The MMU2 Purge Bucket Thread (Mod/WIP)  

Page 1 / 3
  RSS
vintagepc
(@vintagepc)
Noble Member

 

Since this has picked up steam and somewhat derailed at least two separate threads, I'm starting a new one for discussion/progress on the development of the MMU2 purge bucket mod that @gnat and I have been working on.

It started as an itch here:

https://forum.prusaprinters.org/forum/original-prusa-i3-mmu2s-mmu2-general-discussion-announcements-and-releases/0-05mm-layer-height-is-possible-and-im-sharing/#post-150847

and gained some interest around page 7 here:

https://forum.prusaprinters.org/forum/original-prusa-i3-mmu2s-mmu2-general-discussion-announcements-and-releases/i-challenge-to-prusa-to-use-mmu2s/paged/7/

Note it is still very much a WIP. We've gotten what we think is  a proof-of-concept design that has successfully yielded a 2 color benchy. But there are still many improvements to be made, and we're working on those at the moment. Note we are currently still in a fairly early state and so aren't actively looking for widespread testing yet. You're still welcome to play with things but just be aware we may not be as responsive to reports of issues until things like the design settle down and we've dealt with all of the issues we find in our own testing.

The official repository for the mod in its current state with test results is here:

https://gitlab.com/gnat.org/originalprusa/tree/master/devices/mmu2

 

Posted : 02/08/2019 2:03 pm
cyril.m5, troy.p3, gnat and 2 people liked
gnat
 gnat
(@gnat)
Prominent Member
gnat
 gnat
(@gnat)
Prominent Member
Posted by: CybrSage

My big concern is if the filament does not fall off the extruder, but stays connected instead.  A long strang would be drug over the print bed, where the heat would make it stick.

Can this discussion get it's own thread? It is too big a thing to be lost in a rant thread.

Yes that is a concern and we had to do some tweaking to both the code and the bucket (and I think we have some more to do, though I haven't looked at v3 of the bucket yet).

The current process is:

  1. Move rapidly to X 250 (and Z 20 if the current layer is less than 20mm)
  2. Move slowly to X 253 to deploy the bucket
  3. Perform the tip forming and tool change
  4. Disable the part fan if it was running.
  5. Advance the extruder to prime the nozzle
  6. Extrude the configured (PrusaSlicer settings) purge volume
  7. Retract for movement (same as any "z hop")
  8. Turn the fan on for 2 seconds to harden the strand
  9. Slowly retreat to X 250 to release the bucket.
  10. Hop to Z +50mm if we are less than 160mm
  11. Disable the fan again if it was not on prior to the process starting.
  12. Resume normal printing.

Step 8 is where the strand is disconnected. In early testing the return to the part was so quick that it did exactly what you are worried about it. It does need some more work though as it can still leave a bit of a hair trail sometimes. There is also (at least in v2) a significant gap between the bucket and nozzle. Reducing that will help with this issue and allow us to speed things up a bit. Additionally I think I need to be more aggressive on step 8 so that it moves a bit farther away (say X 245) and do it faster. Currently the bucket movement is slow and I think if it was more of a snap back that would also help.

Step 10 is to help the strand clear the bucket to make way for the next purge. It needs work still.

In the above shown Benchy that was 193 tool changes over 9 hours without any issues of strands being drug along with the extruder.

The bigger issue we need to address is waste management as it was clear that I had to babysit it to remove the waste periodically so that it wouldn't foul things up. @vintagepc did add one of our ideas to v3, but I haven't tested it yet.

This post was modified 1 year ago by gnat
MMU tips and troubleshooting...
Posted : 02/08/2019 3:45 pm
CybrSage
(@cybrsage)
Honorable Member

The video appears as a broken link to me.

Posted : 02/08/2019 3:53 pm
gnat
 gnat
(@gnat)
Prominent Member

BigBrain3D's bucket design is definitely doing better than us in regards to waste management (relevant part starts at 1:41 if the forum doesn't respect the time specific link):

Their Kickstarter is done and they are supposed to be releasing that bucket design on Thingiverse, but we are still waiting. Someone did ask on the KS campaign about the time/volume impact of their bucket design, but they haven't offered up that info yet as far as I've seen.

MMU tips and troubleshooting...
Posted : 02/08/2019 3:53 pm
gnat
 gnat
(@gnat)
Prominent Member
Posted by: CybrSage

The video appears as a broken link to me.

Yeah and because of this wonderful forum software I can't fix it now 🙄 

It is shown on the GitLab link I put in that post, but the direct URL is https://gitlab.com/gnat.org/originalprusa/raw/master/devices/mmu2/bucket/tool/3DBenchy/3DBenchy-bucket.mp4

MMU tips and troubleshooting...
Posted : 02/08/2019 3:55 pm
CybrSage liked
Antimix
(@antimix)
Estimable Member

Interesting concept.

What I see is that they "sell" the software, AND the hardware device that take the waste.
So I assume the whole idea is not open source, and may be patented.  Have they registered and claimed something like "any device that is capable to collect 3D printer material waste" ? No idea.

However, in the video I see that there are still pieces of plastic dropped on the plate now and then, and that should never happen.
This means that the device shown is still an alpha prototype.

And what about changing the filament from MMU2 ? This is the real cause of all MMU2 issue we have! Removing the waste tower from the plate is not the first issue to solve in terms of reliability.

This means that even if this device would be at final stage, without any plastic dropped on the plate, and we install it in the MK3 with MMU, we would not be in condition to print reliably because of the MMU2 filament jam/failures at colour change. 😐 

Once we will have solved all the MMU2 issue, that would be the next goal. 😉 

Posted : 10/08/2019 8:36 pm
gnat
 gnat
(@gnat)
Prominent Member
Posted by: Antimix

Interesting concept.

What I see is that they "sell" the software, AND the hardware device that take the waste.
So I assume the whole idea is not open source, and may be patented.  

The KS is for the app, somewhere in there was a mention that the bucket is going to make it to Thingiverse at some point.

However, in the video I see that there are still pieces of plastic dropped on the plate now and then, and that should never happen.
This means that the device shown is still an alpha prototype.

I ran across a different video where the pellets were being collected more effectively. I agree that the waste must be kept out of the print area, otherwise you court collisions and failures.

And what about changing the filament from MMU2 ? This is the real cause of all MMU2 issue we have! Removing the waste tower from the plate is not the first issue to solve in terms of reliability.

Yes and no. The idea of removing the tower is more about saving filament and time, not reliability. It also recovers print space.

It does, however, remove the issue of the tower coming loose. So there is a bit of reliability to get out of it as well.

This means that even if this device would be at final stage, without any plastic dropped on the plate, and we install it in the MK3 with MMU, we would not be in condition to print reliably because of the MMU2 filament jam/failures at colour change. 😐 

Once we will have solved all the MMU2 issue, that would be the next goal. 😉 

We all have different skills we can bring to bear on the ecosystem. I've been watching what you are doing and am interested even if it is not an issue that has caused me much trouble. I've also been in and out of the PrusaSlicer code and dread the day I try to bring this idea into that 😥 

MMU tips and troubleshooting...
Posted : 10/08/2019 8:56 pm
Antimix
(@antimix)
Estimable Member

I have just developed my improved version of PrusaSlicer, so it should not be a problem support a kind if waste device, and drop the material there instead of creating the tower. 😉 

Posted : 10/08/2019 9:46 pm
Antimix
(@antimix)
Estimable Member

Hi here there is another interesting concept. He developed a new MMU, and regarding the waste material, he drilled an hole in the table, and drop everything there, and then use a brush to clean the nozzle 😀  I can imagine he also use a modified slicer.

On thingiverse there is the whole project.

Posted : 11/08/2019 12:04 pm
vintagepc
(@vintagepc)
Noble Member

Yeah, we've actually contemplated a hole in the table with a box/container under it for managing strands.  there's no reason you couldn't do something like that under our bucket either. 

Posted : 11/08/2019 12:32 pm
vintagepc
(@vintagepc)
Noble Member

Apologies for the lack of movement on this; Gnat's printer is down and I am in the process of getting my own MMU working. I had to reroute some wiring which I kept putting off so my X carriage could go a little further right.

If all goes well I should be able to get continue working on this either later in the week or on the coming weekend. 🙂

Posted : 13/08/2019 1:06 pm
vintagepc
(@vintagepc)
Noble Member

Quick update - I had some time to play tonight and ran off some prints while tinkering with the postprocessing script.

Highlights:

1. I finished what was estimated as a 1h10m print in about 44 minutes. Not entirely valid comparison as the PS estimate was based on different ramming times and cooling moves, but there's definitely time saving to be had.

2. Bucket+Chute works very well over the edge of a desk (or hole) with a catch box below. It's very satisfying to watch it casually flick/drop a purge strand into the box. 

3. I question whether some sort of wiper is even necessary. Right now we run the fan to cool the purged strand and harden it before the bucket snaps away; this seems to sever quite cleanly even if the bucket is a few mm below the nozzle. My print had one small goober across about 40 tool changes. I'll likely still revise the bucket design to include some way to add one, though.

- There are naturally still some issues and TODOs with the script surrounding tool changes, custom gcode alterations, and retractions. We're working on those. 

Stay tuned! 🙂

Posted : 14/08/2019 3:16 am
CybrSage liked
gnat
 gnat
(@gnat)
Prominent Member
Posted by: vintagepc

1. I finished what was estimated as a 1h10m print in about 44 minutes. Not entirely valid comparison as the PS estimate was based on different ramming times and cooling moves, but there's definitely time saving to be had.

Awesome! Congrats on finally getting to see your idea in action 😎 

2. Bucket+Chute works very well over the edge of a desk (or hole) with a catch box below. It's very satisfying to watch it casually flick/drop a purge strand into the box. 

Unfortunately not an option for all of us 🙁 I haven't gotten a chance to print or try v3 yet though.

3. I question whether some sort of wiper is even necessary. Right now we run the fan to cool the purged strand and harden it before the bucket snaps away; this seems to sever quite cleanly even if the bucket is a few mm below the nozzle. My print had one small goober across about 40 tool changes. I'll likely still revise the bucket design to include some way to add one, though.

You have a sock right? I did not for my testing so far and had issues with threads wanting to stick. Could be the sock, could be filament being used, other differences? Ideally I'd like to get rid of the fan blast as 2 seconds per tool change adds up (e.g. 2 minutes for a 60 tool change print).

- There are naturally still some issues and TODOs with the script surrounding tool changes, custom gcode alterations, and retractions. We're working on those. 

You are a master of understatement 🤣 

MMU tips and troubleshooting...
Posted : 14/08/2019 3:48 am
vintagepc
(@vintagepc)
Noble Member

Yes, I have a sock. good point on that.

I'm experimenting with altering the fan to be active during the purge itself so the filament cools as desired but without adding as much, if any, time to the end. Reasoning being that once the tip is in the bucket, cooling things will help keep it rigid and moving rather than globbing/spiraling up.

You are a master of understatement 

I try... 😉

Posted : 14/08/2019 1:09 pm
gnat
 gnat
(@gnat)
Prominent Member

Yeah I've gone back and forth on the fan. I moved to the off model as I would sometimes (lower purge speeds?) have issues with it blowing the purge out side of or onto the bucket. At best this kept the bucket from removing it when it retracted and it would then be drug back on the plate. At worst it attached to the block and made a mess.

I expect turning it off to start the purge (say the prime) and then enabling it for the remainder might be a good approach.

MMU tips and troubleshooting...
Posted : 14/08/2019 3:00 pm
Neophyl
(@neophyl)
Noble Member

Mainly posting as I want to subscribe to this topic.

Loving this modification guys, keep up the great work.  It might actually make getting a MMU worth it 🙂

The brush method pictured is similar to what my old DaVinci 1.0 used, it had a small spring plate that the nozzle used to move back and forth over a few times to clear any plastic from the nozzle which dropped into a collection bucket below it.  Of course that only did that at the start of a print during the nozzle priming stage but the concept for cleaning filament off the nozzle is the same.

Posted : 14/08/2019 3:14 pm
vintagepc liked
vintagepc
(@vintagepc)
Noble Member

I have some small brushes left over from a skelestruder build; it'd not be hard to stick one to the back of the bucket and see what happens... I'll have to give that a go at some point. 

Posted : 14/08/2019 6:20 pm
vintagepc
(@vintagepc)
Noble Member

Ladies and gents, here is some purge bucket porn. Probably best not to watch it with your Prusa in the room. . 😆 This print ran completely without user intervention (~40 tool change test print, 3 colours). 

A few observations during my tinkering:

I've modified the script to purge at the maximum possible rate. As a result, the colour change region is actually quite narrow, and usually in the first third or so of the purge strand (with the defaults of all 70s). This has two implications:

1. My theory that we'd be able to cut purge volumes because of the high extrusion rate seems to be correct, less mixing going on. 

2. It will be *much* easier to optimize your purge volume settings with this method. You can simply measure the length of the strand, calculate the percentage at which it has stopped changing colour, and modify your purge volume for that combo by the same percentage.  For example, if my transition from black to grey is completely done in the first third of the strand, I know I can drop my purge volume to 33% of what it was.

This post was modified 1 year ago by vintagepc
Posted : 15/08/2019 2:39 am
gnat and david.a66 liked
gnat
 gnat
(@gnat)
Prominent Member

Sure. Rub salt into my broken printer wounds 😛 

😉

I like the v3 bucket. I think even without the long drop you have that will do a good job keeping things clear.

I'd like to try your updates on the purge speed. I know there was room to improve over what I had, but I kept running into issues where the extruder would grind and skip even at lower speeds. I wonder if that was just tuning for my idler, an extrusion issue, or some difference in how the Skele extruder drives the filament.

I forgot if you told me already, is that the factory extruder motor or an upgrade? If the later, 0.9 or 1.8?

MMU tips and troubleshooting...
Posted : 15/08/2019 3:04 am
Page 1 / 3
Share:

Please Login or Register