating. I tend to build my geometries as regions in TurboCAD by placing circles, ellipses and line segments at specific rectangular co-ordinates then using 2D and/or 3D Boolean operations to create a region. This usually works well unless there is a micro-scale overlap issue which creates a minute spike in the outline.

Anyway, I realised that my 2.5D geometries can be easily reproduced by manually specifying arcs and line segments (except in the case of the elliptical arcs unfortunately) using G02 and G03 codes for clockwise and counter-clockwise arcs.

I had a quick go at this and attach the results for a semicircular plate constructed from three arc segments and three line segments. The code is 10 lines instead of 188 lines.

I attach the code if anybody is interested in trying this for themselves instead of relying on their CAM software to produce simple toolpaths consisting of circular arc segments and line segments.

If anybody knows a smart way of generating an elliptical arc segment, I'd like to hear it. Perhaps this "four centre method" will be suitable?

http://www.cnczone.com/forums/g-code-programing/312772-gcode-ellipses.html

I should point out that in the case of this pickguard outline cut, I use a 90 degree bevel cutter and the point follows the outline exactly, not at a distance equal to the radius of the cutter as with a normal type of cut. Further allowance would need to be made in the calculations with the arc size, etc. for normal cutters.

Regards.

Doug

Hi Doug, I'm off topic but I had this to pass on to you!

Many thanks for that.

My next project will be to attempt to recreate a 1959 vintage Les Paul which has an arch top similar to a violin.

I have been thinking about getting DeskProto due to some frustrations using MeshCAM, for example, it will not generate a toolpath for routing a deep (12 - 23 mm) angled truss rod channel. Part of the frustration is that it doesn't tell you when or why it won't do what you've asked it to do. Some of the issues start at the CAD stage so I can't blame MeshCAM for that.

At current Euro:GBP exchange rate (1.18), DeskProto hobby license for the multi-axis edition works out at about £220. Some older (2014) reviews that came up on Google searches commented that it had a few more bells and whistles (toolpath strategies) than MeshCAM but nobody said it was significantly better. No affordable CAM software packages use arc G02/G03 codes it seems.

I will evaluate it for 30 days (fully functional) and see if I prefer it to MeshCAM. I have better things to spend £220 on than alternative CAM software right now. My credit card bill for one.

Doug

Would be nice to see some pictures of your finished guitars Sounds like to make per day...?

I'm going with what the good Doc says

Doug schrieb: ... routing a deep (12 - 23 mm) angled truss rod channel.

You got me lost there, can you post a screenshot of the problem so I can have a look?

Yes, Deskproto has quite a variety of toolpath strategies for different tasks, and the beauty is that you can chain a number of operations with different parameters for different zones of the same model. I'm looking forward to buying a tool changer to get the complete choreography running without lifting my posterior!

For instance, I recently milled a tank turret where the toolpath for the sidewalls was made a mess of the roof and vice versa, so I had to divide the task into zones with quite different methods parameters. Deskproto took it all in without a flinch. And programming all this was just a matter of adjusting half a mm here or there, it was all quite intuitive.

DocBrown schrieb: Would be nice to see some pictures of your finished guitars Sounds like to make per day...?

Hartmut,

The only guitar I have completed to date (2014) is a 3/4 scale replica of Brian May's Red Special. I designed it but had the body and neck CNC cut commercially at great cost.

I will certainly post pictures of the finished guitars when they are available, but I have so little time to work on this hobby, I fear it will not be for many months.

Doug

MMorao schrieb: I'm going with what the good Doc says

Doug schrieb: ... routing a deep (12 - 23 mm) angled truss rod channel.

You got me lost there, can you post a screenshot of the problem so I can have a look?

Yes, Deskproto has quite a variety of toolpath strategies for different tasks, and the beauty is that you can chain a number of operations with different parameters for different zones of the same model. I'm looking forward to buying a tool changer to get the complete choreography running without lifting my posterior!

For instance, I recently milled a tank turret where the toolpath for the sidewalls was made a mess of the roof and vice versa, so I had to divide the task into zones with quite different methods parameters. Deskproto took it all in without a flinch. And programming all this was just a matter of adjusting half a mm here or there, it was all quite intuitive.

I'll post up images later. I will try again but with a channel in the cuboid stock and see if I can work out what is going wrong.

I like the sound of using multiple toolpath strategies for different zones on the geometry. MeshCAM can do this but not chaining so it gets a bit clunky. I have had to manually edit and build G Code. This isn't such a bad thing because it means I've learned quite a bit, including the G02/G03 arcs code.

I'll download the trial version and give it a try and see how it compares to MeshCAM. Better I should imagine. The biggest flaw in MeshCAM is the lack of a user manual and proper support when it doesn't work. They ask you to rely on the forum for support but only one or two expert users answer people's questions. Not ideal for a £200 piece of software.

Doug

Concerning support, DP state that direct (i.e. email, phone) support is for professional licensees only, but they've replied to my (few) queries without raising any problem.

As for the forum itself, the owner does respond in a timely, effective and friendly manner. I can only recommend.

O.k., I have attached the STL file with the 5 mm diameter, angled, rounded bottom channel in the stock only if anybody wishes to try creating a toolpath on their software.

The intention was to use a 5 mm, 2 flute FLAT nose slot drill for roughing then a 5 mm 2 flute BALL nose slot drill for finishing.

I suspect MeshCAM is unhappy because I am trying to use cutters with the same diameter as the channel I am trying to cut. It doesn't make any difference whether I select the machining margin to zero or not.

I haven't tried calculating it with smaller diameter (e.g. 4 mm) cutters yet.

Doug

I recalculated this toolpath with 4 mm flat and ball nose cutters and MeshCAM successfully produced a plausible toolpath over the full depth. As I suspected, MeshCAM didn't like it when the channel width is the same as the cutter diameter, even when the machining margin is set to zero.

I'll seek advice on the GRZ Software/MeshCAM forum to see if this obvious flaw can be circumvented.

Doug

It's not so much an "obvious flaw", but a limitation of the way that MeshCAM works.

To be honest, if you're mostly doing 2D cutouts, I would say that MeshCAM is the wrong tool for the job.
In fact, using .stl files is the wrong approach. simple 2D drawings with a 2.5D CAM program will provide much more efficient toolpaths, that will typically run much faster.

The reason you can't really get G2/G3 arcs from .stl files, is that there are no arcs in an .stl file.
An .stl model is made up of triangular faces.
MeshCAM can fit arcs and use G2/G3, but results can vary, and depend greatly on the model, and require very high tolerance settings to get decent results.
I've been using MeshCAM since the very first beta, and it's been great. But I'm also the first persson to tell y ou that it's not always the right tool for the job. Robert's added a lot of features over the years to make it a more general purpose CAM program. But it my opinion, it's still best used for it's original purpose, 3D cutting of .stl files. For 2.5D work, there are many other better options out there.
I downloaded your .stl file.
If you were using a program like Cut2D, all you need to do is draw a single line, and create the toolpath for that line. A much simpler process. Fwiw, if you were to draw a 4mm rectangle, and try to cut a pocket with a 4mm tool, most other CAM packages would fail as well. If you made the slot 4.01mm wide, it would likely succeed.

Hi,

I can thoroughly recommend Cut2D - does everything I want from a 2D package and is very quick and easy to learn. Files will run directly in UCCNC.

Excellent tutorials available on the Vectric website and demo versions have virtually full functionality bar generating output Gcode.

I should add that I have no connection with Vectric apart from being a satisfied user.

Peter

Thanks for your suggestions and comments Gerry and Peter.

I'm reasonably satisfied with MeshCAM to be honest. I accept that .STL files consist of small triangles although I don't see this limiting the geometry quality in any way.

I evaluated a number of CAM software options and decided on MeshCAM because it was optimised for 3D work since the key part of my project is the 3D carve of the guitar neck which has relatively complex curved surfaces. It also does just fine on 2.5D toolpaths but clearly is limited by some calculation logic to cutting every feature layer by layer whereas I would save time on retract and G00 travel by milling pockets individually and then go around the outline at the end. To some extent I have enjoyed playing with it to build toolpaths manually from individual pockets and outlines.

Doug