Gimbled laser head for 5 axis 2.X design

[dirktheeng]

I was watching the video for the Dolphin Wire EDM CAM Generator, it looks like we could fairly simply use this program in an ideal way for lasers. You can set thread through and remove positions for the wire wich we can use a post prossessing script to remove those lines and replace them with laser on/off commands. It will also do the 4 axis processing to do tapering and any angle you want off of 2d profiles with loads of radiusing and champhering options. I think this could be very usefull and it's only $995. I think we can do this. There may be even cheaper Wire EDM CAM solutions out there.

[macona]

This project might interest you:
http://www.kickstarter.com/projects/109 ... -machine-p

You get tonnes of diy 3-axis routers but 5-axis versions are very rare mostly because of the software issue. This guy outlines a few options although I don't know how mature they are.

I have seen that router table in operation on youtube. From the operation you can tell he is not using an decent software to create code as the tool is constantly crashing against the walls he is routing.

[macona]

I was watching the video for the Dolphin Wire EDM CAM Generator, it looks like we could fairly simply use this program in an ideal way for lasers. You can set thread through and remove positions for the wire wich we can use a post prossessing script to remove those lines and replace them with laser on/off commands. It will also do the 4 axis processing to do tapering and any angle you want off of 2d profiles with loads of radiusing and champhering options. I think this could be very usefull and it's only $995. I think we can do this. There may be even cheaper Wire EDM CAM solutions out there.

Wire edm is nothing like a laser cutter. Think of a wire edm as a 4 axis machine. X-Y (top), and X1-Y1 (bottom). The wire is fed though guides at the intersection of these two, the offset creates the angle. It will not create useful code.

You would have a better chance using a hexapod table. But still you are going to run into software issues there.

4 axis mill software wont do squat for a laser cutter. The toolpaths will not be useful either.

To build a cutting head look at existing commercially available heads and scale and cheapen it.

[dirktheeng]

I was thinking we coudl set the upper X1 and Y1 to be the gantry position, and set up some simple shift registers to read the output of both X1, Y1 and X2, Y2 and calculate the angle to set the head at. This wouldn't be hard, but I forgot that you would have to move the Z axis up when at a tilt to keep the focal distance right and that is going to move the work piece out of the path defined by the wire edm code unless they have a way to vary the distance of the top wire guide form the work piece. Otherwise, we would have to compensate for the X1 Y1 position on the fly and that isn't easy or simple.

by the way, you would have the same trouble if you used the "conventional" laser head with 2 axis each with a 45 degree mirror and the problem would be even harder because you have 2 axis offsets to worry about. At least with this kind of gimbled design, there is ony 1 axis offset to worry about.

That said, the math to generate a tool path for cutting this isn't so hard once I define the edge shape I want to cut. I can do that with vector algebra. I can calculate the 3D tool knowing just 2 profiles, botom and top and a known thickness. I can project the 3d path onto a horizontal plane to get the X - Y coordinates that the head needs to travel at as well as the x - y angles as a function of perimeter length. The projected co-ordinates can go to the standard cam software and to mach 3. Then we can listen to the steps and get absolute path length and set the z, U and V axis as a function of path length. That may work assuming that the z u and v axes respond faster than the x - y axis.

I don't know, that could be fairly hard to make work right, but it would be fun.

[dirktheeng]

Here is an example tool path drawn up in autocad.

Isometric

Here the bottom curve represents the botom of the work peice. The blue line represents the top. The side profile is a sweep with the top and botom as rails. The orange curve between the green and blue curves represents the focal point of the laser. I can then use anoter sweep to calculate the top curve which is the actual tool path that the laser head needs to have relative to the piece.

Here are the top, right, and front views.

TP Top

TP Front

TP Right

This path could be exported and used to control the 3 axis of the laser. We could calculate the angle at each step point and send that to the arduino which can independantly set the angle of the head when it gets step impulses.

[dirktheeng]

I think I am going to have to give up on this one for now. I really need to write a specific Gcode generator / cam for 5 axis control, but I need more experience to do this. This kind of CAM would be significantly easier to do than calculating tool paths for a 5 axis mill, but still more time than I have right now. Does anybody know of a good open source cam program that generates G Code compatible with Mach 3? I took a look at gcam, it seems pretty advanced but exports code only for EMC2.

[thedudevt]

Open source CAM with 6 axis kinematic engine.

http://cnc-toolkit.com/

I looked into building the 5-axis router plans at www.doughtydrive.com and you can do it for about $5K. The doughty drives hem selves are pretty sweet units.

Kyle

[Fred]

I was thinking of the same thing myself recently. Not too surprised to find I wasn't the first.

As far as toolpath generation goes, I'm sure for the sort of things I'd likely be creating - 45 degree bevels on acrylic boxes, etc. - the g-code could be hand cranked or generated with scripts / post processors.

I was wondering was if anyone actually gone ahead and built any hardware like this?