Sunday, February 14, 2016

Q1 2016 Rocket active stabilization system update, now with fins, nose cone and a mockup

With 2016 well underway it's time for the first update on the state of the stabilization system!
First off, apologies for the massive delays this post has had, was originally slated for mid-Jan launch but between job hunting and other stuff I've been hitting delays and trouble at every turn. Since last semester ended on Dec 18th, I'd been going to TechShop pretty much every day until Jan 18th making massive leaps and bounds of progress along the way. The most obvious of progress is the completion of the demonstration mockup airframe which will be gone into more in-depth after the break. 
The system's current state on the demonstration mockup

The demonstration mockup was constructed to not only have a physical representation of the final assembly but also to familiarize myself with layout and wiring inside a confined space which is new territory to me. It is constructed out of 3 inch ID cardboard mailing tubes and has fins cut from standard cardboard bolted on, nothing special or lightweight due to this not being the final flight model. The overall mass of this version is 1.48Kg of which 452g is the stabilization system module. Testing and flight versions will be constructed out of "Blue tube" for it's superior durability and lighter weight though at much higher cost ($30/4 feet, whereas the tubes I'm using I salvaged from the scrap pile for free) 
The latest revison of the stabilization moduel
On the assembled stabilization module, the most notable additions are the printed fins and nose cone, both of which went through a few revisions during the month. Print issues with bed adhesion and extruder jams were the bane of my existence during the 3 hour prints of the nose cone sections for a total of 4 failures.
The new fin
Fin attachment detail
Some of my longer term readers may remember how much difficulty I was having with figuring out a method for attaching a fin to the servo spline, trying everything from printing a horn to trying integration of a round arm to no avail. In the end, I decided to stop trying to reinvent the wheel and rework an older design to bolt onto a premade servo horn. These are designed in standard diamond hypersonic form with the axis point 3/4 up the mean aerodynamic chord. Due to the mounting method of fins bolting directly onto the horn, installing these after the servos are in the body tube is impossible, forcing me to have a slot-in module that is held in place with bolts. The slots perform an additional function of being mechanical endstops for the fins to prevent an overcontrol instance(in theory). Sadly, two of the scavenged servos finally kicked the bucket and a full new set is going to cost me $72 ($18/HS82MG) which is FAR out of my frayed kitestring budget which has dwindled to nearly zero as of late. If any of my readers know how to get in contact with Hitec for a sponsorship or how to get parts donated to a student project it would be greatly appreciated if you would drop me a line. 
Servo retention plate installed
The integration can for my system needed to be reworked due to being a bit too small to fit in the 3 inch inner diameter body tube. I retained the previous method of servo containment and mounting, just upscaled a bit and added bolt holes for attaching to the airframe, in addition to adding a forward riser to hold the control board in place and provide a nose cone mounting flange. Making sure this has a slightly snug fit is key, as internal movement can instigate a self-amplifying and destructive resonant wobble. 
Board retaining slots on the system upper
Control boards V1 and V2
V1 and V2 wiring comparison
Sensor offset on current layout
The breadboard and connection board of the previous version were not only too hard to set up quickly but also very fragile and annoying to work with. I'd had enough so finally I went and built a nice shield that integrates almost everything while still keeping the Arduino removable so it can be used in other projects. V1 was a first attempt, all wires were red (bad idea) and shoddily laid out which caused a lot of headaches when it simply failed to work when powered up. Debugging when powered up showed the 5V rail hanging around 1.2V and not budging at all even when everything was disconnected. After a few hours of frustration my father walked by and offhandedly remarked to check the ground, much to my chagrin I found that I'd forgotten to connect the ground net to the ground rail which promptly fixed the problems (check for floating grounds). On the second revision the power and ground wires were in standard red and black while signals are in white (I ran out of wire colors okay?) in addition to being MUCH cleaner and better laid out. On the final revison, the IMU will be flipped around 180 degrees since this mount has it offset from the centerline by about 1/4in which may cause control reference issues resulting in the loss of launch vehicle.
APC220 Telemetry "backpack" next to cardboard prototype
Cardboard is top-notch free prototyping material, does the trick to get the basics blocked out quickly
Since I will be monitoring this launch from the ground during ascent I needed to get a telemetry solution figured out. Since I had a pair of APC220 wireless UART transceivers laying around after I was gifted them the year prior I decided to use those since they were not only on hand but had good nominal range (2Km LOS). After an initial prototype of a backpack in cardboard and hot glue, I designed and printed a proper backpack that has Velcro to hold down the module and a captive adapter for power and data lines. This assembly bolts onto the back of the control board with two diagonally placed holes as all 4 would interfere with my...not so great point to point soldering. 

Powered up 90% assembled system (top ring was removed in recent versions)
Trisegment nose cone attached to the system module
The nose cone was quite the interesting design challenge, the total height was 15 inches which surpassed the 6.1 inch maximum vertical print volume of the Makerbot Replicator 2s at TechShop. I decided to print the nose cone in 3 parts that bolted together for modularity and ease of printing. One of the biggest challenges in this was figuring out how to get the bolt interfaces designed on the curved surfaces of the nose cone interior while still being printable without support structure. In the end I opted for simple curved leadup flush connection plates which worked phenomenally with the minimal droop of Hatchbox filament (not sponsored, just freaking love that filament). 
Lower bolt interface detail
Internal detail of assembled nose cone
Lastly, an update that cannot be shown due to those ITAR restrictons; code revisions! I've started a ground up rewrite of the code, this time with full Kalman filtration of the sensor values and PID to finally ditch the modified example code with shoddy division operators and no tuning ability. So far it's been a challenge as I am unsure as to whether the Kalman should go in front of or after the PID control, has stumped me since the rewrite was on the books. In addition to the code upgrade, I've been considering upgrading to a more powerful Arduino or just going ARM, maybe even use the MSP432 that TI sent over for the additional horsepower that will be needed to crunch all the numbers involved in the new code.
On a funny note, the last day I was at TechShop for this month block, had packed everything up and found that the rocket was going to be too hard to carry around in parts. I left it assembled, slung it over my shoulder and walked out the door. Now you'd probably expect raised eyebrows considering it was in a government town but that is not the case on this particular day. It just so happened to coincide with a munitions expo, so while heading out I passed a crew wheeling out a SM3 missile mockup. 
Latest version's cable managed leads laid out on the workbench 
Sleek lines of the business end of a rocket
3AM Rocket surgery on an intermediary version
Intermediary version was pretty crappily built
Gotta love rocket surgery
Rocket standing next to 5'8" me, turns out it's about half the size of an AIM-9 Sidewinder
Failures in the background of the recently slotted body tube
Sneak peek into my workflow and hand drawn references for CAD
Three years between V0.1 and the latest, all self taught skills
Crystal city at 2AM around the holidays after a 20 hour marathon day of work at TechShop

1 comment: