Sally Williams @ Fab Academy 2015

Assignment: 

• Make a machine, including the end effector
• build the passive parts and operate it manually
• document the group project and your individual contribution

Note: Fablab were providing 4 x machine kits (including 4 x stepper motors / spindles, 4 x nodes plus assembly components).

I document my individual participation for the group assignment below and the group documentation can be foud here.

I was tasked with taking photos and essential documentation of the key stages for the group project, so that the process would not be lost nor time wasted on duplicate efforts. This means that this individual documentation page looks very similar to the group documentation page.

Settling on an idea

Over the course of the first few days, we all brought different ideas that might suit the challenge of Mechanical and Machine Design and Build week. These included a 3D scanning bed; golf putting exercise machine; roulette-style game to name a few. 

The one that got the overall approval by the group, was an idea from Philip Hemstead - a digitally controlled ball tilting maze game.
To crystallize the project, we titled it "Fabaryinth".

In this game, incremental movement of stepper motors would provide the control over the tilt level of a plane, upon which the user (through a PC user interface with a simple control console, would try to guide a ball over a maze obstacle course and into a target hole.

This game would include the following areas of learning:

Machine Design
Electronics and stepper motor control (using multiple nodes)
Lasercutting / Milling
Electronics
Assemby of parts
Networking of single node communication system
Programming (input and output)

See tutorial for overview.


Machine Design

In the first instance, the suggestion was to use all 4 stepper motors (probably as it would be stable, and 4 were provided, not sure why).

     

I challenged this as I did not think 4 stepper motors would be necessary - the movement could be established with maximum 2 or 3 if the design was adapted. I could see that in order to acheive a tilt, a certain degree of freedom of movement would be required. I found this an interesting problem. To see how much freedom, I quickly cut three slots in a flat piece of cardboard and placed it over 3 leadscrews. It revealed the range of free movement that would be required between the game-bed and 3 guiding leadscrews to enable the tilt, which, to roll a ball, would only have to be minimal. This is demonstrated below:

 

Personally, I am more interested in design than electronics and programming, and looking back I would have liked for more time to have been allocated for each person to present their ideas on design development and rationale, especially on transferring the tilt to the game bed effectively. I felt we moved too quickly, without asking enough questions before progressing. However, I acknowledge that if my team mates hadn't applied momentum, we probably would not have finished the task as fully.

I also felt that to some degree, the cardboard housing template supplied, hindered our thinking; the heavy framework was unnecessary for our application. I couldnt help but wonder if we should have stepped back and considered giving a lighter touch to the housing and attempting to integrate the links to the game-bed? Again as a group, with a big task to complete under a time limit, this would have been a big ask, so I daren't voice it!

With a broad vision of the extra systems required to make the game work, essentially a custom-made game-bed and a flexible linkage to transfer the movement, some of the initial tasks were taken up by the four members of our group.

Electronics

I looked into the tutorials on the Fabnet Adaptor board, downloaded the file, milled it on the Modela and soldered the components onto the board. The ftdi cable had to be customised to suit the board and the single nodes.
The detail is shown in the photos / diagram below:

 


I joined David to follow the testing of the network for connections to Node 1, PC and  power supply. To help me understand the network and for the purpose of documentation, I suggested placing the network on some cardboard so that we could mark up and identify each part - this is shown below:




Cardboard housing

The main part of this task was taken up by Philip and Michael. This was fiddly; I tracked the process, and they had to make some adjustments and cut some cardboard leafs away in order to create more space inside the housing so that the spindles would be in true alignment with the housing support guides.





Game-bed

This task had been assigned to Kasper, but nonetheless I suggested the idea of creating a simple organic undulating landscape form in Solidworks and then to mill it on the shopbot using a light foam board. In my rationale, this material would offer a lightness to the choice of materials. A light material would be beneficial in reducing the stress, therefore torsion on the linkage between the game-bed and its connection to each leg and relative to each other.


Calibration

Next I joined in with the debate with David and Kasper around the discrepancy between the distance of travel requested in the python code (written by James and  David) versus actual distance travelled. Without stating the word 'calibration' once (!) we worked out that we needed to ascertain the distance travelled by one revoution of the leadscrew and amend this amount accordingly in the code. Currently, the amount defined in the code (6.096mm) was relating to a leadscrew with a different lead / pitch. This lead travelled by the leadscrew in our kit was tested and shown to be 8.0mm. David changed this definition in the code, thus completing the mystery of the discrepancy and performing an essential calibration.

Therefore, following calibration, an instrcution in code to traverse the stepper motor by 10mm would deliver a true 10mm.



For the remainder of the project, I shadowed David Mason and James Fletcher in the following areas (see group page for fuller documentation):

• creating the control console with wx python
• contributing ideas to improve the flexibility of the linkage between the bed and each leadscrew housing.

I felt that whilst it was an attractive solution, we were introducing unneccessary complexity into the design by using a living hinge. In earlier tests, the simplicity of string had proved to be totally effective in providing a problem-free solution to tilting the game bed; one that had no problem with torsion or attachment.  More than once the torsion upon the living hinge was fatal, as shown below (although this was cardboard, it was also very fragile in 3mm plywood):

 

However, with sheer grit and determination the team managed to get Fabaryinth up and running for the presentation to Neil.
David managed to score a live goal in under 15 seconds. I would call that a success.

   


Link to Tutorials:

Fab in a Box

 Tutorial

Explanation of Fabnet

Download files:

Fabnet Board