How to tune Small Area Flow Compensation to improve 3D print quality
Вставка
- Опубліковано 25 лип 2024
- Check out my 2nd channel, TT Racing: / @ttracingyt
Thanks to the community, we continue to squeeze more and more out of 3D printing.
In this video, we use Small Area Flow Compensation to do just that. This slicer feature alters the flow rate of short extrusions to eliminate bulging and mess. As I demonstrate, it’s quick and easy to tune with the help of the right model and supporting tool.
Not every 3D printer will benefit greatly from this. Based on my testing, I would suggest that the thicker/taller the extrusion, the more benefits will be found.
Thanks to Alexander T. Moss, mjonuschat and all other contributors for their great work.
My test model and spreadsheet on Printables: www.printables.com/model/9047...
0:00 Introduction
0:42 What is small area flow compensation?
Post processing script: github.com/Alexander-T-Moss/S...
OrcaSlicer release (version 2.0.0 beta or newer): github.com/SoftFever/OrcaSlic...
2:34 Use in OrcaSlicer
3:10 Use in PrusaSlicer/SuperSlicer
5:02 Designing, printing and examining a test model
7:10 Understanding the flow compensation model
8:15 Tuning the parameters
11:07 Conclusion
Buy quality and affordable filament from X3D. Buy 3, get 1 free and a free sample pack with every order: www.x3d.com.au
Get Quality Resins from 3D Printers Online. 5% off storewide for Teaching Tech subscribers [Code: tech5]
3dprintersonline.com.au/
Take a look around and if you like what you see, please subscribe.
Support me on Patreon: / teachingtech
Regularly amazed with the contributions from members of this community. From the idea, execution, and showcase here in this video, big thanks to all involved!
Brilliant video! You've managed to explain it better than I could! ❤
Thank you for making this. By chance, how did you come up with the default values? It feels like they should come from some function that might be easier to tweak than a table.
Appreciate you and your contribution. This is great
@@maxconser9930 At it's core, the model uses (cubic spline) interpolation. This is the actual mathsy bit behind joining all the flow model points together, there are specific tools to show the result of this type of interpolation, but they are nowhere near as user friendly as a simpler chart in a spreadsheet. The default values were actually taken from the initial approach used for the flow model, which was to use an equation for the model line and not interpolation of points, in the end interpolation was favoured as is more intuitive to change than an arbitrary equation (you'll notice the lengths of the points in the default model are at "nice" values, then the flow rate % was what the original equation model produced at those lengths)
Thanks for your great work!
Only one year into printing and i learned so much from you. Thank you!
I agree. Michael has done so much for the community and has a treasure trove library of information. I wouldn't have been able to do as much with the hobby if it weren't for his videos and site.
OMG your timing with this video is *spot on!* I just started playing with my first 0.8mm nozzle and this small area flow issue is so much more pronounced with larger nozzles! I have been trying to tweak for weeks now and couldn't find a balance between long runs and smaller areas. This might have been the solution I didn't know I needed!
Thank you for all that you and the community, at large, does for the evolution of 3D printing! 💙
An amazingly giving community, that allows the non-initiated, like myself, to benefit from these improvements so that I can focus on making better quality designs to share!
This is am issue i've had for years and never knew what caused my top layers of small parts to be overextrudes. I kept changing the flow rate to no avail. This is a great fix, will definitly give it a try. Thanks for the video and thanks to the community :)
It's about time. This was the whole key to getting Bambu machines to iron properly
You absolute psychic magician. I'm in the process of tuning my Magneto X and this was something I noticed in a test print last night after EM calibration and some top layer quantity and speed tweaks. I was going to test print the same model facing 4 different ways to see the impact, and probably still will, but it looks like this may be what I need to get top surface quality all the way across the finish line. Love your timing, and adore this community and how these open source contributions make life better for everyone.
fantastic bit of work and thanks for sharing with the community
I like the format : focus on a setting, understanding and applying. Thanks !!!
Great! I hadn't looked into it much and I understood the basic principles, but this made the feature extremely clear.
Thank you muchly for the explanation and the coverage. Test print looks good!
A big thanks to the people who created this and to you for sharing this. Definitely interested to see how this works!
You are very good at explaining/wording the ideas and concepts in each of your videos. Thank you.
Thanks, Michael! This is an amazing contribution.
I had already turned this on in orca slicer but it is great to now know how to tune it. Thanks so much for this!
Great work sharing this, thank u, and great work from the developers!
Thanks for explaining the OrcaSlicer settings. Gonna try it now.
Fantastic! Thanks mate, that’s incredibly helpful. I really appreciate you going through it step by step.
Nice work on that spreadsheet!
Wow so this will be the video that pushes me to finally download Orca Slicer. Thanks!
Now this is something I am going to need to test!
this actually improved print quality on my ender 3 pro quite a bit. thanks, very helpful
Impressive feature
Thanks for sharing your experiences with all of us :-)
This was so helpful. I was struggling to tune in my top surfaces and it turned out that while I needed small area compensation, I needed much lower amounts of compensation than the default model provided.
Nicely done, thanks!
Fascinating! Thanks, Michael! 😊
Stay safe there with your family! 🖖😊
I wish we'd had this 5 years ago. My old delta with a huge bowden would have become awesome!
I appreciate that you covered all the slicers and the script version too. I use Orca, so it was a tad annoying, but you still did the internet a service.
There is config to set % of overlap between infill and perimeters on PrusaSlicer (probably something similar on other slicer). This config may help you close the gaps on your final results. As usual, great video, thanks for sharing!
I have always noticed this on prints, but never payed any thought to it. This is awesome gonna try it now.
Great video I like postprocessing scripts, because they enable us to come up with ideas to modify gcode. This way we have a proof of concept and it will eventually be implemented in the slicer, which is quite complex to implement if you aren't experienced and familiar with the code. It's also possible to write some code specific to one project and add a specific use case , which isn't available in the slicer. I've done some work to the gradient infill script of CNC kitchen to adopt to prusa and orca slicer. I also added a maximum volumetric flow. My first project was the implementation for a maximum amount of retractions in an extrusion interval which was already implemented in Cura. This way I was able to add this function to any slicers gcode.
I would love to look at various examples with and without ironing to see how the two features interact. Great video
I recently moved from superslicer to orcaslicer, first being a little sceptical because of all the hype, and eventually love it a lot (i also moved from marlin to klipper at the same time 😅) even if I'm not an expert at 3d printer. Thanks for showcasing this feature!
Did you convert and Ender to Klipper using a Pi?
@@charlesraes8541 not an ender, not a pi, sorry
Well what was it then kid? 🤣
@@charlesraes8541 I3 clone with btt cb1 on a btt manta board
That’s absolutely mahussive!!! I have problems with my corners and around holes since I typically print with an 0.8mm nozzle. My situation won’t be perfect but this will fix the worst of my printing problems for sure!
I actually red this in the release notes for 2.0 and thought it was a much bigger deal (for me personally) than scarf joins, but then promptly forgot about it. Thank you for the reminder!
Just started using some Rapid PETG and having this EXACT problem, can't wait to try out the script, thanks for sharing it with us!
That is so cool. Thank you so much for sharing that!
What perfect timing.
I just had this issue pop up after moving from .4 to .6 nozzle
Wow what a great feature!
I've been using this but wasn't sure how to tune, thanks for the explanation
Great explanation! Thanks
Wow this is gonna be huge!
This is finally something new after weeks, for us retail consumers! Thanks!
Having a Bambu I'm pretty ok with surfaces in general. Wonder if this will solve uneven surfaces in my case since it might be this lil tweak but it looks like more the color of the material changing when slowing to shorter paths, in an infill. How it reflects light, like broad surfaces shine while short track corners are matte, dull. Indeed it's a game changer on objects with many upper surfaces like the first one, because overextrusion in corners is still slightly a thing even on Bambu machines and calibrated flowrate.
very very cool and needed!
Tuning this could be simpler, but again thank you for publicizing this awesome new feature
If only orca would propperly work with toolchangers, then I wouldn't need to use Prusaslicer for my Voron Stealthchanger. Great video, thank you for your service to the community.
Just did this a week ago, the default orca profile got over extrusion for me, had to do tuning but then it worked great!
People's skills and willingness to share and improve on common subjects is a truly incredible means of perfecting an idea. No amount of tuning has ever dealt with the amount of overextrusion and plowing that happens in those corners and my preferred solution was setting a perpendicular top layer at about 10mms - ironing, if patient enough. I'm torn about whether I feel like implementing this and going through another 100 hours of experimenting and perfecting. Would like to see Prusa integrate it.
What a perfect timing ! Im having a weird problem with a fresh roll abs black. I have runned flow calibration 3 times in the last 24h starting to doubt myself as i didnt print in 2 weeks wich is waaay to long
I had this problem with a run of items which needed a mating face finish. I was using a 1.2mm nozzle so the lumps are significant round small features. It's a run of 80 so finish filing would have been tedious. Since I use Cura I found the best way to solve it was to write a script that excised the top skin from the last couple of layers in the gCode and just left the walls.
I need to look at these Slic3r variants.
I think this interacts/influences preassure advance.
May be a thing to look into for an upcoming video
Damn... I think i'm going thrue the rabbit hole this afternoon...
I'll have to give this a try. I designed and use a, what I call, skinning plane to shave off surface bumps.
Bring on the Hilbert Curves 🎉
Excellent
Oh man, we neeeeed this feature in Prusa Slicer too!!! I have particularly noticed this inconsistency on my Prusa XL printing with a 0.6mm nozzle.
That curve looks like exponential curve. F(d)=1-exp(-d/t). In that case you only have to guess one value t. The d is the distance. Maybe a few value to put in your sheet with that.
Thank you for explaining.
I’ll be giving this a go. Hopefully it’ll become native in bambu slicer soon.
This would be really cool to have on your website. Something that spits out multiple of these in a range that can all be printed at once and compared
I’ve been using this feature since it appeared in a nightly build and it works a treat! I’ve been very dissatisfied with how large flat areas turn out using modern slicers (Bambu Studio and Orca Slicer) and printers (X1C and P1P). My very old Wanhao Duplicator i3 Plus with Simplify3D v4 did a much better job in corners and around holes or protrusions, SAFC is definitely a step in the right direction.
I would love to see that oversized part reprinted with this and appended to this video. Again, thank you for your dedication to the subject and hobby. 👍
On Windows you can copy a filename by holding Shift and right clicking the file in Explorer. You will have a hidden option (revealed by the Shift) of Copy Path.
Hey Michael, thanks a lot for your work, great as always. I wonder though if it's worth going through each value separately or better to just define a function? With a little reverse engineering I found that the original values seem to follow f(x)=0.25/(x-0.25) except for x>=10 where f(x) is simply fixed at 1. So I would suggest using the function f(x)=1-a/(x-a) starting with a=0.25 and then adjusting the value of a to tune the model. Increasing a means lowering the values in the model, which makes the reduction stronger. What do you think?
Could this be translated into a spreadsheet formula in excel/google?
The amazing thing about 3D printing is that it's just physics in practice - which means that with enough understanding of math, physics and logic you can continue to improve on and on with just software changes!
Unfortunately most of the user base has a smooth brain
I think this is way better than scarf joints. My z seams are fantastic and barely noticeable anyway. This is huge and addresses the only problem I haven't been able to tune out. I have to balance underextrusion vs correct extrusion, but has bits overextruded and able to catch on the nozzle causing a layer shift in the y axis (300x300 bed running 8k accels on travel moves, that little bit of resistance overcomes my 85mm nema17 w/1.09nm torque) or it pops a peice off of the bed when it hangs up.
this has nothing to with scarf joints
Rather than modify the points manually extract the trend line formula in Excel, or the open source equivalent, and modify the variables there.
This lets us modify the curve in any non-linear way we may need without having to guess at the points values for a smooth line. Using this method the curve can also be broken up into more divisions which should reduce the errors from interpolation.
Or define a bezier curve of some sorts.
you could use a parametric exponential curve, and adjust only a single value, reflecting in all extrusion lenghts while maintaining the curve smooth
docs.google.com/spreadsheets/d/1Rv1aXhaP_OuGJEjUB59luggX0_8VCEHpCYnVCULFAN4/edit?usp=sharing
Sounds like this might improve tolerance test results and i bet it would help print in place models like Sunshine's print in place engines 😁
This is really handy. I've been deliberately setting non-perimeters to underextrude to avoid this issue, just accepting that I'll get small gaps as a result. I accepted that because it's better than the little bulges getting hit by the nozzle and either knocking the print loose or causing a layer shift, both of which have only gotten worse as I've increased my printing speed with the help of resonance compensation and linear advance...
I've never used Orca; hopefully there's nothing I'm using in Cura that I would miss... I guess I could also try post-processing it myself, or see if there's a way to add post-processing directly to klipper, now that I'm running it.
I was curious if linear advance could fix this issue on its own, but you experience seems to indicate otherwise.
@@ALittleSlowest It helped a little, but didn't solve it outright. It was better at preventing over-extrusion on corners, resulting in actual 90° corners instead of columns protruding from the corner.
Looks like I'm going to try orca slicer again 😂
that's pretty cool but shouldn't pressure advance/Linear advance take care of this?
Not really, no, that just changes when the extrusion pulses are given, advancing them by a certain pressure dependent time, to account for pressure lag in the hotend. If you have inconsistent extrusion depending on speed, that fast prints extrude less than slow ones, there's nothing it can do about that, it only corrects the transition from fast to slow extrusion or vice versa.
But something still rubs me wrong about this. It's a good first step to figure out a more pointed more physically based solution though.
I think the same techniques ..
will be helpful to control temperature..
when printing into small areas then large areas
I can see the filament .. being cooked to much- comes out black etc.
This is great, I know you focused on the top layer, but I find my printers could realllyyy use this on the first layer with sharp corners. Every now and then it creates that lumpy texture and there’s a chance that it touches the side of the nozzle on the next movements and drags the print away.
Didn't know what this feature was for, but I'll definitely be enabling it, even if I don't play around with it a bunch.
I just wish that wherever the tiny white gap filling poop command came from would disappear..it ends up just making the surfaces look worse then having tiny gaps, but despite looking over every single variable feild, I can't figure out which variable turns that off..
that Nth degree of tuning further effects clumping of fiber filed filaments. loved it, ...then turned, when you mentioned it performs these adjustments thought the file. After the first layer, except perhaps zits and holes 'thoughout' the printing, why would you not just use ironing on that single last top layer? (much quicker)
...perhaps I'll stack processes. :-)
Hope this gets merged with upstream Prusa Slicer too
Hope this will become standard in Slicers
I hope it doesn't and we figure out a better solution instead. But it is good to have in the interim.
I love when we can improve the product with simple math and code
The hardware is already very capable so now the focus falls on perfecting the slicers
when this was in beta, this had an issue with my printer using CAN, the flow changes were so detailed and gradient, it would cap out the CAN line data, because of the multiple different extruder commands, i will try again and see if that has been fixed
👌
Some should work on extrusion releated to speed. Intead of constant value
That CR10Max doesn't look like mine. And I have never gotten mine to work. Could you make a video going over your modifications and tuning of your machine? Did you change the main board? If you did, how did you handle the proprietary ribbon connector? Any ways, love your channel, keep up the great work!
Note that the copy paste feature of the spreadsheet may use localized separators for values, not the US format. This causes a real mess with the result, since many languages uses the ',' instead of the decimal '.'. I don't know how locale settings affects this parameter in Orca. Supposedly US English, because initial values uses it. Regardless from what values I write (or locale) I don't see any changes in the flow view.
Would you say different colors of the same material type would need different values?
Is it useful with a Bambu Lab A1/ A1 mini? There is a sensor in the toolhead for pressure compensation.
Waiting for an option to reduce the flow rate before reaching the edges...like the coasting option in Simplify3d
is there any High Speed Auto Bed Leveling project, something like Eddy/Beacon??
Could you share your Orca profile for the CR10MAX?
Ok so I've been watching some FDM printing videos just for the sake of learning since I've only been resin printing and I'm just astounded each time how fucky FDM printing is. Like, this video was just posted today but this change of flow rate problem seems like something that should have been instantly recognized and fixed like YEARS ago, back when this modern 3D printing first started. How the hell was something so fundamental ignored?
That's a really cool contribution, however it feels like its a high level workaround for something that needs more like a low level solution.
Wouldn't an accurate model for pressure advance address pretty much this already?
Awesome. The are flow compensation model looks like a logarithmic curve. I created a geogebra to generate one. Let me know if you want the link to it. YT is quite picky with links in comments :|
where can i get that fan duct model at 6:00 ?
I've had similar problems but it's not exactly small areas, but rather the edges of areas (like shown around the borders and middle holes at 1:29 ). SAFC didn't help when I tried it with the OrcaSlicer beta a while ago. Increasing pressure advance helps a fair amount but that produces underextrusion in other areas. Any ideas?
Anyone knows how to get this one on Cura?
is there a limit on the number of points on the graph? and do they need to be spaced equally? like having more data points at shorter extrusion lengths where the difference is bigger
I was just talking about how tuning the extruder while not under load will not give optimal result. And with 35 years of process control experience, I have seen that not all autotune algorithms are the same. Some just do not produce good real world results.
This is a topic I have been rolling around in my brain. How do you put the extruder under a normal load without dumping copious quantities of filament into the air?
Highly interesting! 🎉 On the Bambu X1C the issue is the opposite : large areas (~200mm+ ) the layer infill creates a micro mesh [using default settings]. The layers are not watertight and water leaks (after hours or days)!
Does this feature resolve this underextrusion issue?
It should, if you increase the general extrusion multiplier.
The flow at _full hotend flowrate_ will be less than was asked for. Less speed, more heat or better hotend should help 😉
It could also be quite difficult to achieve watertightness without coating the object 🤔
@@kimmotoivanen Thanks for your reply. I meant: by using the default BambuLab settings. I was not meaning: by pushing to the limits. Sorry for the ambiguity!
Also, I did achieve watertightness by increasing the flowrate. The default one is ok on short distances (under 100mm or so), but not on longer distances; based on my experience.
Plus, by using a magnifier, space can be noticed between the nozzle passes (while using a clean nozzle and the default Bambu settings).
@@LigneDesign I get it 👍
I think Marlin FW had pull request for compensating extruder "slip" at high flow rates.
PrusaSlicer has PR #12186 and SuperSlicer has issue #3854 for similar compensation. So it may become possibility in near future 😉
Fast printing is not as simple to calibrate that _good old_ 40 mm/s everywhere 😅
I wish Prusa Slicer would adopt features at the speed at which Orca Slicer does.
I am just not able to tune my printer as well in Orca as I did in Prusa for some reason (even if the settings are all the same).
I wish PS would add the calibration tests of the others, and I wish orcaslicer would add support for multi extrusion.
@@TeachingTech Yes, the only reason I want to use OrcaSlicer are the calibration test because it makes it so fast and easy to tune for a new filament, nozzle, extruder, etc.
Is this going to be available for cura as a plug-in yet? Is it only orca and should I switch to orca
Can't wait until this is implemented into Cura
I never had that problem in cura.
But I use pressure advance (klipper) and used linear advance before (marlin).