FINDA Filament Selector Redesign to eliminate Sensor inconsistencies  

FINDA Filament Selector Redesign to eliminate Sensor inconsistencies

After receiving my MMU2 I have had overall some of the same struggles that everyone else has had and have used many of the suggestions posted here to get closer to success. I wanted to publish and post a change that I have made that has ultimately made the MMU2 finally usable for me. Hopefully this will help others and if nothing else give some ideas on how we need to improve the way the MMU2 senses the presence of the filament.

Below is really a duplicate of what I have posted to Thingiverse. The link is also below. Give it a try and let me know what you find.

https://www.thingiverse.com/thing:3264138

This is a version of the original Prusa MMU2 Filament Selector that includes some modifications on the Filament feed, Filament cutter blade holder, and most important addition of a fix for the FINDA sensor which includes a fix for the ball getting stuck in the chamber holding the ball between the sensor and the filament.

History -
There has been a mixed back of success with the MMU2 and here are some of my observations and discoveries.
1. One of the issues people have been having is layer skips. There are a number of reasons for this but the primary reason for this is that the filament never reliably or consistently feeds the filament to the bondtech gear in the extruder head. In the stock firmware the success of this path is completely depended on accuracy of calibrating the distance from the FINDA sensor to the bondtech gears in the extruder. There have been a number of attempts to find a way to make this better and less susceptible to error or hardware differences. ZeroBeast rewrote the firmware to include the use of the laser sensor in the extruder so the calibration of the steps in distance did not have to be done. I tried this firmware and there are a number of very good ideas implemented in this. Overall I had good success and it was quick, until the never ending plague of the sensor not seeing black PETG hit again. The filament fed to the bondtech and never saw the black PETG and kept feeding until it completely destroyed the sensor in the head. I moved back to the stock firmware and then tried to understand what was needed to calibrate the steps from the MMU2 to the extruder more accurately. After reading a lot of others experiences I found that there is a very fine line that has to be found to make this accurate. I noticed that some users have replaced the orange bowden tube with white tube hoping to make a difference. I don't think this helps and I think it makes it worse. The reason Prusa chose to use the orange tube was that it had tighter tolerances in the ID of this tube. This made the steps that were fed to the extruder more consistent. I ended up finally swapping out the orange tube with Capricorn Blue bowden tube which has the tightest tolerances of tube on the market today. Of course by doing this I had to make sure that the filament tip was as clean as possible and it took a while to figure this out with ramming, temp, and speed settings. The biggest difference came from the change Prusa made in the MK3 firmware when they disengaged the stepper motor in the extruder during a filament change. This allowed the filament to be a little longer and ultimately fixed the idler in the mmu2 from skipping steps and crashing over time if the filament was calibrated a little long. Having all this the length of the filament in calibration has to be close but not perfect.

2. I mentioned the tip shape calibration and this will be a never ending story of a mixed bag of how to get this reliable and consistent. If you don't get a clean tip without strings then this daisy chains into a whole bunch of other issues that come from getting the filament stuck in the bowden tube to screwing up the finda sensor so that the MMU2 never knows that the filament was loaded or unloaded correctly and never switches until intervention and this is cleared. This has added hours to a print for me and in the end kind of makes the MMU2 worthless if you have to keep encouraging the print to continue.

3. Which brings us to the reason I made all the changes in the filament selector in the MMU2. Even after tip, and length calibration there was always the case where little bits of filament particles build up in the extruder and in the MMU2. Overtime throughout a print eventually these will interfere with the FINDA sensor and cause a print to stop. Usually when this happened I would be able to push and pull the filament to clear the particle and the ball in the Selector would start to work again. Sometimes so much junk was around the ball in the selector that I had to pull the PINDA probe out and blow it out. Prusa designed this Pinda column at about 8mm with a 7mm steel ball which leaves a little bit of room for the ball to move. It does leave a little room for junk to build and then there is a little bit of extra room around where the filament leaves the selector to hopefully clear the junk. The biggest problem is that the ball needs some encouragement to stay down. The ball is so light that even the slightest of change in the tip will cause it to go high and stay high when it needs to drop. I have noticed that people have tried to raise the PINDA up and what this does is allow more room on top of the ball but the reality is that there is only still about 2mm of difference that has to be seen by the PINDA in order for this to work and still sense the filament either being there or not. I tried to add some foam on the bottom of the PINDA probe to create some extra pressure on the ball. This actually helped and made it through a 1000 tool changes without an intervention until the foam eventually slid over the ball and caused the ball to stay up again. I then thought of the change that is included in this upload. I needed a strong magnet to make the ball heavier and assist the ball to stay down and in the off position. You will notice I had to make a few changes to get the 6mm round slot under the column that holds the ball. I originally designed this to take a 8mm x 3mm magnet which worked great but I changed this to hold a 6mm x 2mm to allow more material strength in the selector around the POM nut. Below is a list of what is needed to make this work. Just FYI after making this change I have not had one case where I have had to intervene for a misfeed.

Design changes from Prusa V1 -
1. Chamfered inlet to the selector from the MMU2 idler. (This opens the tolerance needed to succeed a feed to the FINDA.
2. Changed the tolerance to add pressure in the filament cutter to hold the blade better. (I did this before I discovered that I wanted to remove the blade all together).
3. Added a 6mm round chamber to hold a magnet at the bottom of the FINDA ball chamber.

Note: As I mentioned I also removed the cutter blade all together. It is no longer used in the stock firmware anyway and I found that it sometimes just got caught on the filament and caused problems. My recommendation is to remove the blade.

Required materials
1 or 2 6mm x 2mm magnets (get the strongest you can find)
These worked for me.
https://www.amazon.com/gp/product/B01NAL9I5H/ref=oh_aui_detailpage_o08_s00?ie=UTF8&psc=1

You will need a longer M3 Stainless steel bolt to hold the POM nut. I think the original is 10mm. You will need at least a 15mm-16mm. Obviously not too long.