The reason I've not been posting all sorts of pics of things I've made on the 3D printer is that, other than the initial teapot, I haven't gotten anything successful.

For a while it was software issues with settings of Skeinforge, the model slicer I use. It has dozens of cryptic settings. The results are infinitely tunable as a result, but to get good, much less usable results, you have to find the "magic settings" for your machine's speed and plastruder. Skeinforge's user interface... is painful. It was obviously designed by an engineer to be able to do Amazing Things if you have a Masters in Plastic Extrusion Technology. But it's barely usable by folks who just want to print 3d objects. And the app is sloooow. If anything is going to make me spend another year of my life writing a replacement app out of frustration, this'd be it.

I also kept running into meshing issues with the STL models I was using. Blender was great for pointing out meshing errors, but I couldn't figure out how to use it to FIX the problems. The auto face add feature didn't do it's mojo on the model I was working on, and I don't know Blender's Way Of Thinking yet. In fact, it's almost actively hostile for newbies to learn, despite the numerous tutorials available. Steep. Learning. Curve. And with my general lack of memory, I end up having to go through the tutorials over and over. On a laptop without a numeric keypad, or a 3 button mouse, you have extra config setting hoops to jump though to make it even usable. Meh. Supposedly Blender 1.5 will fix a number of the more egregious UI issues. I hope so. In any case, I found another open source app called Meshlab, which Just Works and was intuitive for the most part. It let me quickly fix up the holes in the model that made it non-manifold (Not water-tight). Blender pointed me to where the holes were, Meshlab fixed them easily.

Next problem was my machine itself.. I was having backlash and slippage issues with the belt drive. I ordered thinner belts from McMaster Carr and tightened them down well, and that made a world of difference, but I was still occasionally getting slippage. I've determined that lithium grease is too viscous for my uses here. Switching to some air-tool oil for the sliders made it run much smoother on the X and Y axes. I probably should find a more permanent oil, but for the moment, I can periodically re-lubricate.

I'm starting to build up a variety of failed prints of skulls. I was originally trying to print some for Halloween, but now it's my test object. I will soon have the world's largest collection of replica skulls of people who failed to live through The Philadelphia Experiment. In fact, I seem to have just made another one. Drat.

My first (and only so far) successful print was the Utah Teapot, which tugrik posted a picture of, which I'll re-post here.
( Image behind cut. )

Dusty, you need to come over and play Brütal Legend on the 360. I won't be home this weekend, but you should come over and play it anyways. It's in the Xbox. It's very silly, and very metal, and very fun.

<Palpatine>Now witness the power of this fully Armed and Operational battle 'struder!</Palpatine>

Click here for video of running extruder

I've gotten the 3d printer finished!


But it ran only a few minutes before the extruder head ate itself, and melted it's PTFE thermal barrier. Booo! I have a new one and spares on order.

It's quieter than I expected, which is good. The final build area looks like 7.5" x 6" x 9". I managed to hide away all the electronics underneath, except for the extruder controller. A short circuit nearly fried my extruder controller, but I was able to solder up a bridge past the fried traces. I still need to set up my opto-endstops, but that's mostly just a wiring issue.

( More pics behind the cut. )

Tasks completed today:

• Designed and built X-axis table mk3 for CheesecakeCNC.
• Soldered up 6 opto endstops for same.
• Assembled, updated the firmware of, and tested Gen3 electronics.
• Lasered out the parts for the mk4 plastruder.
• Started designing the Y/Z frame.
• Lasered 50 plaques with artwork for an unrelated project.
• Lathed a car pully for a friend.

...

Okay, maybe it was a productive day. :)

( Read more... )

So far with the screw-based drive for the X table, I've been able to get a speed of 15 inches per minute with the NEMA-17 stepper motor. I was thinking that this wasn't too bad, as my experience with the CNC TAIG mill was that I could get 45 inches per minute out of the larger NEMA-23 motors.

But then I saw a page talking about the feed rates used with the newer plastic extruder head, where they were talking 45 mm/SECOND feed rates, which translates to about 106 inches per minute.

...

Okay, the screw drive just isn't going to manage that. The plastic wants much faster than I could hope to get with the 20:1 gear reduction the lead-screw provides. It might be fine in metalworking, but not for plastruding. So back to the drawing board CAD.

ACE hardware seems to be rather lacking in the supply of belts and pulleys. I'm going to take a look at possibilities of using chains and sprockets from a bike or auto store, but things are looking bleak for using easily found off-the-shelf parts. I may have to settle for merely making a super-sized CupcakeCNC type design, with just mostly off the shelf parts. Well, I can live with that. McMaster-Carr has plentiful options of pulleys and belts, at the least.

So, the project description is now throttled back to:
How to make a 3D printer like the CupcakeCNC, but that can print larger items, using the same plastruder and electronics, and parts from any ACE hardware affiliate store, and a couple parts from McMaster-Carr.

Doesn't exactly roll off the tongue.

On a lighter note, I've decided that I'm going to call my design the CheesecakeCNC. It's bigger than a cupcake, and definitely cheesy in design.

Last night I hooked up a Teensy++ micro-controller, running my G-code interpreter firmware, up to a stepper motor driver and stepper motor from Makerbot, powered by an ATX power supply. I ran some g-code from my CAD through it.

We have Motion! And the motion is correct for all three axes!

*ahem* "MwahahahahahaHAHAHAHAHA! It's ALIIIIVVVE!"

Now, I just need to make a framework to attach the stepper motors to, to drive the three axes of a proper motion control system. I'm currently designing up something on my CAD.

Because I can't ever leave well enough alone, I'm designing something new, instead of using a publicly available design. I want to see if I can make a 3D plastic extrusion printer mostly from parts you can buy at your local ACE hardware store, and maybe some from a bicycle or auto-parts store. The CupcakeCNC design from Makerbot is neat, but kind of limited in the size of parts that it can output, and it seems to suffer from the steppers resonating with the wood, making it louder then I'd like.

I can do better. :)

Some parts of our lawn have been dying, while other parts are green. I know I've got the sprinkler timer running enough to have a full-green lawn, as I've checked it three times now. I know water is flowing, because I ran the sprinklers previously. But apparently not long enough.

Today I ran a sprinkler test again, and watched it more carefully. Apparently one of the sprinkler heads is just stuck, and the other one sometimes hangs up after a couple cycles. So I go and swivel the heads manually, and notice they rotate freely. But the sprinklers still get stuck when running themselves.

About this point I've gotten kinda wet from collateral spray.

While looking at one of the heads, I notice that there seems to be more water than expected flowing down the lawn, and not from the sprinkler itself. I follow the stream and find a burbling hole in the ground a few feet away.

"Okay," says I, "there's an underground leak here. That may explain the weak sticking sprinklers." So I grab a trowel to dig up the leak, so I can figure out what parts I need to buy at the store.

Dig dig dig. No pipe. "Oho! I see. The water is coming from the side of this hole I've dug. Let me follow it."

Dig dig dig dig dig. "Huh. This seems to lead towards the sprinkler head. My there's lots of roots here. Let me get the shovel."

Dig dig SPANG! "Ow! WTH was that? Good lord! That's a big tree root! Let me try digging around the other side."

By this point I'm covered in mud. SPANG! "Dangit! Another tree root!"

Spang Spang! "They completely surround this sprinkler! Big roots! Let me whack them with the shovel like an axe."

Spang Crack! Sprinkler parts everywhere. "Augh! Well it needed replacing anyways. Maybe I can use the water hose to excavate the roots so I can cut them with the hacksaw."

Excavate excavate. "Good grief! This is one solid sleeve of root around this sprinkler! Augh! Get the power tools!"

Which, of course are back at the shop, and I am covered in mud, head to toe, and don't want to get my car muddy.

"Grr! I know! Gasoline and a blowtorch! Maybe dynamite! I'm sure Homeland Security would understand!"^[1]


[1] No, the Department of Homeland Security would not understand. Luckily, I regained my senses before immolation began.

Today, I finally received in the mail the $30 CNC relay circuit board I've been waiting for. So I ended up spending hours shuttling between Home Depot, Radio Shack, and the shop, getting and assembling electrical parts and cables. The final result is that I now have a pair of electrical sockets in a sealed box on the side of my mill desk, that can be turned on and off by the mill computer. This lets me finally (Finally!) turn the spindle and flood coolant on and off from my milling programs.

Now, when a job finishes, the spindle and coolant will stop by themselves!

I just got to witness Skippy the Wonder Electrician involuntarily attempt self immolation by short-circuiting our shop's 200 Amp 3-Phase mains, while wiring in a new circuit:

BLAM!

*Lights go out*
"You still alive over there?"
"I'm good."
"Ooookay."

Yeah. Starting to wonder about this guy.

Well, converting from simple boolean geometry code to Compound boolean geometry code wasn't quite as simple as I first thought. BUT, it wasn't too bad either. Basically, you just have to do the same inside/outside/shared/unshared tagging that you do with simple paths, and apply it to all subpaths as well. There's some fiddly bits about making sure you are inside an odd number of the opposite path set's paths, but it wasn't too bad. The result is shown below:

Hey, triggur, check this out!

I've been working on getting my CAD/CAM program up to snuff with basic stuff like, oh, Grouping/Ungrouping objects, and getting SVG import/export working. Part of this has gotten me to revisit some code that I had stopped working on out of frustration.

My Boolean Geometry code.

This is code that is supposed to let you take two or more polygons and join them into one object, or subtract one from the other, or get just the regions that are common to both polygons. This is incredibly useful code when you are trying to make milling paths in CNC work. But until now, my code was just plain broken.

As with many complicated tasks, a few months working on other code has let the design simmer in the back of my brain, and as I finally come back to it, now the design is both obvious and simple to me, so I coded it up, and it now works for simple closed polygons. (Compound polygons are slightly more complex, but with the basics in place, won't be a big deal to finish.)

( For folks on the net searching for the solution to this topic, I give an explanation of my algorithm behind this cut tag. )

Belfry.com and belfrycomics.net are down at the moment. Our ISP did some scheduled maintenance tonight, and when they brought my server back up, the hard drive made godawful noises and wouldn't mount. Luckily we have backups, but it may take me a day or two to restore from them.

I'll post more info when I have it.

My ghod I work in a beautiful business park! It even has a small hedge maze!