AFC add on, a few observations

[swift comet]

rsarwar: Hi, recently built a nightowl, first time working with I am a potato.

when it works, it just works perfectly, but there are a couple of failure modes that i like to highlight:

1. sometimes, when doing a toolchange, AFC pulls past the load switch, and ejects the filament altogether
2. if there is a clog due to some hardware jam, i am using BT_LANE_MOVE to clear the hub, fix the clog, then Tx to load the filament followed by BT_RESUME. when i do this, AFC loads the filament, does a poop, whipes the nozzle starts to move towards the last position and whilst doing so, extrudes filament leaving a trail of filament that tagged to the print objects.
3. i am not sure what or why it happens, but it seems AFC loses its calibration for some reason. I had instances where,
   a. upon toolchange request, AFC sent a request to klipper to move extruder by 5.8m. leading to a fatal print. it was suggested that this could be related to exclude object?
   b. on different instances upon tool change request, it decided to retract the filament so much that the tip went above the cutter leading to a deformed tip. once it happens, it keeps happening till klipper is restarted.

klipper configs are here: https://github.com/rsarwar87/voron2.4_250_settings/tree/master/AFC

[icy canyon]

AFC add on, a few observations

[frank crest]

re: 1 - reduce tool_stn_unload if it’s going far past out the toolhead, or otherwise set afc_bowden_unload_length in the lane config

3a - bug with with exclude object in how klipper handles it

3b - i have a patch to try to fix this that i need to PR

(fix in that it will give up)

re: 2

are you using a prime tower?

[sinful prairie]

For 2 was it a lot of filament that was extruded?

3a I'm starting to think it's not a klipper problem but AFC the more I look into the problem, same reply that I had to you in the voron discord

3b did the cut still happen? Also sound like your filament sensor still showed filament present when it wasn't, but like thunder said he has a fix for that

[icy canyon]

thanks for the replies.

1. ill try that, seems like the right thing to do.
   3b, cool, can you point me to the PR please.
2. no prime tower. it works fine if i dont have to hit RESUME

@sinful prairie
2- about 3-4 cm strand, extruded at very high speed, probably 5-10 mm of filament. almost like it wanted to poop for the second time but decided "meh"
3a. i dont understand why exclude oject would make AFC loose its calibration.
3b. the cut still happened, but the tip had already gone past the cutter. i don't think it went past the extruder, just went past the blade and got retracted out after the cut.

The latter two points fix themself after a reset of klipper. so, are the calibration values being altered during a print, based on how much it had to drive before certain switch was triggered?

About the excluded object, does the 5.8 mm of filament correspond to the filament that would have been used in making the excluded object? -- If you get me started on which bit of the code triggers it, i can have a look in the weekend.

[sterile inlet]

at this point it is not recomended to use exclude object with AFC, doing so will likely cause you to miss tool changes as all of the slicers have those changes inside an object

was this a 1 color print that it failed on? or one that had a large volume of 1 color at a time

[icy canyon]

It was multicoloured, with the first tool going through 20 layers before the first change. definitely more than 5.8 m.

[frank crest]

it could be just 5.8m in retracts

anyways - Jimmy is the one to best speak to on this as he's looked into it some -- but we have seen this behavior with print start.

Another thing you can check as a potential workaround, is if you are doing a retract in PRINT_END, make sure you zero the extruder before doing that, something like this

    CLEAR_PAUSE
    M400                           ; wait for buffer to clear
    G92 E0                         ; zero the extruder
    G1 E-10.0 F3600                 ; retract filament

even without this though - and if we fix this if it is an afc bug - depending on teh object you exclude you may miss tool changes / not get correct prints

#1328943069399093298 message

has some more details on that

[sinful prairie]

Yes basically

[icy canyon]

i am already zeroing it, which probably explains why i dont see the issue at print_end.

also i think i found a possible reason why i see the problem 2. my pause has a retraction command, which is reversed in my resume macro by extrution. I assume AFC_PAUSE is called when AFC hits an error, and then it subsequently calls the main PAUSE macro? ill have to test it to confirm.
https://github.com/rsarwar87/voron2.4_250_settings/blob/master/klipper_macros/pause_resume_cancel.cfg#L111

[sinful prairie]

I think it's actually an issue in the AFC software

[icy canyon]

which issue were u referring to, the 5.8 mm retraction or the AFC_RESUME creating a trail?

[sinful prairie]

Well maybe both are tied to the same problem..... 😑

[icy canyon]

now that i have a hang of the afc add on codes, ill have a look when i get some time. maybe in the weekend 🙂

[sinful prairie]

I made a change for exclude object bug in a branch, going to try it tonight to see if it fixes the error

[sinful prairie]

@icy canyon can we call this thread solved now?

[icy canyon]

Brilliant. Please do close if the bug has been fixed