Conduct comprehensive analysis on various servo motors from different manufacturers

Currently, there are 2 versions of pib in planning; pro and education. The education version is stable and in market and the pro version is in planning and initial development phase.

In both versions, we need to have a good grasp on motors in market and their suitability with our product. Before diving into comprehensive analysis, here are some ground rules for motors in the short term.

1. Motors used must offer a complete embedded solution, not needing separate encoders and control algorithm, i.e. servo motors.

2. Motors for education version must be relatively cheap.

3. Motors for education version must use PWM control and only 7.3 volts.

4. Motors for pro version must not be very different than education version, dimension wise to eliminate redesigning parts.

5. Motors for pro version should be much cheaper than competition since price is our main differentiation factor.

6. Motors for pro version should offer higher strength, accuracy and repeatability than the education version.

Education version

According to analysis and experimentation done in Analysis and documentation of motors used in pib . MG996R will not be changed in next versions in short term. However, DS3225MG needs to be changed with more powerful servo motors with a better core DC motor and not higher reduction ratio.

According to the ground rules set at the beginning the options for upgrading DS3225MG are few. Towerpro, which manufactures MG996R do not offer higher powered motors. The 2 major manufacturers that offer stronger motors are FEETECH and DS servo . They offers the following motors for upgrading:

Pro version

The pro version should be stronger, more accurate and repeatable. After thorough research it was found that common robotic actuators that use CAN bus and a brushless core with a circular exterior like MIT open cheetah actuator are quite expensive on themselves and on control side, using them will eliminate our price point differentiation factor.

Open

Creating our own actuators with inhouse manufactured parts are out of our resources currently. However, we are working on it long term by designing cycloidal drive systems and experimenting with position control encoders.

The last option left is usage of smart servo systems. Those are servo motos using high quality strong DC motors, usually brushless, better gearing system, magnetic encoders providing better accuracy and less friction, better control circuity using UART instead of PWM.

1. Magnetic encoders provide much better accuracy than potentiometers and eliminate extra friction in movement

2. UART control provides much better feedback metrics which enables dynamic movement similar to humans instead of constant movement to an end position. Additionally, unlike PWM control it enables reduction of speed without reduction of torque which causes serious trouble. Lastly UART wiring is much cleaner than PWM wiring since motors are connected in series and not all to one source

3. Usage of better DC motors, usually brushless ones mean providing higher power without straining gears.

Usage of smart servos in humanoids are common and one brand stands out; DYNAMIXEL. However this brand is overpriced according to internal components and defy our ground rule of maintaining low price.

Shown below are options for motorizing pro version: