Tricopter: single-axis test rig and some notes on balance
I thought it would be useful to have a way to test control of a single axis of the tricopter in a safely constrained manner, so I built a rig to do so:
The rig is made of scrap aluminum I found in the room bolted to a cutting board that is clamped to the table. One of my XBees, my laptop, and the joystick are to the left on the table. My battery is charging to the right, on the adjacent table. The tricopter is skewered through a steel rod at the top of the rig.
I tested PID control on this rig (instead of the PD control I had before). The I term seems to help a lot with making the tricopter both responsive and stable. I’ll talk more about this in the next post.
Some notes on balance:
First, it’s important that the skewer passes through the midpoint of the tricopter, at least in the X/Y body plane of the tricopter. When I first built a test rig (before I disassembled it and made this one), I skewered it where the tricopter naturally balanced. That balance point ended up being slightly towards the back due to the weight of the tail servo, but this created an effectively shorter lever arm for the tail rotor and an equally longer lever arm for the right and left rotors. Thus, a throttle increase for the tail rotor was overwhelmed by the same throttle increase in the other two rotors, which meant my PD controller (no I at the time) couldn’t work.
Second, a PID controller for a multirotor like this one technically needs to be able to linearly control the thrust of the rotors, not their rotation rate. Because the RPM-to-thrust ratio is variable, the PID gains will work for some throttle level but not as well for others. However, if that ratio is linear enough (which is what I’m assuming for now), I could get by with a linear throttle control. Either way, I will run some tests in the next week or two to determine what that curve looks like.
Third, a low center of mass is good. Theoretically, a robust PID controller should be able to stabilize the system regardless of whether or not the center of mass is displaced in the X, Y, and/or Z axes. Even so, the controller works better if the center of mass is centered and is lower in relation to the rotors, because the tricopter should essentially “hang” from the rotors in the air like a pendulum (thus the relocation of the battery to the bottom and the aluminum counterweights at the front). On the other hand, if the center of mass is located above the rotors, the tricopter becomes a sort of inverted pendulum, which is unnecessarily difficult to balance.
More photos of the setup: