I think it’s fairly common in engineering to give beat-up prototypes and vehicles a warrior’s sendoff by making it do things you would never do with a new one. Tinycopter was pretty much a wreck already going into this “test”. The carbon fiber booms were splitting, the props were again cracked and taped together, and the four silicone tubing standoffs had mostly sheared through, causing the whole board to become very wobbly. It was almost impossible to control, a fact visible by the constant pendulum-ing seen in the video.
Since I was going to have to rebuild it from scratch anyway, and everyonearoundme was building tesla coils, the choice of final voyage was clear.
Except it survived fine. The damage in fact was all mechanical – it flew fine after the first strike, and was just too broken to do so successfully afterwards. Avoiding the Ship of Theseus problem, Tinycopter’s control board will be retained and used for the frame rebuild.
And by that I don’t mean that I can’t land it without everything exploding – in fact, now I can!
I’ve been making little usability mods to Tinycopter to aid in my flight attempts. One of the biggest problems I have as someone who has been well trained to operate vehicles on 2D surfaces is that I lack the 3rd dimension spatial awareness (at least in hardware – I have to think and emulate it) that model pilots have. This means I lose track of which way something in the air is pointing pretty easily. Especially for something like Tinycopter, which is 90 degree rotationally symmetric from afar, all it takes is for it to turn 90 degrees while I’m attending to keeping it level, or making sure it’s not about to hit the ceiling, etc. for me to just totally lose orientation and fuck everything up.
Which is why I added little LEDs to the end of the arms. One red LED on the front-left arm, and one green on the front-right, in accordance with regulations. This made a world of difference, since I can now positively orient it at any time. I can now reasonably hold Tinycopter within a one meter diameter circle without much trouble, assuming everything is working.
I’ve also played with the P and D gains some more, but it seems that the pair I have now (P = 2.2, D = 0.6) is about where the margin is between stable but responsive and limit-cycle oscillations. I’ve tried P up to 2.4 and D up to 0.8 with similar outcomes.
And now, more video!
Hey, what’s with the onboard camera? I threw one of these little things, which I bought some months ago when they were actually in stock, on Tinycopter using a rubber band. It seems to be able to handle the offset weight without much trouble, but I definitely have to trim the elevator channel back significantly when it’s installed. The video is a little grainy, there’s no steadicam or vibration isolation, and watching it makes people seasick, but whatever – it’s a little derpy cheap camera. It’s not a Cinestar 8 Revolution with a gimbal-mounted Sony NEX-FS100.
I keep saying I’ll keep Tinycopter working for Techfair by not destroying it, but chances are it will get totally rebuilt before then since the frame is becoming a wreck, I now have smaller ESCs, and the silicone pneumatic tubing is starting to come apart.
