I've had an epiphany. So in the winch in place pulley system I was working on before, my concern was that when winching in the string things would be taught and reliable but when the motor reverses and releases string, that is when any snags in the system could cause the string to not be taken up rigidly and tension on the system is then lost and the motor is then unspooling string which isn't being taken up which will result in a spaghetti mess of string spraying everywhere out of control and getting all tangled up. The solution I had was a constant tension spring attached to the turn in place pulley output that would ensure that always keeps the string in tension as the motor unwinds. However, that was a extra cost and complexity and volume taken up by yet another thing and when you multiply that out by 300+ motors that's a LOT of springs added taking up a ton of extra space. That is why I moved to a belt based system instead of string and winch based for the first pulley. So the epiphany was this: it hit me that I can simply have the spring that does the extension of the final finger joint be what puts tension on the whole system and then if at any point in the system a snag were to happen, rather than tension being lost as the motor blindly unravels, not detecting the snag, I could have the motor NOT actively unwind anything at any point! So the motor, when unwinding is to occur, will simply turn OFF, rather than actively drive the unwinding electronically. It can pulse width turn off just acting as a brake to moderate speed of extension but at no point do any counter clockwise release or unwinding of the string. This way, the system only itself pulls string off the motor output shaft and if the system at any point snags, the extension stops and the string is all still under moderate tension but just no further advancement takes place and the motor does nothing further but blindly turning on and off but not actually spraying out thread everywhere at all. Eventually, the potentiometer measuring the joint angle of the finger joint would detect things are not moving and the system would KNOW it has a snag somewhere and at that point it would perhaps try to contract then attempt extension again hoping to dislodge the snag. If this did not work, the system would go into a troubleshooting routine like notifying the user (myself) to fix it or fixing it itself or w/e. But no damage would occur in this setup involving a unraveling mess or tangled mess. Simply the snag itself would be discovered and addressed but no catastrophic series of failures would result in theory under this new setup.
So with all of that said, and this solution in place, I am ready to return to the turn in place style winch style first pulley setup I had before and then the Archimedes pulley will do the rest. So the first pulley will be 2:1 downgearing and the Archimedes system will do 16:1 for a total of 32:1 downgearing. No constant tension spring needed anymore! Much simpler now. Everything I was concerned about is then solved now.
The belt based system fix ideas I was going for may have worked but as of right now I'm abandoning that course. I prefer the winch style and think belts would be higher maintenance and slippage would perhaps be an issue even with all the changes I had mentioned to improve on it. The fact is, belts only have so much surface area to grip onto so they don't scale down too well to tiny pulleys IMO. Large pulleys are better due to large surface area and more for the belt to grip. So my super miniature belt idea was a bit doomed from the start even if it could have worked (and it may well have worked) it just isn't ideal theoretically and I'd rather go with something I trust more intuitively for now.