Tuesday, June 14, 2016

Side projects update: Optical/radio telescopes, CAD, new rigs and custom sensor breakouts

Apologies for the radio silence on this blog, work has been continually crazy and some projects have been sidelined for temporary holds due to funding reallocation, however the backlog of completed short term projects has been full for a while and I thought I'd get some posted to tide you over until the next flagship update (Coming soon! Major redesign underway). Details on projects after the break, with a teaser for the flagship redesign at the end. 
8 inch Dobsonian telescope with craptastic eyepiece
First up is the camera eyepiece for my Dobsonian telescope which was born when I needed a way to show multiple friends the views but their glasses made looking into the eyepiece a difficult affair that had problems with getting the optics to be perfectly aligned and seeing smears of light. As I had a cheap security camera on hand for a different project, adding a C mount to telescope bore adapter made this a no brainer upgrade. Add in a 7 inch component screen and a lithium 12v battery to complete the setup for portability. To use the system all the user has to do is remove the eyepiece, add in the camera w/ bore adaper, plug into power and the screen, focus and get amazing images (proportional to the sensor quality). There is a way to use this design with certain USB webcams but that's a project for another day.

Craters on the day/night terminator

Mountain range!

 Looking at the moon through a telescope had me thinking, with the versatility of the new SDR I bought (that has standard SMA connectors instead of the crap MCX) if it would be possible to use it as a sort of rudimentary radio telescope. A bit of googling and scrounging later, I obtained a disused satellite dish from a friend (neighbors weren't keen on me borrowing their dishes sadly). Initial testing trying to receive the sun's radio emissions were unsuccessful using either antenna that came with the dish, unsurprising since the left antenna has a downconverter and I was not supplying any power to it, right antenna was a cavity antenna tuned to around 24Ghz which is WAY out of the RX range of this SDR that peaks at 1.7Ghz. This project will likely be in progress for a long time with the end goal of listening to Jovian radio storms.
New SDR with wideband antenna (WiFi)

World's crappiest radio telescope
Strapped to a fencepost for support

Another side project I've been working on that will be available to purchase soon is a custom breakout for the WiiMote IR camera sensor. This sensor is special because it packs 1024x768 tracking resolution for up to 4 points in X/Y/Size at 150Hz max speed over I2C into a fingertip sized monolithic package. This project was born after I was looking for a tracking camera but all the solutions were either full machine vision requiring a computer (bulky and heavy), or too complex to work with my limited skillset. While doing the research I had stumbled across the WiiMote IR sensor and it caught my attention, prompting me to go searching for a library and somewhere to buy a breakout. Finding the library to make it work with the Arduino proved to be quite easy, however it was abandonware that didn't work with IDE >1.0. A few tweaks to the code and calls made it compatible and I uploaded it to my GitHub under the new name "IRVision" so it would be easier to find for those who wish to tinker with it. Breakouts were another story entirely, being practically nonexistent aside from a few that are no longer available for sale, student projects, or a terribly messy and bulky. This lack of what I wanted prompted me to start designing a breakout to fulfill my needs. There remained a major problem to design this breakout; the datasheet, part number and consequently the precise footprint information are not able to be found anywhere I've looked (even a few greynet contacts had nothing), much less sensors available for sale(through the usual suppliers). These problems relegated the project to a back burner until I had a breakthrough, one of my net crawlers managed to find someone who had a reel of factory fresh sensors available for sale at $14/sensor shipped. With this news, design of the board began with the footprint. From an image of the WiiMote board and a reference dimension the footprint was extrapolated(Basic reverse engineering, tutorial coming soon), boards, components and a sensor were ordered. A week and a half later, with all parts in hand I excitedly sat down to assemble a test unit only to find out my footprint was wrong. Not only were the pinholes and support shield holes too small but the plastic pegs were misaligned. Undeterred, I tried redesigning the footprint but ran into the problem of overlapping holes violating DFM rules. This had me reach out to Laen of OSHPark asking what I could do, to my surprise he somehow managed to find an existing footprint of the sensor, made it a component and the latest revision of boards is ordered. Once this breakout is known to work and the beta testers(already chosen) and I come up with some cool example projects, a larger run of 10 will be ordered, assembled and put up for sale on Tindie.
Custom PCB design
Still feels weird holding something I designed
PixArt sensor scaled to finger
The joys of making footprints for a component that has an impossible to find datasheet
 After spending way too long doing all my CAD only on the computers at TechShop DC (only had Macs at home), I'd needed a computer for school so I bought an Acer Aspire 573G packing a GT940M GPU and windows. For nearly 2 years it became my daily driver and allowed me to work on projects at home. However, the taste of power and true PC gaming made me realize I needed a replacement for my weak Mac Mini. That journey began one day at TechShop while taking a poke through some new scrap that arrived I popped the side panel off a computer and immediately noted the newer generation stock Intel heatsink and i7 labeling on the motherboard. Slightly giddy, I removed the heatsink, wiped off the thermal compound and discovered an Intel i7 920 CPU in a matched motherboard. Upon testing, the motherboard fully POSTed and it was carried home that night, where it sat on my shelf for a month or so. While helping a friend ascend to a god-tier rig I noted the prior rig was mostly gutted and asked if he'd be willing to part with some components to which he told me $40 for the lot. In there was the GTX 550Ti, 8gb of Corsair RAM, and a OCZ power supply. On top of that I ordered a cheapie $35 Rosewill Blackbone case off amazon to complete the system as I had a few hard drives laying around. After a few months of use and late night gaming and CAD sessions, the bright blue fan started to grate on my nerves and retinas as it lit up my entire freaking room at night. This annoyance and the lack of USB3 started to drive me nuts so it was back to case shopping which led me to the Corsair Carbide 100R. At $50 it was a bit more expensive than the prior case but it was a WORLD of difference. Higher quality plastics (matte!), Side facing drive bays with sleds instead of rearfacing bolt bays, side window, dim white power LED and MUCH nicer cable management spaces. While migrating my computer to the new case, I was given an i3 2120 CPU, GT610 GPU and 8Gb of ram by a colleague at work. Listening to my brother's lamentations about their terribly slow iMac I built a rig out of the parts I was given in my old case with a cheap $50 motherboard to end their complaining and make house LAN night possible. Meanwhile, I've been planning upgrades to my rig including triple monitors and a better GPU for increased productivity.
Old case with a new computer thrown inside on left, new case on right
Old case w/ i3 2120, GT610, 8gb RAM (Built for $50)
Current rig: i7 920, GTX 550Ti, 12GB ram (built for $95)

The last side project was a challenge I took on out of boredom and to hone my skills, to reverse engineer a full CAD model from a single image and dimension. I'd remembered reading a Wired article about a new missile in development that mentioned a single dimension "...25-inch..." and had the image below. Based on figuring the stated dimension was the longest axis, I scaled it in CorelDraw and began to turn it into CAD parts. Complete work time to go from image to individual CAD parts and assembled model was about 18 hours. I'll probably 3D print out all the parts and hang the model in my workshop, possibly even make a flyable model rocket version for laughs. This foray into reverse engineering was a heck of a lot of fun and I'll probably make a tutorial for my method down the line a bit since it is a useful skill and a good challenge.
The source image
Completed and assembled CAD model
Below is the teaser for the new rocket stabilization system design. The current design is full of problems that the newer version fixes most of but brings in a whole new batch of problems. For the full details, you'll just have to wait until the post goes live. Until then, keep tinkering!
Quick teaser for the new system design on the left, current on right.