Late Night in the “Office”May 19, 2013
Getting close to transferring an image. Below are image captures on a 500 MHz oscilloscope with matching LVDS probes. The data in the images represents the slave FIFO buffer on the FPGA going over the GPIF lines into the Cypress FX3 microcontroller. Thanks to Agilent for loaning this equipment to us for a few weeks, this project would have been at a stand still with out them.
ReflowMay 7, 2013
Last week, our prototype PCB was populated and baked in our DIY reflow oven. This reflow oven was built from an off-the-shelf toaster oven, retrofitted with thermocouples, solid state relays, and an FPGA as a P controller with Matlab running on a Lenovo uploading new temperature set points.
To properly apply solder paste to the pads, kaptop stencils from OHARARP were utilized for the top and bottom of the boards.
The image sensor was hand soldered after the power supplies and all nets were validated.
Sensor CommunicationMay 5, 2013
We have successfully baked our board(more info to come about this) and our initial communication over SPI is working. Upon the inception of our design we had planned on using the hardware bases SPI, however, after turning on 32 bit GPIF mode on the Cypress FX3 we have lost our DMA tunnel from one side of the chip to the other…. Bit-Banging it is! I am issuing a read command of 0x7D and its returning as expected 0×20. We are seeing some fall out on our LVDS lines but I am happy so far.
Extruded AluminumMay 3, 2013
Or aluminium depending on what side of the pond you are on. Aluminum is a great material for product design as it is easier to machine than steel as well as easier to finish. The specific blend of our 3″x3″ square bar is called T6061. Alcoa defines it as 95-98% of pure Al and that contains less than 1% of the following; Mg, Si, Cr, Mn, Ti, Cu, Zn and Fe. This formula is tweaked to produce different grade aluminums that have different tensile strengths. While we are not concerned about strength, this specific blend will accept anodizing very well so hopefully we can finish it with a nice color or colour that will not oxidize.
Camera Body Design TessellationApril 25, 2013
While waiting for a few parts of the project to develop, I decided to poke around with the design of the camera body. Lacking product design talent and CAD design, I turned to mathematics in aid of our design. I came up with the idea of machining aluminum into an asymmetrical diamond pattern. After digging through mathematical algorithms, I found that the Voronoi Tessellation was exactly what I needed. It turns out this algorithm is used in everything from epidemiology to computer machine learning.
I approached this design by generating random points on the face of the housing. I tuned the point generator to give higher probabilities to the area around the lens and then decay as it approached the edge. This would generate larger faces and also ease the machining process. These points were then piped into the Voronoi algorithm to make the tessellation. You can view a live random version of it here. If you hit refresh, a completely new design will pop up as the points are generated on the fly. You may need to zoom out with control+”-” as I still need to figure out how to convert these units into inches rather than pixels.