As you can see, I added front guides to the "lower units", as I've taken to calling them. I had front guides in all my drawings of this mechanism as I developed it, over several years; but as I built it this first time, I ended up not fitting front guides at first, and then I thought I'd see if it could survive without them: simpler. But no. The spring force from the rollers below tends to drive the lower units upward until they rub against the lower machine rail. I had to grind away some of the inner surfaces of the machine rails, so they wouldn't rub against the shafts of the stop levers (overall, this mechanism should be built with a tad more clearance, next time); but grinding wouldn't help here, the spring force would just push the lower units higher until they contacted the new surface. Nothing else in the mechanism provides the necessary stiffness and force to resist this. Fortunately, though unplanned, it works out that exactly two thicknesses of my plywood (i.e., the guide itself and then a plywood mounting block) is just the right depth, so that the end of the lower unit dowel is just retained in its full travel, but the nearby fourth actuator-wire hole is not blocked by the guide. Thus, the dowel-holes of the front guide do not have to be notched, like the ones on the back guide, to clear the actuator wires. So, they have more "guiding ability" -- which is important because the most-crucial top of the circle is necessarily notched anyway: there's not clearance for a fully-closed circle on top, it has to be cut off, again because of my too-tight clearances. Anyway, the guides, such as they are, seem to help adequately.
The other change, to not-yet-implemented aspects, is that I have decided to put 2:1 levers in between the machine rails and the sliding trapwork with the inclined plane, which moves with the pedal. The force on the rails will be quite large, and therefore very strong return springs will be needed. Some mechanical advantage from the levers will help the sliding planes, which I call the "hills", do their job more smoothly.
Meanwhile, I've benefitted from this long break in working on the keyboard action itself. As often happens, my mind churned over a problem that was bothering me, and somehow it arrived at a solution "in the background", as my conscious focus was on other things (the stops). The problem was, too much friction in the damper system (with the springs and monofilament going over "pulleys"): I had assumed that monofilament making a close 90-degree bend over a smooth metal surface, was close to frictionless, but far from it. I could possibly have mitigated the issue by using actual, moving pulleys, but then things just got 98 pulleys more complicated, where I thought a straight piece of wire was going to do. And, 98 pulleys of unknown design, from an unknown source. Me, fabricate? All this bothered me.
But then I realized a better solution, going back to something I had discarded before, is to use levers. There is not room for damper levers above the keylevers, as in some early designs of mine; but there is room below. The change works out to be smoother than one might think. Same damper guides in the upper touch rail. Same monofilament adjustment screws in the tops of the keylevers. But instead of springs and pulleys, the monofilament directly lifts up the front end of the damper lever. The back end goes down, by a smaller amount (about 2:1), taking the damper wire down with it. The weight imbalance of the asymmetrical lever is, hopefully, enough to return the mechanism by itself, without added weights.
Not only is this likely to move with less friction felt at the keyboard, but also it provides a nice low-force way to implement "sustain", by lifting the front ends of all 49 levers at once. I have been worried that the force on the stops mechanism from pulling down 49 damper springs, was going to be too much and would make the machine pedal interface unworkable. There are still many worries there, but this big issue is at least significantly mitigated with these gravity-levers.
