Following Tim's advice, I now have a layout and functionality similar to a standard Filawinder, but I decided to make a simple CPU-less unit of my own design.  While a simpler design could have arguable benefits, the truth is that I could scrounge up most of the parts from leftovers I had lying around (hence it would be cheap to make) and I thought it would be a fun challenge to do it myself.

Since I don't have any suitable wall space where I would be allowed to hang a winder, I came up with the idea of building it into a handy portable case that could house both itself and the Filastruder, keeping them free of dust and out of the way when not in use.

I made a simple plywood box, roughly 20 inches square and 6 inches thick.  I stained it and added a leather handle and some brass hardware, fancying that this gave it the appearance of some bizarre 19th century science experiment or instrument of medical quackery.

The toughest part of the design (as I suspect is with any winder) was finding a way to sense slack in the filament without disturbing the path of the filament itself.  I found that even the lightest micro-switch I could find would transmit too much force back up the filament and cause a kink near the nozzle opening.

I settled on a lightweight 3D-printed "see-saw" rocker that I weighted down slightly on one side.  The filament moves in a loop down the left side of the box, across the bottom, and up the right side.  When the filament develops too much slack, the bottom of the loop pushes down on a curved acetate sheet glued to one side of the rocker. This lifts up the other side, raising a shutter that exposes a CDS-cell light detector to a single LED light source.

The cell is connected to a solid state relay that drives a small gear motor and the take-up spool.  I also added a cheap $5 PWM motor speed control board to allow fine-tuning of the motor speed.  This would give only simple on/off control to the motor (no PID), but I found in testing that this was sufficient.

For a filament guide, I used a short length of tubing mounted onto a hinge.  As the hinge moves, the filament comes out at a different place on the reel, keeping it from bunching up in one place.  The looser white filament underneath was from before I added an additional tensioner.

The hinge, in turn, is linked to a printed carriage that moves up and down along a worm gear (reall just a 4-40 threaded rod) driven by a slow gear motor.  Two momentary limit switches keep the carriage from moving too far in either direction. Keeping with the CPU-less design, they simply switch on and off a DPDT latching relay, wired in a way (one momentarily on, one momentarily off) so that the motor reverses direction when either switch is hit.

After my initial tests, I also added a small, spring loaded clamp to keep the filament under more tension.  Unlike passing the filament through a restricted path, the clamp has a small benefit in that I can engage it later after initially threading the filament.

Here is the completed winder case.  When in storage, the Filastruder fits neatly inside with extra hanging space for an empty or full filament spool.

When I'm ready use it, I simply set the extruder on top of the case and pass the new filament down through a hole in the top and loop it round to the winder inside.

I was surprised how well it worked right off the bat.  The rocker assembly tends to stay right on the edge between on and off, moving ever so slightly to periodically activate the take-up motor.  The movements are so small that the filament path moves very little, leading to the most consistent filament than I've ever made before.

I've already successfully used it to create two one-pound (half-size) spools (one black one white) and will put it through more paces in the coming weeks.