Wednesday, May 30, 2012

WhiteAnt Build- Parts A,B,F,G and H

I'm using the plans which I downloaded from the site and the book "Printing in Plastic: Build Your Own 3D Printer" to build my version of the WhiteAnt. I'm going to reference the parts as they are in the book, but I am not going to follow the build order that it outlines. Since I'm modifying the size, I think it's better to start with the main structure. This will allow me greater freedom to fine tune the X-axis rail in relation to the bed (Y-axis). I have a greater concern for keeping the planes level with this printer since it's highly dependent on the accuracy of drilled holes and cuts in wood.


All the parts are basically as the plan calls, except for part H. This is the
X-axis Rail Support*. The plan calls for a 10 13/16" length, but I have extended
it to 18 3/4". This will allow the printer to support a bed size of 12 x 17".
The Prusa Mendel has a bed size of roughly 8" square.

* Note that in the book and plans the x-axis support rail is labeled as Y-axis. There are other oddities, mistakes and omissions through out the book and plans. Because of this, I find myself triple checking everything.

Comparison of size to the Prusa Mendel

Friday, May 25, 2012

3d Printing Natural Forms III

The third sample series compares a low and high resolution save of the same
model. I created a simple terrain with Bryce Pro using a cauliflower fractal
filter with erosion. When I exported from Bryce, I saved  a low polygon count
.obj file and then a high polygon file.  In preparing the files for printing, I
kept them consistent to ensure  an accurate comparison.

Here is the pertinent data for this test:

Skeinforge settings are the same for both print runs:
Layer thickness: 0.4mm
Perimeter over thickness ratio: 1.55
Infil solidity: 0.1
Feed and flow rate: 40.0mm/s

10,712 vertices
21,420 faces

3mm ABS filament / .5mm nozzle
Size: 182mm x 174mm x 43mm
Time for Skeinforge to compile: 25 min.
109 layers
149,962 lines
Estimated filament length for print: 37,941mm (41.5 yards)
Print time: 5hr 57min
Object weight: 280 g (9.87 ounces)
Ambient temp: 23 degrees C.
Bed temp: 63 - 89 degrees C.
Extruder temp: 200 degrees C.

294,466 vertices
588,928 faces

3mm ABS filament / .5mm nozzle
Size: 182mm x 174mm x 43mm
Time for Skeinforge to compile: 56 min.
109 layers
215,827 lines
Estimated filament length for print: 35,501mm (38.8 yards)
Print time: 8hr 19min
Object weight: 259 g (9.12 ounces)
Ambient temp: 29.6 degrees C.
Bed temp: 99 - 110 degrees C.
Extruder temp: 200 degrees C.

Close-up comparisons:


low resolution

high resolution

Monday, May 21, 2012

DAVID LaserScanner- Completed Setup

It's taken a lot longer to refine my setup for the DAVID laserscanner software than I expected. It's been more than a month since I experimented with the free edition of the software and a temporary rigging setup that I assembled (with an inexpensive red line laser and NEMA 17 stepper motor). Since then, I've purchased the Pro USB version of the software from a U.S. reseller (mydigitaldiscount- their stock goes fast. Mine was on back order for awhile). I also refined my rigging, components, and assembled an electronic control console.

Here's an overview of the components:
I designed all the structural components using 3/8" aluminum rods and plastic
parts modeled with Google SketchUp and printed on my Prusa Mendel 3d printer.

Calibration Panel
The plastic brackets can slide along the rods for the adjustment of larger
calibration panels. The rods themselves can be replaced with longer ones for
even bigger panels.

The turntable is basically a round platform attached to a NEMA 17 stepper motor.
The structure can be adjusted in height by replacing the rods with longer ones.
Also the round platform can be switched out with one that's larger.

Laser and Camera Rig
I changed my original 2 rod design to a 3 rod design for the additional
stability. I replaced the 10mW red line laser with a <5mW green line laser
(available at Innovam). This is a more expensive device, but will give much better results (the camera will utilize 50% of the available pixels opposed to 25% that the red laser would). I also replaced the NEMA 17 stepper with a NEMA 08 planetary gearbox (available direct from factory through This gives a 1:64 gear reduction and will drive the laser with a much smoother motion. The camera is a Logitech Webcam Pro 9000 with Carl Zeiss optics which is supported by the software.

I kept a triangular design to all the component bases in order to help
facilitate in lining all the components together. Bisecting a 90 degree angle
and drawing out the line provides a clear setup for all the pieces.

The electronics driving all this is a lot simpler than it looks. It consists of  basically 2 Arduino Duemilanoves, stepper drivers and a simple voltage regulator for the laser. The electronic design is from WalterMo on the David LaserScanner Forums. The schematic and arduino programs can be found here and here. (I’d like to thank him for sharing his work). The only changes I made was to combine the two circuits, removed the 10k resistors from the STEP and DIR inputs and added a switch to ground the MS2 pin on the stepper driver running the laser movement. This latter change gives me a choice of 1/8 and 1/2 step movements and I feel a little more variance.

The controls for the turntable allows for precise rotational setups by degree in both 
forward and reverse.
Laser control speed is adjusted by a potentiometer and my additional 1/8 or 1/2
step switch. Movement up or down is also controlled by a switch.

Monday, May 14, 2012

3d Printing Natural Forms II

The second sample I made, I think is the largest Prusa Mendel print I've seen. I
wanted to test the limits of the print envelope by creating a model that would fill the print bed. At the same time, I wanted to make a model with  graceful curves with sharp delineated edges.

Base with 10% infill.

About 47% complete.

Completed print.

Here is the pertinent data for this set up:

3mm ABS / .5mm nozzle
Size: 185mm x 187mm x 81mm
Time for Skeinforge to compile: 1hr 10min
203 layers
Estimated filament length for print: 50,058mm (54.75 yards)
Print time: 11hr 16min
Object weight: 375 g (13.25 ounces)
Ambient temp: 25 degrees C.
Bed temp: 85 - 100 degrees C.
Extruder temp: 200 degrees C.

Skeinforge settings:
Layer thickness: 0.4mm
Perimeter over thickness ratio: 1.55
Infil solidity: 0.1
Feed and flow rate: 40.0mm/s

At about 27% completion I noticed warping at three corners. It continued to warp
slightly and then stabilized.

I wasn't aware when the crack appeared since I left the latter half of the
printing (about 5 hrs) unattended.

Overall I am very happy with the results. Aesthetically, I am very pleased. The
forms are very well defined and interesting. Even the layers add to the overall
appeal. The scalloped relief with the texture of the print layers remind me of


Friday, May 11, 2012

3d Printing Natural Forms I

I decided to start a series of samples that would explore different textures and
how these textures will translate by 3d printing. The easiest way I found to
generate the models I need was with a 3d program popular with designing
environments, Bryce 7 Pro. Within  minutes it’s easy to create mountains or
other landscape elements and export the model as a mesh.
Unfortunately, the ease ends here, or at least for me since my background with
working with 3d programs is very limited. Through my trial and error, I
eventually developed a workflow that provided me with a model that could be 3d
Here is a summary:

1) Bryce 7 Pro, basically start up the program and click on the mountain icon
and a wire frame mountain appears. Playing around with the edit features will
model the mountain with an infinite array of features. I started with a typical
projection with fractal eroded surfaces. I exported the wire frame as a wavefront .obj file.

Bryce working environment.

2) Meshmixer 07, is an open source program and a very important one for me since it's relatively easy to learn and it was my best option of capping the mountain I made in Bryce. For 3d printing, a model must be a solid. The mountain I made
in Bryce is basically hollow with an open bottom, essentially a shell. There is a great YouTube video showing how to enclose a model such as this here. The modified mesh was exported as an .obj file.

Meshmix working environment.

Under the mountain model before it's capped.

3) Rhino 4.0, is a very popular and powerful 3d design program. I would have
used a simpler program like Google Sketchup, but my file was too complex and it
couldn't handle it. Rhino has an evaluation copy that will only allow 25 saves,
this is what I used. This program was important for 2 reasons (which I couldn't
solve with Meshmixer): I needed to scale the object and I needed to orient it
properly within the Cartesian space coordinates. I quickly discovered this problem
after trying to print; the model was on it's side and only about 8mm square in
size. Rhino was able to correct all of this and export as an .obj file.

Rhino working environment. Here the model is shown positioned upside down in the coordinate system. It would slice and print this way if left unrotated.

4) Meshlab 1.3.0a, is an open source program used for merging and fixing meshes
for printing. I ran my model through this and exported as a .stl file. But once
I started to slice my .stl file with Skeinforge, I found that I didn't quite fix
all my problems with Meshlab. I still had a hole in my model somewhere.

It took an hour to slice and I didn't want to go back and try to fix it, so I
printed the model with the errors. The only problems I could see
with the print was an irregular start on the base and the print printed solid
even though I had my fill ratio set at 10%.

Even though base was capped square, the print started irregularly.

I also had a slight problem with warping. The print on one side lifted enough
that it actually cracked the glass, but the print overall is quite flat.


Aesthetically, I'm not too happy with the overall look. It kind of reminds me of
the mealy texture of paper pulp. Also the "architectural" resolution is poor,
but this is probably due to the 3d modeling.

Tuesday, May 1, 2012

Warping of ABS Prints

I had issues with warping early on when I was in calibration mode with the Prusa Mendel. I solved it by using Kapton tape and In extreme cases with rafting. What I found to be the most helpful was giving the Kapton tape a fine tooth by sanding the surface with swirls. This has given me almost flawless results until now.

Presently, I'm working on pieces that I need for my laser scanner system. One
piece in particular is a thin right angle support that I made as large as I
could fit on the print bed. It spreads 165mm along the x and y axes. It's primary width is 4mm and overall it's 15mm high.

Printing an angle brace.

After 3 print runs with warpage, I realized I was back to this old problem. I decided I needed to take the time to solve this. And since I think this may be useful information for others, I'm going to document my 8 print runs.

Prints 1-3
Bed temperature is 72 degrees C.
Ambient temperature is 25 degrees C.
1st print - failed at about 22% completion.
2nd print - Sanded bed slightly. Print failed after about 60% completion.
3rd print - Aggressively sanded more in problem areas. Print failed at about 70%
I always run my bed with a set temperature of 110 degrees C., but it rarely reaches 83 degrees C. I'm not quite sure why. I double checked the temperature reading of the bed thermistor by checking with an IR thermometer and the readings are close and  consistant. Maybe, if I could get my bed temperature up higher it would work?

Print warps at the end and becomes distorted.

Print 4
Bed temperature is 72 degrees C.
Ambient temperature is 25 degrees C.
I initiated a raft. Print failed after about 75% completion.

Printing on a raft dosen't prevent warping in this case.

Print 5
Bed temperature is 76 degrees C.
Ambient temperature is 27 degrees C.
I calibrated z axis home tight on the print bed so that the filament squeezed
wide. Initiated a raft. Print failed after about 90%
completion. This time only 1 end lifted.

The bottom raft in this picture is from print 4. It's weave is open. The top raft is from print 5 and is squished tight by the extruder nozzel.

Print 6
Bed temperature is 81 degrees C.
Ambient temperature is 27 degrees C.
Coated glass with blue tape. Coated the tape with Ace ABS Solvent Cement.
Initiated raft. Print failed due to printer malfunction. After about 35% of  print completion I lost a thermistor contact and my extruder stopped, so I had to abort the print.

Blue tape coated with ABS cement.

Print 7
Bed temperature is 78 degrees C.
Ambient temperature is 29 degrees C.
2nd attempt with using ABS glue. This time I decided to turn rafting off. I feel cofident that glue is the solution.
Printing is a success! Print holds well to bed with no warping and without the
use of a raft.

Print finished with no warping.

Close-up reveals very tight bond with bed.
Part removed and ready for use.

Print 8
Bed temperature is 27 degrees C.
Ambient temperature is 29 degrees C.
I ran one more test to see what would happen without a heated bed and only ABS
glue on blue tape.
Printing failed!
Print warped after about 60% completion. It adhered strongly to the tape, but
pulled the tape off the glass. It's quite possible that if glue was used on
Kapton tape it would work. That will have to be tried some other day.
Blue tape lifted.
In conclusion, for large printing: ABS glue on blue tape  with a heated print
bed may be sufficient for all prints within a Prusa Mendel print envelope. Adding a
raft should increase this success rate. Later I plan to explore this hypothesis
by maxing out my build envelope with a solid print.

ADDENDUM 6 May 2012:
I increased the height (thus increasing my print time) for my remaining corner braces. This time I did get a problem with one end slightly pulling up the blue tape. I changed over to using Kapton tape with a sanded surface and ABS glue. This has worked for 3 more prints. No warping.

ADDENDUM 25 May 2012
Now that I am printing models which fill the entire print bed, my warping issues have become more problematic. I have not successfully kept the corners from lifting on my prints (averaging 180x180x50mm in size). To minimalize the warping, my system has been to use Kapton tape on glass, sand the surface and coat with ABS glue. Unfortunately, for the size of the print, I am unable to remove it without breaking the glass. My next option (and more favorable) is to use plexiglas instead of glass. I've eliminated the Kapton tape, sand the plexiglass surface and then coat with ABS glue. The plexiglass is more easily removed, but unusable due to warping. In summary, my next possible solution to prevent warping is to build a heat chamber for the printer.

ADDENDUM 20 June 2012
The ambient temperature has increased dramatically this summer and now my bed temperature has occassionally reached 110 degrees C.  This has not seemed to help much in preventing warping.

A final note: The ABS glue can be manufactured.
The ACE brand ABS cement contains ABS resin, acetone and MEK.
My homemade ABS glue is the same, but without the MEK. I chopped fine extruder
waste and added it to acetone:

8 grams ABS waste
40 ml acetone
Within several hours I had a nice mix that could be easily brushed.

ABS scrap
Dissolving ABS scrap in acetone.