For those of you who are not familiar with existing aerial photography gimbals, or the recent brushless motor controller breakthrough, here's a brief explanation that I gave someone in the "Quadcopters" Facebook group:
Most stabilized gimbals use servos to control each axis. The problem with servos is that the brushless motor which drives it is always connected to a series of gears to increase mechanical advantage (generate more torque), which inevitably introduces backlash or "slop" into the system. This shows up in footage as 'jitters' or jerkiness.
So, if you eliminate the gears, and instead use just the motor to directly drive each axis, you can achieve a much more smooth effect from the gimbal. The problem is that brushless motors are not meant for very slow, high torque applications (they're sweet spot is usually several thousand RPM). But, they can be altered to have increased torque by disassembling, stripping the copper windings off of the stator, then re-winding with much more and much thinner wire. This creates much more magnetic flux per stator arm (which is the driving force of all brushless motors), which can increase the output torque enough to be usable for gimbals.
Using this sort of direct drive brushless motor on gimbals will yield results light years more smooth and clean than with servo-powered gimbals.
After a few weeks of designing, prototyping, manufacturing, and headaches from malfunctioning components motors/controllers, I've finally managed to put together a properly functioning 2-axis camera gimbal using brushless motors and the recently famous "AlexMos" brushless gimbal motor controller. We're calling it the Rotorpixel Gimbal Lite (for now).
In its current form, it fits under the (ridiculously) popular DJI Phantom, but requires a slight landing gear extension to keep the gimbal from hitting the ground. I'm having some mounting brackets made specifically for the DJI Phantom, which should eliminate this problem, and reduce overall weight.
Some rough tests that I've done show that the motors are capable of comforttably stabilizing a load of around 400g. But, the mass that I used wasn't camera-shaped, so this number could be off, since it's not just the mass, but the overall required torque that matters. I guess I should just measure the max output torque of the motors to be as accurate as possible, but that'll come later when the design is completely finished.
It's important to have the camera's center of gravity (COG) balanced around each of the motor axes, so I designed the camera plate, pitch-assembly, and roll-assembly each to be fully adjustable to achieve perfect balance. Doing so reduces the output required by each of the motors to preform the stabilization.
Designing the frame was relatively simple, and this project would've been complete a long time ago if it weren't for several glitches in the process.
1. Custom wound gimbal motors didn't exist, so I had to order suitable motors which I could disassemble and rewind by hand. This is a huge pain in the butt, and by the time I was almost finished doing the rewinding, motor manufacturers caught wind of this gimbal motor craze, and started manufacturing them!
2. I ordered a few new gimbal motors, and when they arrived, one was faulty, so I couldn't get a fully-functioning system together.
3. The gimbal controllers were (and still are) very difficult to get your hands on, due to demand. But, I got an order in for one early, and had it shipped over from Europe... by lettermail, which took forever. Then, I tried it out, and it was faulty! So, I put an order in for another couple, and waited patiently. They came sooner, but one of them was also faulty, leaving me with only one functional controller.
4. Yesterday, I shot this basic test video with a Phantom, expecting that I'd go shoot some pretty stuff today. I did, but promptly crashed it into a shallow river, so now my Phantom is out of commission until I can get it working again.
I'm sitting on a bunch of gimbal motors, finalizing the gimbal design, and waiting for a bulk order of gimbal controllers to arrive. In the meantime, I'm going to spend my time trying to fine-tune the design, experiment with the PIDs (I'll explain those later), and, of course, shoot as much footage as possible... the results are sometimes so buttery smooth, that it can give even the most mundane subject matter a cinematic look.
Photos of the gimbal and the initial vibration isolation system will follow soon! Stay tuned.