I’ve made a lot of experiments on the filament track switch in an attempt to reduce its friction. These are the results of everything I tested so far so you don’t have to 
. I dismantled the switch so many times I now can disassemble and reassemble it with my eyes closed
Some of the changes had some effect, some had none, some have just resulted in so slight improvements that it was not worth the pain.
I’ve started by changing the orientation of the switch to make it vertical so that it could adapt to the positions of my AMSs (on top, below and on the left side). This has been the best change I made but although the filament would flow without triggering any warning, I was still not fully happy as I could still feel some resistance.
Although I initially determined the main culprit seemed to be the switch lever part of the switch, I did not find an easy way to improve it so I proceeded instead to try to improve the 2-in-1 track part of the switch by making then longer and curved smoother. The result was worse than the original eventhough I printed them in PTFE filled filament from Igus.
I then thought, if longer tracks do not work, why not try to make them shorter and straight pushing the 2-in-1 farther from the switch. This gave a visible improvement but too minor to be worth the trouble.
I then attempted to reduce the pressure from the magnet/hall sensor. For this I tried different spacers between the magnet spring and PCB. I tried different thicknesses (0.6mm, 0.8mm, 1mm). 1mm was too thick and the hall sensor was not triggering, 0.8mm was not reliable, 0.6mm was perfect. I cannot say it improves the friction or not, some tests seem to make it better, some seem to make no difference. I also tried some softer springs with same result, again not worth the trouble.
The only thing remaining was to finally take an other look at the lever part. Checking it more closely, there is a restriction around the lever shaft where the width of the track is smaller there which seemed it could be improved. I then gently filed the exit corner of the lever track to increase the width and reduce the length of the lever contact. The improvement was significant and the friction was greatly reduced (I may be exaggerating but the improvement was substantial).
If you want to do that, do not overdo it to not risk the filament hitting the plastic corner behind it at insertion. I may have done that on one side of one lever and I could hear sometimes the filament blocking somewhat before getting through again.
This was really the only mod that made sense although I wouldn’t really recommend it as if you do it wrong the filament may get stuck there and render the switch inoperable and certainly no (or difficult) way to fix it. I bought a second switch to do all this so I didn’t care if it was bust in the end
I also experimented again with my switch mount as during my many experiments I was sometimes too lazy to mount and dismount the switch from its support and when it was free to move just kept in place by the PTFE tubes, the flow was better. I then just designed a small rotating adapter between the switch and its support to allow the switch to sort of self align rather than constraining the PTFE tubes to a forced orientation and further improve the flow.
This adapter can be installed on any mount, bambu’s or alternates.
In the end, my final configuration is (until I find more things to tweak):
1. Vertical support on the left and down from the printer inputs to keep a certain length and nice curves
2. Filing of the switch levers on both sides
3. Rotating adapter between switch and support
I’m now satisfied, the filament now flows freely (at least I’d like to think so) with only minimal more drag than without the switch.
Thanks for reading 
. I did not even troubleshoot it (cut the filament to leave untouched end and retest, I was too upset with that thing that I don’t actually really need 
)
