Nearly Empty spool issue - possible workaround

Bambu Lab AMS
1

Nearly empty spool misfeeds is the only real issue I ever get with the AMS system after 100s of hours and 10s of thousands of successful colour changes. The AMS usually recovers itself, but occasionally needs a helping hand,

I have tried weighing the centre with desiccant or nuts and bolts but haven’t really found gravity reliable enough.

So have been looking at solutions that either pull or push down the reels in the AMS.

The best one I have come up with so far is adding a 3d printed screwed centre to the spool with a small bearing on each side.

From that bearing I created a very thin holder to rests just below the AMS lid and when closed causes the lid to hold the spool down onto the rollers.

The holder is kept in the right position by it resting slightly on the AMS divider too,

One particular spool tends to misbehave about a 1/3 of the time normally - but it doesn’t seem to fail anymore in my testing so far.

Will publish the stls once i have refined it and given it some more extensive testing.

But thought I would share a short demo video now in case anyone can suggest any improvements.

Current design sized purely for BL spools - but it should be hard to make it work with other makes.

Update 23/9 - edited video slightly better after negative comments from wife!
Demo Video

2 Likes
2

How weird I never have this problem with my two AMSs. I use polymake ABS with Bambu Lab refill spools. They never kick up like that.

Only at the last 100 grams does it not want to spool back after a finished print. But even that is rare.

I wonder what the difference is?

I do like your solution tho! :raised_hands:

2 Likes
3

Its fairly rare for me too - plus i haven’t really tracked or tested whether it is isolated to one of my AMSs or occurs on all of them.

Although I suspect it probably happens more often than i realise - but the AMS probably just handles it.

I will probably do some more testing on different spools and AMSs to try and find the most common causes, plus to see whether i ever get problems on unload too.

As far as I have been able to work out:
On loading - the rollers are not driven - so the loading pulling power only comes from the internal gears in the first stage feeder and/or the AMS internal hub. Having an almost empty spool causes the spool to spin faster as the wound filament has a smaller diameter - this together with the lower mass means the spools sometimes try to climb up at the front. My workaround hopefully physically hold the spools down stopping then from climbing and getting jammed by the lid.

For unloads - the front roller in the AMS is powered and does provide part of the pulling power.
Less filament on the roll means a smaller internal diameter - so a slower more powerful pulling force. So a more powerful pull, combined with lower mass could well cause slipping on the rollers I guess but don’t think it would cause any jumping, - I think my workaround might also address unload issues by pressing the spools down a bit harder on the front roller.

I might investigate the logs on the printer or the HMS next to see if any of the issues i can recreate are reported in a readable form on there.

1 Like
4

Had the problem again for the first time for a while on a spool without my proposed solution in it - caused a print to stop eventually - and have to be restarted after a 3hr delay.

So decided to print my spool hold downs for the 12 AMS slots that I haven’t done already - which is taking a while.

Ironically while printing one of ‘bearing’ the spool inners one of my BL spools separated and caused a jam. Whilst I couldn’t get the spool halves to slot back together again properly I did manage to use an already printed version of the threaded inners that I am trying to print to hold the two halves of the separated spool together long enough to complete the print.

Will probably have to unspool and respool the whole roll to get the two halves together properly - but did think it was quite funny to use to spool inners I was trying to print to work around the problem.

2 Likes
5

My AMS is doing the same with nearly empty spools. I have to babysit color changes and start/end of single color prints. I saw a suggestion where someone added “D” batteries inside the spool for added weight but I haven’t tried that yet. Still researching the problem when I found your topic - thanks!

6

I use spool desiccant inserts in my almost empty spools, with a D size Battery inside it, works for me.

7

For me I haven’t found gravity enough to stop near empty spool jumping problems.
As a belt and braces approach decided to re-model my spool middle to have enough room inside it for a D Cell battery too - so that, together with a physical “push-down” of the spools should hopefully be enough to cure my problems - (unless the issue is caused by the rollers locking when they shouldn’t).

Currently printing a few test versions and will then test them out on a few multi colour prints before releasing on Makerworld.

1 Like
8

Did some further investigation of the issue today - and I think I probably understand what is happening now.

In the first stage feeder the are two separate driven parts - the roller for unloading, and the extruder type gear near the yoke for loading.

There is a clever set of gears within the first stage feeder - which swing around depending on what direct the motor is turning.

What I think is happening is that sometimes when loading a channel when the yoke gear engages the gear for the roller doesn’t disengage - so powers the roller for a few seconds, as well as the yoke gear powering the extruder type gear and pulling the filament off the spool.

Because the speed of the roller isn’t fast enough to keep up with the spool being spun by the other gear pulling on the filament the roll starts to climb ahead of the roller.

I have filmed the issue occurring and a normal load - and slowed it down to 10% to demonstrate the fact that the roller starts immediately the yellow gear starts turning when the problem occurs. But on a normal load the gear correctly swings out of the way.

I don’t really understand why weighing the spool down is helping - I can only think that this makes the spool more likely to slip than climb up the roller.

Fixes like Papsjundi’s may also be helping lift the spool off the roller in a more controlled way.

But ultimately I think the solution is to either better control the spool when it does climb to stop it getting stuck on the lid, or maybe putting some sort of one way bearings into the roller or edge of the spool.

It is possible that the firmware could be changed to joggle the gears a bit to try and release them, or ultimately I suppose a hardware fix to stop the gears from getting stuck.

To doubly determine that my understanding is correct I think I will try creating a reusable spool with a very large inner section and then rewind my nearly empty spool on it and run a few more tests.

Link to video here - sorry if it is not completely clear - but I am just learning how to do these sort of YouTube videos.

Update - made an extended size inner to change the gearing - it weights 71g - so quite a lot less than the nearly 200g that a D cell battery screw in inner weights.
No failures so far - will try about 20 times.

52B2D4D6-36F2-4A07-8692-50553D45E846
52B2D4D6-36F2-4A07-8692-50553D45E8461219×734 82.2 KB

11034053-A5FC-4534-A517-017D9DFF04C6
11034053-A5FC-4534-A517-017D9DFF04C61031×886 90.4 KB

Other ideas are trying the make one of these - https://www.youtube.com/watch?v=UIhCPl8eb7s. - then creating some sort of thin outer casing to full escape the rims of the spool - and then have either one of Angus’ type one way bearings or a meta sprag bearing in the middle - to make the spool rim rolling one way. This will probably be better as it won’t require re-spooling. Might also help with things like cardboard spools.

Update unloaded and reloaded over 20 times - no problems so far - might try creating a lighter version of the inner.

Plus I am favouring trying to create a test 3d printed roller clutch - that sits on the edge of the reel - probably with the actual mechanism setting inside the reel. Or just sitting on top of each front roller between the roller and the reel - or maybe something that replaces the whole roller - and includes some needle bearings inside the actual rollers.

1 Like
9

I think you may be onto something with the roller not disengaging. When putting a new roll in the AMS, the rollers won’t spin in one direction until I spin it just a fraction backwards. I forget which direction is which but it implies that the gear isn’t backing away as you mention and is binding the roller. A loose weight like a battery in the spool core may jostle the spool enough to release the gear. Actual hold downs may simply force things to slip.

1 Like
10

Agreed - when the last thing it has done is unload then often the roller gear will be left engaged. Due to the worm gear on the motor shaft this will lock the roller.

What normally happens is then when the motor runs in the load direction the roller gear immediately disengages which allows the roller to run whatever speed it likes.

I guess that even for fully loaded spools it must always eventually disengage otherwise they would have a problem too when the first stage motor stops.

I guess it is possible that heavier spools also try to go faster than the bound up roller - and because they maybe slip less they drive the roller forward more and put more pressure on the gears to disengage.

Will try to do some filming of heavy spool loads too - to try to see how long before they disengage

1 Like
12

Update - did some more filming today - and have created a frame by frame comparison between and ok changeover and a problem changeover - where the roller gear doesn’t engage - causing the spool to climb.

Also tried to film the cog not engaging on a fuller spool to see what happens - but after about 20 tries it always disengaged - so maybe the weight of the spool is helping it to disengage in some way - so maybe weight is the answer in a lot of cases.

Think I will leave this for now.

Yesterdays was filmed on an iPhone - today’s was filmed on a Wyze Cam 3 - which I did the modification today to allow closer focus - so will be handy as a backup camera on the X1C - plus will see if I can fit one inside an AMS and leave it running during a long multi colour print in case it picks up anything strange.

2 Likes
13

I have this problem with one feeder, where it is obvious that the gear mechanism is not dropping down to disengage from the driven feed roller. In short the gear mechanism is binding on this one feeder, where the other three don’t seem to have this problem (yet).

My take is this is a Bambu design or manufacturing issue. I ordered a replacement feeder where my intent is to disassemble the problematic feeder to see what I can be done to cause the mechanism to properly disengage while keeping the AMS running while I play.

1 Like
14

I agree that it is a design issue with the gearing in the first stage feeder. Ultimately I think it needs either a gearing redesign, a one way bearing - or maybe magnetic solution like this one for example GitHub - pure100kim/WinchRack_Type2: WinchRack_Type2 for ERCF buffer

In the meantime my own problems are pretty much solved with 500g weights in the spool centres once they get a bit low - I use this little adaptor to hold them in. 500g AMS Spool weight holder - fix for nearly empty Bambu Lab Spools by Ukdavewood - MakerWorld

1 Like
15

The issue at the end of spool is more due to the angle the AMS is pulling on the filament than the weight of the spool (increasing the weight can wear the AMS faster). There are several models on makerworld (look for AMS saver, AMS guide or AMS helper) to solve this.

I use a mix of these 2 to make a set as light as possible (although I modified the button so the PTFE can screw in it rather than just slide in) and I never had a problem with any spool even at end of it and even with spools that have a smaller inner diameter.

AMS Guide
AMS Guide1920×1440 219 KB

2 Likes
16

Implemented you solution though with a different feeder coupling. It appears that you have hit on the answer from my limited testing to date!