Stronger Compact Spool Auto-Rewinder

Original model was working wonderfully, but just on almost empty (light) spools. Reason was I used ASA as material, which is much softer than PLA, under the weight of the full spool the axle deformed a bit, and instead of resting on the bearing, it rested on the friction clutch - which this way lost its function. In this remix the main support bearings are on the outside of the spool interface, supporting the weight much better than before.

One additional “bearing holder” part was added, and another C-clip needs to be printed (3 per spool holder now). This is the most reliable solution I've come across, tried many buffer designs, but none of those worked reliably enough to leave any multimaterial print without babysitting and a lot of manual interventions. In my test print I made 170 tool changes without any intervention, or sign of any issues. 

I'm confident this is the most universal solution of buffering/rewinder mechanisms. Why I think it? I use ASA as my go to material and no matter how I designed/adapted also my buffer design, sometimes it would just not want work, other times it worked - while still on the same spool. This design is able to work also with flexibles if needed. 

I also added little handles to the other side of the spool holder part, so it is easier to tighten/loosen the nut. Printable without supports, but I used them either way retain clean resting surfaces for the bearings. 

Assembly instructions are basically the same as original design, however I found it is much easier to first thread the spring through the axle, then by rotating the axle pull in most of the spring. The last part is a little bit tricky, you need to hold with one hand the spring holder body, spring outside end, and axle against the tension, and with other hand introduce more tension into the spring, so it pulls away from the spring holder walls. Then the end of the spring can be inserted into the hole. And now you can carefully release the tension.

How to assemble the rest is self explanatory from the section. 

Some more tips: 

To adjust the tension, do not put in the last C clip on the threaded side of the tool holder. Try out how many rotations does it take to slip, take the insides out, adjust friction, insert back, and repeat until satisfied. In my case I need 1m of filament to be rewinded, on full spool it means 1.5 rotations, but on empty spool it is closer to 3. I set my friction to around 5 rotations - a little extra tension helps to fight the inertia of a full heavy spool.

Only insert the toothed end pieces on the axle, when the friction is set up to your requirements, they are difficult to get off. 

If the bearing holder fits too tightly, you can file down the little ribs I put on the outside diameter - be careful, do not introduce gap here, or the spool will sit on the clutch, not on the bearing. Here a nice snug fit is preferred.

If you do not have a brake in the MMU / ERCF / Tradrack, sometimes the rewinder can pull back more filament by itself. Since the spring is weak, it takes only a little friction in the reverse bowden path to solve this - just bend the PTFE tube once or twice. Even with this bend + constant spring pulling I have to pull the filament with less force, than with any of the buffers I tried previously.

Since the spring is weak, you will need to adjust the acceleration and speed of your MMU unit, so that the spool has time to spin up. You do not want to push more filament back, than your spool can take, take it slowly - reliability and repeatability is much more important, than speed.

Thank you Oleksii for this design, if I didn't find it, I probably give up on this ERCF experiment!

Moved middle bearing to the outside, greatly mitigating effect of heavy spool bending the axle, then increasing friction on the clutch. Added handles to the screw part of the spool holder for easier tightening/loosening.