I apologize for my frustration ahead of time. I am posting this in this group to see if anyone has any solution:
This is in relation to Issue #516 that never was addressed or solved by the developers. I would like to note that I did not write the original issue. I am having the same problem though.
I have spent dozens of hours and days and a lot of money converting to Klipper only to discover this issue that does not go away no matter what. It seems that the developers try to explain it away rather than actually fix the problem which is very frustrating because it renders this firmware useless when you are trying to obtain an accurately printed part (which can be achieved through either Repetier firmware or Marlin in my case).
The problem is very simple. The print head pauses at the end of every layer before starting to retract and then travel or begin the next layer. This can not be mitigated by turning up the speed of anything and it can not go away through direct drive either though it will not be as apparent because there is less pressure build up in the filament tract. As soon as you start printing something like TPU on your direct drive, say goodbye to any quality finish because this problem will be very apparent.
I have tried all the possible settings options as per issue [#516] and nothing has helped. As you increase the pressure advance, what happens is that as the head is heading towards the end of the layer, it thins out the extrusion but then it pauses like it always does and leaves a wide blob.
Can someone please take this seriously and address the problem that hasn’t been fixed for the past 4 years. I will provide further detail and evidence if needed. Thank you in advance. If for some ungodly reason I am writing about firmware issues in the wrong place, then please direct me to where this will indeed be read and addressed. Thank you again
I do not know your printer but I’d be very surprised if a delta is able to handle:
square_corner_velocity = 50
max_accel = 5000
at a printing speed of 130 mm/s.
Especially the square_corner_velocity is way beyond any reccomendation, which typically is 5.
The described symptoms perfectly match to a physical motion system (which includes the extruder) that is not able to cope with the requested acceleration and junction speed. This in combination with a bowden extruder looks like a sure recipe for failure.
Return to sane values in the range
square_corner_velocity = 5
max_accel = 3000
and try again. If needed work up max_accel until problems arise but leave square_corner_velocity alone.
According to input shaper, my max accel should be 5000. This is a fully custom delta with highest torque NEMA 17s that you can find.
I did actually have my square_corner_velocity at 5 but I was turning it up because certain people who have experienced the same problem stated that increasing the speeds somehow helps. The person in the original issue thread has also tried all these things only to find that it does not help. So I am at a loss.
I will say that I have not seen any difference in the blobbing whether at 5 or 50.
This is NOT a statement of the acceleration your machine is able to achieve. This is a theoretical value from which on the chosen shaper will produce an unhealthy amount of smoothing, i.e. deviation from the intended traction path.
General remark:
Except for the GH issue you mentioned in your original post, there are near to none complaints on pressure advance.
This can be seen as an indication that the problem is not on the PA side but rather in your machine / settings.
And seeing settings like your square corner velocity, acceleration, extruder retract speed (especially for TPU), I personally have my doubts.
That having said, I do not know your machine or slicer settings (there are quite some highly detrimental when using PA), so unfortunately I cannot contribute any more.
Understood. I can try your settings and I will let you know whether they have any impact.
I agree, this is not a PA issue at all. Some would suggest though that PA should be adjusted in order to help fix the issue and my whole issue is that this is simply not the case; I have tried all sorts of settings with PA and nothing helped. It is something else. Based on my testing, it leads me to believe that there is a pause when a layer ends and the next one begins. This causes a huge seam line down any part I try to print. Whether I print it using PA or with PA turned off completely.
I mentioned TPU as an example for a direct drive. I have a direct drive delta as my other printer that I converted to Klipper and I am experiencing blobs there too (specifically with TPU because it is the only material I print on it) which makes all of this that much more frustrating.
Based on my research and troubleshooting, I have eliminated the potential troublesome settings in the Cura slicer such as coasting and wipe. If there is anything else, then I am happy to test other options.
Never seen such, except for misconfigured slicers that introduce a pause on layer switch, e.g. Cura’s “Pause at Height” and IIRC correctly a bug back in Cura 4.1
Check your sliced gcode if there is anything funky on a layer switch. You can also attach a problematic file here.
Could you send a gcode file that exemplifies this problem?
Reproducing the issue by simulating the print in batch mode might be the best way to analyze the issue. Also this will clear any suspicions of something fishy happening at the gcode level.
Are you printing from Octoprint or from the virtual sd-card?
Thank you for taking a look. I will be working on replacing the PTFE tubing and ensuring that there is no excessive friction there. Perhaps that’ll help as well. PLA Star Pattern R01.zip (235.9 KB)
Since I am working with a bowden setup, I have set the unretract to 0.3mm to compensate for any slippage. However I have tried also with 0 unretract and it didn’t change anything.
I also tried reducing the accel to 3000, square corner velocity to 5, and max extruder accel from 1500 to 1000 and nothing affected the problem.
So to do a sanity check, I printed the same model on my other Delta with direct drive and it does not exhibit the pause at all. It just gets to the end and immediately starts to retract and it shows in the actual finish.
The conclusion I am drawing here is that it could be the old SKR v1.3 that is not keeping up with Klipper versus the new Mellow Fly Gemini with an integrated “rpi” on the direct drive delta.
Could this be a valid potential issue with the SKR v1.3 only having 100MHz chip? I understand that all the computing power comes through the RPI but is it possible that the SKR is becoming a bottleneck? If not, could it be the USB cable for some reason?
Top plot shows the toolhead velocity in purple (all directions) and in the z direction (blue). Bottom plot shows the extruder velocity.
I see no dwell time between the last extrusion of the layer and the retraction. Layer change goes with z-hop up, move XY, move Z up, then Z hop down. Then, it underacts immediately before the first extrusion of the next layer.
I haven’t looked at every layer changes, but they seem similar. Is the blobbing happening before retraction at each every layer change?
I don’t know much about the specifics of your printer, but your retraction settings seem a bit wild. Maybe they are normal for a long bowden, I don’t know. All I can see is that the retraction could be taking more time than expected since the extruder have to accelerate a lot to reach that velocity of 70mm/s.
I could be wrong, but I’m starting to suspect that the pausing is caused by the long retraction by itself. So rather than cranking the speed and accel up, I’d suggest to start fresh with no retraction at all, only PA and slowly increase the retraction (eg. on a retraction tower test). If I’m right, with a retraction length equal to zero, there should be no dwell time during layer change.
To be certain that this is not caused by retraction, you could try to uncheck “Retract at layer change” and Z-hop in the Travels panels of Cura and reprint your star pattern. If it helps, then maybe try increasing max_extrude_only_accel even more. But be aware that increasing it too much could cause the extruder stepper to skip and exacerbate the issue.
Quite unlikely. Klipper would react in such situations with a hard stop
Fully agree. Also such fast retracts are detrimental IMHO because they rather separate the molten plastic from the non-molten filament, so you retract only the filament but not the molten blob in the nozzle.
