Lab LogoFletch's Fab Academy 2014 Blog

07. Computer Controlled Machining

This weeks assignment is to 'Make Something Big' on the shopbot.  So we have available a 3-axis CNC milling machine with an 8 foot by 4 foot bed and as much 12mm plywood as I need.  As I'd designed my press fit 'Castle Construction Kit' from week 3 parametrically I decided to simply scale it up for this weeks task, so I'm aiming for a castle construction kit that a 5 year old could use as a playhouse.

07.01 Design Tests.

My first task was to get to grips with the tools and workflow for using the shopbot.  So I used this as an opportunity to do a couple of design tests of joints hinges and sizing.

07.01.1 Hinge Test

One of my fellow students was testing a hinge design on the shopbot for his project.  The initial test was very stiff and had less flex than he had expected.  As I wanted hinges to allow my castle towers to wrap around into shape, I decided to examine the design, try to understand what was wrong and then improve it.  Below is a picture of my improved hinge design marked to show where the flex in the hinge comes from.

Mikes hingeMy hinge

Ply Hinge Flex

By maximising the length of the section marked in red you can maximise the flexability of the hinge.  I designed this test part completely in 'Partworks3.5' as a test of this tool path as the rest of my design was in kokopelli.  CRV file here.

Ply hinge design in partworks

07.01.2 Koko Toolpath Test / Scale Test

Moving on from this I decided to do a quick test in kokopelli.  I wanted to see what the work flow was to get the files from kokopelli onto the shopbot and check what scale my final part would come out at.

Kokopelli sizing test
I would like to test using the CAM panel direct from kokopelli, however our shopbot computer doesn't currently have fab modules or kokopelli installed, I've started to rectify this, but more time is required to get Ubuntu updated to a suitably current version.  I also want to test Neils prefered method of CAD->PNG->Shopbot vai fab modules.  My kokopeli test file is here.

  1. So I decided to export my files from kokopelli into Partworks and go from there. 
  2. I exported to svg from kokopelli.  Resolution default of 10 pixels/mm.
  3. I imported the svg into Inkscape.
  4. Export from Inkscape as a pdf with no modifications.
  5. Open this PDF in Partworks.
Once in Partworks I noticed that the vectors in this file were 'fragmented' into smaller sections and that I had some extra vectors across tabs etc that I needed to delete manually.  I created the tool paths as before but when I previewd the part some vectors in the origional design were not being cut.  I had to add another step and 'join' all of the vecotrs into one continuous loop once I had imported the PDF into Partworks to fix this.

Bad export vectors

It took a couple of iterations to make sure that I had the scale factors correct and that the part was going to come out at a sensible test size.  Once I was happy with this I cut two copies on the shopbot.

Castel wall test

My first test part was to check scales and tolerances of the end tabs, so it had no dog bones on internal corners and the joints (particularly the center slot) wouldn't work. 

07.01.2 Joint Test

So I went back to kokopelli, added dog bones, and re-cut to test the joints.  I could have added these dog bones manually in Partworks.  However I wanted to reduce the number of additional manual operations required in each design itteration, so I made them part of the kokopelli design.
The external / internal dimensions of the tabs work OK, but I'd like to cut a series of test slots to correctly gauge the tolerances required for a good interferance fit.  I also noted that my 12mm ply was actually only 10.8mm thick.

Castle wall with dog bones on joints

07.02.3 Joint Test 2

I then moved on to creating a variant of the MTM snap joint in 10mm ply.  This uses the same design file as week 3 'Press fit kit', just set the machine type to 'shopbot'.
MTM joint

This joint works, but in order to get the flexibility required the 'tangs' need to be longer than in the 3mm ply version.  Also the 10mm ply used is lower quality than our 3mm ply.  It's full of discontinuities in the internal material layers that create weak spots.  This joint design regularly snaps in this material.

07.03 Cut Real Thing

Again the same design file was used as week 3 'Press fit' but with the machine type changed.


Above left - tower section before assembly.  Above right - wall sections.

Tower folded

Above - tower section folded into shape.


Above - The big and little versions of the castle together.

Further Unordered Thoughs...

Cut a 3D profile in foam, to test the 3D tool paths.
Try one in thicker laminated wood?
Fab modules and Kokopelli onto shopbot PC or drive shopbot direct from my mac.
Test CAD->PNG->Shopbot vai fab modules.
Tabs are great, you can easily remove them with a craft knife.

Test cut with different slot sizes. Koko export error...
Test cuts with 6mm down cut and 0.25inch compression cut
  no space to ramp in?
  tool gets very hot  and whilst no problem do get 'burning smell', also cut surfaces are 'blackened'.  Tool blunt?
  really glowing embers, see photo, so stopped job.  turns out speeds and feeds for tool into tooldb were wrong!  Tool going too slow!
  corrected and cuts really quick and well.

measured with calipers:
  test tab cut with 6mm down cut 50.1mm
  test tab cut with 0.25inch compressiion 51.5mm, even though tool size was correct.  Surface internal to material is much rougher!