brainwagon

Comment from Anjin
Time 11/1/2010 at 11:13 pm

I have built a similar machine from bits for a bit more cost, but with high reliability and accuracy. Having a hardware based backlash system is a must when using threaded rod. As well using bearings are required IMHO. And not using flexible couplings is a serious mistake. The first unit cost me about 300 bucks to build a working machine with three axis including NEMA 17 stepper motors, cast backlash system, and guide bearings and standard shaft couplers. Another couple bills to pay for a good cutting spindle (foredom tool with a flexshaft and 1″ handpiece with jacobs chuck), and the handful of electronics to make the stepper controllers and parallel port based PC interface.

This Mantis design is based on the rockwell cnc system that has been pushed public domain… open source I suppose is the PC term today. Sadly I saw no references to the original design, but perhaps he didn’t have issue with copying a design without referencing prior work. I found a fault with this design in handling variable weight objects on the Y controlled table. It would mess with acceleration and braking timings, which might only impact one step out of 100,000… But you can drive the average stepper motor 14,00 steps every quarter inch when using threaded rod, and the only way you know position is buy counting steps. This means losing a few steps an hour, turns into big errors in cutting. My second build uses a moving gantry, with a nearly fixed mass. The only differential is the size of the cutting head, which might only vary a few grams and causes no issues. Having all three axis controlled by the gantry allows you to cut on nearly anything you can slap the machine on.

I also have built a couple x/y table designs that are easy to replicate and can be cobbled together from an old office sized copier machine for parts. Keeping the Z in the same place allows use of a big arse router for the cutter though, as well extruders from the reprap project.

Regardless, when you are pondering jumping in, I would recommend having the frame members cut to spec using another mill or laser so they are dead on accurate. Which solves the issues he is still trying to solve in his post. If you cant find anyone, ping me and I will cut some out for you using either PVC (think 12mm thick printed circuit board material made of PVC and fiberglass) or extruded aluminum t channel in trade for materials costs and another Pixar christmas card 🙂 I solved the backlash problem using a segmented brass drive bushing for the threaded rods. The first set I cast, the second I milled with the machine I built from castings. But I can make up a set of those too, or point you at some commercial solutions.

Cheers
Anjin