Hello, Im very much into experiments with overhangs right now. It seems PLA tends often to curl, even with filaments Im using on other printers without problems. Ive tryed different ways in reducing flow, slowing down the prints, playing with temperature and so on. Yesterday I laid my finger on the tip of the nozzle. It seems that the airflow is concentrating a bit too deep and also behind the nozzle instead of its center. The auxfan wont help, because its directed only to one side. Is that maybe one crucial part why overhangs on the X1C are worse than on other printers? Many people say the P1P will print overhangs better because its not enclosed. It seems this is only the half truth, because the airflow of the 5015 is pretty strong, so it should not matter that much. My assumption is now, that the direction of the air is the reason for all that. Can anyone confirm this?
I never had issues with PLA curling in the X1C, so I was surprised by this issue. Maybe I am not pushing the machine.
Is the P1P and X1C air flux direction different? I thought fans and air ducts were identical, but maybe I was wrong. The only way to assess how effective is the airflow direction is through CFD simulation, which I am sure that the Bambu Labs team did that work; however, for PLA, there is a significant influence from the AUX fan that cannot be disregarded. I did not find any published study on the topic, but you may get a basic idea of X1C air flux in this video, that seems to agree with your assessment:
Note: Regarding filaments, most (if not all) of my filaments outcome in bad prints if not perfectly dry, even though they work flawlessly in my EnderS1Pro. I think that the X1C is very strict regarding the filament condition.
The airflow is powerfull, no question. But try a different approach. Put your finger under the nozzle while letting the fan run at maximum speed. I got 2 X1Cs and see the same behaviour. The flow is centered about 5 mm below.
For bigger parts it isnt a problem, but even if you slow down to 30 mm/s or slower for small parts it tends to curl, while there is no curling on my X-Plus 2 with faster speeds (Icarus Mod).
I will try to dry the filament over the weekend.
The question is, why should the Bambu be more sensitive? Its still a hotend with a meltzone and still the same filament. Even with a Volcano (Vyper), I didnt have this kind of problems. Thats why I think the design of the cooling is flawed.
I don’t use PLA very often, but it has never curled on my X1C, using the cool plate or the textured PEI plate… unless I fail to use glue stick, as recommended.
I dont speak about warping.
This is my calibration tool for retract, temp and the final flow. The legs start to Curl. The first layer is not a problem.
I will check.
I am not refuting your statement; I am only trying to understand the cause, and I am sure there is always a margin for improvement.
You made a good analysis regarding whether it is a small or big print, or even better, if the part’s smallest xy area is large enough to receive the airflow from the ducts. Your test print’s highest layer is not.
Assuming, as you mention, that the airflow attack angle is so that it hits the xy nozzle position at 5 mm lower in the z position and that during printing, the nozzle is between 0.2-0.4 mm from the print.
In that case, it is clear that if the printing layer area (xy) is relatively small, the air flux will hit the surface at a lower height than the printing layer; in that case, the auxiliary fan and the minimum layer time come handy to ensure that the part is cooled before printing the next layer, and thus avoiding the curl.
Someone checked for me already 
The fan is blowing offset, but it seems the slicer is playing a role too. Ive noticed that printing with 0.16 mm height is getting the best results while printing on lower heights does make it worse. This should physically seen not happen.
I will try his suggestion and dry my filament over night and see if the curling can be reduced. Still there is something odd and working differently as on other printers. If we can identify this, Im sure we will find a solution. I found on printables a front cover for aftermarket nozzles and LEDs. Im not that good at CAD but maybe the guy is nice enough to recreate the cover for optimizing the attack angle.
Still there are people who have problems with bigger overhangs, but maybe there are more factors playing a role each on its own. Slowing down the print for overhangs while having the same temperature is a problem in my eyes. The cooling not being exactly centered comes in too. Maybe there is a sum of all this creating the common overhang problems. In the post Ive linked ppl say old slicer variants tend to work better. To evaluate this I`ll try to setup a profile on Prusaslicer tomorrow.
With a bit of luck and good teamwork we`ll find the ghost in the machine 
Edit: The aftermarket nozzles seem to be 2mm longer. I`ve ordered one to see if I can improve and to test brass nozzles for better heatconducting
I made some tests tonight and this are the results.
Printersettings are the same I just changed layer height and aux fan settings.
PLA SuperFilament 210° C, 30mm/s speed, accel 2k
0.1mm - Aux off
0.1mm - aux on
0.1mm - aux on - overhangs directed to aux fan
0.16mm - aux off
The results of 0.16mm compared to 0.1mm with aux fan on are similiar. 0.1 without aux is getting the worst results. My first theory is that through extruding more Material, gravity is pulling the molten plastic more downwards. Extruding less material leads to higher meltrates and the partfan is not able to cool those layer quickly enough, resulting in bad overhangs even on slower speeds.
The second theory is the use of a steel nozzle instead of brass or plated copper. A steel nozzle will lead to cooler extrusion. Getting the temps up sometimes seems to help. After extruding, the molten plastic wants to return in its old form. Colder plastic likes to expand more, while the hotter one is less viscous. A brass nozzle will hopefully arrive around 15. August to verfiy this. Getting the temps up on the pyramid reduce curling but induce stringing. 
Many thanks for sharing.
Until now, I did not have time to catch up on all this, but I just printed a pyramid similar to yours and didn’t like the overhangs look.
I am surprised as I never noticed this in the past. My typical prints do not involve aesthetic pla outputs, so that I may be wrong.
I will check my prints and compare them to older ones as soon as I have time.
I need first to dry the filament and calibrate it again, as I replaced the part cooling fan and want to be sure everything is working.
No Problem, im already on fixing this.
The angle on overhangs increased as I made smaller openings in the duct (raised the bottom of the duct 3mm).
It increased way more while closing the sides about 5mm and I raised the bottom of the duct even more to also 5mm.
Give me some weeks and I will recreate the whole duct/shroud. Got my hands on an good model from chiz on printables. He was so nice and gave me his sketchup file, but I cant convert it with free Fusion. So I created a workaround. For the final release I will create a whole new model in Fusion. Since I dont know much about airflow its trial and error design for me. What I already found out is:
- The openings are to big so the air is not cool enough for big overhangs.
- The openings are to wide, air is flowing in areas where its not needed.
- The angle is bad and leads to curling on sharp overhangs.
On my old Anycubic Mega S I used the MK4 Mod. This duct was the best Ive ever seen. One Sunon 5015 fan was enough to blow ice cold air only because of its design. It seems the 5015 Bambu has even more power. On a V400 we can see that its printing 70° overhangs with ease only using 2 4010 radial fans. So 2 sides are more than enough to get sufficient cooling and a complete new design can be ignored. I dont like those ducts using 3-4 sides to blow because the splitted flow wont be cool enough.
Totally unrelated to this thread, I ran across this while poking around aliexpress for other 3d printer accessories.
You can buy this file on cults3d for 1/3 of the price. I dont trust the design. Looks like the airflow will sink even deeper below the nozzle. Give me some time. The last tests Ive shown are printed without aux fan support. I had many printers now (2 Qidis, 2 Anycubic, 1 Creality, 1 Sovol) and the best design was this duct.
I hope to achieve 70° without the need for the aux. After reaching that angle I will release the model on printables as remix from chiz’s design
I’m excited! Thanks for working on this!
Commenting to follow along and print the new duct when released!
I made some progress using a CHT Brass Nozzle. I can print 5° higher. The reason for this supports my theory that the (hardened) steel nozzle conducts heat not very well. So the plastic will cool too rapidly. When it cools too fast, the plastic tries to revert in its old form resulting in edges curling up.
My prototype-duct is also nearly finished. Since im pretty new to desigining in F360 I had to use many sketches and extrusions. If someone knows a way to create something like that with less steps, I would appreciate to know. 
Hey there!
Its been silent for a while I know. So here a little update. I made it up to 95% clean 65° using the standart speed profile 0.12mm without the aux fan. So the duct IS flawed or not being perfect aligned at least. 
Some minor adjustments and we are ready to go.
Here some pics:
Since Im short on time I will stop at those 65° and maybe start over again in a few weeks. This should be enough for most of the parts being printed.
And yeah I used pink PLA for prototyping 
Edit : Ive used the High-Speed settings this time. The older ones were printed on 30 mm/s for walls. So maybe 70° should be possible if you slow down.
nice! ive been having issues with curling and cooling too. excited to see your fan duct design finished!
have you completed your duct?