This post is pretty long, mainly because this part of the project has been ... difficult. I have designed, built, thrown out, redesigned, rebuilt, and thrown out the eye mechanism several times. Movements binding, linkages are too complex, the motion isn't smooth ... lot's of problems.
The basic design goal has been for the eyes to move side to side and the eyelids to at least blink. If the eyelids can move to intermediate positions (wide, narrow, squint, etc.) then so much the better. I originally wanted the eyes to be able to move up and down as well, but decided to remove this capability in order to make the mechanism simpler and lighter. Besides, up and down is handled by the neck/head mechanism.
The basic side to side movement has been part of the design for awhile. I haven't automated it yet, but the movement is so smooth that I'm not concerned.
It's the eyelid mechanism that's been the struggle. I intend the visible skin of the eyelid to be made of silicone infused spandex to give it a rubbery texture, sort of like a toad's skin. I've made a sample based on a technique I learned from Ted Haines at the Stan Winston School.
The movement of the eyelid is structured around a leading wire that the skin will attach to. That leading wire will lever up and down over the eyeball. The upper and lower eyelid will mirror each other.
To make that leading wire, I used a 12 gauge gauge aluminum wire. I took one long piece and bent it into the shape I needed using two pairs of pliers and a hammer and anvil. The basic shape is a half circle over the eyeball, but there is a pivot point at each end of the curve, and on one end the wire extends out into a lever arm.
That shape took a lot of experimentation and changed a bit as I was testing it. The main problems were associated with clearances as the mechanism moved, but the fundamental direction of the lever action changed as I tested the drive train.
I also tried a couple of retraction/tension designs for the lid. I started with a spring, and then tried an elastic strap to hold the eye open. The actuator would pull it closed and release for open. I never found a good balance of force between what was required to pull the eye open and a solenoid to pull it closed.
I finally settled on a push/pull design, where a single solenoid is connected to both ends of the lever arm, Rotating the solenoid in one direction pulls the eyelid closed, and rotating the other way pulls it open. The eyelid is always under a bit of tension, but the resistance to a pull decreases along with the rotation of the solenoid. The amount of force required is quite a bit less than the spring/elastic method for any move.
I put the solenoids (the actuators) down at the bottom of the head. This kept the weight away from the eyes, which is at the end of a moment arm. The more I can keep the weight closer to my body, the easier it will be to control the movement.
However, the push/pull mechanism requires two 'tendons' for each eyelid. And with an upper and lower eyelid, that means four tendons for each eye. Those tendons need to be apply force to the lids in specific directions, and they need to run down along the 'spine' to the solenoids below. The angle they attach to the solenoids is important as well.
All of these requirements meant that I need to direct the tendon cables through specific paths, and those paths would be unique for each tendon. I tried using bicycle cabling, a traditional method used in animatronics, but the angles were too tight and there was too much resistance. Instead, I settled on fishing line fed through thin brass tubing that I could shape as needed.
3/32" brass tubing is easy to find and fairly inexpensive. It's soft enough to cut with a wood saw. I use my band saw and clean up the edges with a sander. I also use an awl to gently open up the ends, knocking off sharp edges.
I shape the tubing with pliers. Any curves are gentle with the largest radius possible. That means a lot of small bends over what will become the circumference of the curve. I could bend the tube over a hard form with the radius I need, but that limits the range of radii I can achieve. I prefer to hand form to exactly fit the shape I need.
Mounting the tendon paths to the skeleton of the head took some trial and error as well. As I was shaping the tendon paths, I could use duct tape to hold them down, but that wasn't rigid enough for the forces applied by the actuators. I'd experimented with using screws as hold-downs, but they they didn't give the kind of coverage the tubing needed. I finally ended up building my own clamps using c-channel aluminum extrusion, cut to 3/8 inch and drilled for a 4-40 screw clearance. It's easy for me to drill and tap holes for the 4-40 screws in the skeleton, so the clamps ended up being a pretty easy fix. The mounting screw served as a guide separating the two brass tubes and the channel wraps around the tubes and the skeleton. It's a firm and removable mount.
Finally, there are the tendons themselves. I decided to use 15 lb monofilament multi-purpose line (fishing line). It's flexible, strong, thin, and cheap. I can cut it to whatever length I need, and it's cheap enough to cut away and re-thread as desired.
I created two attachment points on each eyelid lever arm by wrapping a thinner gauge wire around the heavier wire, leaving a small loop as a tie-on point. I spread the coils of the wire so that I could cover the wrapping with epoxy, firmly mounting the attachment point to the lever arm. The heavy wire of the lever arm had enough pitting and gouges to provide the rough surface needed for a good epoxy joint. The positioning of those attachment points is tricky, because it determines the length of play needed for the push and pull in order to get good eyelid opening and closing.
The monofilament is tied to the attachment point using a fishing knot (yet another crazy skill I have which gets used for cosplay), then the tendon is routed through the brass tubing paths, and finally loops are tied in the end to fit over the brass fitting on the solenoid arms. The length of these tendons are critical to get the right push/pull tension on each side of the solenoid arm. I position the initial loop so that the tendon is a little long, and then I add knots in the line on top of the loop knot in order to gradually shorten the tendon until it is at just the right length.
Here's a shot of the (almost) final configuration. The position of the lower eyelid is a little off, but I went ahead and did a full operation test, and the solenoid works great. I have a blink!
Next, I want to go back to getting skin on this thing. I want to see how it operates inside foam and silicone. I also found some great attachment connectors for the solenoids to the tendons, so I'll switch to those when I reposition the eyelid tendons. And I need to add the actuator for side to side eye movement.
But for now, this basic design challenge is done.