Also the rail screws (two screws on the left side close to the fan duct)
OK, I wouldn’t describe any of them as super loose, but most of the screws I was able to snug up a little bit. I’ll re-run the left and right tests and see if there’s any difference. I’ll report back afterward.
I’m surprised Bambulab didn’t use some kind of Loctite on the screws…
Probably because some of these parts need to be replaced after a couple thousand hours and some of them are really fragile
Theres lots of reports from all over the forum regarding badly tightened bolts all over the machine, wether that means loose or overtightened.
That was it! Not sure if it was those screws exactly or the other screws you mentioned later, but wow, night and day difference! Now the left nozzle print looks like the right nozzle print. 
Thank you!
It looks better than the printer cam suggests, but you can see by comparison to the OP picture the improvement.
1 Like
Yes, my experience with these screws is that if you can still drive it easily, even if just slightly, it’s loose enough to cause issues. But then again you don’t want to over tighten it as well…
These screws are very fragile and brittle, they might break in the hole if you overtighten it (or Loctite then try remove it), quite a nightmare. Which is why I suggested to have spare heating assemblies prepared…
1 Like
Now that it’s finished, here’s the final “after” photo that makes for a better apples-to-apples comparison with the OP photo.
What may look like a print defect in the front near the camera is just moiré effect from the camera. i.e. camera artifact that isn’t really there.
Well, yet another reason to run a “perfect first layer test” to find this kind of problem before it bites you during a real print.
Final proof of success:
Both left and right nozzles print the “perfect first layer” on the same print.
3 Likes
Well lucky you :D, im still trying to figure why certain spots WILL print perfectly on a similar test, but with actual models it looks like theres a lower spot that is not compensated. Maybe due to auto flow calibration?
For the benefit of anyone reading this, there is an auto dynamic flow calibration (Bambulab’s term for pressure advance), but there is only manual flow calibration.
Now, on to your question:
You’re ahead of me, because I haven’t yet noticed that. Also, it’s hard to know without seeing pictures exactly what you mean. However, based on the above, I’m assuming you mean auto dynamic flow calibration. You could test that theory by doing a manual dynamic flow calibration and turning off the auto dynamic flow calibration.
Sorry I can’t be more helpful, but at least this gives you a plan of attack to answer your question if you didn’t already have one.
My hunch though is yes, because what else could be interfering? Maybe there’s some way to affect it even if it’s “auto”? I’ve never looked into it. Have you tried running some of the pressure advance calibration models to see how well the auto does or doesn’t do on them? Not sure what else to suggest. I think the H2D may use some kind of pressure transducer to do its PA adjustment, so maybe yours is failing? Perhaps its worth gathering the evidence to see if that’s the case.
Its honestly hard to put it into pictures because the only way to realize it is watch the first layer as its laid down. Practically imagine it like theres a pothole, 2x2cm in couple spots, where the filament is not squeezed down as it does in the rest of the bed, resulting in it not being a layer perse but couple lines not fused and Iphone is betraying me in trying to photograph that.
Regarding the Auto Flow Calibration its really murky waters because I cant understand if the Auto/On/Off when starting a print, reffers to flow RATE or pressure advance or both.
IF it means flowrate (ffs bambu, just name them accordingly and properly, come on ) it would make sense as Ive noticed that those “potholes” coincide with the small period that the nozzle camera light is ON.
Keep in mind I am printing ABS and exactly like others have stated, in high chamber temps the nozzle camera seems to only activate for small portions instead of all the time watching.
My thought after noticing this is that somehow the AI is trying to compensate for what it thinks is Under/Over extrusion , resulting tin the layer not fused in the spots .
Now if Auto Flow calib (again , talking about the option before printing, not the “manual” auto calib of pressure advance or flowrate from Calibration tab) means its only pressure advance checking, yeah I dont think it would make much difference (given I ran first layer at 30mm due to complexity and fine details )
Both PA and Flow Rate in my filaments is calibrated manually too and the values are saved .
Ive noticed though , on the AUTO selection of the option, it says “if the filament has been calibrated recently, no calibration will occur” but that doesnt specify what recently means. Also I seem to remember a mention somewhere saying that the nozzle camera, servoextruder and AI will actively monitor and alter said rates during the print, but im not very sure as to wether I really saw that or I was hallucinating from tireness
Οk so , ive made a mistake BUT i have resolved that issue.
The mistake is I took 2 troubleshooting steps simultaneously, and I cant be sure which on was the solution.
So 1st, I set the auto calibration on OFF
and 2nd, checking physically the machine I noticed that A) my rollers had black rubberlike residue on them , enough for it to not be only an imprint but to be considered debris that alter how the belt steps on the roller. CLEANED .
Ive seen the same residue but in dust/particles form on the rods. CLEANED.
The current task running is covering I believe all the spots ive noticed to behave like potholes, but in the meantime ive realised that I have several belts that are out of allignment with their rollers, resulting in running/rubbing on the lip and causing wear on the top of the belts. I can already see the fibers.
AFAIK, “Flow Dynamics Calibration”, as shown in the picture here, refers only to pressure advance. At least for me, it works so well that I never bother doing a manual PA calibration anymore. I always set it to “ON” when starting a print because it takes so little time.
If I had to guess, I think what you are noticing is that the flow rate you get when you do the manual flow rate calibration is “wrong” for doing the initial layer print. It works fine for subsequent layers, but for whatever reason it just not right for the initial layer. Fortunately, the slicer allows you to compensate for that wrongness by giving you the option of an initial layer flow multiplier:
Here you can see that I set mine to 0.9 in order to get my “perfect first layer”. What does it do? It takes the flow rate in the filament profile (the one that you manually calibrate) and multiplies it by 0.9, and uses that as the flow rate for the initial layer. All the other layers revert to the flow rate set in the filament profile.
I think what this means in practice is that the z-height that’s automatically set by my particular H2D is actually about 10% lower than it theoretically should be. However, we do not have access to either setting or amending z-height, so this is the workaround. It maybe does have an implication for the final z-height of your model. Maybe it will be too short by, say, 0.02mm if your layer height is 0.2mm. However, that amount of deviation is within the prescribed accuracy of the machine, so all is good.
Yeah that would be the answer, If i had the issue everywhere in the printed area, but I only got that in certain spots, and always the same, and not in the form of wrinkling but rather like underextrussion.
All good though, after cleaning the debris my first layer is almost as good as if I used Ironing , which is also what my “standard” for my production is , since in almost all my products the first layer is what the customer sees and im strict enough to redo incase of even slight imperfections. Call me crazy but i strive for 100% perfection everytime !
But… solved so all good, in that aspect.
All tha remains is to run a full bed 40-50 hours print, multiple parts , to see the reliability of it.
Question is does it compensate in z-height? cause i believe flowrate refers only to the amount of material used, not the X-y-z position/dimensions of it.
Infact im almost sure that what you are reffering to, is controled by the X-y-z Compensation settings
Before dialing in the initial layer flow multiplier, I had some areas that looked like over-extrusion and other areas that looked like under-extrusion. I didn’t see how adjusting the flow rate could ever remedy that, because in theory improving one would make the other worse. However, as strange as it is, when I zeroed in on the right mutliplier, it all came out fine. Go figure. I don’t understand why it works, but I’m glad it does. 
1 Like
Those type of issues I call ghosts. BTW how are you belts? any sign of wear on the edges?
Hi, can you help me on identifying the rail screws? I am having the same issue and have tried everything other than the rail screws.
Вы смогли найти эти винты направляющей? У меня проблема с левым соплом, всё перепробовал, но возможно тоже не затянуты какие-то винты… А тут, к сожалению, топикстартер не указал и другие люди, какие именно винты направляющей?