I’ve seen this reported elsewhere, with different ways of describing it, but after dialing in a perfect “perfect first layer” on the right hotend, I switched the same spool of filament to the left hotend and tried printing using the exact same settings. Well, it didn’t match!
It’s hard to see it from just the photo, but the basic problem was over-extrusion. However, assuming the extrusion rate is actually the same, as it should be, then it’s as though the z-height on the left nozzle was too low.
I’d settle for separate “initial layer” factors in BS slicer for the left and right nozzles.
I don’t see a solution given the current state of the BS slicer, other than arriving at “less bad” output for the left nozzle that would also worsen the first layer output of the right nozzle.
I think this will require some kind of bug fixing by Bambulab.
I’m running the current stable release precisely because it had been reported that this problem existed in the current beta release. Well, it exists in the most recent stable release also.
And, yes, I did all the calibrations and the bed leveling too.
I just started my first 8 color (abs again) print on H2D .
Im watching the leveling process as it goes, and apparently after meshing the full bed , it switche to the other nozzle, probed in zigzag, and then started again full mesh.
Thing is , im noticing it pauses on certain points, which I have failed to see if they are the same all the time for now.
My experience with klipper and several meshing methods (klicky probe, inductives, eddies yada yada) is telling me that those repeats are due to out of margin measurement so it has to repeat until its within that margin .
Anyone has any clue why it stops at a point and probes like 10 more times? (or a bit less , its so subtle i cant really count, but for sure it stays longer)
That was my experience with klipper as well. We can’t know what the situation is with Bambu, since it’s black box, but I would think it has to be the same.
Side note, my print failed from less than first layer, about halfway there.
Nose clump detection is a straight lie at this point, I had a whole flake almost a finger wide stuck on the active nozzle like a flap, and AI was like “no bro all good go sleep”
I mean what does it take for the AI to notice? a whole 350x320 model to attach on the nozzle?
pics
I have all my settings set to high and have only experienced two false detections—every other instance has saved me significant time on cleaning and related issues.
I don’t use Textured PEI, only Smooth PEI, but I haven’t noticed any difference in first layer performance between the two nozzles.
For reference, I’m currently on the latest beta version.
@L0rdS474n Just to be clear: you’ve run “perfect first layer” tests that cover the entire build plate for both nozzles, and
you’ve observed no difference at all between the perfect first layer prints given by the left and right nozzles?
Both really are perfect “perfect first layer test” prints?
As noted by @NothingHam , I have observed the bed leveling sometimes taking an unusually high number of measurements per test point, and this suggests there is some kind of repeatability issue going on. IIRC correctly from how it worked in klipper, if there was too large a deviation in any of the measurements from either the median or the arithmetic average, then it had to keep trying until either it hit its retry limit (resulting in a FAILed bed leveling) or else the deviation amount eventually fell below an allowable limit after retrying.
Yes thats exactly how it goes down.
offset_margins: -2,2 #The minimum and maximum margins allowed for the calculated offset. #If the offset is outside these values, it will stop! #The margin can be defined as “min,max” e.g. “-0.5,1.5” or by just one #value e.g. “1.0” which translates to “-1.0,1.0” (which is also the default).
So if the offset is any other number apart the ones between 2 and -2 , it will probe again using:
#samples_tolerance_retries: default from “probe:samples_tolerance_retries” section #The number of times to retry if a sample is found that exceeds #samples_tolerance. The default is from the probe’s configuration.
and you can give it a bit of tolerance with #samples_tolerance: default from “probe:samples_tolerance” section #The maximum Z distance (in mm) that a sample may differ from other #samples. The default is from the probe’s configuration.
these lines in particular are for Z-offset, but it goes down similarly for meshing.
Yes, I have printed a first layer that covers the entire build plate using the dual nozzles.
The left section corresponds to the left nozzle’s working area, and the right section to the right nozzle’s area.
I’ve also printed first layers that span the overlapping region accessible by both nozzles, in order to compare performance at the same position on the plate.
Both results are excellent — not a single flaw or defect on the first layer.
The only time the printer re-runs bed leveling is if I’ve removed or physically flexed the plate.
In most cases, I don’t need to remove it between prints, as the parts usually release from the plate once it has cooled.
That said, when using a Smooth PEI Plate, poor adhesive bonding between the PEI layer and the metal base can cause it to lift slightly, creating microscopic height differences that bed leveling cannot compensate for.
This can lead to various print inconsistencies, which is why I generally prefer the Carbon Fiber Plate — although it is not yet available on the market.
@L0rdS474n Thanks for clarifying. So that suggests it’s maybe something about my machine that’s fishy then if yours is not experiencing it. I guess I’ll have to pull a ticket. Maybe the strain sensors are somehow off and aren’t registering the nozzle probe attempt correctly.
it wasnt a buildup, it was literally a whole piece from the first layer that it knocked away and it stuck there like a veil between camera and nozzle, yet nothing from AI , and it was set at max.
If the issue is on the left side, it could be caused by the filament guide getting stuck, which can also lead to various print defects.
In my case, that part got stuck in the upper position and caused filament to leak out between the nozzle and the extruder.
Check whether the guide is stuck or not sliding smoothly as it should.
Refer to this guide and inspect the spring-loaded section for any debris or obstruction:
Also make sure that all components are properly fastened.
It’s difficult to say for certain, as first layer issues can stem from a wide range of causes.
My gut says this is more about z-height than filament guide, as I don’t see how filament guide could lead to over-extrusion. My hunch is that the strain gauges are sensing differently just based on the two inches apart the nozzles are, and that they are reading in error, or at least not very repeatedly, which explains the multiple retries during bed leveling. It shouldn’t need to do a bunch of retries.
Honestly, when I first got it, I don’t remember it doing a bunch of retries, like it does now, so maybe something degraded between now and then.
That said, I’m going to pay more careful attention to bed leveling between the left and right nozel and see whether only one side or the other requires a lot of bed-leveling re-attempts, or whether it is equal on each side. About as fun as watching paint dry, but it must be done. Normally I barely watch at all but instead go off and do other things while it bed levels, but now I have a reason to pay close attention.
I just watched the new print (dual color) that i started while leveling.
It probed mesh with left nozzle, then it went back to right, started meshing in diagonal throw the plate, halfway stoped and now its meshing it all again with the left…Im so confused and honestly this secrecy from bambu doesnt really help us to understand the machine, evenless troubleshoot…
Are you on beta firmware? When I was on beta it happened to me as well (either too high or too low), I did the calibrations over and over it doesn’t work, even factory reset doesn’t work. Reverted back to production firmware then it works. Strangely I can’t reproduce this issue with beta firmware now.
@hotellonely That’s very strange to hear! I was on beta, and then I read about your problem and how it went away after reverting to the stable release, so before even running the test in beta I backed up to the stable release. Hence my disappointment when it didn’t work as expected.
Actually, before reading about your tribulations, I had just assumed it would work without problem on the left nozzle after I tuned up the right nozzle. It was your report that made me wonder and wanting to check if there was a problem, as indeed there seems to be! So, thanks to you for the heads up.
I could probably limp along with things as they are, but it doesn’t seem up to par.
I could tune a different “initial layer” flow factor for the left nozzle, but then it would be off for when it switched to the right nozzle, as both left and right nozzle share the same “initial layer” flow factor in the print process settings. Actually, that does seem like a fatal flaw all by itself in the slicer, because what if there were different materials in the A and B nozzles? It’s conceivable they might need different “initial layer” flow factors.
For that reason, an acceptable solution to me would be separate “initial layer” flow factors for the left and right nozzles in the slicer. Maybe Bambulab simply didn’t think far enough ahead yet to consider that possibility.
