MAS 863- How to Make (almost) anything- Final Project- Sergio Araya- Ayah Bdeir |
kiNET
A dynamic, flexible surface
18.10 Proximity detecting structure: kiNET version 1.0
The first prototype of kiNET we made is a mobile structure on the Stratasys, kiNET. The structure consists of a grid of flexible nodes that can propagate movement.
We want to make kiNET responsive to human interaction and presence. Ultimately the surface would detect the proximity of a hand and move accordingly. Here we use capacitive sensing to create a light response to the location of a hand on the surface. We implement efe, a microcontroller board that calculates human capacitance between a plate on the board and a person's hand (The PCB layout, schematic and assembly code for efe can be found in Neil’s website: Efe- button+LED). We modify the assembly code so that when the capacitance value crosses a certain threshold, the ATTiny15L lights the appropriate LED.
KiNET's first prototype as a grid of nodes, 1.0
We mount these circuit boards (here 2) on the flexible 3D structure and use a 6V power source to drive them. As you move your hand the light follows it.
As you move your finger close to the plate, the total capacitance crosses over a threshold and the LED lights up (actually, the plate is not very sensitive so you actually need to put your finger on it.. :) )
A part of the proximity detection 3D structure (imagine there are modified efe circuit boards at every articulation) . In the previous assignment, we used a microcontroller board that controls a stepper motor to create the moving surface.
11.08 Motor controlled node actuation: node version 1.0
The main idea is to achieve smooth movement as a response to human interaction. We need to create an outward movement on each of kiNET's nodes. Using stepper motors might be a good idea. We create a structure (we call it the Terminator) that translates rotational movement to linear movement in the z direction. A Terminator has a microcontroller board in the center that controls 4 motors on each of its 4 axes.
A node on the kiNET surface that will be controlled by stepper motors
One axis of the Terminator is explained here:
On the rod of the stepper motor, we insert a screw thread drilled in its center. On the terminator is attached a bolt, and the thread is inserted in it. Depending on the direction of the stepper motor rotation, the thread is screwed or unscrewed from the bolt, and the terminator legs are separated or brought closer ->creating a linear movement.
One part of the terminator
Rotation transformed to linear movement through the terminator structure
With four stepper motors, the Terminator can fully control and move the kiNET node.
The terminator with its four axes of controlled motion
In order to control the linear displacement and it direction, we use a microcontroller board based on Neil's stepper motor board (found on the fab website: motion: stepper- hello9). Then the delays are shortened, we step more to each side so that the structure moves in and out considerably, and we maximize the torque of the motor by pulsing simultaneously rather than sequentially.
The motion control board (here it controls one stepper)
11.22 MicroNodes: node version 1.1
As an alternative to actuating and moving the nodes using stepper motors, we consider using micronodes. We make a flexing surface out of thin metal material for each node and then by actuating on that node, transform movement in the x-y plane to movement in the z direction. Below are some flexors Manu showed us. These create movement in the xy plane. We, on the other hand, would like to create movement out of plane.
Manu's flexor surfaces
We make several macromodels, in paper and metal in order to better understand how the actuation and motion control will be performed.
MicroNode Paper Model
We design a freestanding rectangle, connected by thin strips to the rest of the structure. When the two U shaped forms on the left and right of the structure are pushed inwards, the middle rectangle is lifted.
A paper model of the micronode. the middle freestanding rectangle moves up and down by pushing towards the center.
We experiment with different lengths of strips, size of the rectangle, and shapes to get noticeable movement.
An array of paper micronodes with different lengths of strips and different sizes of the freestanding rectangle.
MicroNode Steel ModelUsing the waterjet cutter, we make a large prototype of the micronode in steel. The sheet is thin but not flexible. Probably steel is not the best material. It is too brittle.
Each node is a micronode made by flexor metal surfaces, and the nodes themselves are connected to each other using flexors. This will help create a smoother movement and some degree of motion propagation.
A large steel prototype of the node. the steel is too brittle but the concept seems to work.
The idea of using micro flexors is very attractive because the nodes wouldn't need a bulky motor and power connections. Also KiNET could be a thin surface rather than having a thick electronic backbone.
A prototype of the KiNET structure, an array of interconnected mobile nodes.
MicroNodes do present a few problems, the main one being we may not obtain a large range of movement.
Layering of micronodes might be a good idea. Such that the movements of each layer adds to the one next to it, thus making the movement more noticeable.
ActuationHow can we create the horizontal inward movement in the xy plane?
On each U shaped structure we could mount a magnet. The magnets could be addressable by a microcontroller and the actuation done by polarizing the magnets to attract or repulse depending on the desired movement.
kiNET consists of two layers of surfaces separated by a small gap. The inner layer is like a base. It is fixed. The second layer is the outer surface and it is flexible. On nodes of the outer surface are permanent magnets. On nodes of the inner surface are some electromagnets. We control the electromagnet with ATTiny26s. We determine the direction and polarity by changing the current applied to these electromagnets. Accordingly, nodes on the outer surface will be pulled towards or away, creating motion on the outer surface.
12.01 Flexor surfaces: node version 1.2
We design a set of flexor surfaces. All of these surfaces can be actuated on on a node. These are the nodes where the permanent magnets will be placed. The node is like a unit for the whole surface. By repetition, and transformation on the interconnected units, we get the below designs.
We play with different designs to see what parameters can give us optimum displacement: the length of the beams, their thickness, the condense aspect of the surface...
We cut some of these flexor surfaces on the waterjet. Using different materials, thicknesses of materials. Thin aluminum seems to be the best choice.
A potential kiNET unit, which will be actuated
12.03 Electromagnet driver circuit: board version 1.0
In order to energize the solenoids properly, and to create the desired movement, an ATTiny26L is used. Below is the driver board. Mosfets are used to provide the magnets with Vcc when the corresponding pin on the Tiny is pulled high. A protection diode is used to avoid back emf. We just use PORTB because we are hoping to be able to implement a sensing mechanism. Ultimately, you would move your hand in front of the sensors (infrared reflection?, capacitive sensing?) above a certain threshold, the movement is created.
We had a few problems with over heating. Too small a resistor (10 ohm) heats up way too much when the magnet is energized. With larger resistors, the magnet is too weak. We need to play around with that in order to get the desired result. We should also get rid of the LED to avoid their excess voltage drop. And the Mosfets should be connected to unregulated voltage, meaning in htis case 12V, not 5.
EM driver, 1.0
12.06 Flexor Patterns: node version 1.3
The waterjet seems to prefer curves rather than right angles. Also, the longer the beams around the point of actuation, the wider the range of movement. We make some changes to the above designs.
The flexor surfaces become sort of patterns. We are essentially fabricating our own material, with flexible properties and determinable behavior.
12.08 Planar coils: node version 1.4
What about using planar coils instead of rods and windings to make electromagnets?
By trying different pitches and spacings, we may be able to achieve a considerable amount of turns. Below are the board designs for planar coils with 5 mil, 10 mil and 25 mil pitch, generated by Eagle's scripting language.
We try both, the Vinyl Cutter, and the etcher to make the planar coil boards.
For the Vinyl cutter, 25 mil doesn't work. The turn wires are too close together and they conduct. We also tried larger pitches, and more turns, but we don't have any pictures of them.
Copper Planar coil done by the Vinyl Cutter
This is the etched planar coil, it has a pitch. It has a pitch of 10mils and a very weak magnetic force.
10 mil pitch planar coil
Planar coils didn't seem to work too well. The problem is there should be enough distance between the turns so they are not connected, but to get a noticeable magnetic force, we need too many turns. Therefore the coil has to be large to be effective. Too large to be useable in kiNET.
12. 09 Electromagnet driver: version 1.1
Triage:
We will not have time to implement the sensing mechanism. For now, a hardcoded motion patern will be loaded onto the processor which will drive the solenoids. The electromagnet driver 1.1 controls 12 solenoids, has zener diodes to protect against back emf and gives unregulated voltage to the solenoids.
Each board will be used on one module of kiNET. Each will control 12 solenoids.
We used 12 power resistors of 10 ohm each so they could withstand the heat dissipation.
the final electromagnet driver schematic
electromagnet driver board version 1.1
12.10 Actuation by solenoids: node version 1.5
We decided we need a total of 24 solenoids because the piece is going to be a 11 by 44 inch rectangle that can be hung on the wall. We test different lengths of rod, thickness of wires, number of turns...
The final solenoids are made using the bobbin winder and 2000 turns of 28AWG insulated wire. The rod is a 0.5 inch diameter iron core.
A subset of the 24 solenoids we need for kiNET
A small prototype of the structure is shown below. The surface material of choice is a thin ABS plastic. The material is flexible, but the nodes retain their position after being actuated on.
kiNET 1.1, a small scale model of the final structure
The surface is mounted on an acrylic frame, below which is the bed of solenoids (not seen here). Below the nodes are small magnets which are attracted to the solenoids when these are energized.
The node actuated by the solenoid (some of the traces broke so they need to be thickened in the final design)
We had to change the surface material we had selected because it was sold out. We started trying out different materials including: steel and aluminum. This was possible because kINET doesn't depend on the material but rather on the design of the pattern. Minor variations to the pattern have to be made depending on which material we are considering. These variations are the thicknessof the leads and the length of the beams.
The thin aluminum worked the best. Nodes and beams would bend noticeably, but would still return to their original position once the actuation force is released.
BUT of course, something HAD to go WRONG! At 7 pm, Saturday night before the deadline, the waterjet broke down. Which means we couldn't cut aluminum. We continued our search for the perfect material: acrylic, PETG, glossy paper... we tested virtually everything. Finally we decided on
kINET: final specifications:
3 Modules: SIZE?
Surface Material: PLASTIC
Frame: ACRYLIC
Back structure: ALUMINUM
Actuators: Solenoids
Number of solenoids per module: 12
With the deadline approaching, we had to do some triage and cut down on the initial requirements of kiNET. We ended up making 2 modules instead of 3.
2 front pannels of the modules
Thin metal coins were attached on the nodes of the surface, in front of each solenoid. A gap between the solenoid and the coin allows for the movement of the node.
The front surface, 6 nodes are seen here
Once the module is finished (front and back pannel mounted, solenoids inserted firmly), the microcontroller circuit is placed on top and the appropriate connections are made to each of the solenoids.
The microcontroller circuit provides power to each of the solenoids to energize them
Wach of the solenoids draws about 1.2 A when energized, so we had to use a computer power supply: 12 Volts, 15 Amps for each of the modules.
12.13 How to make (almost) anything: Open House
Below are some pictures and movies of the Open house. More information and updates still to come.
kiNET Pictures
Videos (all avis)