There was a joke that one could take the gimbal from a quadcopter, flip upside-down, put it on a handle, and it becomes a handheld gimbal for use on land. It was an intriguing idea. I would like to share our interesting experience in the design process.
Popular Gimbal Controllers Available
The most popular controller board for DIY gimbal seems to be AlexMos Brushless Gimbal Controller (BGC). It ranges from early 8bit model that uses ATmega328 with expensive L6234 3 phase motor driver IC to later 32bit model that uses STM32F303 with lower cost DRV8839 dual-1/2-H bridge driver IC.
Other available gimbal controller boards includes 32bit STorM32-BGC, 32bit EvvGC-BGC, and 8bit Martinez-BGC.
The gimbal controller boards use 6-axis gyro/accelerometer IMU chip such as MPU-6050 to determine orientation as in quadcopter autopilot, and adjust gimbal motors to form closed-loop control. Earlier designs use single IMU, later designs use dual-IMU for better gimbal stability.
Variant Design
Some gimbals use special brushless DC motor with magnet mounted at the center, angle sensor such as TLE5012, and MOSFET for motor control.
Shaky video is simply unbearable. Seeing cinematic videos taken using quadcopters with camera gimbal, and handheld gimbals were quite expensive, we set out a goal to design one that could be affordable at $99, based on early estimates we derived 10 months ago.
Criteria for the gimbal motor selection is easy, one that offers highest torque w.r.t current consumption, to offer longer battery operating time and able to work with larger heavier loading. We checked 6 available ones on the market that is suitable in size, and ended up picking a most expensive one which offered best torque over current consumption performance.
Seeing none of the available gimbal controller boards on the market could meet the performance and cost objective we wish to achieve, we decided to design our own.
Checking performance of several 6-axis gyro/accelerometer IMU available on market, we settled on Bosch BMI160 for best low noise performance.
32bit STM32F104 is selected to drive DRV8839 half bridge driver, standard attitude sensing technique using gyro/accelerometer IMU on camera holder is used to form close-loop control. A second IMU on the handle motor is used for better overall stability.
Gimbal arm cut from thick acrylic sheets was used to make a gimbal prototype for initial firmware development. Due to long length of the gimbal arms needed to hold a phone and less rigid structure of the acrylic material, there seem no solution to the mechanical vibration and hum encountered. So we decided non-metallic material is not viable for gimbal arm design.
We then made a gimbal prototype out of aluminum tube. It was more stable than previous.
Afterwards, 3 iterations of CNC-milled gimbal prototypes & PCB electronics were done over 4 month period and additional 2 more months on firmware development to arrive at current good functional state. Several iterations was needed due to (a) first time designing a new product, some design issues could not be foreseen before holding a real prototype in hand. (b) Some changes were necessary when later considering production issues. If production engineer were involved from beginning, might reduce to two iterations.
The firmware design for a handheld phone gimbal is much more complex than for a GoPro gimbal. A GoPro gimbal only has limited size and weight variations. A phone gimbal need to cover a wide range of sizes and weights. Thus firmware dealing with wider range of variations becomes more complex.
Balancing is important in gimbal design. More balanced it initially is, less power needed to maintain gimbal lock, and more power can be devoted to maintain locked state under more stressful operation. If initially unbalanced, more power is needed just to maintain balance when holding it static, less power is available to maintain lock under stressful operation. Due to this, the design only aims to cover mid to large size phones ranging from 130g ~ 210g, to give optimal performance to latest large size 4K phones such as iPhone 6S Plus; opposed to most phone gimbal on the market that tries to cover phones of all sizes and perform less well with large heavy phones under more stressful operation.
Although the development time is only around 10 months, we are proud that the result is quite good as below performance tests show.
And this is the classic elevator test:
Different from most designs on the market that use USB interface, this design uses Bluetooth interface. It opens up possibility of automated video shooting like GPS Follow Me function found in aerial photography. Below is proof of concept demo of what can be done. First it shows WiFi streaming of video recording to remote phone that could control camera direction. Next it shows Follow Me tracking using GPS of both phones. It’s rapid prototyped code without any optimization, so behavior is very sluggish, proof of concept only.
Later with further firmware development making this function stable, it would be interesting to see it mount on ArduRover application, enabling low angle camera shots never possible before. Imagine what video footage can be made with a 4K phone!
With aerial photography hardware becoming quite affordable nowadays, we believe there should be similar high-performance stabilizer for casual video taking affordable to most people. Believe me, flipping over the quadcopter gimbal and stick it on a handle is not the way to do it.
Development of this exciting handheld gimbal is near production ready. It’s launched on Indiegogo with early bird perk of $125. Moving beyond shooting amazing video from air that everyone here is doing, it enables casual phone video recording into cinematic footage. Put it on APM Rover vehicle, one could make never before seen creative low angle shots. We hope you are excited about this new product as we are and would be interest to check out our Indiegogo project: http://bit.ly/1Pv1OvE Thank you.
