Contraptor

I've built my own controller, based on GEN7 and GEN6 RepRap designs.

I use the Teacup firmware with the following parameters.

These translate to 4x stepping:
#define STEPS_PER_M_X 629921
#define STEPS_PER_M_Y 629921
#define STEPS_PER_M_Z 629921

And I found these to be the most gentle to the mechanics:
#define ACCELERATION_RAMPING
#define ACCELERATION 20
#define LOOKAHEAD

It seems, I am not too far from the ideal speeds (600 vs 750 on XY, 450 vs 600 on Z), maybe I just have to tighten up my tolerances…

If I may bore you with my pride of accomplishment, please take a look at it in action at youtube.
http : / / youtube . be / pd1dOlA69FU
http : / / youtube . be / R_yfV9PzO9A
(I am not allowed to post links, please remove the spaces above).

http://www.youtube.com/watch?v=pd1dOlA69FU
http://www.youtube.com/watch?v=R_yfV9PzO9A

Wow, nice work and great videos! How are the tolerances/accuracy you're getting?

On feedrates, if you're on 4x stepping, you might be hitting a limit of the particular Bresenham implementation on Arduino in terms of how fast it can generate smooth sequence of steps. I haven't tried Teacup so I might be wrong - the numbers I quoted are from Contraptor fork of Arduino GCode Interpreter (you might try that to confirm/disprove this theory: https://github.com/AlbanetcSr/GCode_Interpreter, don't forget to change settings in the _init file).

Also, Grbl firmware can reach higher step rates on Arduino - in my tests, I could run it about 30% faster before stalling mechanics.

EDIT: All in all, this is mostly due to high pitch leadscrews (20 TPI). If you want to significantly speed up XY, replacing leadscrews with 4-start ACME screw might be the way to go. This reduces transmission ratio by 5x and can bump up the speeds to around 100 IPM (YMMV). Here is an example of such an upgrade: https://www.youtube.com/watch?v=3shXHOgNlRg

Thank you for the compliments, the video quality can be attributed to my Canon EOS 7D… ;-)

Well, when I was talking about tolerances, I was referring to the parts, mostly how accurately I could drill and tap. Not very much, but thanks to the tolerance tolerant design (pun intended), and the assembly instructions, the lead screws are not moving much, hardly at all, actually. There seems to be no tension, at least, no matter which end the lead nut is closer to. I think I could improve on the ball bearing mounts, they are a bit sloppy. I can see (and hear) them wobble a bit between those three 6-32 screws holding them from three sides.

As much as the end accuracy is concerned, the only way I so far "measured" it was that I drilled those holes on that copper clad board, and then tried the top solder printed transparency on (all originated from the same Eagle design). By the naked eye, the holes and the pattern on the transparency perfectly matched! :-) I know it doesn't mean much, since the board is small (about 2"x2"), neither my eyesight is as good as when I was 20, but I'm happy with it. This was my primary objective when building this machine. I, and my eyes, were tired of drilling those 0.024" vias, on the small boards you can design with the free version of Eagle. There were 127 of them on the control board…

I have an idea about how to measure the hypothetical top speed of the control board. I will hook a scope on the step pin of one axis (motor not connected), and start increasing the feed rate with G1 commands.

About the ACME lead screw. Yes, I was thinking about them, but boy aren't they expensive! :-( And of course, you would want matching lead nuts, equally expensive.

The only reason I am trying to increase the speed is that I want to fabricate aluminum parts, and with the minimum RPM my Dewalt 611 router can do (16000), even with a 1/8" Solid Carbide One Flute Upcut O Flute for Aluminum from Onsrud, I am close to the speed limit. I guess I will order one of those bits, and try it out. Maybe 600mm/min will eventually be enough.

Turns out, I didn't nearly need that speed. With my Dewalt 611 router, I can cut aluminum at 200mm/min, with the lowest router speed (about 16000RPM according to specs), and a 1/8" single O flute end mill (Onsrud 63-604), with 0.4mm depth increment. There is no chip weld, and the edges of the cut look smooth. I'll try higher feed rates with higher RPMs later, I guess it will be the test of how sturdy the structure is. The first useful thing I made (after a test cut) were the mounting brackets of the router.

http://www.gyalogkakukk.net:8081/~gal/dwp611.JPG

I guess now I can call it RepRap? ;-)

Pretty impressive! Do you have closeups of the cuts and/or the video?

Also, could you share how you made those brackets end-to-end, with hold-down techniques, more pictures and possibly source files? I take it that you dropped Dremel due to problems with runout, and since you've already done the work of figuring out how to attach DW611 to mini-CNC, I'm sure it will be valuable to others. Ideally it would be great to publish it in a separate project page "DeWalt 611 bracket for mini-CNC" or something like that.