The intertwined waves pattern in 3D printing

Part 3 of my experiments with 3D printer extrusion issues that have been bugging me for months. This intertwined waves pattern is very faint and only shows up clearly once your printer can reliably print within 10-20 microns of accuracy, so if you got yours this far, it's quite an achievement.
Extruder test models (print pattern number 8) mihaid.com/extrusion-test
Chapters:
0:00 Intro
0:12 The intertwined waves pattern
2:11 Solving the waves pattern
Part 1 of this series • What I've learned abou...
Part 2 on the wood grain pattern • What I've learned abou...
Patreon / mihaidesigns
Buy me a coffee mihaid.com/coffee
Instagram: / mihaidesigns
mihaidesigns.com/​
#mihaidesigns

[@Vez3D]

Just print fast 😉 it prints better fast hehe

[@joshua43214]

I know. I almost started coughing when he said go faster than 40mm/s. I call this Prusa Speed lol

[@DavidGarcia-nx2gj]

@@joshua43214 weird but I've always print at 80 m/s and have never seen those patterns or maybe is because i'm using a bowden system

[@ameliabuns4058]

gotta go fast *sonic noises*

[@RubixB0y]

Seeing this response really made my day 😂 So on brand

[@JTskates]

Helical gears might solve these texture problems. They have a more consistent tooth engagement which makes for smoother transmission of power. It might help to use these instead of straight gears for dual gear extruders. Just a thought…

[@daveansell1970]

If they were helical enough it would also average out the bumps going into the heater as the bumps would be angled.

[@williamaycock579]

If toothed drive gears are the issue, the only thing I can think of to remove that would be rubber or some other type of material with a high coefficient of static friction covering the drive wheel. If it's just smooth there'd be no grip to push. Definitely a nice puzzle to ponder. If only I had the resources to test. Ah well, I guess that's why I'm subscribed to your channel!

[@JJFX-]

I think any smooth surface material would become more susceptible to build up between it and the filament unless there was a self-cleaning solution in place. Perhaps a very fine knurled pattern would be an improvement but the finer the pattern the faster it would need to be cleaned as well. It would be interesting to test with rough surfaces like that of sand paper, a diamond file, etc...

[@quinnobi42]

I feel like i have come across such an extruder that used a drive wheel with rubber flaps or something like that. Unfortunately, I don't remember the name of that extruder.

[@SkorghDk]

@@quinnobi42 Flex Wheel Extruder? reprap.org/forum/read.php?94,840959 . I've also seen a few designs using the smooth, rubber side of a GT2 belt running over a large diameter idler to "sandwich" the filament and push it. Both solutions are quite bulky and need gearing, which will likely introduce some sort of cogging, which again would lead to periodic variations in extrusion flow.

[@olafmarzocchi6194]

You could also use toothed gears with a bigger diameter, so that several teeth grip at the same time.

[@notsam498]

@@SkorghDk this is exactly what I was thinking.! I think though we'd need a costum belt to make it work well. Mainly the contact surface with the filament would probably need to be mildly abrasive. Possibly just rigid segments... That however as far as I know doesn't exist. With a very small belt that has a surface like that I'm guessing the size could be reduced, but it would always be bigger than gears.

[@imGeeyo]

Can you please explain how to solve rollers problems?? Thank you

[@pirobot668beta]

Think of it like a blacksmith working on a metal rod; hammer blows make the rod thinner and longer! Each tooth is another hammer blow. Gear-tooth cold-forging (nibbling!) reduces the cross-section while increasing the length. The gear movement could be be perfectly smooth, but the 'nibbling' effect is still with us. As you have pointed out, it is a small and subtle thing. Bowden's see less of the effect; the 'springiness' of the long tube absorbs some to the length-wise pulsations. Direct drive, the pulsations in length are passed on to the nozzle, making pressure spikes. Two possible fixes: 1) Subtle changes in feed rate, timed to match the teeth of the hobbed gear! Call it "Nibble Compensation" or something cute. Maybe fixed in firmware? 2) Planetary screw feed to replace gears in extruder. Rotating screws in a rotating cage that orbits the filament, driving it onward. But unlike gear teeth, the screw-teeth make a spiral impression around the filament. This impression depth is much less than with hobbed gears, due to the larger contact area of the screws. Step-wise gear nibbling is replaced by a continuous faint spiral. No more bumps.

[@salesjunkie8309]

Option 2 like a fluid pump?

[@marsgizmo]

Amazing findings! 👏😎 Super valuable! I’m looking forward to some better gears on the market, curios who will do a better design for them.

[@MihaiDesigns]

You and me both!

[@mact.26]

We use a fortus 450mc for industrial prototyping. And every print taller than 0.400 inch gets these repeating patterns as well. Not nearly as severe as your examples, but we've always got surface finishes like that.

[@JJFX-]

Do the Stratasys gods basically just utilize the same style of extruder gears?

[@Tedlasman]

@@JJFX- afaik they do, but the use a bldc and encoder instead of a stepper

[@DavidGarcia-nx2gj]

@@Tedlasman so there's a market out there for better extrusion gears

[@ethaneveraldo]

I would love it if you could please partner with a manufacturer and make your own 3D printer for others to benefit from, you clearly are passionate about FDM printing so the Mihai 3D printer is going to be a beast that can only rock the market. We need a decent FDM printer that doesn’t cost half a million dollars. As for your question, an idea I was thinking of was to use an array of several silicone or rubber gears that would apply pressure and push the filament without damaging it. The array of multiple gears (maybe 3 or 4 counter facing gears, so 6 or 8 total) is to make sure that if one rubber gear slips, there are several backups, ensuring continuous flow. This concept could be tested with 3D printed TPU.

[@frankbauerful]

Interesting. Maybe a feature in the slicer to add some small random variation to the print speed of outer perimeters would be useful to improve the visual appeal.

[@Thomas_Lemmey]

Just a few variables come to mind that would need to be tested... Roller grip patterns (smooth, fine texture, toothed), Roller material (Metal, hard rubber, soft rubber), Roller spring pressure, print speeds, print temps, filament type. But i'm just happy when I get successful prints.

[@stratosdrix1091]

thank you for taking your time to figure this out for us.

[@jamespray]

I think the fact that bowden tubes mitigate the effect would tend to rule out the indentations on the filament as the primary cause (though I'm sure they do have some impact -- I've heard the E3D guys bring this up in talks with Tom Sanladerer, I think), because the indentations are still present at the meltzone regardless of the feed length. What the spring effect would mitigate would be variations in feed pressure, and we still need to look to the hob for that. The teeth of the hob are going to push unevenly on the filament as they rotate because they don't have a linear motion, plus they're deforming the plastic, so my money's on the hob profile itself; seems like finer teeth, herringbone teeth, knurling, or no teeth (soft surface?) would be the answer. Whether indentations or hob profile or both, though, I also think the viscosity of the molten plastic in the nozzle would have some damping effect as well -- which would increase with feed rate, consistent with the tests showing faster print speeds giving more even results. I've observed this on my machine, which can do single-wall prints more cleanly at 100mm/s+2400mm/s2 than at 50mm/s+1600mm/s2.

[@markthompson5983]

This is quite an interesting video, I run dual hotends that use dual gear extruders which leave the little tooth marks on my filament. E0 is direct drive with a short bowden tube, E1 is bowden and has a long bowden tube due to my 400mm X build volume. I've never noticed these kinds of patterns on my prints and I've printed very low wall speeds at low accel settings. I'll have to do some further testing on my machine because my vrollers are stock from when my printer could still be called an ender 3 pro.

[@FrenzyIncarnate]

Channel is doing great man.. Been here since the PitStop Extruder videos..

[@MihaiDesigns]

Much appreciated!

[@Jobobn1998]

Oh, wow, that actually makes a lot of sense to me! I think it's a very solid hypothesis.

[@genau14zeichen]

Interesting observations you have there.

[@medyk3D]

At this point Prusa should just hire you 😂

[@esotericbear9829]

I just had a thought. What about a quad gear extruder that uses softer tires to drive it? No chewing filament. The flexibility of rubber tires would put a more consistent pressure on the gears, taking up some of the backlash variability.

[@salesjunkie8309]

An orbiter of some sort?

[@YanickT]

Awesome time consuming work. Bravo

[@GeoffIX]

The angled hob teeth might need to have opposite angles for the two extruder gears to avoid inducing a twist in the filament...

[@komradekat2000]

At first I thought that the motion of the printer and how the slicer makes the g-code files was the problem for that happening on curved walls ( a curved line in g-code is just many points of a normal line , that makes it look curved ) , but yeah , I think the lines caused on the filament by the extruder are the problem . Thanks for the video , I will try that !

[@3DPrinterAcademy]

You should make a filament comparison video, and link the filaments with Amazon affiliate links. Great vid as usual! -Steven

[@Acestrex]

Nice job and thank you!👍

[@RaoulEvilD]

Now that you exposed your findings it seems to makes sense: the tiny teeth of the cog wheels combined with the spring pressure induces a cyclic alternating parasitic torque back to the extruder motor (like driving a bicycle with square wheels). Given that we generally use stepper motor, without position feedback, this torque creates an slight alternated over-shoot and undershoot of the intended position, therefore generating over and under extrusion. If true lighter gear pressure will reduce the amplitude of the defect. And "helical cog wheel" should almost get ride of it.

[@garagemonkeysan]

Great video! Excellent research. Mahalo for sharing! : )

[@GeoffIX]

Taking inspiration from your suggestion of bevelled/chevron gear teeth to align and reduce backlash the dual extruder gears, imagining what the area profile of the filament looks like as it passes through the hobbed gears may offer a viable modification. Imagining both hob teeth squeezing the filament together would reduce the cross-sectional area of the filament the most, and the space between hob teeth reducing the filament cross-section the least, perhaps having the hob teeth angled would spread the deformation over a greater length of filament and reduce the problem. Just my $0.02 :-)

[@Kyrazlan]

Another great video. I wonder how well a gear with a beadblasted like texture for the grip would work to mitigate this.

[@GeekDetour]

That’s it!

[@possiblyneil4978]

Huh, I thought from your first video you were talking about the drive gearing and not the hobbed gripping parts. That explanation still makes more sense to me because extrusion amount is a function of pressure. The smaller nozzle will increase the pressure. The length of filament in a Bowden will act as a spring to hold tension and keep pressure even for longer. Going faster requires higher pressure. Though I suppose those bumps in the filament would also cause higher pressure because there periodically more then less material What I found so compelling about the gears explanation was that it is difficult to transfer torque perfectly linearly using gears because of their profile and meshing. I'd love to see an experiment where the two hobbed gears are bound together with a belt and pulleys to see if that helps

[@MihaiDesigns]

It was the drive gear for the wood grain pattern in the previous 2 videos. This is after you solve all the other problems including the wood grain. I'll be testing more gears.

[@b-side.]

Good video, good explanation.

[@MihaiDesigns]

Glad it helped

[@eroc1944]

Interesting, never thought about the pulsing caused by the hobbed gear.

[@bett0diaz]

Awesome videos!!!

[@nathanbuster8428]

Okay so one of the major things you noted was the rollers biting into the filament, maybe if a different material was used that was not metal (I dont think nylon but maybe pc?) It may not bite like the metal does.

Thanks for the analysis! After installing a Printermods MDD kit and Trianglelabs bimetallic heatbreak to a CR-10, I was getting these patterns and they were quite prominent. According to your math the interval for me was 0.4mm. I changed hobbed gears, idler tensioner springs, nozzles, used a different 1.8 degree stepper (wish I had a 0.9), uncoupled the stock fan shroud from the gantry to reduce vibrations, increased speed by 50%...nothing. In the end I removed the bimetallic heatbreak and the surface finish of prints went back to being acceptable (although this problem in its exaggerated state allowed me to see that it was still there afterwards, although suuuuper subtle). This gives credibility to the assertion that Bowden systems buffer against this effect. I want to try a CHT nozzle to see if I can get back the increased flowrate the bimetallic heatbreak was giving me, or even if it could dampen these wood grain artifacts.

[@MihaiDesigns]

That's super interesting! Having played with some hotends to push them to the limit I also noticed some strange behavior where the extrusion flows in bursts rather than being uniform (for a simple uniform purge command). Not that it had anything to do with your observation, but that there are still things that happen inside the hotend that I don't quite understand. 0.4mm should be the distance between the teeth of the hobbed gears I believe? I still do get those. Does depend a lot on the filament though. They're always there, but barely visible with some filaments and very accentuated with others. I'm also curious about the CHT nozzle. Placed an order for it and I ended up ordering everything but the nozzle 😅.

​@@MihaiDesigns Thanks for replying! The stock Creality gear I have is one of the 11mm OD 40-tooth gears which would be ~0.86mm/tooth, so if there is some sort of reciprocating deformation (like compressing a straight object in a tube it might turn wavy) perhaps that could explain it. I'm sorry I don't have an alternative extruder to test. Your comment about "burst" movement kind of makes me get this hunch that each extrusion-hotend setup has its own signature, influenced by the sum of its internal friction. I get this image in my head of a kitchen blender with silicone feet that you're trying to pull across a countertop. Your arms/hands being the extrusion force, the blender does all these "micro hops" as you move it due to interactions between the silicone feet (soft, elastic) and granite (hard, rigid) happening in intervals as the static friction is overcome. It would be the analogy of the softened filament moving past metal heatbreaks/nozzle surfaces, I suppose. I am no master in tribology or fluid dynamics, just musing...but I am curious about the impact that machined surfaces would have like drilling vs EDM or what coating or plating methods could help or hinder the situation...perhaps the CHT with its higher surface area could make things worse! Anyway, hope you can get to test it and if anything notable happens that you'll share! Thanks again!

[@Kyrazlan]

I wonder if the stepper motor stepping patterns make a contribution to this. Definitely seems like a mechanical artifact though. More investigation and experimentation is needed!

[@alfredanil]

Ultimaker uses a knurled feeder roller with very fine rough surface, maybe this can solve the problem. Great investigation by the way, a real eye opener. Keep up the great work and all the best. Greetings from India.

[@photocromax]

Maybe a gcode post processor could slightly vary the speed of a long segment breaking it in different shorter segments. The final result will be a more fuzzy (maybe looking like a sparkle filament but without the clogging risk) but it should not have any pattern if the speed change is random enough. A similar trick is used with CNC spindles that continues to slightly vary the speed to avoid resonance (most of the time)

[@svenhansen9618]

Great find, I tested with a .4 nozzle and had the same result. At .3 layer, I could easily see the wave, at .2 less and at .1 no wave at all.

[@ibeat]

same was observed throughout the years in 3D printing, it was related to the use of micro stepping and s-curve acceleration or interpolation in the stepper driver itself, IIRC

[@girass]

Great video's, youtube algorithm worked for once! 😁 I'd be interested to see what effect the Bondtech LGX might have on the last few issues you've explored. 🤔

[@JonS]

Using alternating loop direction printing (where even layer layers go one direction, odd the opposite direction) might hide this a little. Craftware has this option (not sure about Prusa Slicer, Cura, etc.). I will give it a try soon.

[@salesjunkie8309]

Seems that faster print speeds streches the gnarled filament to even out the patterns when heated if we go with the theory of extrusion marks that effect the outcome

[@hapskie]

Sounds very plausible indeed. Would be cool if you could test this hypothesis by maybe raising the temperature a bit and use smooth rollers instead of teethed gear.

[@jeffm2787]

Can also be caused by stepper drivers, motors without enough micro stepping holding torque that snap into steps , etc. Bowden setup with a really powerful single gear extruder has worked well for me in the past. 0.9 deg large bowden stepper works wonders. Geared extruders of course add their own artifacts. Good reason not to go cheap on direct drive.

[@joman104]

I wonder if it is possable to make a duel real thats made of some type of high friction rubber?

[@robinte98]

I already said that in part 2 but what about making the gears larger so that the distance to the filament is higher and rotating the second gear around the one on the stepper until it squishes the filament? The gears would always be close together and every filament can be used. I don't know if I explained my idea clearly, but in my head it's a simple solution.

[@tallman2186]

Fascinating! Just thinking out loud... What if there is a rubber band around each of the extruder gears where it touches the filament? Similar to a baseball pitching machine. Pitching machines have two counter rotating "tires" and the ball is grabbed and shot out by the spinning tires. I realize it's a tiny rubber band. However, possibly no marks on the filament???

[@KLP99]

I'm switching to direct drive on both printers. I'm also switching out mainboards and trying new firmware from the manufacturer. Once they are back together and I've finished the test prints for the manufacturer, I'm going to print a few things and see what happens. Oh, and what about the surface of cylinders? Have you tested that? Different sizes? Your videos are always informational gems. Thank you!

[@MihaiDesigns]

I do include a few cylinders as models to show the same patterns but I've only got a few sizes so this particular pattern could come out very squished.

[@JoeMalovich]

I remember screw extruders being a thing, do you think they would have a similar problem? What about larger diameter wheels?

[@MihaiDesigns]

These could both help. I don't have them to run tests myself.

[@KeithSachs]

The ascender extruder I was using doesn't appear to have this problem either and it is not a Bowden extruder. I'm not super convinced this is an issue with ALL bondtech style gears

[@MihaiDesigns]

That could be true. More data could help, even though most printers have more serious problems than this, so just a part of the data received is useful for this test.

[@H34...]

I remember "salmon skin" and the "moire" extrusion patterns similar(?) to this being a hot topic a while ago, some people blamed it on one of the stepper driver types, but I can't remember which. Does it happen to align to the step/microstep of the extruder motor?

[@thanhavictus]

3:18 suggestions for wax like filaments? Will any "natural" PLA do?

[@MihaiDesigns]

I guess, as long as it's a bit matte.

[@johnthomasjacobs]

I wonder if it’s because your using a Bowden feed extruded instead of a direct drive extruder. I recently got a ender 3 s-1 with the sprite extruder and I haven’t noticed anything like that. I’ll have to double check though, I wasn’t really looking for waves in my prints lol

[@jimbob1er]

I believe it has to do with the profile of the teeth. Distance to the center of rotation is smaller at the center than at the edges, of course it slip at every teeth. Straight and sharp teeth should help. Larger diameter of the hub gear also, because it will increase the number of tooth in contact with the filament with less variation than small radius. There is also another pattern, the coaxiality of the gear hole with the motor shaft: the set screw push it on one side to compensate the play and lock. (for this too a larger dia compensate better)

[@alexanderrobinson3219]

Hey Mihai, you haven't posted in a while and missing your content - hope everything's good keen to see what you do next :)

[@wuzoe]

It would be interesting to see if a drive gear with knurling (little diamond-shaped teeth) would be better than the horizontal cuts that are common. Also, someone could modify the firmware to add noise to the extrusion to cover up any small periodic defects. That would probably be more helpful than using lower-quality filament.

[@xxportalxx.]

Or perhaps using something like a Dremel grinding wheel with a shallow groove cut into it

[@glabifrons]

@MihaiDesigns The gripping teeth causing this does make sense as it is elongating the filament as the teeth are going into it. As each tooth goes in, a slight elongation occurs. That said, the angle at which those teeth are going in can be a factor. Do you have any data/feedback from users who used the larger (8mm ID/12mm OD) dual-gears, like those used in the Orbiter extruder? The larger radius of those gears means that not only are more teeth engaged with the filament, but the angle of attack of the teeth engaging the filament is less. Another thought is: Are the extruders simply adjusted to grip too tightly? Does loosening them to the point just before slippage help? It should. This also would be partially mitigated by the larger radius gears, as the same pressure over a larger surface area would mean less distortion of the filament. FYI: I thought for certain, from your last video, that this was caused by the jerky meshing of the teeth in the meshing teeth. I've noticed that if I have mine cranked tight, simply rolling the gear with your fingers with no filament gives a tactile feedback that has very strong steps to it.

[@SianaGearz]

I can rebuild my printer between 8mm and 12mm direct extrusion but I want to get a 0.9° stepper as well, currently 1.8° but it doesn't make a lick of sense. I'll let you lot know how it all turns out.

[@MihaiDesigns]

I wonder the same thing. Only one sample from an Orbiter, but they forgot to slow down the print 🙁 I haven't noticed any difference when varying the grub screw tightness, but that was back when I was focusing on the wood grain pattern. Perhaps I should run more tests now that my extruder are better tuned.

[@Plumpkatt1]

I have been researching how to fix this problem for a long time. Most people don't know what I'm talking about lol. Anyhow I've been experimenting with how tight the tensioner level is versus how much bytes into the filament. I find it better with the tensioner looser but then there's more chance of slippage.

[@MisterkeTube]

Would it not help if the teeth of a dual-geared extruder would not be opposite of each other, but at 180 degree phase offset ... That would mean you still need a dual-geared extruder while fixing the previous problem (which I think should be feasible by adding some extra gears in fixed position and have the two gears engaging with the filament then each driven by one of those 2 gears and mounted so that their shaft can rotate around the center of the gear driving each of them (and pulling those 2 axis together with a spring). It's so much easier to draw than to explain. Anyway, that would ensure all the gears are always interlocking at the same fixed distance and that should solve your first problem while retaining dual-drive. Then you would need 2 different Bondtech gears, with the actual filament ridges at a 180 degrees phase offset (or you somehow need to ensure your other gears ensure such a difference - although I haven't thought this through - perhaps the way the gears interact makes the phase between both Bondtech gears variable depending on the distance between the gears, i.e. the filament thickness). Anyway, just some quick brain-farts from my end ...

[@MihaiDesigns]

Interesting idea, needs some studying and testing. I haven't looked too closely into how the two hobbed gears align their teeth

[@seanday6127]

Wonder if you could design a little "give" between the extruder and hotend on your direct drive design. Maybe put a spring in there or some rubber that would dampen out the supplied pressure to the hot end.

[@JJFX-]

Apologies if you've addressed this but I'm curious, have you tried delving into Klipper? Resonance tuning isn't designed for this issue but I'm wondering if input shaping is a potential work around for issues as consistent as these.

[@MihaiDesigns]

I have no idea. I assume we'd need a way to "home" the extruder to know the offset of the pattern.

[@joshua43214]

Input shaper will not do anything for this, but pressure advance might. Pressure advance is very responsive to changing demand, so it might disrupt the pattern a bit by adding some noise. The clue is printing at a reasonable speed, not the silly 40 or 50mm/s Prusa people seem to always use.

[@JJFX-]

@@joshua43214 Currently it won't help, the extruder isn't even supported. It's probably not practical but I'm just wondering it could potentially be applied to extruder movements in a similar way to resonance tuning. I've tested with levels of pressure advance on and off and it doesn't make much of a difference. There MIGHT be small improvements near the edges but very minor.

[@JJFX-]

@@MihaiDesigns Good point, it's probably not practical but the wave pattern, at least on mine, is very small and surprisingly consistent so it would be an interesting test regardless. Each wave oscillation repeats every ~2-3mm horizontally, ~4-5mm vertically and aren't visibly intertwined.

[@JasonHartsoe]

how about at 1:50 we rubber coat it? Not the gears, just the extrusion area for the filament? Would that not help keep from digging in and maybe solve the pattern issue?

[@Shred72]

Flywheels have been used to smooth out power delivery. No idea if it would be possible to incorporate or practical but who knows.

[@Metalistforlife]

Could a pulley system (with no toothed gears or belt) work instead of gears ? Pulley + some kind friction drive ?

[@medyk3D]

Maybe this really faint wood grain pattern comes from the hobbed gear. The screw we use to tighten it to the motor shaft can set it a little bit offcenter. Similar issue Stefan has in his Cetus machine: kzbin.info/www/bejne/l2TSgqxsq5iAqKM

[@MihaiDesigns]

No, that by itself would produce a very very elongated pattern which would look like misaligned layers.

[@Architectpremsolanki]

try using flex wheel extruder. there are no gear in gripping the filament. and its was very much surface area of grip to filament.

[@sindrejl]

Hmm ive always had an assumption that the waves was caused by frequenze made by the DD unit causing the molten filament to have a small kind of ripple effect that will give theese artifacts / "uneven extrusion"

[@MihaiDesigns]

I'll have to run more tests to see for sure.

[@onewa712]

Wasn't that the "ringing" effect? Correct me if I'm wrong

[@sindrejl]

Well, the ringing effect is mostly caused by vibration of the entire unit because of the extra weight added by the DD unit. Ringing can often be cured by going slower with less accel and jerk.

[@wuzoe]

Stepper motors do step! That does seem like a likely culprit. Configuring the controllers for microstepping can result in smaller steps, but they are still steps, and even microstepping depends on the voltage available and torque required.

[@hillfortherstudios2757]

You can add recirculating ball screws as your drive system like a cnc mill. This is clearly a very expensive solution but it is far more ridgid and abosrbs the vibrations/counter forces.

[@Smaugarth]

Could a small o-ring be put in the theethed part of the gears so instead of theeth biting the filament its pushed by friction?

[@Tomaskom]

Seems like both issues, the spur gear teeth and the grip teeth impact, could be solved by the Nextruder on the Prusa XL. Single, very big gear, where many teeth engage the filament at once, meaning one less or more tooth engaging would have absolutely negligible impact on the filament movement.

[@MihaiDesigns]

It's what I'm hoping for as well. Can't wait to see what people think of it once it's out.

[@neur303]

Just to verify the issue, have you tried modifying the teeth? Something like grinding them to leave less indents or filling up the gaps in-between? Is it more pronounced with more pressure on the filament?

[@spock81]

I wonder if the Prusa Mk4 with the Nextruder has these artifacts

[@henninghoefer]

1:25 It can't really be the displaced filament reaching the melting zone - because then, bowden setups would be affected as well?

[@LameCorvette]

Was about to say this too

[@MihaiDesigns]

Oh, I see your point. Didn't think of that. On the other hand, there's a very limited number of samples with bowden setups showing no other issues. Something to be further investigated.

[@sindrejl]

@@MihaiDesigns i have printed alot with Ultimaker um2 and 3, and I always got the salmon skin on the prints there too. (what you call intertwined waves). But with the UM machines I always printed very slow

[@MisterDeets]

Very good point. I am wondering if these printers are all gear driven or have noisy stepper drivers pushing the filament. Could be the meshing of the drive gears or the noise from the stepper driver translating to imprecise and not very smooth prints. Bowden printers would hide THAT barely perceptible pulsing with the spring action quite well. I wonder if direct drive extruders with no gears and TMC drivers have the same issues?

[@joman104]

I likely haven't noticed these patterns because both my printers are Bowden and I print at 70mms. I feel that most people don't really get the most out of there printers by not trying higher speeds.

[@MihaiDesigns]

Yup, I do test print at high speeds as well to see how far I can push it and what happens when surface degrades.

[@joman104]

@@MihaiDesigns have you experimented on adjusting interior and outer wall speed to reduce the patterning? Perhaps if your make them quite different you could reduce the pattern. Almost like phase cancellation. Keep up the good content😎👍

[@DJ-Sellout]

If it's the dents themselves, then perhaps larger-diameter filiment would reduce the effect?

[@KaidoLP]

A cross type diagonal knurling might help. It is on average much closer to a consistent diameter als the axial knurling.

[@jobotheclown]

Could it also be the extruder stepper motor quantization errors?

[@MihaiDesigns]

I'd expect the pattern to be much much smaller in that case (using geared motor) I doubt I'd be able to see it. Doesn't mean it doesn't have an effect though.

[@AirsoftAbominations]

could you try covering the hobbed teeth with some very very small rubberbands or heat shrink, either way something super rubbery/grippy so its not biting on indexted teeth but instead on a rounded grippy surface, bonyus points its a non perminant modification

[@BadVoidShip]

A resin printer would have no problems printing some herring bone gears (hell I can do M3 threads on my Mars 2, and probably even M2 if I messed with the exposure times more). They wouldn't be metal but modern resins are fairly tough so it should be good enough for testing (maybe even for general use, but I'm not sure how much wear those gears actually get.). Could even look at replacing the teeth with some kind of a finer alternating grid pattern. Or maybe using some kind of a rubber coating instead of teeth (unless it would rub off to quickly)?

[@deathpony698]

What if the teeth biting the filament were out of sync so that the effect would get canceled out? Or would that just half the amplitude and double the frequency?

[@hardwareful]

Try to replace the cut teeth with knurling / herringbone pattern - same bite, but spread out along the length of the filament

[@willthepill251]

Interested i am good video I have came across.

[@techromancer6702]

if bowden hides it by acting as a spring, how about adding a spring-loaded toothless roller system before the extruder to pre-tension the filament?

[@3dtagtraum]

what about the public release of pitstop2?

[@ChiralSymmetry]

Semi-technical comment (you'll have to do some research on your own and/or watch the videos and think about it to understand): A few years ago, a 3d-printing company demoed a new type of extruder that avoided most of these problems. They said their extruder was "axial-symmetric" or "axi-symmetric". Instead of a pair of perpendicular grooves (like for BondTech geared extruders), it made a *helical* groove all the way around the filament. This means the cross-sectional area of the grooved filament is constant. (Sorry, for the life of me, I can't remember the company's name, and wasn't able to Google for it.) Their idea isn't mechanically new; it's based on the idea of "axial thread-rolling" devices used to manufacture large bolts. Video (actual thread rolling starts at 0:20 into the video): kzbin.info/www/bejne/b3zRZ6GOaNeWqa8 Fixed rod, spinning head: kzbin.info/www/bejne/eWTZhKx-n6ehhbM Spinning rod, fixed head: kzbin.info/www/bejne/d56xZ2SghK2LpLc Notes: I couldn't find a good KZbin video, but the idea is that you have three rollers with a thread profile that "roll" around a cylindrical workpiece. The three rollers have axes that are slightly tilted with respect to the central axis, and they have thread profiles. (Best to watch videos to see this.) Sometimes, the rollers have planetary gears so that they roll together, but sometimes they do not have gears. Sometimes the rod spins and the head is fixed, sometimes the rod is fixed and the head spins. For 3d-printing we would want the spinning head version, so that we don't twist the filament. A company, LMT Fette, used to have lots of really cool videos of this on KZbin, but I'm having trouble finding the company's KZbin channel. But other re-sellers have videos. www.lmtfettetools.com/catalogs/Thread%20Rolling_Turning%20Heads/2023%20Rollkopfkatalog%20deutsch%20englisch.pdf Of *course* the original purpose of these "axial thread rollers" is *threading* metal rods. However, they also have the property that they axially advance the rods they work on. So a 3d company (I forget the name) designed an extruder based on all this. Instead of having the filament "rotate" while it is advanced, they instead rotated the "threading head" so the filament wouldn't have to turn. I think they used a planetary-gear set-up with the rollers. Sadly, I think the use of this for a FDM extruder may have been patented by that 3d printing company. However, I wanted to point out that we may be able to use a helical thread in different ways, other than "thread rolling." For example, Bondtech gears make threads on the "left" and "right sides of the filament. If we add a second set of Bondtech gears, then we can make threads on the "front" and "back" as well. This would make threads that almost wrap-around the filament. In theory, this could be used to make a helical-thread all the way around the thread, although it wouldn't be completely uniform; it would have some "square-like" shape. However, it would be an approximation of a perfect-helical thread/groove. And it is likely to have a cross-sectional-area that is more uniform. If you like, add a third pair of bondtech gears in a hexagon arrangement for an even better approximation. Hope that makes sense.

[@MihaiDesigns]

Yes I remember that 3d printer extruder, I can't remember the name either, but there's been no news in over a year on that.

[@ChiralSymmetry]

Not directly related, but I find it interesting that the Bondtech LGX extruder does not use any tensioning springs for gripping the filament. It appears the LGX just sets a rigid distance between the rollers?

[@vazgeraldes]

Hi Mihai, Have you tried using 0.9 deg steppers on the extruders, these should provide a smaller level of amplitude on the patterns? Great posts by the way, loving the channel.

[@MihaiDesigns]

Yes I have 1.8, 0.9, and 3:1 reduced motor. Same result.

[@vazgeraldes]

@@MihaiDesigns Thanks, I imagined that if the stepper/reduction would have an effect it would be expressed in slight reduction in the amplitude of the pattern, modification in its wavelenth may not occurr because the pattern may be generated by a forcing which is a lower common multiple of the system (step, gear teeth,filament teeth). I am trying to move on from the 8mm outter diameter (OD) BMG gearing to the 12 mm OD BMG gearing on my extruders but your last post really made me thing about going back to Titan style extruders. Another sugestion (difficult to implement though) is to cut new filament teeth that have different multiples and therefore generate different patterns that would surelly validate your thesis.

[@vazgeraldes]

Different prime numbers on each side of the BMG gearing and teeth come to mind... you's still have multiples to contend though.

[@robbinkoot2154]

i'm using an orbiter extruder om my Saphire plus, and i dont have this problem. it's probably because of the bigger gears. i'm using Colorfabb Economy pla red.

[@alessandrogiuliani2915]

It's all great, but in the end what's the point of getting a perfect print for you? Do you do it because you want to achieve perfection even for fun? or do you need perfection for a type of parts?

[@MihaiDesigns]

I'm just a bit obsessed with getting perfect flat faces, because many of my projects look very mathematical and when I print them I see these patterns and I get annoyed.

[@alessandrogiuliani2915]

@@MihaiDesigns I understand, amazing

[@Raiscan]

what if you use some sort of soft/hard/? rubber washer to 'grip' the filament slightly without leaving a mark? Love the quest for perfect prints, you mad man!

[@simonmadsen3540]

First of all that could introduce slipping, BUT there're sensors for that exact issue. Secondly it might act as a spring (like the bowden setup) so you might be onto something :) I was thinking maybe a helically cut hobbed gear, although that would introduce a lateral force on the filament, which would require good filament constraining or a dual gear, with the hobs going opposite directions

[@luciusdante4254]

Have you ever tried belt driven extruders, it is said that there are nearly no wood gradient on surface

[@86abaile]

I wonder if the problem would go away if the knurled part were at 45 degrees, like a helical gear.

[@GrimmChristopher]

Mellow does offer BMG with helical gearing now

[@sed6]

It seems that most of the problems you discover and solve are on high end or modified printers. You should address that roller problem on the Ender 3, the most popular printer in the world.

[@MihaiDesigns]

Yes, it is the high end. I guess most people won't get here. I'm hoping to not have to dive into the roller problem, at least not for a while. These tests take a ton of time.

[@Shred72]

@@MihaiDesigns I have an ender 3 pro and definitely appreciate all of these tests and solutions because they are showing all of the factors that go into a print. It is all transferable knowledge.

[@justchill4886]

i will modify one extruder project from thingiverse the "powerFlower belted extruder" it uses mk8 hobbed gear but i will modify it to "lerdge mt" gear its belted but single drive also second gear which most uses bearing , i will use ender 3 stock extruder gear to increase grip so there is not a backlash or lag with gear system caused by teeth , i think it will be good.

[@meky0]

are u sure this occurs on 0.9 steppers with gears?

[@MegaMaking]

this is actually not caused by the dual gear... evidently, print speed will cause it shows up or disappear. you should take a look for my video explaining why the stepper motor and the stepper driver is the root cause.

[@muradmalik]

Try putting a little rubber band on the gear to grab the filament and maybe it won't cause the pattern?

[@greenveg42]

Perhaps a belt to drive the second pulley? You should be able to prototype that with standard parts atleast!

[@deimuadda1387]

How can it be that the pattern has obviously different distances from one another if the gear wheel is even? Or is that deceiving me and the distances are always the same?

[@Tarex_]

Wasn't this the kind of issue solved with TL-smoothers on 8 bit boards? But anyway it goes to show that FFF/FDM is inherently so flawed that if you use perfect filament, it still messes up, or as stefan @cnc-kitchen showed, just the bed heating and cooling by 1 degree in normal operation gives surface flaws

[@SianaGearz]

TL Smoother only works on DRV8825 high current driver, it compensates for its inherent engineering issue that this chip cannot emit any current at small drive voltages. The TL Smoother is a diode assembly which always consumes a fixed voltage at zero current, mitigating this issue, and allowing this chip to drive current into any phase angle. This issue of the chip is well known so it's not widely used. The most common chip is A4988 which doesn't have that much of a maximum drive current but is very well behaved, and dirt cheap too. Trinamic chips used on silent boards are also well behaved.

[@Tarex_]

@@SianaGearz have been using TMC 2208 and 2209 for a long time now, the only printer with A9908 drivers i had was my first i3 plywood knockoff, my CR10S got a new board because of the temperature issue, ever since all my printers have SKR boards and 2209s as soon as they arrived, so never had to use TL smoothers, the ripples just reminded of the fishscaling issues lots of people had

[@MihaiDesigns]

Yeah with Trinamic drivers there's no need for TL-smoothers. And indeed, I initially also thought that my filament quality was causing issues, but turned out to not be the case. Of course some filament can still make it look like the surface is smoother, but that's another story.

[@Tarex_]

@@MihaiDesigns yeah, ive tried quite a lot of different filaments and depending on the quality and consistency of the diameter it reflects quite well in the print, the firmness of the material plays a role, softer (ABS/ASA/PETG) can be shaped more by the extruder which evens out the diameter, carbon fiber in any combination makes for smooth prints because the filament is stiff and consistent. And direct drive and dual gear, unlike BMG and titan etc. Are affected by the heat of the stepper shaft heating the gear and melting the filament or expanding for more pressure.. and all within 1-10 degrees fluctuating

[@dragonskunkstudio7582]

What if the biting teeth were helicoidal, wouldn't that solve the problem?