I’m getting poor overhang quality on this model using Bambu matte filament.
Here is what I have tried:
Slow down for Overhang is on at 60, 30, 10 and 10
Default Bambu profile involves overhang cooling
Belt tensioning
cleaning the carbon rods
re-orienting the model (an overhang test showed worse overhangs in certain axes)
a variety of support options including trees and default mode
Various OrcaSlicer settings aimed at improving overhang quality
I am printing at 0.08 layer height
Theories:
The elbow where this is an issue is often prone to vibration and prints previously failed at this point. But it also occurs on the knee and I cannot explain that.
The overhangs look the same regardless of the type of support, it seems and even in unsupported areas
I am using 10% gyroid infill and wondered if this may be playing a role
It could be the filament but I only have Matte PLA to test at the moment
That is what I would consider a difficult print. Not sure my experiences solve your issues but there may be some insights that may help.
My first complex print was a couple of unicorns from Printables. 4 very thin legs, a large belly and a humgous tail needing to be printed in thin air before connecting to the legs & body. First, I had to stop the spindly legs from snapping. So I upped the infill. You may want to go for honeycomb infill and go quite a bit higher than 10%.
The second issue was with the tail and belly being both nice and smooth after support removal but staying in place during printing. Fiberlogy BVOH was a great interface material for that PLA print.
That did not however hold true for my second complex print. A fairy also from printable which had 2 of those elbow problems. I printed that in PETG so BVOH did not bond sufficiently to the model to really support the elbow mechanically. It tended to break off due to bending with a large moment arm to the build plate rather than vibration. In consequence, I differentiated between the two functions of support.
A) To provide a stable platform for “Island” printing and B) to provide a snug platform for high angle overhang surfaces.
I did not find a single solution for this so I reverted back to the ancient method of manually building support structures in the slicer from primitives for stability which would be cut away and backing that up with manual tree support interfaces from a dissimilar material (I used BVOH but would try PLA for PETG in the future) to prevent high angle overhangs from drooping with a zero mm z-distance.
It is a complex print so there’ll be quite a lot of experimentation. You may want to just focus on the problem areas first by moving the print down to only look at the elbow for example. It’ll not mimick the moment arm, but should give feedback on support design and interface effects.
and good luck,
Eno
PS: If you want to do it all in a single material, the “Islands” and long lever arms force you to use a 0mm interface z-distance, i.e. full soldering of the structure to the support. It’ll be a pain to remove even when using a coarse classical support.
However, you could also try to cut the model twice:
At the waist and print the lower part upside down (maybe at an angle due to overhanging feet),
At the shoulders and print upside down to always ensure the left arm to have a stable load path. The head should be OK with a bit of tinkering for the chin support.
Actually, you had a very good point. Islands are aggravated by the drastically increased warping in curves. Hence the lateral forces on the print which tend to first bend the Island and then quickly break the supports/interfaces.
I will say I have had similar issues but really when it comes to radial overhang not linear. When it radial overhangs I noticed it tends to be more of a warping issues for me.
Exactly A straight line is one dimensional, so warping has a (mainly) one-dimensional effect. On curves and edges, we have a 2D situation so the effect is squared. Adding layers makes this cubed!
That makes total since, complex, but simple. Are they general techniques on how to combait this? Different orentation of how your printing the part, but it seems that even chaning how fast or slow a print happens wont necessary affect the warping affect because that material cooling properties.
I want to provide a brief update. I agree this is a difficult print. Both Triangle and Honeycomb infill failed. The supports were knocked off (in one case I even added 10mm brims to supports). I think the problem is the nozzle knocking the print which I have observed before. I have changed to 20% Gyroid infill which mitigates this problem. Of great interest however, the left knee which previously showed some poor overhangs was quite smooth. This does suggest that the problem is partly the print vibrating with high speeds causing poor quality layers on overhangs. Will post my updates.
I was surprised that Honeycomb also did not work but I guess it does make sense when compared to gyroid as locally it has twice the thickness, aggravating local warping. And you do have exceptionally large moment arms from the shoulder to the support interface and the support interface to the bed.
Did you also enable an infill for the supports? You have fewer options but that can help in getting extra cm out of them.
I usually associate the type of overhang drooping and wall defects more with moisture, calibration, flow tuning and of course too much of a support interface distance rather than vibration but it is certainly not impossible.
My first guess (without actually being there) would be, that you have been particularly succesful in combating the wobble resulting from deposition on warping surfaces.
In my case last autumn, the (much lower) moment arms were still sufficiently long to neccessitate custom supports with solid interfaces against island wobbling. I used manual tree support (5mm diameter, 2mm honeycomb infill) with a BVOH 0mm concentric interface (would go for PLA nowadays with the PETG print) against drooping. While my solid supports came of easily with a sharp cutter, the manual tree supports came right off themselfes.
I don’t really want to use 2 filaments. The print takes 10-12 hours as is and I don’t want to lengthen that. Besides, if you see my next post, it probably won’t fix the issue.
Ok, I am becoming a bit frustrated with this model. After having 5 or 6 successful prints, I now cannot get it to complete with the new BL roll of filament. I tried gyroid 20%. The overall quality is higher but the print fails.
Yes the supports have had infill the whole time.
I think the issue is due to the nozzle hitting the print which I have recorded previously but I cannot upload video here (I was using weaker supports at that time). Currently, the support is being dislodged despite having quite good bed adhesion (the model is not that easy to remove from the bed). My infill speed is 400. This is bothersome. I will contact support when I have time.
Actually, looking at where you were when you started this thread and where you are now, you have made more significant progress than you may realize.
Most importantly, you seem to have managed to get to the point where the left arm attaches to the main body. And with very high speed at that!
There are a couple of points that may help you a little bit from my experiences with my fairy print.
I found that it really pays to go slower. It is nice that the Bambu can go so fast, but it really shines in terms of quality when going slower. That allows for better cool down (and hence lower forces on fewer warped areas) and gets rid of a lot of dynamic effects. Since you have good results until around the shoulders, you may want to add a modifier to the upper part of your model where you lower the speeds. Since the long moment arm leads to a quadratic (!) increase in base moments, you may want to consider a major speed reduction for the upper body. That high up and close to the frustration level, I’d hesitate to go to tripple digits.
It may take longer, but I have found it to be much more rewarding to get one acceptable print after a couple of days than to get 6 throw aways during a week. Especially when a good print followed 6 bad ones
A second noticeable item are the spagetthi on the torso. You may want to add a cleaning tower to reduce that. That is also added when doing a timelapse. Takes a bit of filament, minimal additional time but it sucks up most of these noodles.
A third item is the very high bonding between your supports and your model. While that is probably a key contributor to being able to go as high as you have, the strong bond suggests that it’ll be a real pain to remove post-print. I find PLA-PLA bonds annoying already and with substandard quality, but when doing PETG-PETG, I have found that even with a large interface gap the support fuses strongly enough with the model to tear the model surfaces out cleanly.
You may want to give support removal a try on this print and think about staying with a single material print or if you are OK with the removal process and subsequent model quality.
Don’t loose hope! You have come a longer way than you may realize!
the spaghetti is only due to the print failing, if it gets past the elbow, there’s no spaghetti.
I think speed probably plays a role but do I slow down everything? I will look into how modifier works. Really, I would mainly need it for the elbow. Once it connects to the rest of the print, it usually succeeds. I would want the outer wall speed to remain the same to have the same finish, however.
The supports actually remove extremely easy. This is a matte pla and generally the supports come off with minimal force. I might use pliers to get rid of the last few pieces but that’s it.
No, you do not need to slow down everything. You can right click on your model in the prepare tab and just select Add modifier → cube. You then move it in XYZ to encompass the upper part of the model. Maybe from the elbow interface or even the armpit? Then, with your modifier selected and having the “Objects” selected in “Process” you can change speed settings only for the volume in the modifier.
That is good to know. I was never happy with my overhangs as even with a 10% z-distance increase, I found bonding too strong while also giving enough room for drooping. Maybe some more reading for me…
I used this modifier and it still failed. It did not even print any of the elbow and the support looks identical to previous. Given that this print succeeded with the previous roll, I am wondering if I have a bad roll. I did not calibrate it as it is BL filament.
New rolls of filament are not neccessarily dry. They had all moisture removed during extrusion, but were probably water cooled and may have sat around in moist atmosphere before vacuum sealing. Nowadays, it is not a given that new rolls are in fact dry. For a complex print like this, you’ll want to both dry and calibrate that roll.
The big challenge with this print is the very high moment arm between the upper torso and the base. That is just inherent and is height driven. So from on a little below the level of the last failure, you’ll want the modifier to encompass the entire model all the way to the tip of the hairs. The current modifier may make life a little easier in that area, but once print height goes above it, the same things will happen as before as there, you have not actually changed anything.
Dare to slow down even further with the modifier. Maybe as low as 100. Or around the head even 80. And also halve acceleration. It has a dual effect:
It leads to better colling => less local warping => lower dislodging forces from the nozzle.
The nozzle itself will accelerate slower. This will directly reduce the excitation forces.
Remember that you have time in printing. Almost always a slower print will be much quicker than having to redo a failed print. It may strain your patience, but it cuts down on filament waste, frustration level and is easier on your nerves. Once you have achieved the quality that you want, you can then always try to speed things up. It does not work the other way round.
I will dry the filament. The thing is, I had success without any modifier previously. I actually recorded the entire print yesterday. Somehow the nozzle is slowly detaching the support from the base.
It’s true. I only want to get one good print at this point, so I’ll take a different approach.
Yes, exactly.
While the slight forces exerted by the nozzle on the model due to it slightly pressing its extrusion onto the model are in themselves fairly minor, you are likely also struggling with edge curling/warping on the elbow. While PLA is very forgiving, it too shrinks during cool down. This gives rise to warping and curling. Essentially edges and changing radii (such as a cross section of an elbow) curl upwards as a result. Much less than for example PETG, but it still occurs with PLA. It is cumulative and means that your nozzles forces on the printed piece increase with distance from the bed. At the same time, the moment arm increases and the print fails by support dislocation from the build plate.
Slowing down will reduce heat injection per layer while improving cooling and of course also directly reducing nozzle forces (F = a*m => Reducing speed reduces aceleration on the part). Looking at your initial speed graph, I would recommend halving all speeds (i.e. 80mm/s) and accel’s from just below the elbow upwards. It may also be useful to half speeds (i.e. 40mm/s) and accel’s again from the neck upwards.
Keep in mind that for complex prints such as this, you first need to learn to crawl. Once you are standing on solid quality ground you can walk and then try to run.
You can either go much, much slower for the entire model, or use one or more height (!) driven modifiers to lower your speeds significantly. Probably from just below the elbow.
and good luck,
A straight line is one dimensional, so warping has a (mainly) one-dimensional effect. On curves and edges, we have a 2D situation so the effect is squared. Adding layers makes this cubed!
