Digital Assembly

Keywon Chung, Marcelo Coelho and Scott Greenwald

final ball diameter = 1/16"
part width = 6/16" = "1/3-inch part" (0.375")
thinner: 0.02" = 1/3 of the ball diameter
thicker: 0.04" = 2/3 of the ball diameter

- How to keep parts in place? Do we need tabs?
- Spacer + tabs?

 - Conductive tiles and insulating tiles
 - Triangle vs. cross: zig-zag, no shear lines?
 - Balls vs. pegs: chamfered edge
 - Cut them in sheets, connected by tabs
 - Alignment rods for multiple layers
 - Flow spheres over the layers
 - Final dimensions: 50 mils? 
 - Tool for cutting shim stocks: alternating teeth?

Motion sequence:
 1. Button press
 2. Move pen down
 3. Button release
 4. Move pen up
 [move x-y]
 5. Move pen down
 6. Button press (extra force down here would be useful...)
 7. Move pen up
 8. Button release

To do:
1. laser cut and assemble conductive and non-conductive parts
   a. can we cut a single material by etching smaller different depth holes?
   b. can we bake the parts together so they will stick?
   c. what kind of glue would work and at what scale

2. figure out suction cup mechanism
   a. valve 
   b. suction cup itself
   c. how will the suction cup press the part into place
   d. pump

- what process(es) will it use?

For robust design with that reduces the number of motions necessary per part assembled, we use three principles:

1. Periodic Juxtaposition: Use a sewing machine-like head in combination with a platform moving at constant speed (as opposed to a jerky-arm model)


2. Part Aperture: For small numbers of parts, we propose columns of feed stock parts with individual apertures that release parts in coded order. For large numbers of parts, we propose a cylindrical array of part columns with a fixed aperture that releases parts in a coded order


3. Funnel Queueing: Once release from their apertures, parts slide into funnel that queues the parts for placement by the pecking element. Introducing a funnel eliminates the need for a 2-d coded motion that selects the desired part from an array of feed stock.

- what will (and won't) it be able to do?

In this case, "2-d" refers to the motion of a platform which supports the part. We assume that snap, adhesive, or other bonding can be initiated mechanically by the pecking element.

- what are the benefits (and disadvantages) over current practice?

By separating oscillation from translational motion, we achieve a robust design.

- what are the system components?

Feed stock, apertured array, funnel, pecking element, 2-d platform

- which parts exist, and which will require development?

Aperture, pecking element, and funnel need design.

- what will the workflow be from design to output?

Geometry file in, aperture coding out.

- what consumables will it need?

Ordered feed stock parts.

Two-part process Make multi-layer pieces: Operand layers (glues, resin, magnet, copper etc.) and coded dimples optional Assemble the pieces: Achieve multiple configurations using robotic arms and syringed functional materials
Part 1. Make multi-layer pieces: Combination of printer/copier + iron + bandsaw
Part 2. Assemble the pieces: Achieve configurations A-F. Robotic arms pressfitting + injection syringe

• Feed
• Cut
• Layer
• Glue
• Coded dimple
• Grab (2-spot or 4-spot)
• X-Y-Z movement and rotation (assuming an innovative gantry)
• Press fit
• Melt and frost-dispense materials
• Sort and store

• tba

• We can repurpose many coper/printer functions like sorter trays.
• Syringe injection of melted materials (solder etc) exists.
• Robotic arms exist, but we need to investigate their capabilities and alternatives.

• tba

• Assembly alone could be one piece per second. It is the first part of making multilayer materials (like copier) that could be slow. If this part is replaced by active extrusion it may become faster.