Contraptor

So this is an experiment to see how quick/easy something useful could be built from Contraptor. I've gotten several emails regarding possibility of building a small CNC machine from Contraptor, so looks like this might be a good test. I had previously built the mini-router and it worked, but it was pretty cumbersome to assemble and adjust, hence another attempt.

Yesterday I did this napkin drawing.

A CNC machine can be built in many different ways, this is just one of them: fixed gantry, moving table on X (or is it Y?), moving head on Y. Entire Y stage moves in Z.

Then I spent about 3 hours in Sketchup building the structure and some motion assemblies. The Sketchup file is on 3D warehouse, it is draft and unfinished: http://sketchup.google.com/3dwarehouse/search?q=contraptor+mini+CNC I think I'll change a few things and finish the design, hopefully today.

Why is the Y/Z stage not centered over the range of motion in X (which is constrained by the locations of the bearings holding the table in place)? It looks like only one half of the table can be brought under the Y/Z stage

Why design the Y stage to move in Z? It seems like you are unnecessarily overloading the Z stage bearings/belt/screw. Instead you could put the Z stage mechanism on the Y stage, so that the Z stage only moves the tool, while the weight of the Z stage is supported by the Y stage.

Worked some more on the design, changed some things, now it's ready for trial assembly. Sketchup file is at the same place: http://sketchup.google.com/3dwarehouse/search?q=contraptor+mini+CNC

Specs are becoming more or less clear - axis travel X=8", Y=6", Z=2" (though it will probably be 1.5" in Z with Dremel flex shaft). Pics including assembly guides are in the flickr set:

Spent about 5 hours on assembly today. A few hiccups but mostly done, only leadscrews/bearings/belts/motors left. About 90 assembly pics in the flickr set:

It was kind of an off day today, so I only spent about an hour in Sketchup doing some design modifications. Real life Y/Z stage had come out fairly heavy and it will be even heavier with Dremel on it, overloading the Z bearings (the problem pointed out above by sparr), so I'll add a pair of thrust bearings at the top of Z lead screws:

Speaking of bearings, I had also planned to use Dremel flex shaft for milling, but after taking it apart yesterday in attempt to reduce radial play and discovering a pair of really tiny bearings, I think I'm going to use Dremel itself.

Spent about an hour today; pretty much assembled the Z axis motion mechanism and realized that I don't have the right size belt for it. Bummer!

Because I didn't have the belt in the 3D model, this problem sneaked in when I made the design more narrow which changed the distance between Z axis rails. Once again I'm reminded of the importance to work everything out in 3D before assembling.

I had encountered this problem before. From here, the options typically are: a) redesign, b) buy right size belt. This time I'll go with (b) because the belt is only $5.39 on Mcmaster and because I was going to order some stuff from Mcmaster anyway.

I will also see if there is any way to c) mount tensioner pulley somewhere around there ^, though this would require some Contraptor acrobatics.

Highlight of the assembly:

While I'm waiting for the belt to arrive, I decided to make a Dremel mount out of scrap square tube left from making early sliding elements. The process was fairly straightforward, except for 3/4" hole in 1/8 aluminum. I ended up drilling a bunch of 1/16" holes along the hole circumference and hammering the center out with a metalworking chisel. Then I cleaned the hole with large 5/16" high speed cutter until Dremel fit into the hole. Here is the mount:

Here it is, with Dremel, installed into mini CNC:

SVG file should be attached to this page; material is 1.5 x 1/8 aluminum square tube.

On Thursday night I added X and Y motors, lead screws/nuts and belt:

Then I had to take a break from mini CNC till late today, due to family matters.

Today I added stepper boards, connected them to Arduino and performed a few basic motion tests. This one is a circle:

Limit switches and X table still need to be added. I could make the table from 1/4 acrylic, but I'll wait for lasercut one from Riley.

A few hours of work on Thursday night. Added a table from HDPE cutting board and after reading/research about stepper resonance and inertial dampers, decided to try simple flywheels on X and Y screws. The difference is clear - mini CNC can reliably do 20 in/min, straight lines and circles, effectively doubling the feedrate. Noise and vibration are significantly less. Don't know what the performance will be under milling load, hopefully not worse. Without flywheels, straight lines could be performed at 20 IPM but circles were jagged at four points where X speed = Y speed, on any feedrates past 10 IPM. Here is the videos of flywheels in action:

• Test pattern at 20 IPM
• XYZ motion
• G-code fragment from earlier tests, run at 20 IPM (video is 4x realtime)

The lasercut table from Riley arrived yesterday along with some templates and bunch of prototype plexi "skin" parts. It looks very cool comparing to the cutting board :)

Yesterday I also started testing motion around the perimeter. With the exception of corners where the screws are sticking out, I get 6" of motion in X and 6" in Y. The X travel will be 8" when the whole gantry is moved 1" back - my mistake, fixed in Sketchup. The last 1/2" on either end of Y are a problem - it becomes increasingly difficult to rotate the lead screw and the motor stalls at 20 IPM. The reason is slight misalignment of Y lead nut and Y shaft mounts, which becomes pronounced as the lead nut approaches shaft mounts. I removed Y axis yesterday and pretty much took it apart in an attempt to fix it.

Hi Albanetc,

I've been a curious follower of your mini CNC thread since the beginning.
Cudos to your mechanical engineering skills - you practically built that thing in no time!!!

However, since the mini CNC is just another prototype application of the Contraptor concept/kit, I'm starting to wonder if it's actually possible to use the Contraptor package for creation of a larger CNC.
During the past year, I had to do pretty time-consuming woodwork several times.
Unfortunately the parts I had (and will probably again have to) work on are dimensioned roughly 20"x15"x4" (LxWxH).
Do you think it's possible to use Contraptor for building a CNC prototype which is capable to work on workpieces that large?

Thanks,
-Schoeftinger

I also think a larger CNC is "doable" using Contraptor, but given the requirements for a larger motor driver board per axis, you might want to consider using a Cubespawn (www.cubespawn.com) 1 meter cube - as Albanetc, Ril3y and I have discussed, Contraptor is ideally suited to building prototypes, due to its flexibility. But in a production machine its flexibility may be a mixed blessing, as the high fastener count may be susceptible to problems with vibration, which a router produces a lot of…

A Cubespawn solution will cost more than Contraptor as well, although Ebay is a good source for aluminum 80/20 structural material, you could still use the Arduino for control, but drivers might be this:
http://www.parkermotion.com/products/Stepper_Drives_and_Motors__4003__30_32_80_567_29.html
or
http://www.geckodrive.com/
motors like this
http://cgi.ebay.com/GP54760-Parker-OEM83-93-MO-DS34-Nema-34-Stepper-Motor_W0QQitemZ330346167160QQcmdZViewItem?rvr_id=&itemid=330346167160
and so on

If you should decide to scale Contraptor up, the same drivetrain considerations would apply to it as a Cubespawn solution.
you might consider scaling the structural components up, as in: 2 inch angle for long spans…

Either way, building a CNC is like all learning expieriences: i.e. slightly painful at first, but ultimatley rewarding,
Anyhow, theres my 2 cents! ;-)
Good Luck!
James

This is what I made tonight. This is just a wheel. I wanted to test how it would come out before cutting the inertial damper wheels for steppers. It took about 10 minutes to cut, using tungsten carbide bit, speed setting 2, pass depth of 0.02" and feedrate of 12 IPM. I got these parameters from earlier mini router experiments/tests.

This is the cutting process video. The noise is very manageable, but the mess isn't. I need some kind of vacuum to get the swarf out as it forms, otherwise it'll get on the leadscrews, bearings and cause trouble.

So the object is great but can it be useful other than the piece of art? I'm talking about the accuracy of the cut.
The outside cut seems to be right on the spot. The inside cut seemed 1/64" bigger than it should be but then I noticed the gap on my caliper. So far so good.

Areas for improvement?
1. Speed - I'd like to go 60 IPM +. 15-20 IPM is probably as good as it gets with all-thread leadscrews. Also not sure how will Dremel take faster feedrates and whether pass depth needs to be decreased proportionally.
2. It's pretty inconvenient to attach workpiece to the table. Ideally it should have tapped holes.
3. Vacuum

Discovered an interesting problem with Arduino and GCode Interpreter - it pretty much hangs after series of very short moves are commanded. This problem first happened when mini CNC was going through the GCode output from Cambam - I exported Sketchup drawing of the inertial damper disk into DXF and used Cambam to generate GCode in a shortest possible way, without recreating the circles. As a result, all circles, including small ones, consisted of a bunch of very short lines which interpreter/Arduino apparently couldn't handle. Reset helped to a point, then I had to reload the firmware a couple of times.

I also reproduced the problem when trying to engrave a small text in the acrylic piece above - see unfinished "Hello W". The text height is about 0.1", or about 2.5 mm. The V bit I used is really sharp, I think the groove should be finer with a larger angle. Nevertheless, the text seems to be fairly accurate.

I'm using older (and buggier) Arduino software (0012) because it's the latest software with which GCode Interpreter actually fits into Arduino (Diecimila) on Ubuntu. Apparently newer software has larger libraries. The interpreter comes with Reprap related extruder code, I guess I will try removing this code to see if the interpreter will fit on a later version.

The good thing is that the problem appears to be easily reproducible.