Scott Zitek - Fab Academy 2014

Computer-Controlled Cutting

Week 3 Assignment

The assigngment for this week is to design, make, and document a press-fit construction kit. After spending all of my time over the past two weeks sitting in front of a computer, installing software, troubleshooting, learning HTML, Mercurial, and several CAD packages, I was anxious to make stuff on the machines.

Having made many press fit projects in the past, I know first-hand how important systematic test cuts are. I also know how tedious and time consuming it is to individually adjust the width of every slot in a project design to make it the fit just right. You have to change the design to compensate for variations in material thickness and material types. Ordinarily, in non-press fit applications as long as the settings are close enough for the laser to cut through the material without catching on fire people are happy. With press fit, the laser power, speed, and frequency settings can be critical. I have even seen where just changing the color of acrylic required the slots to need re-sizing. I knew all along that there had to be a better way to make these dimensional changes. After spending time the past few weeks learning Inkscape, FreeCAD, Rhino and more I could see the potential of some of the advanced features they offer.

So my goal for this week is to make a press-fit construction kit consisting of equilateral polygons with various numbers of sides along with some generic connector pieces. Not an original idea I know, but I was most interested in making the design parametric. I wanted to be able to adjust values for the length of a side, slot width, and slot depth and have the change propagate through the design accordingly.

Trials and Tribulations

I spent some time testing this approach with SolidWorks to see what it would take. I was able to draw a “master” side design with a slot in it. By setting constraints, it was possible to edit the parameters and have the different aspects of the master object adjust. I found that I could then “clone” the master side and by linking the key dimensions, have all of the clones change when I changed the master. It is even possible to use formulas to determine a dimension.

However, SolidWorks is very expensive and I don’t have access to it when I am not at the college. It also has a steep learning curve and I was looking for something that the typical lab user would understand. So I decided to use FreeCAD. That is when my incredible frustration began. I spent hours drawing the design different ways trying to get it to work. I started out just sketching in 2D because I wanted the advanced capabilities of parametric and constraints but I didn’t really need 3D drawings for the laser. Whereas FreeCAD is similar to SolidWorks, it seems to be missing some of the more important capabilities that I want for this project. I read the help files, watched videos, and followed tutorials on line. Eventually I thought maybe my problem was I was using software designed for 3D to draw 2D. So I tried drawing it in 3D. I still ran into similar problems. It was hard to designate constraints that worked when I rotated the object. If I had limited my press fit parts to squares and rectangles this would have been much easier because then all of the sides would be horizontal or vertical. By default that is how most constraints and dimensions work. If I only wanted square and rectangle parts, I think InkScape, CorelDraw and FreeCAD would have worked to some extent.

For the same reasons as the angled sides, the chamfers on the slots seemed to make this much harder. It was hard to define the chamfers in a way that translated when the sides and therefore slots angle was changed. I did get closer when I started using construction lines and symmetry constraint to automatically center some features. FreeCAD also seems to be limited to what it copies, duplicates, and clones. So I eventually gave up on drawing one side and tried to draw an entire shape. To make matter worse, FreeCAD would crash at the most inopportune times. It would also lock up for long periods of time trying to find the solution for some impossible combination of constraints I had attempted.

I am persistent and didn’t want to give up. So occasionally I would stop and see if any other software I had access to would be capable of what I wanted. I tried Inkscape. The cloning tool seemed like it would work well. Unfortunately, Inkscape is better suited for free style illustration that the precise CAD drawings I was trying to make. It was not dimension driven and did not allow for constraints to control how it should react to changes.

I then went to my old stand by CorelDraw. I didn’t know that CorelDraw had a clone tool until last week. However, it does not work very well. The command is in a pull down menu and only works once. To use it again, you have to reselect the master and find the command under the pull down menu. The biggest problem is that if you change a clone, it disables the link to all related attributes. For example, in my case if I rotate a clone of the master side object, changes to the master no longer effect that clone’s rotation or side. It defeats the purpose of what I wanted to accomplish. Interestingly, changes to the master’s fill color, line thickness, etc. still change the clone.

I think Rhino could do what I want, but I don’t have Rhino where the laser is and the demo version will only let me save a limited number of times. I didn’t want to get it done with Rhino and then not be able to use it or save it. Another option would be Kokopelli but I haven’t been able to get it to run due to an OpenGL or video driver issue.

So I spent an entire very long day working on this and have nothing to show. By now I could have easily saved time and effort manually redrawing the changes several times instead of making something parametric.

Persistence and Success

After a frustrating day of not accomplishing what I wanted using Inscape, CorelDraw, of FreeCAD, I used SolidWorks 2012.


  1. I started by sketching just one side with a slot. SolidWorks makes this easier by displaying dotted lines when a line is horizontal, vertical, across from another node, etc. and then snapping to align with these. I used a centerline and then mirrored it to draw the other half of the geometry.
  2. SolidWorks also automatically adds constraints as you draw. It added horizontal, vertical, tangent, symmetrical constraints etc..
  3. I added dimensions lines to both indicate the dimension of an entity and to allow the dimension to be modified. It is also possible to add dimensions that are only displayed and are driven by some other dimension.
  4. One side with slot (constraint icons not displayed)

  5. I then added and deleted constraints. This was the hardest part. The side would look great until I tried to rotate it. Then SolidWorks would try to solve it within the constraints. So I needed to get rid of the vertical and horizontal constraints and replace them with perpendicular constraints. I also had to replace the default dimension lines (which are typically horizontal and vertical) with non-angle specific dimension lines. The chamfer was also surprisingly challenging. I would have been done in half the time if I didn’t include a chamfer. I ended up defining it using one dimension and an angle. Otherwise, it would get messed up every time I rotated the side. This forced me to dimension the depth of the slot from the chafer down rather than the entire depth of the slot.
  6. I was able to get one side to work properly. I could change any dimension and rest of the design accordingly. If I changed the length of the overall side, the slot would stay centered. If I rotated the side to any angle, the chamfer would stay at the proper angle and the slot would stay perpendicular to the side. This was much harder to do than I expected.
  7. Once I got one side working properly I had to duplicate it several times. I found that if I copied and pasted a duplicate of the side it would become separate. Changes to the dimensions of the original would not change the copy. Luckily SolidWorks offers the capability of linked dimensions. I was able to name each of the dimensions on the original and then link them as needed to dimensions on the copies. Once set up changes to dimensions on the original or the copy would change all related linked dimensions.
  8. Unfortunetly, even with linked dimensions, if I copied and pasted a duplicate of the side it would become separate. The dimension links did not copy. So I found that I could make a block using all the original geometry, constraints, dimensions, and dimension links. This allowed me to save the block, copy the block, and move the block as if it was a single entity. This worked great except I discovered that the dimension links did not work while the geometry was in a block. So I needed to explode the blocks. All of the details made the drawings look very complicated. There ended up being so many constraints that Solidworks started asking to not display them so that the software would not get sluggish.
  9. Example of connector piece showing constraint icons (green), driver dimensions, driven dimensions (grey) and linked dimensions (red).

  10. To make the various shapes, I used the polygon tool to draw polygons with 3, 4, and 5 sides. I converted the polygons to construction lines and then just snapped my side “blocks” to a corner and rotated them. This went pretty fast and it would not take much more effort to make more polygon variations, connectors with different angles, etc.
  11. Example of pentagon drawn using five side blocks.

Laser Cutting:

  1. I started by finding a pile of cardboard that was all the same. In this case, it was nice three-ply cardboard re-purposed from computer monitor shipping boxes.
  2. Using a digital caliper, I measured the thickness of a couple of pieces of cardboard. They were all about 0.17 inches thick.
  3. I then used SolidWorks to create two identical test strips with slots ranging from 0.18 inches to 0.14 inches wide.
  4. I attempted to laser cut the test strips on our 40-watt Epilog Helix 24”x18”. The started with the recommended settings for cardboard for this machine: Raster- speed 10/power 30 and vector- speed 26/ power 100/frequency 5000. It wasn’t even close to cutting all the way through the cardboard.
  5. The first test strips did not get completely cut out.

  6. So next I tried speed 10/ power 100/frequency 5000. At this setting, although it barely cut through, it caught on fire a little bit a couple of times and in general scorched the edges. This is not normal. I did a little trouble shooting and determined the air assist tube was way out of position. I rotated it back until it was pointing at where the laser is focused to cut. This greatly reduced the flames. Now, I was able to use a setting of 7/100/5000 to cut through and it would not burn. This still is not right. It should not take that much power. Maybe the lens is dirty or the mirrors are out of alignment. So I went to our other laser.
  7. A problem with the air assist caused the cardboard to catch on fire and burn the edges.

  8. I was able to laser cut the test strips on our 25-watt Epilog Mini Helix 18”x12” without any problems using the recommended settings for cardboard for this machine: Raster- speed 100/power 30 and vector- speed 20/ power 100/frequency 5000. I think there was a typo in the raster setting. It was readable but very light. The vector cuts were just right.
  9. The chamfer in my design makes assembly easier and since the cardboard is compressible the tolerances are more forgiving. So, I decided that I liked the tight fit of the 0.14” slot.
  10. A test strip makes it easy to determine just the right slot size without a lot of trial and error.

  11. I considered making several different slotted connectors each with a different angle. Instead, I decided to try a generic round connector piece that could connect at any angle. Thanks to the tight press fit, the round connectors work but do not hold nearly as well as the slotted connectors.
  12. The rest was easy. I used the parametric ability of my SolidWorks design to set the slot width at 0.14” and the overall depth of the slot at 0.5 inches with a 45 degree chamfer. Each side was set for 8 inches. I nested the parts to make the most of the 18”x12” bed size of our 25-watt Epilog Mini Helix. The pentagons ended up being slightly bigger than 18 inches, so I ended up needing to cut them on the 24”x18” bed of our 40-watt Epilog Helix.
  13. Just the right laser settings for a clean cut all the way through without any burning.

    This is the carcasses of cardboard from all the parts I cut.

    The parts fit together snuggly and it was fun to see what 3D shapes I could assemble.

Project Files:

Back to index