This is the third major concept change, I think it will be the last one for a while.
This version stays with the horizontal racks driving the leg, then they're driven with gears from servos. The servos are geared up to increase the speed of the leg.
There was a lot of iteration to get to it's current state. I had quite a few issues.
The first layer of the gears squished out a little bit, preventing them from properly meshing, so I chamfered the outside perimeter.
Origingally I had one servo flipped around, as it took up less room, but it made printing and assembly much harder.
The servo's jittered and moved inconsistently. This took ages to figure out, but it turns out the screws holding the gears onto the servo horns where scratching the tops of the servos.
Currently it works by setting a path, which is made up of points, defined with one rack position and an angle. This is probably unnecesary because the path can be generalised, which will also make it easier to work with in the long run.
The Arduino then interpolates between each point, ensuring it moves the rack that's the most wrong, and the other proportionally, so as to not tell one servo to move faster than it can.
I had tons of bugs :( including instantly generating nans ( three seperate times ) and trying to over travel the servos. These have all been worked out ( fairly sure at least ) so now it's all working!
To be clear, the leg is the diamond moving, it's about as long as it will be.
The current module weighs approximately 55g, with ~30g Parts, ~20g servos, and ~5g of hardware. I can cut 15g ( and maybe more ) from the 3d prints, but the others will be harder to reduce.
I also need to improve the code so the leg can work as part of a pair. If the two legs got desynced, the robot would fishtail wildly, so I'll see what I can do to avoid that. I have a plan, it's just not implemented...
There's also a problem with the racks, they rotate a little bit, then the leg can pop out. To fix that I'm going to switch to a square shaft.