[Treeline]

Does anyone know if Madcam 5.0 has added using G2/G3 arcs rather than straight lines?

[svenakela]

I don't think so.
But for what purpose?

[ulexprecision]

There are still a few machines about that use meters of cardbord strip with holes in it as the metric to measure memory.
I have a machining centre build by Mazak in 1980 that has a 24 tool change
and extended memory of 240m of punced cardbord strip.
Many of the early Fanuc controls did not have much memory.
So it is an advantage to have a "frugal"use of resorces.
I hope I did not hijack this post, but that is my 5 cents worth anyway.
Cheers,
Dirk

[svenakela]

Lets say pure hypothetically, you justify the use of arcs to be able to use a machine from the early 80's when the hardware requirements for Rhino itself requires hardware only a few years old?
I use to run a machine with a 25 years old controller. With a direct feed dongle straight into the Heidenhain box serial interface we could execute any program, any size.

[MetalZilla]

What!?

[svenakela]

What!?

What?

[wdaehn]

That's actually something I wanted to know forever now. I mill an elliptical shape and it looks as if it is built out of straight segments. Depending on the position, some lines are up to 1cm long. It is not that bad that I cared much but always wanted to know if that is because of MadCam, the Processor or the Steppers. Now I get the impression it is because of Madcam. True?

[svenakela]

Yes, there are line segments, like most contact finding CAM programs make the toolpath (and that's about all who makes 3D paths). If you don't get the expected tolerances you have probably not gone through the Rhino Mesh settings as mentioned in the help file. You should also do that for your template file otherwise you have to make it all over again for each new model you make, it's saved per file so old models can have less tolerances configured too.
Also set the tolerance for the tool low.

We mill holes for bearings within very tight tolerances.

To me it's logical (I've been programming in another CAM program specifically made for surfboards), arcs can't be used to follow a shape mathematically, unless you have a super computer and can precompute the the outcome of an arc compared with the surface that is going to milled. Lets say a flat mill is running on a surface that slowly and progressively bends upwards. Lets say either milling one half of a mold for a projectile or a foam core for a surf board. How is the arc going to help out? It will not because when the arc radius is too small it will lift from the surface and when it's too big it will cut down in the surface. This will happen on all arcs computed. And what happens if the surface is bending upwards in the perpendical direction as well, like the projectile mold will when we reach the tip of the projectile? We have to precompute a radius many times before a radius is found that will not hit the surface on the sides nor on the straight line.
You never now what happens in front of a mill when flying above the surface (unless you have that supercomputer).

[wdaehn]

Thanks for the hint, I will look into these settings to get better in Madcam.

[hughesbros]

To me it's logical (I've been programming in another CAM program specifically made for surfboards), arcs can't be used to follow a shape mathematically, unless you have a super computer and can precompute the the outcome of an arc compared with the surface that is going to milled. Lets say a flat mill is running on a surface that slowly and progressively bends upwards. Lets say either milling one half of a mold for a projectile or a foam core for a surf board. How is the arc going to help out? It will not because when the arc radius is too small it will lift from the surface and when it's too big it will cut down in the surface. This will happen on all arcs computed. And what happens if the surface is bending upwards in the perpendical direction as well, like the projectile mold will when we reach the tip of the projectile? We have to precompute a radius many times before a radius is found that will not hit the surface on the sides nor on the straight line.
You never now what happens in front of a mill when flying above the surface (unless you have that supercomputer).

I'm putting this reply on 2 threads where you've stated arc fitting is difficult and you would need a supercomputer.

Well I have another CAM program that does arc fitting and it is very fast. I've attached 2 screen shots showing one file with all G01 moves 76274 lines and one with arcs fitted only 19335 lines. The conversion took less than 2 seconds!. The program first calculates the tool path with G01 moves then you can apply arc fitting.
Attachment 185466
Attachment 185468
I think the bigger challenge is the post processor getting the G2/G3 output correct in three planes wouldn't be easy.

Personally I'd be happy if just the 2d tool paths could be out put as lines and arcs, then I could use it for my wire eroder and not have to use another program.

Mark

[svenakela]

I'm putting this reply on 2 threads where you've stated arc fitting is difficult and you would need a supercomputer.

Well I have another CAM program that does arc fitting and it is very fast. I've attached 2 screen shots showing one file with all G01 moves 76274 lines and one with arcs fitted only 19335 lines. The conversion took less than 2 seconds!. The program first calculates the tool path with G01 moves then you can apply arc fitting.
Attachment 185466
Attachment 185468
I think the bigger challenge is the post processor getting the G2/G3 output correct in three planes wouldn't be easy.

Personally I'd be happy if just the 2d tool paths could be out put as lines and arcs, then I could use it for my wire eroder and not have to use another program.

Mark

Are you willing to test your CAM program against one of my models?
I see no 3D pathing in your screenshots. I see a toolpath that makes profiling in Z steps. Can you make the same post processed toolpath (because I guess you do a post analyze on the path) with a X/Y directional cut instead of a Z-level profiling and how much precision is lost (output fault tolerance)? If that's the case, make a model with a surface with a radius in XZ and a progressive curve in YZ, if you can do that with a sweet output i will be impressed.

Besides that, my old Heidenhain couldn't make G2/G3 in Z, only helicals with Z data. So it wouldn't help if the program is supposed to be arc based to save lines.

[hughesbros]

I just grabbed the fist model that came to hand, here's one that shows arc fitting in the ZY plane goes from 57581 lines to 4091 lines. My old Heidenhain TNC2500 will give far higher feed rates after arc fitting. Yes there is a fitting tolerance but you can generate the first set of tool paths to a higher accuracy then fit the arcs to an accuracy that gives the finish you require.

By the way I have mainly switched to madCAM now and I'm not knocking it at all, it's just it would be better for me if it could generate G02/G03 codes.

Mark
Attachment 185518Attachment 185520

[svenakela]

That was much better of a case! Really sweet.
How much do you loose in tolerance after the post analyze?

In the CAD/CAM I was programming in once upon a time there was a prototype of a similar function, but it was skipped as the curves never proved to give good enough output compared to the tolerances needed, or the computer went into bizarre-arcs-generation-mode and the output wasn't much smaller...

Hmmm... I just got an idea, must do some Rhino scripts and make a test...

[hughesbros]

Here's an example with before and after(Dark blue after), using Rhino's length command on the red line give a length of 0.00mm using What gives a line length of 0.0023. The line count went from 7000 to 855.
Attachment 185576

It's not just file size where it helps recently I had a small graphite electrode to machine is was 30mm wide and 60mm long. I had to machine it with a 1mm ball nose to get into some of the finer detail, one pass across the 30mm direction produced 140 lines of gcode. The control struggled to get a feed rate of 200mm/min in places, at 250mm/min thats over 1000 lines of code per minute and the drip feed could not keep up. Arc fitting reduced 1 pass across to 34 lines.

Mark.

[svenakela]

Mark, could you check your parameter 60 and 61 in your Heidenhain controller?
Depending on their setting you can speed up the machine a lot.