I haven't worked with one in a while, but I was thinking about the standard advice on applying rubber bands to a DR4B. I think it's a little bit off. The whole triangle thing is clever, and I'm not saying to avoid that. I'm saying the targets seem to be incorrect.
Why? Maintaining uniform force on the lift is not desired, nor is uniform torque. Aiming for this is incorrect to begin with. What we really want is to provide the most torque when the bars are horizontal and less as the bars are more vertical (toward the top or bottom). And, unfortunately, we always end up with the most force from the elastics at the bottom since all the elastics are trying to lift the mechanism the whole time.
My thought? It's a pain to try to get something to really fit those sine curves well, both mathematically and because we would use lots of connection points. Using the triangles accomplishes what we want pretty well and more simply. But we could still target a maximum force when the bars are horizontal. Set up triangles of elastics on the upper and lower bars. On one set of bars do them normally. The key to my idea is the other set of bars. Instead of doing the same triangles, we aim for something different. We want to force the bars away from somewhere a little under horizontal, both up and down. This way, when the lift is low the elastics counteract each other so there could actually be less lift at the bottom and more in the middle; this would allow for more strength assisting the motors without the problem of the lift not staying down to start. Also this way, when the lift is above horizontal and the regular triangle of elastics is getting weaker, it gets augmented. So the elastics on this part would need to be the most stretched when the lift is horizontal.
Mathematically, really roughly speaking, we're dealing with the left half of a concave up parabola for the potential energy right now (positive x being upward). Then add a gentler concave down parabola with its vertex a little bit to the left of whatever x is when the bars are horizontal. Ideally this could flatten out the curve some when x is smaller than this vertex while making a bigger drop-off afterward. As the negative of the slope is the force, that would give less positive force early on and boost it a bit around when the bars are horizontal.
Is it possible? The first issue is finding a good way to pull away from horizontal. I think this could be done with a triangle again, just making sure the smallest-perimeter triangle is formed when the bars are horizontal. The second issue is if this can actually be matched well against elastics trying to lift the whole thing. Varying elastics could certainly help, but I'm more worried about things skewing than about trying to find the right relative forces. The third is I'm not sure how well this would really create the most torque when the bars are horizontal, but I would think it would help move things in that direction. Fourth, there might accidentally be a bit too much boost when the lift gets really high, which would need to be avoided as best possible.
Anyway, just thought I'd post this to bounce ideas off of others.