Welcome to Tribbeck.com

STLCutter 2 - Day 5b

I'm not really including this as a "build" day, as I'm not really building the machine.

I've been developing my own software for taking the 3D model I've got and converting it into CNC machine code.

The software was started about a year ago, when the machine was going to mill the whole car in one go - later on, it was modified to take the 3D model and split it into blocks, each block milled individually.

The 3D model is in a format called "STL" (STereo Lithography), which is a very basic format - it just consists of triangles to make out the object.

This gives the name of the machine - STLCutter.

Anyway, I did a bit more of the software, and modified it to produce G-Code instead of the proprietary format I'd done myself (which a PIC understood).

I then ran it through EMC2 to see how long it took - it was about the same time as the previous version. I think the reason is that the machine now accelerates to the right speed, rather than moving off as fast as possible (the PIC code didn't do acceleration, which is why I chose EMC2).

The software is similar to the image to gcode converters, except it has more knowledge about the tools I'm using (well, compared to the one in EMC2).

The algorithm it uses is fairly complex - it knows the maximum mill depth, and cuts down to that depth using a quick cut, where it'll only do every 4mm (the width depends on the size of the tool - I'm using a 5mm diameter cutter, and I've allowed a bit of overlap).

Next, it will find the nearest bit that needs to be cut, and start cutting from there. This continues until there's no more on this layer. It then does a quick cut a layer below, and repeats the process until there's no more left to do.

I decided that the find-and-cut method may not be optimal, so I changed the algorithm to a more classic scan over the model - although with the quick-cuts being performed in one go.

This didn't change the time by that much (it might even be a bit slower), so I then started work on a contour approach - this is still being developed, and I've no idea whether it'll be any faster. The hope is that I'll be able to get the machine doing more fluid movement, which allows it to keep at a fast speed for longer.

« Back to index

Updated: 2011-06-08 20:41:45 | Comments: 0 | Show comments | Add comment
© Copyright 1997-2018
Tribbeck.com / Jason Tribbeck
All trademarks are the property of their respective owners.