Problems after Bowden length adjustment

[narrow lily]

Quick question because I’m not finding much in the search function about this. Last night I had an issue where a few tool changes in on a 3 color print the print paused. lanes 2 and 3 were registering as unloaded in the gui. I thought about it and realized that a few days ago I had replaced my CAN wire to the toolhead and had actually shortened the Bowden and had subsequently forgotten to recalibrate the Bowden length. I think the lane’s unloaded themselves right after the first time they rewound from the toolhead. I recalibrated Bowden length (it came in at 1630mm) and reran the print. It failed in exactly the same way. I then recalibrated multiple times from all 4 lanes and averaged each lane and got anywhere from 1630 to 1605, i then switched to 1605 for my Bowden length and now it’s failing to load all the way to the toolhead. Anyone seen something like this before?
<@&1304550334839918672>

[north dock]

When you are running the calibration, are you holding the buffer closed, or do you have a rubber band around the buffer?

[narrow lily]

I’m holding the buffer closed and releasing it right as the filament reaches the toolhead filament sensor.

I also ran the calibration without holding the buffer at all to compare the difference in output and it varies by about 10-20mm, lane 4 is exactly the same no matter how I run the test oddly enough.

[upbeat heart]

are you getting good cuts? if it unloads twice then it might unload too far.

[narrow lily]

Solid cuts every time

I switched to an A4T with a crossbow cutter about a month ago and after the initial tuning it’s run like a champ. It wasn’t until I had issues with my original CAN cable and replaced it that I got these problems.

Also, I ran a 3 color purge calibration before running this print and it changed filaments multiple times for each lane and had no problems. It also had no problems when I just clicked the load button in the gui for each lane one after the other a few times. It only does it while printing.

[upbeat heart]

if its only during print then likely there is a setting wrong in slicer
Otherwise I was going to suggest that its most likely an issue where we do the unload length twice

you can also set a shorter unload length to try to combat this

[narrow lily]

That’s what I did, I also adjusted the Bowden length in Turtle1.cfg. Now it fails to load all the way. Should I only adjust the unload length?

I doubt it’s a slicer setting since I’m using the same slicer settings I’ve always used.

[narrow lily]

Also, I don’t think it’s trying to unload twice. It’s unloading just far enough for the LEDs to turn red but not enough the fully clear the extruder gears in the BoxTurtle. Like 10mm shy of a full “eject lane”

[upbeat heart]

I meant at the toolhead unload distance

[narrow lily]

Can you clarify what you mean by this?

[upbeat heart]

If unload-stn is too short and the sensor is still made we do the whole length again. So often that’s an extra 60 or so for a lot of people

[narrow lily]

Hmmm, that’s interesting. My previous testing indicated that unload_stn was setup correctly, and the only thing that has changed was the overall Bowden length. Why I tend to think it’s Bowden calibration related is that there is such a wide range of results for the exact same test run multiple times. That and the fact that the unload sequence visually runs exactly as it should on lanes 1 and 4 and only fails on lanes 2 and 3. A problem with unload_stn I would expect to see issues on all 4 lanes.

[upbeat heart]

Except lane 1 and 4 have longer dist_hub

[narrow lily]

Isn’t that accounted for with the lane calibration?

[upbeat heart]

It really only matters during lane load from a calibration standpoint

[narrow lily]

I wish I understood what was going on under the hood more with this thing. As annoying as the pico was by the time I was done tuning it I knew exactly what every variable did and the order of operations for each movement.

Okay, so I’m running A4T with dual sensor WWBMG, crossbow cutter, and Rapido 2 UHF. The recommended tool_stn is 40 and I’m using 35. Recommended tool_stn_unload is 65 and I’m running 65. I’m currently re-running the bowden length auto calibration before I mess with anything else.

[upbeat heart]

So do this, load, extruder 30 run afc_cut then retract until you see your upper sensor come unmade note how far that is

[narrow lily]

Extruder 30? You mean load 30mm?

10-4, gonna ignore the 30 part, load all the way, cut, then retract a mm at a time until the switch indicates it’s cleared the sensor.

Also 2 auto calibrations in a row is indicating 1605 is the correct Bowden distance for lane 2 and 3 which is what I was using last night so I’m leaving that alone.

[upbeat heart]

No I meant extrude 30 after the load to ensure you’re filament is melted back into hotend

[narrow lily]

Gotcha, thanks

Does it not engage the lane extruder motor when I’m using the extrude/retract GUI in Mainsail? I’m currently at 50mm of retraction and I still haven’t cleared the tool start switch but if I release the ECAS connector and pull the Bowden tube out the filament is sitting right on the switch, has cleared the toolhead extruder gears, and I’ve retracted an extra 10mm and it’s not moving.

I think moving in 1mm increments might be the problem? I had to do 10mm increments to see any movement from the lane extruders. Total of 70mm to clear the sensor, I think it had to collapse the turtleneck before I saw any movement.

[narrow lily]

I’ve run this test 3 times now and I’m getting 78-80mm to get a full unload, that’s bizarre.

[upbeat heart]

Well change to that and see how things go

[narrow lily]

Just to clarify, it clears the tool_end sensor in 10mm, then it clears the extruder gears within the next 20mm, then the rest is spent collapsing the turtleneck before any movement is seen in the filament.

[upbeat heart]

🤷🏼‍♂️ you can also start running a rubber band on the TN if you want

[narrow lily]

What does that do? It would cause it to collapse faster and take longer to advance I’m guessing?

[upbeat heart]

It makes Bowden length more consistent

[narrow lily]

And no cons, just pros?

[upbeat heart]

You really don’t need to reply to me every time,
But yes IMO

[narrow lily]

Sorry, I’m not great at Discord, does it ping you or something? I’ll give it a shot when I can get a rubber band. Thanks for the help.

[upbeat heart]

Yes I get a notification every time

[narrow lily]

Sorry about that, I’ll avoid that in the future.

[jagged stag]

But...

[narrow lily]

12 tool changes in, 339 to go. I think this worked, thanks again mc-red.

[narrow lily]

Print is nearly done but has failed to load 5 times now in a little over 300 tool changes. Filament is always loaded, just a mm or 2 shy of triggering the tool_start sensor. Anyone know a way to tune around this? Would a rubber band on the TN help with this?

[upbeat heart]

Yes

[narrow lily]

When I look for tips on the rubber band thing I only see it mentioned in conjunction with the buffer ramming method. It’s still helpful even without using the buffer ramming method? Sorry for all the questions I’m just not finding a lot of info on this topic.

[eager idol]

It’s not necessary at all when not doing buffer ramming

[narrow lily]

But is it helpful? MC-red seems to think so and he’s been right so far. I’m unsure as to why it would be helpful, other than requiring more force to expand the buffer and less force to compress it. What would you recommend to alleviate my new failure to feed problem?

[upbeat heart]

it makes the bowden length more consistent

[crude pecan]

When it fails to load, can you describe what the turtle is doing?

[narrow lily]

I haven’t physically watched it fail, as it doesn’t do it consistently, it’s extremely sporadic. From what I can tell, the filament cuts and retracts properly and then proceeds to the loading stage. the new filament travels all the way to the toolhead but not quite far enough to trigger the tool_start sensor, the buffer also doesn’t appear to be fully advanced when this happens. To fix it I pull the Bowden tube and tug the filament forward a few millimeters until it hits the advance switch in the TN and the reinsert the Bowden tube into the ECAS fitting and manually extrude before resuming the print.

[crude pecan]

295+ successful swaps probably means your configuration is pretty good. I was hoping for more information on what happens after the initial long move doesn't get to the toolhead. If the initial long load comes up shot, it usually it tries a series of short moves before giving up. Does this happen with all lanes, or just 1? It might be pointing toward tight bends and friction in your toolpath, the tension on one or more of the BT extruders not being tight enough, or it could be too tight and griding divots into the filament and getting stuck on said divots.

[upbeat heart]

Otherwise wait for DEV push to main it should fix this

[narrow lily]

Again, I haven’t seen it fail, but the ones that I have watched usually go all the way, trigger the switch immediately, and then load and purge. If anything i usually see it hit the extruder gears and then keep feeding a bit past that. Out of 5 failures on a 3 color print I’ve seen 2 failures on lane 1, 2 on lane 3, and then 1 on lane 2.

When I recalibrated the Bowden length I used lane 2 and it came out to 1605mm which is my shortest length of all lanes. I’m wondering if I added the rubber band how much it would change that figure. Maybe a few more mm of Bowden length would increase the reliability.

[upbeat heart]

Possibly but it should be going into slow moves if the fast one doesn’t get there

[narrow lily]

Regardless I’m probably going to add a rubber band tonight and run a calibration print and see if I get any failures.

[narrow lily]

I’m wondering now if maybe my sensor just isn’t triggering? I just walked into my printer dungeon to find yet another failure to feed, but the filament was pushed into the extruder gears and the filament switch actually showed “detected” in the gui. All I can guess is maybe the switch didn’t trigger until the pause macro was called and the printhead moved from the back left to the front middle and relieved some kind of stress on the filament path? Shit is weird.

[upbeat heart]

possbile

[upbeat heart]

@narrow lily can you post your extruder section from the hardware file?

[narrow lily]

No problem, gonna be a little while though, I’ve been working on other projects all day. It’s finally good garage weather around here. Also, I picked up some rubber bands today so was finally gonna give that a go.

[jagged stag]

also might as well and update AFC 😉

when you get the chance

[narrow lily]

here it is, should be pretty standard

[upbeat heart]

cool, just wanted to check something. if we have more issues need to verify that device names all match up correctly

[narrow lily]

the boxturtle is named Crush and the Hub is named Squirt, everything else is named the way the installer set it up. I didn't mess with much.

[narrow lily]

new update come out today? what does it address?

[upbeat heart]

#afc-updates

loads

[narrow lily]

oh, awesome!

im gonna do that right quick and then slap a rubber band on and recalibrate my bowden length

[upbeat heart]

you'll need to calibrate each lane distance as well

[narrow lily]

do i need to update the macros? i usually skip that

[upbeat heart]

if you haven't modified them I would update there have been fixes

[narrow lily]

I really only use brush, cut, and park. poop and kick are turned off because I use the goose belt purger. I'll go ahead and run updates on all of it.

updates are done

anyone got a good picture of how the rubber band should be connected?

also, do i need to hold the TN shut while calibrating or will the rubber band handle that?

[narrow lily]

thank you

[upbeat heart]

and holding it really shouldn'e be necessary with this

[narrow lily]

you weren't kidding, the buffer never comes off of the compressed switch during the entire load sequence now

also, im still getting different bowden lengths when tested on each lane. lane 4 is always the longest, all lanes are now shorter than the shortest distance from before though.

lane1= 1570

lane2=1568

lane3=1559

[jagged stag]

That shouldn't really matter anymore with the newest update

[narrow lily]

lane4=1597

im not really concerned with any of them until we hit lane4, it's wildly different in comparison

and the shortest length i had from before the update and the rubber band was 1605

[narrow lily]

okay then, so just send it and see what happens?

[jagged stag]

So you can run AFC_TEST_LANES and choose all 4 and see what happens

This will got though and unload/load all the lanes

And then you can choose to do it x many times

[narrow lily]

no, im not listing lane test numbers, im listing the bowden lengths when tested from each lane

[narrow lily]

oh okay, i see what youre saying now

[upbeat heart]

that almost seems like lane 4 needs a bit of tension added, maybe its slipping

[jagged stag]

Also documentation for AFC_TEST_LANES if you are curious to what it does https://www.armoredturtle.xyz/docs/afc-klipper-add-on/klipper/internal/calibration.html?h=afc_test#AFC_functions.afcFunction.AFC_TEST_LANES

[narrow lily]

just looked at the lane4 extruder and it appears to be tensioned exactly the same as the rest of the extruders, i can add a few twists after this test is done running. i saved the bowden length as the longest one and then ran afc_test_lanes on all 4 lanes with 5 iterations.

[upbeat heart]

I’d give it like a half or check the anti squish hasn’t backed off into the arm

[jagged stag]

just watch the other lanes closely, if they hit the extruder before the move is done, add load_undershoot to AFC.cfg and increase it, this variable defaults to 20