7x CNC lathe first part

[sansbury]

Based on the JFettig Machines plans, I did this on a lark as a parts-bin conversion. The 125oz motors are underpowered but actually manage to squeak through. I have some bipolar drivers laying around that might give enough extra oomph to make this work reliably.

Right now the biggest problem is the backlash in the X axis. It's about .010" and for some reason I can't seem to program around it (the Z axis is worse but I'm able to program around it easier) so my parts are coming off the wrong size, but for this part it wasn't a showstopper.

Next up is to try rigging a spindle encoder to see if I can get threading to work.

[nomedia="http://www.youtube.com/watch?v=w9KLGe7fS58"]YouTube- ‪7x CNC lathe first real part‬‎[/nomedia]

[blades]

Youtube blocked @ work.

[knudsen]

Nice swarfbarfer!

Aesthetics is for pussies. I like the dyechem on the motor coupler.

[sansbury]

Over the weekend I managed to get home switches wired on the X and Z, which are real useful since it's not unusual to get a stall with these very marginally-sized motors. Today I got the spindle index pulse fully connected, using a single-slot disk with a photoswitch.

As with the rest of the lathe, it was a bit of a hack job that will need some better enclosure design before I get too serious with it. Right now there's wires dangling everywhere just waiting to get caught in something and pull itself apart. But I wanted to first make sure the electronics actually worked before getting too fancy...

Much to my amazement, they worked right the first time! The photoswitch is an Optek OP 821 type, and I used this simple circuit to drive it, using power drawn from the PC via a USB cable:

http://www.ee.washington.edu/circuit...CIISCHEMIR_007

               o +5v     o +5v
               |         |
               |         |
               \         \
    220 ohms   /         / 4.7K
               \         \
               /         /
               |         |
               |         +-------> Vout
              _|_        /
              \ /      |/
             -----     |\ NPN
               |         \
               |          |
               |          |
               |          |
              GND        GND

                | air gap |

As a tach, it seems to work well, needing around 5-10 seconds from a stop to lock on and stabilize. I've only barely tried threading--I just ran a few light passes to see if it responded as expected, and so far, so good.

I need to code up a proper program to really exercise it and see if it produces a usable thread. My biggest concern right now is whether the spindle speed will be stable enough to give good results at low RPM. Since my max speed on the Z axis is around 18IPM, that gives me a max of 360 RPM for a 20-pitch thread, which is the coarsest I use regularly. This one concerns me a bit as I've often noticed some cogging when threading in steel or larger diameters before, so we will see if the extra speed is enough to keep things in sync.

[sansbury]

So, a little update on this project. Since the last post, I replaced pretty much everything except the motor mounts. Current setup looks like:

- 2x Keling 381oz motors
- 2x Gecko G251 drivers
- EMC 2.4.6 from the latest LiveCD on an Intel Atom D510-based PC

I decided to switch to EMC for a bunch of reasons including what seemed like more robust lathe support and because I've been working on a very sophisticated EMC-based pendant as can be seen here:

[nomedia="http://www.youtube.com/watch?v=ONxVoAXadTk"]YouTube - Arduino EMC Pendant Prototype[/nomedia]

The setup with EMC was mostly simple. The basic stepconf wizard set up everything except the spindle encoder, which required a little manual editing of the HAL file. This was mostly a cut-and-paste job so not too bad with a little hand-holding. I also used the ngcgui wizard framework with some of John Thornton's lathe subroutines to give a nice semi-conversational framework for basic operations like threading, tapers, turning, and boring. These are like Mach's wizards but I find them a little more flexible and powerful.

Anyway, long story short, the picture below shows my third attempt at putting a 1/4-20 thread on a piece of random screw-machine steel. This was a hard-case test since most of my threads are smaller and/or in softer materials. With only a single pulse encoder, there is a risk of pitch deviation if the spindle slows down too fast. I ran this at about 500RPM, which is a lot faster than I thread manually, since my spindle tends to cog at low speeds.

http://i.imgur.com/lcwDM.jpg

The last Big Thing I have to work out is how to set up the tool table properly. That is still a bit confusing. I also want to tweak John's threading subroutine to allow the user to pick standard thread sizes without having to input all the parameters individually. At that point I think I will finally have my lathe back and ready to make parts.