How I got my MMU2S to work reliably
I have to admit that the MMU2S hasn't been easy to use for several reasons but rather than advocating for improvements I want to tell you how I got mine to work reliably. My first attempt was to configure a stock setup using the materials provided and the instructions in the manual, when this didn't produce satisfactory results I started experimenting. In retrospect many of the problems may have been due to bad or marginally functional electronics but that is a condition that was hard to pin down since I had only one unit and no experience with a properly working MMU.
The first thing I did was dispense with using the buffer. Since there seemed to be a problem feeding the filament through the MMU it didn't make much sense to add the drag that the buffer contributed to the problem. I subsequently found that some of the PTFE tubes had smaller inner diameters than others and though Prusament fit well other brands of filament didn't, so I stopped using the smaller tubes, though this didn't help much. To replace the buffer and filament dispensers I made some filament hangers that held the spools through the center rather than by the edges. These have roller bearing axels to reduce friction but I later found the bearings weren't necessary. To account for the slack produced when the filaments are exchanged I allowed the filament to fall in the space between the spools and MMU. This works well and even though the filaments tangle they don't get stuck.
Next I reprinted parts of my extruder using polycarbonate to take the flex out of the Bondtech gear assembly to help cure a problem where the filament was getting gouged and would no longer feed. This didn't have much effect but it made me feel better.
Then I noticed that the bearings in the MMU used to pinch the filament against the Bondtech rollers had quite a bit of axial play causing them to wander and fail to contact the filament with a perpendicular surface. This situation was exacerbated by a loose fit of the bearings in the slots of the drum allowing them to wander even more axially. I fixed this by putting washers on the sides of the bearings to remove the play. I used paper washers that were designed to provide electrical insulation but anything suitably thin would work. This turned out to produce a significant improvement in the ability to feed the filaments through the MMU.
Next I replace both the MMU and Einsy controllers. I did this because the MMU seemed to act erratically in that it didn't always calibrate when powered on and in general seemed to act differently from time to time. In particular I was unable to make any sense out of what the three buttons on the MMU do.
Finally I realized that I had failed to recalibrate to account for changes I had made to the length of the filament path so I ran the setup procedure as a remedy.
Honestly I can't tell you which of these finally fixed the problem but my suspicion is that replacing the electronics and improving the axial runout of the bearings were the most important steps. After these changes I was able to print a five color model that required changing all five filaments for each layer. It required 22 hours to print and completed without any problems.