[deadlykitten]

hello zero

Linear X axis moving towards and away from the spindle (perpendicular to the spindle) : near Z travel end, the X axis is approaching smoothly; when Z travel begins, X axis is departing rapidly; thus X axis movement is not simetrical

But possibly you are trying to point me to similarities in threading and this operation : i try to show the differences between a predictible ( kind of a pattern ) movement, and a more random movement; that machine is pretty basic for what it does predictible movements are more easier to code, while random movements can be treated as real 3D

would this mean edit a gcode file that has been generated by CAM software : no, not quite; the file is the output; i am talking about editing the CAM source, so to alter the output

kindly

[ZeroBacklash]

x-winder is using a custom interpretor, for example G5/G6 rotate the feed head and axis acceleration seems to be part of the code is well.
Cadwind has a post for Mach3 so if you can find an example of that output it may give some ideas.

I thought Cadwind produced Gcode and Gcode could be read by a standard Gcode interpreter built in to Mach3 or other such programs. So I didn't quite understand the "post" part.

[ZeroBacklash]

I would expect that a 4 axis CAM program would map the axis correctly. Fusion 360 is supposed to have a 4th axis option, but I have never used it. When you get to 5 or 6 axis, the cost goes up exponentially.

That sounds interesting probably I can test it out today.

However, my opinion is that you are going to need some purpose built software to really get what you want. Really all that's required would be to enter the parameters of the wind, such as: length, layers, rotational speed, stepover/pass, and a number of other parameters then press ''Generate Code'' No reason to have G code as an intermediate step, outputting the native controller code would make more sense.

That would need a G code library wouldn't it?

[Jim Dawson]

Cadwind or any CAM program runs its output against a control specific Post Processor to create the G code output. The G code has to be formatted so that the G code interpreter of the chosen CNC program can read it. All G code is not created equal.. Mach3 is one of the more forgiving programs for reading G code. Most CNC controls require very strict adherence to their own proprietary format. You would have to look in the Mach3 support forums to see if Mach3 even knows what to do with a G5 or a G6 for instance.

The Cadwind Mach3 post might convert a G5/G6 to G1 moves in the output G code. Changing acceleration on the fly might be handled with a series of small G1 moves, each with slightly greater feed rate than the previous move.

[Jim Dawson]

That sounds interesting probably I can test it out today.


That would need a G code library wouldn't it?

No, the motion controller has no idea what to do with G code. The G code needs to be translated into a command string that the motion controller can understand. This is normally done in the CNC program. In the case of Mach3, it is a motion controller, but it still has to convert the G code into step & direction commands to drive the axis motors. Motion controllers understand pulses (for positioning), pulses/second (for speed), and pulses/second^2 (for acceleration).

So a G1 X1.00 Y1.00 F10 would be translated to what the motion controller can read, maybe something like LI100,100<1000>1000 So rather than make the translation, just output the native machine code directly.

[ZeroBacklash]

No, the motion controller has no idea what to do with G code. The G code needs to be translated into a command string that the motion controller can understand. This is normally done in the CNC program. In the case of Mach3, it is a motion controller, but it still has to convert the G code into step & direction commands to drive the axis motors. Motion controllers understand pulses (for positioning), pulses/second (for speed), and pulses/second^2 (for acceleration).

So a G1 X1.00 Y1.00 F10 would be translated to what the motion controller can read, maybe something like LI100,100<1000>1000 So rather than make the translation, just output the native machine code directly.

What you are trying to suggest is output the step / dir pulses directly from software instead of first converting it to g code right? That's an amazing idea. The issue I face here is how best to co-ordinate all the axis together, will a simple state machine do the job here.
GRBL and other such controller's do a good job of getting co-ordinated motion from multiple axis, but I like what you say and it seems doable.
Will a plugin for Mach3 do the job here I mean insead of feeding Mach3 gcode we can directly tell it to output the desired step/dir pulses.

Are there any existing libraries out there that do this kind of thing that is bypass intepreting G code and output step/dir signals on parallel port or in theses days on the USB port?

[Jim Dawson]

If you want to use Mach3 or similar, then you will probably have to create G code that your chosen CNC software and/or plug-in will understand. I don't know enough about Mach3 and other similar platforms to know what is possible with them and how you can manipulate them. I normally work with more industrial motion systems, but I'm pretty sure that Arduino or something similar could be programed to do what you want. What I am suggesting is really much like CAD/CAM/CNC all wrapped up in one package.

As an example, winding an electrical coil on a form is much easier to describe textually than trying to draw it in CAD and then going through the process of trying to generate a ''toolpath'' from that. An electrical coil is easy to describe by number of turns, width, wire size, ect. Winding carbon fiber on a tube is similar and can be described with a few more parameters. The textual description can be easily converted to motion commands that the motion controller can understand.

[bostosh]

Your task is not impossible
Generating the path is no problem converting is.
Check out Tow Placement or Automated fiber placement.
I had a Phd in math and computer science do the postprocessor from C/L x,y,z,i,j,k,f data to 7axis motion 0, 90, +45, -45 with cuts and adds
Just like rolling paint.....................