This experiment began early this week when I worked on the case parts for a new electronic test rig. I didn't have any industry standard parametric modeling tools at the moment, but that was not enough to stop me. After working with 3D models so long, I was able to create the case using the traditional polygon pushing skills that I've developed over the years. Before creating the parts for the test rig case, I went through and measured all the dimensions with calipers. The objective was to make a strong case that made room for the internal circuit boards to sit firmly in place. To achieve this, I created a lip area where the edges of the boards would sit along the edges of the case, and two raised regions that would run along the length of the case to support the boards in critical areas. Since this was a prototype the dimensions of each circuit board fell within the 77.35±0.20mm X 68.27±0.02mm. The test rig consisted of 4 identical circuit boards that would function as nodes using a TWI to communicate button events to a "brain" microprocessor that would gather the information via a ribbon that ran between the nodes, and deliver it to the PC via USB. Due to the indefinite dimensions of the boards length and the space between them, I would create the parts so that there would be longer depressions in the lower part. The depressions would be extended along one axis to ensure that the header pins, reverse mounted LEDs, and microprocessors would not displace the boards from the desired fastening plane. This would be important for the second machined part that would leave little room for error. If the boards would not lay down properly, there would not be enough space between them, the silicone button layer, and the spacers to fit under the lid.
In this project, besides the test rig case, I was responsible for toner transfer, acid etching of the majority of the double sided copper clad boards, drilled vias, bridged vias, and attached an assortment of components to the boards.
Placeholder, I will be making a proper demonstration video for this semester project soon.
Motion sensing method one.
Motion sensing method two.
This is not my first 3D print. However, this is the first 3D print that I designed hoping to achieve a dynamic mechanical system. The tolerance on the machine was 0.007". I had used a buffer of 0.008" when designing the plans for the object. I made the buffer space between the parts a hair bigger (0.008") in anticipation of a one piece print where everything would come out of the printer and function without and assembly or post processing required. Plans changed when I went to slice the object in half and laid the objects out in printing space to save support material. Since post processing and assembly would be required in this scheme, I should have went back to change the buffer size to 0.007" since I would need to clean everything up anyways. There was also an issue with the overlapping hulls that I created. The printing software filled the overlapping areas with support material. I was not expecting this because of how I had done prints with a makerbot in the past. Overlapping hulls were filled as if the two objects went through a union Boolean operation. That was not the case with the 3DS systems that I was printing on. The support material replaced the stronger ABS plastic in key points that needed to be reinforced.
There were a few reasons why this design failed, weak joints in the mechanics, large tolerances, and no stops on the pegs that held the moving parts in place. I'm glad I had the experience; now I know what I could do to ensure that the final product will function as planned.
Here are some mockup renders of the model.
It's possible to get pressure readings by measuring how much light is passed from one IR emitter to an IR detector. After experimentation we realized that the light from one receiver could be split into different channels and detected by multiple IR detectors. Before realizing that the signal could be routed along a 2D surface like circuit boards, I had some interesting days thinking about how the signal could be transferred through fiber optics, and how they could be merged/split.
Eventually I devised a system that could read multiple regions of the button and report an x coordinate, y coordinate, and pressure for each button. The beauty of the system is how the emitter channel panel could be flipped upside down and rotated 270° to be reused as a wave guide that could direct the light to the detector components.
Regardless, I hoped that I could create a one layer system to reduce production complexity. Along that line of thought, I created a complex waveguide that hoped to guide the IR by changing how the pipes branched off of the source channels. Unfortunately the design was one layer so a large number of 90° intersections was unavoidable. The probability maps showed that the light leaked to different light channels in this scenario.
I've stopped looking into this sensing method. I may revisit this later.
Copyright © 2013 Richard Hale. All rights reserved