ฝัง
- เผยแพร่เมื่อ 28 เม.ย. 2024
- Previous extrusion force testing highlighted some ripples on the graphs that needed some further investigation so that's what I'm doing today; investigating the origins of extrusion force ripples.
Patreon: / vector3d
Twitter: @Adam_V3D
Join the Vector 3D Discord: / discord
Vector 3D Shop: vector3d.shop
Tools:
Wera Hex Keys: amzn.to/3gIYdNV
Wera Ratchet and Socket Set: https: amzn.to/3gH0OrY
Knipex Cutter: amzn.to/2Wylf2M
Ifixit Tool kit: amzn.to/3BdJXES
Wire Stripper Tool: amzn.to/3kC7hoW
3M Applicator Tool: amzn.to/3jqc8ud
0:00 Are Clone Extruders Ruining Your 3D Prints?
0:50 Can we Find a Pattern?
3:16 What could be the cause?
5:24 Inspecting the Dual Drive Gears - วิทยาศาสตร์และเทคโนโลยี
MihaiDesigns has a similar video discussing the same problem and some fixes, he derived the length and found the problem a little differently than you but came to the same conclusion. Titled "What I've learned about dual gear extruders and how to patch them, maybe". Interesting problem, his assessment was it was the inherit flaw of dual drive system.
I came here to say just that, thank you.
Besides the ripple, there's also layer artifacts (layer width slight inconsistency) under extreme harsh light (parallel to the vertical print direction). I also see this on prints coming off a Fortus 400mc but less prominent. I suppose this is just from very small variation in extrusion and retraction from pressure build up.
@@foureight84 @foureight84 PID tune heat bed. I heard that 0.1 degree Celcius fluctuation of the heated bed can contribute to the layer width inconsistency. I tried this myself and it helped but not eliminate all issue.
I often PID tune one temperature and use that for all ABS filament. Of course, thank to klipper's config, it makes saving multi-config so easy.
Though, my PID tuned hotend isn't that great. I often see it fluctuate around 0.4 degree Celcius even though I PID tune it a few times.
A potential solution is to have the idler which does not moves away from the main gear in a parallel fashion (as in every dual gear system I have seen), but which "tilts" away, using the its gear-end as pivoting point. This would allow for a better (not perfect) meshing of the teeth in different use conditions. Or the teeth could be replaced by hard rubber, maybe.
@@nhozdien5058 I've seen people comment that Klippers PID tune is not that great, I've never messed with it but my duet keeps the bed at the target temp within 0.1 degree Celcius. Building a klipper voron so I guess I'll find out
You could try the latest CW2(SteathBurner beta), they added a screw to limit how far the 2 drive gears get close together, I feel like adjusting this could change the data considerably.
2:23 Déjà vu
I swear that wasn't there before 😅 oh well, these things happen.
@@Vector3DP glitch in the matrix
Can’t wait to see more results of this. Very happy to see the VFA/salmon/woodgrain issues we’ve known about for a long time starting to get more mainstream attention and thorough testing. Of all the methods I’ve seen, your load cell setup is by far the best in terms of test iteration time and accuracy/repeatability. I don’t think anyone else has been able to actually measure the magnitude of the inconsistency.
Great Job! I hope you continue researching the issues you have found and can propose some solutions to address this problem!
Exactly what I predicted in your last video! I made a test.. printed an object where the extruded amount per layer is extremely close to one rotation of the extruder gear.. you can clearly see the variation in extrusion width wandering up the object and count the teeth on it. Transparent filament is working very good for this. After seeing this, I really think this is a substantial problem we have with fdm printing . I´m working on a sensor system to make this visible and later correcting while printing. Or closed loop extrusion in other words. This will take some time, but I did my research, pcb and hardware design, building the first prototype in a few days. When I have something working I´ll send some out for research, maybe you´re also interested? Keep on with this series, we need to have more data on this stuff:)
'After seeing this, I really think this is a substantial problem we have with fdm printing . I´m working on a sensor system to make this visible and later correcting while printing. '
Just to save you some trouble, all you really need to do is keep the drive gears fully engaged at all times. IOW don't let minor variations in filament diameter push the exactly parallel shafts apart. Push harder on the bottom to keep the drive gears together, and let the top float a bit to take up filament diameter differences. The tiny change in angle won't matter.
May still be some artifacts anyway, the teeth are relatively large for the drive diameter. But the variation in pattern from changing filament diameter is from changing the drive gear engagement. Just press them together and stop over constraining the shafts to be exactly parallel is all that's needed.
If you have one affected by this, simply open it, heat up the idler arm a bit, and twist toward closing the drive gears a degree or few and let it cool. A tiny bit of skew in favor of the drive gears engaging is probably the only thing needed for most of these..
Look at Adam Hollywood over here with all the cliffhangers!
Looking forward to the next one, i love this. Awesome stuff Adam!
There are a few other people coming to the same conclusion with different methods (the vertical ripples in vertical walls for instance). This is however the most "scientific" proof that gear backlash is causing this.
I would love to indeed see the comparison with the original BMG but also with the larger gear extruders like Orbiter and LGX with its plastic gears which might well dampen the backlash more effectively.
I've experienced this when i was developing remotely driven flexible shaft extruder a few years ago. I couldn't thought of a good way to test what was the root cause... your load cell extrusion force method combined with the ideas in this video just gave me a big inspiration on how to solve some issues i've experienced or at least find out what went wrong. good to see people are getting more scientific about consumer FDM 3d Printing....
@@jamesmccaul2945 haha i kind of stopped working on it... and yes you are right about the performance compromises which most flex shaft driven designs/products i've seen didn't address or even mention.
What a great video. I can’t wait to see more. I feel a lot of commenters don’t realize this is just a part of a bigger story.
Great work!
The extruder of my V0 was skipping a lot, until I ground off the teeth of the dual feeder gears. Much better now. Prusa probably has the best solution with the XL.
Absolutely stellar video! Perfect timing too(no pun intended) - currently trying to find an alternate extruder for the absolutely crap one that came with my new FLSUN SR. Looking forward to next video. 👍🏻👍🏻👍🏻
Time to make helical geared extruders.
How about a system like the Bondtech LGX? Would the larger dual gears affect this positively or negatively?
My initial thought is the tooth profile not quite being linear with rotation. Involute gears have a contact patch that moves in and out. Changing ratio slightly and sinusoidally?
Might be nice to try belted gearing as I'd imagine multi tooth contact would have an averaging effect.
I'm wondering if we csn fix this using different type of gear engagement. Maybe that's only the issue with simple gear?
had a deja vu at 2:21 (same clip 2x)
great video. I was just wondering if the clone that I ordered will handle my next build...
From what I've tested and seen on MihaiDesigns videos the dual gear extruders will always produce this pattern. It can be more or less visible but I couldn't get rid of it completely.
I replied to Vector 3D's Tweet about his previous video with a link to MihaiDesigns' video on topic. As MihaiDesigns points out, this problem is inherent to dual gear extruders. As long as the distance between the two gears is allowed to move a bit, this problem will persist. The extruder really needs to be redesigned to solve the problem.
@@ddegn Yes, you're right. This is a design problem. Bad poor quality clones can make it more obvious/visible.
It takes some energy and force to imbed a tooth into the filament. Do the ripples change in amplitude when the tension is changed?
I'm so glad you asked this. I've already tested, answers should be in the next video.
If the teeth bind in the plastic the gears mesh better because there is less space between them. My guess is that periodicity would likely not change that much.
Design Prototype Test did a video about the LGX extruders and why those larger wheels would exhibit less of exactly what you are seeing here. I'd like to see how the LGX extruders behave and if they have much lower magnitude ripples.
LGX didnt make inconsistent extrusion any better nor any worse for me. I'm keen to find a Voron compatible single gear extruder to see if that does eliminate the issue.
this reminds me of the issue that Stephan (CNC Kitchen) had with a single hobbed gear extruder where the center hole was offset (i.e. NOT in the center of the gear). All the talk of backlash (???) doesn't make sense, nor does the synch of the teeth. However, the difference when offset, means the gear actually has a higher speed at the "circumference" because the different places are closer or further away from the center. Different "speed", means different force!
So, is the center hole (on all the gears, including stepper) actually in the center of the gear????
Hi, very nice video, super interesting investigation! Sorry for the offtopic question 😅😅: Could you share the file of the Little 3D printed exteuder gears? Appreciate!
same, i definitely want those extruder gears as a desk toy
Had issues on my Voron 0.1 and replaced the main gear + shaft assembly from the BMG one to a cloned version and my issues were almost gone. I assume my main gear is slightly off-center and was causing these issues
I had exactly the same experience but in my case fake BMG gears kit solved wood grain inconsistency completely, compare to original BMG gears.
I had a very cheap clone with a roller bearing on the idler that failed and gave me very obvious artifacts. Changed the bearing and shaft and it's good. I think one of the ballrace bearings is getting a little notchy on a clockwork extruder and it's doing similar.
But with good bearings the difference between cheap 1.8 degree steppers and e3d/motech 0.9 degree steppers in my XY is much more apparent than extruder induced artifacts. Specifically moving one axis whist the other is stationary on a weak microstep can give some jitter thats visible with shiny filament in some lighting.
You should also look into StealthChop and how it affects stepper motor inputs. When you turn it off on the extruder driver, you get much more stable stepper inputs, and better prints.
Use of helical/double helical gears should mitigate this problem ensuring smooth engagment of extruder gears. These gears are allot more difficult to produce though
I agree. I think even 2 single helical gears would help the issue a lot, though this is just a hypothesis.
I think you would get the same result with genuine bondtech gears. I would like to see the results from a genuine titan, any chance?
I'm looking forward to the followup video. Hanging in suspension....
...and now you've opened a can of worms. Great analysis.
So when do the LGX and ProperPrinting belt drive get tested? (and the new screw drive extruder)
Gr8 job m8!
Unfortunately, but this problem occurs on all dual gear extruders. It does not matter whether it is a cheap chinese one or original BMG. Issue manifests itself as unevenly printed layers. On small prints it can be easily mistaken as "z banding", on larger ones horizontal banding lines occurs slightly at an angle. That's the reason why I bought OMG V2s.
Let's settle this once and for all, dual gear extruders are just bad.
Interesting take. So should a LGX perform better?
@@NathanBuildsRobots I don't own one, so I can't be 100% sure about this, but I think it will probably be better. It is still a dual gear extruder, just with bigger gears, so the issue might be still there. IMO stick with single geared.
You sent me searching for this thing. Serialized single-gear feeders? Makes me wonder if anyone has tried to use TWO synchronized small steppers, set face-to-face, so that both sides of the filament are guided under force, but not mechanically geared together.
@@GnuReligion That's an interesting idea, but let's stick to the KISS rule and make it simpler! What do you think about individually driving both extrusion gears from the same stepper, like in V2s?
@@real-insan3 I like the extra traction from dual gear though. I think it should be less of an issue with larger gears too.
Sooo...when do we get the answer ;) :D looking forward to your next video on this topic!
I've actually put a lot of thought into this one. My prediction is that a filament test is in your future. The gear teeth would have to be very small if you don't want this artifact, to the point of being useless. My intuition says that it's the filament hardness (compressive and shear strength) that is the cause. If you were to use a different filament, you should get a different result. Temperature will make a difference of course, but since you have your fancy filament-force-o-meter, which I nominate for its name before you do the obvious next thing, it should be a breeze.
I almost forgot. My prediction is that the softer the filament, the less of an artifact because the teeth are really in there, and the radius of movement? is smaller. This could be exaggerated and compensated by adjusting your spring tension
Another prediction is that smaller diameter hobbed gears will have a more noticeable artifact, than larger diameter hobbed gears. More teeth to do bites
My logic support:
In civil engineering you calculate the volume of asphalt for a road by multiplying the centerline length by the width, by the thickness. You don't have to account for curves because the centerline is in the middle of the road.
I used to think the line was on the outside surface of the filament. It is not. I feel like you want the average radius of the gears to be as close as you can to the radius of the filament.
@2:22 and @2:29 I got a dejavue 😂
When will the video where you test all the Extruders come? 😊
At the moment I feel that i'm still working out the right questions to ask. When I know the questions, then I can research the answers.
What if the filament is softer or harder?
Uneven layers will be more noticeable with harder materials when this issue occurs.
So... what are the results of testing the original?
Instead of having 2 drive gears mesh, why can't it be designed with a worm gear between the two drive gears ?
i wonder is a worm drive gear wont have that problem? if not be better as at all times the teeth be in contact at any point
Helical gears with their 'infinite' mesh should surely solve this issue
You cannot just end a video like that... :)
Look at Prober Printing ( or something like that) a Dutch guys take on an extruder.
Also an extruder like the VDE would be the future
I don't understand the tie-in with the statement that clone extruders are bad as you state in the title. If they are bad, as you claim, I would like to see the test data that bears that out.
You have a great test rig - use it!
Pretty sure that will be his next video.
Solution?
I've got about 6 of these in rotation. And to be honest, they are pretty trash unsurprisingly lol
After a few hundred hours of printing the roller bearings die and everything goes to hell. I've had much more consistent luck with the even cheaper titan clones being a simpler design.
Been wanting an authentic bmg, but eh the price is tough when you have 24 machines lol
This is called periodic error, I told you that was the cause in your first video. It will exist in all machines that have rotating parts. In theory it could be calibrated out in software. This is an issue that astronomers deal with with their star tracking mounts and it can be removed in software but they also use a process called guiding where they track the position of specific stars which also removes it.
Your E3D extruder will have some periodic error, as will your stepper motor. the error you see in the extrusion will be the sum of all of the errors in the devices in the chain. It's really dishonest to try to put the blame on chinese clones as its simply not possible to remove via machining alone.
What a nasty comment to leave, I hope your ok. Allow me to address your concerns.
Firstly, not all of the problem is the periodic teeth there are other features of the extrusion that are not periodic. Also, getting some information from random youtube comments is not how I do my research.
Also, I have not blamed this on anyone and i'm not being dishonest. I asked the question in the title, isolated where the issue comes from and concluded more investigation is needed before deciding if its clones or a design issue.
I feel you've really missed the point here regarding the issue too. The existence of the error, while not ideal, is not the main problem. The magnitude of it is. Good design and manufacture will improve it, I know this, because I've tested it.
An 8min video, but the source of the problem is only mentioned in the last 30 seconds... and it's not even explained why that is
WTF! You're gonna leave us hangin'!!!
Unsubscribed!!!!
jk ;o)
it's not pitch, its diametral pitch.
Conclusion: Cheap gears are cheap.