As another commenter mentioned, it’s the result of having too much acceleration for the rigidity of the machine. The belts used on low-cost 3D printers are kinda springy, resulting in the attached components vibrating when suddenly accelerated. If you can’t upgrade the rigidity of your machine, the only other parameter you can adjust is to turn down the acceleration. Adjusting the speed may seem to help, but what’s going on is that the ringing is still present, but it’s just been compressed into a smaller area because the head isn’t moving as fast. Turning the acceleration down will actually reduce the root problem; not just mask the effect. BTW, this is why CNC milling machines and lathes use ballscrews for driving their axes - they are much more rigid than belts.
They also use ballscrews due to the fact that they have to go through many more forces linearly. 3D printers have very low force witch allows them to use belts to save weight and costs. But yes they are more rigid than belts.
CNC Machinist and design engineer here, you're correct about the ballscrews, linear guides are even better! But no budget 3d printer will have those on their machines, I've found that adding 4 tungsten weights to my Elegoo Neptune 3 dampens vibration quite a bit. Also setting the machine on a paving block helps quite a bit too.
Lowering the perimeter or wall acceleration settings in slicer has a huge effect on removing ghosting. Available at least in Slic3r (Print Settings -> Speed -> Acceleration control -> Perimaters) and Cura (Speed -> Print Acceleration -> Wall Accceleration -> Outer Wall Accceleration). I use 2000mm/s² as the default acceleration and have outer walls printed with 310mm/s² acceleration.
Thanks for sharing, Angus. I print lots of miniatures & terrain for tabletop gaming, and lowering speed is what has worked best for me. Will share with others who get ringing as well. Cheers!
A 4 year old video!!! still completely relevant today! I’m new to 3D printing, and Angus has all the answers... plus I love his challenges and test prints
You could also try adjusting the timing belt tension. The direction change isn't transferred as quickly to a loose belt, so the ghosting is at a lower frequency but more pronounced. A tighter belt will help transfer vibration to the frame faster and hence dampen it.
The belt tension can also make a big difference. If your 3D printer is belt driven you kinda have to find a good balance in belt tension. A loose belt dampens vibration quite well, but may result in higher tolerances. While a tight belt decreases the tolerances but also increases the springiness of the belt, potentially 'causing things to bounce back and forth a little after a sudden change in speed. Similar to a weight hanging on a rubber band when the other end of the rubber band is suddenly moved up or down slightly.
Thanks, you helped me confirm that careful assembly with strict tolerances will eliminate yet another 'common' problem. That and a double Z screw. Like others, I want to switch all belts to drive screws. Get them within 10,000/in in the XYZ and never have any of these common issues that stem from 'minimum engineered products'.
Just the other day I did a basic search for Ghosting tips. Everyone was arguing about settings. Then Angus swings in and explains things better. I'd rather go with things everyone can agree on and easily understand. Thanks Angus. You're such a good resource for people that just want to know without having to take a master class or sift through all the internet arguing
Ghosting can also be reduced by fixing your printer down to a rigid base. I screwed mine down to a 3/4" plywood base and saw an instant improvement. I still get ghosting when printing fast, but generally it's acceptable. Also it reduces the frequency I need to re-level my print bed. I'll look at improving the rigidity of the X-Z axis next with braces
It does distract. He's supposed to be a college graduate. "Graduated Sydney's University of Technology (UTS) in 2013 Bachelor of Industrial Design First class Honours"
Small correction: moment of inertia is an object's resistance to changes in angular acceleration. The moment of inertia is just a property that comes from the mass and geometry. It's not 'caused' by the motion, but is rather the property of the object that makes it more difficult to make quick changes in direction. The moment of inertia of the extruder is constant regardless of whether it is moving or not.
Cool. Small imperfections aren't usually a concern since my prints get post-processed but it's good to know exactly what's going on and ideas to fix it if necessary. Thanks for the info.
I thought the the same, then I became obsessed with having perfect prints at .2 layer height and only having to lightly do post processing on whats left of the start stop seam.
Also, if you change the orientation of your printed part by 90degrees in the slicer, this may help reduce ringing further after all hardware optimisations you can realistically achieve have been performed.
One tip: Putting your printer onto a foam board or put it on springs for vibration Isolation does not reduce the ghosting. It just changes the eigenfrequencies of the whole system so the ghosting will appear at different movement speeds / changes of speeds. Putting the printer on a vibration isolation only reduces the vibration that the printer gives to the surface it stands of. In general the printer on it's own will "wobble" more.
Stiffening and reducing mass both have the effect of raising the natural frequency, and dampening like with shock absorbers as you mention has the effect or reducing the amplitude but widening the range of that response. Likewise slowing down the print speed lowers the working frequencies of the system. The goal is to push the systems resonance response high enough that it's outside of the working range. This method can of course reduce resonance but it can also introduce new resonances that weren't there before. There is an alternative in that you can increase mass and attempt to push the resonance outside of the working frequencies by decreasing the resonance frequency substantially. This isn't necessarily a good fit for 3d printing but it does well with someapplications. Ultimately though no simple solution like this will do as well with as little as a control system in the firmware with proper resonance compensation. Traditionally for mechanical systems you use a a state observer (a low order Luenberger observer) to inform a feed forward controller which deliberately offsets the resonance peak from the control signal. This is part of why higher end 3d printers despite similar construction standards to other cheaper printers, can at times get better results. A lot of thought is given to system rigidity, resonance, and the control logic itself.
Moment of inertia is the resistance to changes in rotation and is a property of the geometry of an object. It can effect the rigidity of an object, but moment of inertia isn't changed by the movement of an object. It's momentum that is effected by velocity.
Are there any engineers watching that actually know what moment of inertia is and how to calculate it? If yes, then you probably thought: _"wtf is this guy talking about?"_ Moment of inertia is not caused by acceleration, it's the property of s structure that determines how easily it will deflect based on the force that's acting on it. And then there are 2 separate "moment of inertia"s, the first one and the second one. Or the mass moment of inertia and the area moment of inertia. In order to calculate at which frequency your printer will resonate you will need both as well as the material property.
Agree that Angus doesn't understand moments of inertia, and shouldn't have used the term in the video - but this small misinformation doesn't greatly detract from the overall content. The TLDR of his video is basically "it's a vibration problem, so can be rectified by the usual means". His explanation was somewhat lacking based on his knowledge of the field, as you noted. All in all I liked the video and your comments, but the second comment whilst humorous is a little harsh :P
Yeah, that made me double take. I think there was confusion with the common meaning of "moment" and the use of the word in physics. Understandable. People make similar mistakes with phrases like "begging the question" all the time.
Great video Angus! You're explaining some pretty advanced concepts in your video and know they are not always the easiest to distill down to just a few words or phrases. Just a subtle correction: Changes is acceleration and deceleration as well as speed DOES NOT result in moments of inertia. Moment of inertia is a geometric property. Nothing to do with the material or forces/accelerations acting on the object.
Moments of inertia are to rotation (not really, but kinda also bending a bit) what mass is to linear movement. That said, velocity, acceleration and any movement won't change the moment of inertia. However, higher accelerations cause bigger spikes in resonance vibration. Increasing inertias and masses is a good thing for the frame - as you correctly showed, but it's a bad thing on moving parts, since these "masses" store engery and vibrate as long as they do. So: inertia is mass and accelerating that mass leads to swinging masses, which causes trouble. So make things light and rigid (rigidity helps to shift vibration frequencies to so high bandwidths that they can be ignored).
one of the quick ways to sometimes get rid of ghosting is to change the orientation of your print and try to have the widest parts lowest when possible and have the print head print less and less per layer as it travels up. Doesn't always help, but some times it does.
I tried some aproaches with mediocre or no success: 1. I improved regiditty. This made the prints much more accurate but induced evan mor ringing. That is counter intuitive but logical because the slack I had before acted as an damper. 2. I enforced belt tension. This changed the pattern and helped a bit overall. Increasing the force of the steppers by changing them and cranking up the current did pity much the same as higher belttension. 3. I than tried shock absorber under the printer. The impact on the ringin was not noticeable, but it reduced the accuracy. 4. The next attempt will be stronger belts. From an physical perspective it is quite simple: if you put energy in the system it must has a way to come out again. This happens by dampening only. Higher regiditty only changes the frequency of the resonanz. This helps a little, because higher frequency means the existing damtening is applied more often. Realizing this in practice turned out to be much more complicated.
Here are some further suggestions for reducing ghosting. 1. Take into account the geometry of the printer. So, for example, on the Cetus, oscillations in the cantilevered arm contribute to ghosting, so try rotating the print 90 degrees. This will reduce ghosting on one surface (but increase it on the other) which may give a better result for your print. 2. Place the print on a different place on the print bed. So, for example, on the Cetus, moving the print to the left will mean the print is nearer to the cantilevered arms fixing point, so the magnitude of the oscillations in the arm will be smaller at this point. 3. Design your prints to reduce ghosting. In particular avoid sharp corners, if possible. So fillet your corners if they are not required to be sharp. This will also have the benefit of marginally reducing the print times (since the print head does not need to slow down as much going around corners).
One thing I never see mentioned and I see a lot of people doing is mounting the filament spool onto the top of the printer. Putting all that weight high up is just exaggerating the swaying that the printer is doing and increasing the ghosting.
Adding more rigidity will not solve it, just change the shape. Basically, the problem is that the system has a resonance frequency which gets excited by the G-code. Slower speed reduces the forces involved, and to some extent the acceleration - that's why it helps. More tension does essentially nothing. Preloading a spring does not change it's spring rate. Reducing mass increases the resonance frequency, which may help The real solution is jerk and acceleration. You have to view them as low pass filters. Basically, they will greatly reduce the forces acting on the frame during direction changes - essentially reducing print speed whenever changing direction. The issue is that properly tuning them is not trivial; of course you could just greatly reduce them and it would help, but it would slow down fine detail prints a lot...
Have you considered using a tuned mass damper fixed on the head to reduce vibrations? Such as a small weight mounted on a flexible wire or a spring for example. This is commonly used on tall buildings to mitigate the effect of wind or earthquakes.
This is what happens when you count on rigidity to keep things stable instead of feedback. And with rigidity, there's not much you can do, and it's a black art. Giving it a whack with a hammer might help, or it might make it worse. Same with attaching a weight or a spring. The advice to increase rigidity ... I'm not saying it can't work, but it will never fix it. And totally overdoing it might make things fall apart or worse: suffer high stress during operation and then suddenly fling a metal arm across the room.
Hello , I am happy to found your video for how to fix the problem on 3D printing , here I got a problem hope you can help , my problem is when I start printing every thing is OK , but after around 50% then the print head will cooling down , I have use a new set of heat sensor , but same problem got .so hope you can sharing your experience , thank you very much , regards John
Yeah, dampeners probably was a bit overboard. But belts are at least something that is pretty easy to swap and as far as I know these can make a big difference if the frame is kind of rigid and the belt is springy.
Ghosting can be eliminated 100% without changing acceleration and jerk with proper firmware/stepper drivers. There currently aren't any solutions for DIY 3D printers that I'm aware of, but there are many for commercial CNC machines. Basically the algorithms will anticipate when a change in direction is approaching and modify the instantaneous jerk and acceleration appropriately in order to eliminate the extra "spring"/slop upon completing that motion or change in direction. I'm sure Marlin will eventually implement something like this, although tuning for each machine individually would be tough which I'm assuming is why nothing like this has been implemented yet. An auto-tuning script would likely be needed which would require stepper drivers with accurate positioning feedback. Here's an example of this algorithm, not my video! th-cam.com/video/eUZVLMCmSxI/w-d-xo.html
exactly :) there are some works being done. constant acceleration and deceleration on most firmware hinders the ability to print at high acceleration/deceleration.
Angus I was actually going to ask you if you had a video like this and you went and made one before I had the chance to ask. I really appreciate the videos dude, thanks a bunch. :{J
really? I never knew that. I started watching these channels right around the time I got a printer, I don't know how else you could fall into that rabbit hole lol.
Interesting - I thought I'd had just about every 3D printing fault but I've never seen this one! Perhaps because I'm always printing in slow motion! Thanks, very interesting.
I tried your ghosting test on my TEVO Michelangelo. It came out nearly perfect, with only slight ghosting around the text and Makers Muse symbol. I was using WYZ white PLA 1.75mm filament. The Gcode had the default printer settings. It might help that my printer is on a sturdy wood desk.
To me, ghosting is somewhat of a "perfect storm" of issues. Frame rigidity, vibration absorption properties of the base, acceleration/jerk settings, and print speed. I would also add nozzle wear, printhead weight, drive belt tension, etc. Unless the ghosting results in an unusable part, it's really not a huge issue, especially if the part will be post-processed. As far as small details and such, if you want good, fine details, you really shouldn't be using an FDM printer in the first place. Great video overall. Still yearning for more Fusion and Meshmixer tutorials... ;)
Hello man your explaination about echo is correct but I understood that the main reason of ghosting is a movement trasversal to an axis over which the nozzle is moving. The eco is not caused by x or y themselves but by the bar at the top of z for 99%. Is a butterfly effect, where the z axes starts to vibrate trasversally to the x or y axes. The eco or lines u see fading out are the translation of the fading out of the vibration of z rod\brace. Because of that if u check it, the slot at the very top of the z rod has a bearing whose socket is very loose (for cr-10), this in order to let the rod vibrate without transmitting the vibration to the brace. So the issue is not the relationship between x and y but with x that turns into y while z vibrates. The question of mass is true but it cant be the main reason of ghosting here. U have not 30kg of mass to brake in 2cm that justify the inertia over the x and y movement relationship only. The z rod is the criminal here. Make a test and try to completely isolate z im order to fight ghostin 👍👍👍
Why not just go to your local sporting goods outlet and get stick on vibration dampners. they can be found any where compound bows are sold. they are soft latex shapes with a double sticky tape adhered to one side. Print out 4 of the muse sample, one every 90 degrees... and stick the damper in the direction of the ghosting... in the video example with the cardboard the damper should be stuck on the side of the bed the cardboard hung off. they also come with screws molded into them if you'd like to firmly mount them.
My ghosting problem with my creatility cr-10 s5 is from the absolutely massive weight of the bed. 500mm square aluminum plate with the frame underneath and same size glass plate, the heater and the aluminum plate for leveling screws. The whole thing probobly weighed over 10 lbs. I modified it to only weight about half as much.
Great video and cool desing! My problem right now is retraction. i cant find a way to fix underextrusion in parts like you're tolerance test. i have tried small amounts slow , fast, big amounts fast, slow etc. this leaves practically unuseable parts.
Printing slower is a quick fix, but also the brute force worst case for fixing bad acceleration and jerk settings: print slower when you don't need to during straight paths but possibly still cause resonances around sharp moves. Getting acceleration and jerk right can fix this without sacrificing straight-line speed A much better order to address ghosting would be: 1. Reduce moving mass to the extent you can 2. Improve rigidity, mounting, and static mass to the extent you're willing 3. At normal speed, tune in acceleration and jerk 4. There is no 4. Don't touch speed, ideally. It would be more fair to say: reduce speed only if you don't have the time/patience for #3. I think people are intimidated by acceleration and jerk, but let's face that unknown.
Would it be possible to use a normal 3D Printer in some sort of non-conductive oil (Like the stuff LinusTechTips used to dunk an entire PC into) And if it is possible at all, would it be able to prevent the need for putting Supports on your prints since the oil might provide the support it needs.
I'll be honest it's a way complex phenomenon and difficult to address completely but definitely something that way too many modern 3D Printers suffer from!
Sometimes just changing the orientation of the print on your bed helps. Just try and rotate your piece 45° around the z-axis if it has a square shape and flat faces. Of course, this does not help that much if your design has organic and rounded shapes. And about the acceleration and jerk settings, most firmwares allow you to adjust these settings via g-code commands. So with a bit of g-code hacking it's easy to make quick adjustments and find the best settings for your machine... (Check the M201, M204 and M205 G-Codes on reprap.org/wiki/G-code )
to fix my issues with blobs and ghosting i just lowered my jerk and accel settings. i worked my way down from stock. im not at home but i have done this on all of my printers and leave my print speed set to 100mm/s on the ft5. i believe i am set to 1100 and 8
Something I'm not seeing in the comments. You mention acceleration and jerk settings, I'm not sure how well known this fact is, but "jerk" is the derivative of acceleration, or how fast you're changing your acceleration. A piece of trivia worth knowing if you're going to start messing with said settings.
Also, side note, the derivative of "jerk" is "snap," then followed by "crackle" and "pop" because we're nerds. (thanks Tom Scott for teaching me this useless information)
Jerk and Acceleration solved my ghosting. Default Marlin has them set far far to high for most printers. I would say this is something to look at more than frame rigidity.
it's weird for me. i get nice outer walls on my prints (i print molds for casting), but when it prints the inner "bowl" part, i get some layers that stick out, and some that are ok. imagine printing a box with a half sphere cut out of the middle. layers 1-40 are fine, layers 41 through 59 stick out like a band, layers 60 through 82 are perfect, then 83-100 are sticking out like there is a band again. it really messes up my casting, since those details show up in the casted object.
+Maker's Muse what if I have a slight offset on the horizontal axis. Its shows some lines clearly more inwards or outwards. I've a Prusa Mk2, can you please tell me how to fix it? Edit:I'm not quite sure, but maybe at around 5:00, you can see this on your mk3 print as well. Some layers are clearly more outwards than others.
I´ve some questions regarding 3D printers, and I hope someone in here could help me out Can you print your own transformer or action figures with ANY 3D printer or do you need an especific type of printer? What´s the best material to use if you want solid toys: ABS or PLA? How solid is going to be the toy? Like Hasbro´s solid, like a Lego´s brick or is going to be like a really cheap plastic? Do you need to sand the pieces or they´re already smooth as soon as they leave the print? What´s the best software to use to create toys for 3D printing? How durable is going to be the toy? Is going to break as soon as it hits the floor? And what´s the best way to deal with tolerance´s issues?
I would call it resonace, but when I show our chief engineer a component we're trying to machine I tell him, Malc, I've got wobbly bits again! Rigidity is key. Hobby grade 3d printers are basically low tolerance machines, so good positional accuracy and smooth surfaces are mostly controlled by either better quality components and, or speed of X/Y moves. We get this in our milling machines once they are a few years old. So, effectively it's caused by wear in the lead screws and the CNC not being able to deal with it. Yes, slowing down the X or Y axes moves will usually solve it, but of course, time is money. It doesn't happen in Z to the best of my knowledge. . I hadn't read this before today. Quote from Wiki. ------------- The running speed for a leadscrew (or ball screw) is typically limited to, at most, 80% of the calculated critical speed. The critical speed is the speed that excites the natural frequency of the screw. For a steel leadscrew or steel ballscrew, the critical speed is approximately[17] N = ( 4.76 × 10 6 ) d r C L 2 {\displaystyle N={(4.76\times 10^{6})d_{r}C \over L^{2}}} N={(4.76\times 10^{6})d_{r}C \over L^{2}} where N = critical speed in RPM dr = smallest (root) diameter of the leadscrew in inches L = length between bearing supports in inches C = .36 for one end fixed, one end free C = 1.00 for both ends simple C = 1.47 for one end fixed, one end simple C = 2.23 for both ends fixed So if your machine, whatever it is, already had issues, ie: low grade parts or any flex in the chassis, any time you hit the critical speed of your lead screws, X or Y, all the other probs only make it worse.
i noticed some ghosting on a reprap model 3d printer. which happens when you put support between prints (overhang) this only happens on a few layers tough. the layer where the support layers start and where they end. the file is a really simple file with no text and less than 15 sides with 1 half circle.
hi I did a print of it no Ghosting at my normal print speed, so I up the print speed till I started getting ghosting,eg 15mm/s to 25mm/s so now print at 20mm/s with no ghosting, THANKS MAKERS MUSE
Ghosts in the machine. Gotta print this on the MK2 and Sigma to compare. My printers are so competitive, that why I can't leave them on overnight ... they fight lol
I downloaded this stl, sliced it in Ideamaker for my Flsun Q5. It is printing now but the print head is moving so slowly you have to touch it as to make sure it is printing. Is this test only for Cartesian style printers?
Nice video! But it's just inertia, not a moment in my ominion! A moment means it has something to do with a rotation which is not the case because the head only move (rapidly) in the x and y axis.
I "upgraded" my lower tray (the structural one below the heated one) on mine to a metal one and I guess it increased the weight because it ghosts horribly now. Need to change it back.
As another commenter mentioned, it’s the result of having too much acceleration for the rigidity of the machine. The belts used on low-cost 3D printers are kinda springy, resulting in the attached components vibrating when suddenly accelerated. If you can’t upgrade the rigidity of your machine, the only other parameter you can adjust is to turn down the acceleration. Adjusting the speed may seem to help, but what’s going on is that the ringing is still present, but it’s just been compressed into a smaller area because the head isn’t moving as fast. Turning the acceleration down will actually reduce the root problem; not just mask the effect.
BTW, this is why CNC milling machines and lathes use ballscrews for driving their axes - they are much more rigid than belts.
They also use ballscrews due to the fact that they have to go through many more forces linearly. 3D printers have very low force witch allows them to use belts to save weight and costs. But yes they are more rigid than belts.
And the tool head moves slower.
Since I upgraded my frame for a solid metal frame, I have a lot of ringing..
CNC Machinist and design engineer here, you're correct about the ballscrews, linear guides are even better! But no budget 3d printer will have those on their machines, I've found that adding 4 tungsten weights to my Elegoo Neptune 3 dampens vibration quite a bit. Also setting the machine on a paving block helps quite a bit too.
Lowering the perimeter or wall acceleration settings in slicer has a huge effect on removing ghosting. Available at least in Slic3r (Print Settings -> Speed -> Acceleration control -> Perimaters) and Cura (Speed -> Print Acceleration -> Wall Accceleration -> Outer Wall Accceleration). I use 2000mm/s² as the default acceleration and have outer walls printed with 310mm/s² acceleration.
I posted a copy of your comment on the top comment for all to see. I think Angus should make this top comment as well.
Good to know! I had no idea the slicer even had such a setting
Well Jonni, aren't you even gunna thank UFO? It's been a year now and we're all waiting.
Thanks for the tip., im try it
@@jameslaine2472 yuo need to activate the options
Thanks for sharing, Angus. I print lots of miniatures & terrain for tabletop gaming, and lowering speed is what has worked best for me. Will share with others who get ringing as well. Cheers!
A 4 year old video!!! still completely relevant today! I’m new to 3D printing, and Angus has all the answers... plus I love his challenges and test prints
You could also try adjusting the timing belt tension. The direction change isn't transferred as quickly to a loose belt, so the ghosting is at a lower frequency but more pronounced. A tighter belt will help transfer vibration to the frame faster and hence dampen it.
>timing belt
The belt tension can also make a big difference. If your 3D printer is belt driven you kinda have to find a good balance in belt tension. A loose belt dampens vibration quite well, but may result in higher tolerances. While a tight belt decreases the tolerances but also increases the springiness of the belt, potentially 'causing things to bounce back and forth a little after a sudden change in speed. Similar to a weight hanging on a rubber band when the other end of the rubber band is suddenly moved up or down slightly.
Thanks, you helped me confirm that careful assembly with strict tolerances will eliminate yet another 'common' problem. That and a double Z screw. Like others, I want to switch all belts to drive screws. Get them within 10,000/in in the XYZ and never have any of these common issues that stem from 'minimum engineered products'.
Just the other day I did a basic search for Ghosting tips. Everyone was arguing about settings. Then Angus swings in and explains things better. I'd rather go with things everyone can agree on and easily understand.
Thanks Angus. You're such a good resource for people that just want to know without having to take a master class or sift through all the internet arguing
Ghosting can also be reduced by fixing your printer down to a rigid base. I screwed mine down to a 3/4" plywood base and saw an instant improvement. I still get ghosting when printing fast, but generally it's acceptable. Also it reduces the frequency I need to re-level my print bed. I'll look at improving the rigidity of the X-Z axis next with braces
I think you mean momentum instead of moment of inertia which is related but really a different thing. Doesn't detract from your excellent video.
It does distract. He's supposed to be a college graduate.
"Graduated Sydney's University of Technology (UTS) in 2013
Bachelor of Industrial Design
First class Honours"
Small correction: moment of inertia is an object's resistance to changes in angular acceleration. The moment of inertia is just a property that comes from the mass and geometry. It's not 'caused' by the motion, but is rather the property of the object that makes it more difficult to make quick changes in direction. The moment of inertia of the extruder is constant regardless of whether it is moving or not.
Cool. Small imperfections aren't usually a concern since my prints get post-processed but it's good to know exactly what's going on and ideas to fix it if necessary. Thanks for the info.
I thought the the same, then I became obsessed with having perfect prints at .2 layer height and only having to lightly do post processing on whats left of the start stop seam.
I really like the idea of putting the paper onto the printer to show the vibration. Great visual aid. Thanks Angus.
Thanks Angus. It always helps to be pointed in several directions when troubleshooting complex problems.
Also, if you change the orientation of your printed part by 90degrees in the slicer, this may help reduce ringing further after all hardware optimisations you can realistically achieve have been performed.
One tip:
Putting your printer onto a foam board or put it on springs for vibration Isolation does not reduce the ghosting. It just changes the eigenfrequencies of the whole system so the ghosting will appear at different movement speeds / changes of speeds.
Putting the printer on a vibration isolation only reduces the vibration that the printer gives to the surface it stands of. In general the printer on it's own will "wobble" more.
Stiffening and reducing mass both have the effect of raising the natural frequency, and dampening like with shock absorbers as you mention has the effect or reducing the amplitude but widening the range of that response. Likewise slowing down the print speed lowers the working frequencies of the system. The goal is to push the systems resonance response high enough that it's outside of the working range. This method can of course reduce resonance but it can also introduce new resonances that weren't there before.
There is an alternative in that you can increase mass and attempt to push the resonance outside of the working frequencies by decreasing the resonance frequency substantially. This isn't necessarily a good fit for 3d printing but it does well with someapplications.
Ultimately though no simple solution like this will do as well with as little as a control system in the firmware with proper resonance compensation. Traditionally for mechanical systems you use a a state observer (a low order Luenberger observer) to inform a feed forward controller which deliberately offsets the resonance peak from the control signal. This is part of why higher end 3d printers despite similar construction standards to other cheaper printers, can at times get better results. A lot of thought is given to system rigidity, resonance, and the control logic itself.
Would you say it's a ... Ghost in the Shell?
I wouldn't.
Only if printed in vase mode.
I'd say that this makes him a "ghost buster"...
*badum-tish*
Nice one. It’s ghost is whispering to it.
Curse your sudden but inevitable joke!
You could also get a bowden setup. I have a cantilever design and don't have this issue at all.
Moment of inertia is the resistance to changes in rotation and is a property of the geometry of an object. It can effect the rigidity of an object, but moment of inertia isn't changed by the movement of an object. It's momentum that is effected by velocity.
Are there any engineers watching that actually know what moment of inertia is and how to calculate it? If yes, then you probably thought: _"wtf is this guy talking about?"_
Moment of inertia is not caused by acceleration, it's the property of s structure that determines how easily it will deflect based on the force that's acting on it.
And then there are 2 separate "moment of inertia"s, the first one and the second one. Or the mass moment of inertia and the area moment of inertia. In order to calculate at which frequency your printer will resonate you will need both as well as the material property.
"Less speed means a lower moment of inertia" *lol, nope.*
Scientific knowledge of a lot of people is completely hopeless.
Agree that Angus doesn't understand moments of inertia, and shouldn't have used the term in the video - but this small misinformation doesn't greatly detract from the overall content. The TLDR of his video is basically "it's a vibration problem, so can be rectified by the usual means". His explanation was somewhat lacking based on his knowledge of the field, as you noted. All in all I liked the video and your comments, but the second comment whilst humorous is a little harsh :P
Yeah, that made me double take. I think there was confusion with the common meaning of "moment" and the use of the word in physics. Understandable. People make similar mistakes with phrases like "begging the question" all the time.
yes I'm a big fan of Maker's Muse but the terrible physics in this video is actually so painful to listen to
Yep. Pretty much. I am not sure why he didn't just say "inertia". I guess "moment of inertia" sounds cooler.
great, explaining printing issues is much more helpful, than reviewing printers ;-)
this is gold hahaha
Great video Angus!
You're explaining some pretty advanced concepts in your video and know they are not always the easiest to distill down to just a few words or phrases.
Just a subtle correction: Changes is acceleration and deceleration as well as speed DOES NOT result in moments of inertia.
Moment of inertia is a geometric property. Nothing to do with the material or forces/accelerations acting on the object.
Great video! I just experienced this after sorting out my tpu setting to the point where I have drastically increased my printing speeds.
Moments of inertia are to rotation (not really, but kinda also bending a bit) what mass is to linear movement. That said, velocity, acceleration and any movement won't change the moment of inertia. However, higher accelerations cause bigger spikes in resonance vibration. Increasing inertias and masses is a good thing for the frame - as you correctly showed, but it's a bad thing on moving parts, since these "masses" store engery and vibrate as long as they do. So: inertia is mass and accelerating that mass leads to swinging masses, which causes trouble. So make things light and rigid (rigidity helps to shift vibration frequencies to so high bandwidths that they can be ignored).
Spooky scary vibrations send shivers through your print...
one of the quick ways to sometimes get rid of ghosting is to change the orientation of your print and try to have the widest parts lowest when possible and have the print head print less and less per layer as it travels up. Doesn't always help, but some times it does.
Also spring tensioners can allow bouncing. Replace them with solid tensioners
I tried some aproaches with mediocre or no success:
1. I improved regiditty. This made the prints much more accurate but induced evan mor ringing. That is counter intuitive but logical because the slack I had before acted as an damper.
2. I enforced belt tension. This changed the pattern and helped a bit overall. Increasing the force of the steppers by changing them and cranking up the current did pity much the same as higher belttension.
3. I than tried shock absorber under the printer. The impact on the ringin was not noticeable, but it reduced the accuracy.
4. The next attempt will be stronger belts.
From an physical perspective it is quite simple: if you put energy in the system it must has a way to come out again. This happens by dampening only. Higher regiditty only changes the frequency of the resonanz. This helps a little, because higher frequency means the existing damtening is applied more often.
Realizing this in practice turned out to be much more complicated.
All Glory to the Hypnotoad!
No
Also: check belts are properly tensioned if you get ringing
Angus,
Thank you so mich for these videos! Your channel has saved me alot of time and trial and error.
Keep up the good work!
Here are some further suggestions for reducing ghosting.
1. Take into account the geometry of the printer. So, for example, on the Cetus, oscillations in the cantilevered arm contribute to ghosting, so try rotating the print 90 degrees. This will reduce ghosting on one surface (but increase it on the other) which may give a better result for your print.
2. Place the print on a different place on the print bed. So, for example, on the Cetus, moving the print to the left will mean the print is nearer to the cantilevered arms fixing point, so the magnitude of the oscillations in the arm will be smaller at this point.
3. Design your prints to reduce ghosting. In particular avoid sharp corners, if possible. So fillet your corners if they are not required to be sharp. This will also have the benefit of marginally reducing the print times (since the print head does not need to slow down as much going around corners).
The surface reflections contributes as well to the visibility of the ghosting. With Glossy filament it's much more visible than with a matt finish.
For some reason I find it hard to take my eyes of this fascinating content.
All glory to Hypnotoad.
One thing I never see mentioned and I see a lot of people doing is mounting the filament spool onto the top of the printer. Putting all that weight high up is just exaggerating the swaying that the printer is doing and increasing the ghosting.
Adding more rigidity will not solve it, just change the shape. Basically, the problem is that the system has a resonance frequency which gets excited by the G-code.
Slower speed reduces the forces involved, and to some extent the acceleration - that's why it helps.
More tension does essentially nothing. Preloading a spring does not change it's spring rate.
Reducing mass increases the resonance frequency, which may help
The real solution is jerk and acceleration. You have to view them as low pass filters. Basically, they will greatly reduce the forces acting on the frame during direction changes - essentially reducing print speed whenever changing direction. The issue is that properly tuning them is not trivial; of course you could just greatly reduce them and it would help, but it would slow down fine detail prints a lot...
Have you considered using a tuned mass damper fixed on the head to reduce vibrations? Such as a small weight mounted on a flexible wire or a spring for example. This is commonly used on tall buildings to mitigate the effect of wind or earthquakes.
Thanks you do a great job. Cheers and have a happy new year.
This is what happens when you count on rigidity to keep things stable instead of feedback. And with rigidity, there's not much you can do, and it's a black art. Giving it a whack with a hammer might help, or it might make it worse. Same with attaching a weight or a spring.
The advice to increase rigidity ... I'm not saying it can't work, but it will never fix it. And totally overdoing it might make things fall apart or worse: suffer high stress during operation and then suddenly fling a metal arm across the room.
Great walk through of the issue easy to understand and learn from great job as always , will have to try these tips now on my ender 2
Hello , I am happy to found your video for how to fix the problem on 3D printing ,
here I got a problem hope you can help , my problem is when I start printing every thing is OK , but after around 50% then the print head will cooling down , I have use a new set of heat sensor , but same problem got .so hope you can sharing your experience , thank you very much , regards
John
Another way to reduce ghosting would be to upgrade to a bowden extruder setup, but it will make printing flexible filament harder.
Nice video Angus! I was a little bit disappointed that you didn't cover stepper driver dampeners and variations in belt material and tensions.
It's such a complex issue... yeah, I need to play more with dampeners but my experience is they round out detail in expense for 'smoother' prints.
Yeah, dampeners probably was a bit overboard. But belts are at least something that is pretty easy to swap and as far as I know these can make a big difference if the frame is kind of rigid and the belt is springy.
Ghosting can be eliminated 100% without changing acceleration and jerk with proper firmware/stepper drivers. There currently aren't any solutions for DIY 3D printers that I'm aware of, but there are many for commercial CNC machines. Basically the algorithms will anticipate when a change in direction is approaching and modify the instantaneous jerk and acceleration appropriately in order to eliminate the extra "spring"/slop upon completing that motion or change in direction. I'm sure Marlin will eventually implement something like this, although tuning for each machine individually would be tough which I'm assuming is why nothing like this has been implemented yet. An auto-tuning script would likely be needed which would require stepper drivers with accurate positioning feedback.
Here's an example of this algorithm, not my video! th-cam.com/video/eUZVLMCmSxI/w-d-xo.html
exactly :) there are some works being done. constant acceleration and deceleration on most firmware hinders the ability to print at high acceleration/deceleration.
Angus I was actually going to ask you if you had a video like this and you went and made one before I had the chance to ask.
I really appreciate the videos dude, thanks a bunch. :{J
Great video Angus. Clear and concise. A great help 👍
I love how most people who watch these Chanel’s don’t have 3D printers
I'll buy one after I get my first microwave. Can't do all the high tech stuff at once.
really? I never knew that. I started watching these channels right around the time I got a printer, I don't know how else you could fall into that rabbit hole lol.
It’s more about going into a purchase with an idea of what it can and can’t do. All large purchases should be researched first.
Interesting - I thought I'd had just about every 3D printing fault but I've never seen this one! Perhaps because I'm always printing in slow motion! Thanks, very interesting.
I found putting a couple of kilos of filament on top of the lid ( Flashforge Dreamer) improved print quality a lot.
I always set my Outline speed to 50% and my Solid Infill to 95%, works very well on my Anet A8 [Stock frame]
What fkd slicer is that. Speed is not measured in %.
I tried your ghosting test on my TEVO Michelangelo. It came out nearly perfect, with only slight ghosting around the text and Makers Muse symbol. I was using WYZ white PLA 1.75mm filament. The Gcode had the default printer settings. It might help that my printer is on a sturdy wood desk.
To me, ghosting is somewhat of a "perfect storm" of issues. Frame rigidity, vibration absorption properties of the base, acceleration/jerk settings, and print speed. I would also add nozzle wear, printhead weight, drive belt tension, etc. Unless the ghosting results in an unusable part, it's really not a huge issue, especially if the part will be post-processed. As far as small details and such, if you want good, fine details, you really shouldn't be using an FDM printer in the first place.
Great video overall. Still yearning for more Fusion and Meshmixer tutorials... ;)
Hello man your explaination about echo is correct but I understood that the main reason of ghosting is a movement trasversal to an axis over which the nozzle is moving. The eco is not caused by x or y themselves but by the bar at the top of z for 99%. Is a butterfly effect, where the z axes starts to vibrate trasversally to the x or y axes. The eco or lines u see fading out are the translation of the fading out of the vibration of z rod\brace. Because of that if u check it, the slot at the very top of the z rod has a bearing whose socket is very loose (for cr-10), this in order to let the rod vibrate without transmitting the vibration to the brace. So the issue is not the relationship between x and y but with x that turns into y while z vibrates. The question of mass is true but it cant be the main reason of ghosting here. U have not 30kg of mass to brake in 2cm that justify the inertia over the x and y movement relationship only. The z rod is the criminal here. Make a test and try to completely isolate z im order to fight ghostin 👍👍👍
Love these reviews, especially when you use the Cetus, since it's the only printer i own so far and I love this printer.
You can also rotate the print +/-90° and use the other axis.
Why not just go to your local sporting goods outlet and get stick on vibration dampners. they can be found any where compound bows are sold. they are soft latex shapes with a double sticky tape adhered to one side. Print out 4 of the muse sample, one every 90 degrees... and stick the damper in the direction of the ghosting... in the video example with the cardboard the damper should be stuck on the side of the bed the cardboard hung off. they also come with screws molded into them if you'd like to firmly mount them.
Very informative. Great video Angus!
Tks A. I have learnt so much from your channel.
My ghosting problem with my creatility cr-10 s5 is from the absolutely massive weight of the bed. 500mm square aluminum plate with the frame underneath and same size glass plate, the heater and the aluminum plate for leveling screws. The whole thing probobly weighed over 10 lbs. I modified it to only weight about half as much.
HypnoToad !!!! Great channel Agus, I love your explanations and opninions. Cheers from Québec City Canada !
Great video and cool desing! My problem right now is retraction. i cant find a way to fix underextrusion in parts like you're tolerance test. i have tried small amounts slow , fast, big amounts fast, slow etc. this leaves practically unuseable parts.
cheers . my a8 is as rigid as it can be without stiffer x-axis rods. I overlooked dampening
Printing slower is a quick fix, but also the brute force worst case for fixing bad acceleration and jerk settings: print slower when you don't need to during straight paths but possibly still cause resonances around sharp moves. Getting acceleration and jerk right can fix this without sacrificing straight-line speed A much better order to address ghosting would be:
1. Reduce moving mass to the extent you can
2. Improve rigidity, mounting, and static mass to the extent you're willing
3. At normal speed, tune in acceleration and jerk
4. There is no 4. Don't touch speed, ideally.
It would be more fair to say: reduce speed only if you don't have the time/patience for #3. I think people are intimidated by acceleration and jerk, but let's face that unknown.
Maybe a small vibration damper will help. Yes it will add mass, but overall it could decrease the jerking n ghosting
hi im new to 3d printing...and i giggled at JERK setting
Excellent teaching video. Thanks, Angus
Would it be possible to use a normal 3D Printer in some sort of non-conductive oil (Like the stuff LinusTechTips used to dunk an entire PC into) And if it is possible at all, would it be able to prevent the need for putting Supports on your prints since the oil might provide the support it needs.
Interesting idea, but i think oil would severely mess with layer adhesion.
At that point why not just go full SLA
Might work though, try it out and tell us how it goes.
Thanks for the what's what on 3D printer Ghosting Angus... Was new to me.
I'll be honest it's a way complex phenomenon and difficult to address completely but definitely something that way too many modern 3D Printers suffer from!
Sometimes just changing the orientation of the print on your bed helps. Just try and rotate your piece 45° around the z-axis if it has a square shape and flat faces. Of course, this does not help that much if your design has organic and rounded shapes. And about the acceleration and jerk settings, most firmwares allow you to adjust these settings via g-code commands. So with a bit of g-code hacking it's easy to make quick adjustments and find the best settings for your machine... (Check the M201, M204 and M205 G-Codes on reprap.org/wiki/G-code )
I was able to reduce Jerk speeds in Cura which eliminated ghosting.
to fix my issues with blobs and ghosting i just lowered my jerk and accel settings. i worked my way down from stock. im not at home but i have done this on all of my printers and leave my print speed set to 100mm/s on the ft5. i believe i am set to 1100 and 8
Would love to see more troubleshooting videos like this :)
Something I'm not seeing in the comments. You mention acceleration and jerk settings, I'm not sure how well known this fact is, but "jerk" is the derivative of acceleration, or how fast you're changing your acceleration. A piece of trivia worth knowing if you're going to start messing with said settings.
Also, side note, the derivative of "jerk" is "snap," then followed by "crackle" and "pop" because we're nerds. (thanks Tom Scott for teaching me this useless information)
Jerk and Acceleration solved my ghosting. Default Marlin has them set far far to high for most printers. I would say this is something to look at more than frame rigidity.
Use a Bowden feed tube. It will also help.
THIS WAS VERY HELPFUL!!! THANK YOU SO MUCH!!!!
Good info as always Angus. Cheers
Why would you change jerk and acceleration in the firmware rather than in the slicer?
it's weird for me. i get nice outer walls on my prints (i print molds for casting), but when it prints the inner "bowl" part, i get some layers that stick out, and some that are ok.
imagine printing a box with a half sphere cut out of the middle.
layers 1-40 are fine, layers 41 through 59 stick out like a band, layers 60 through 82 are perfect, then 83-100 are sticking out like there is a band again.
it really messes up my casting, since those details show up in the casted object.
+Maker's Muse what if I have a slight offset on the horizontal axis.
Its shows some lines clearly more inwards or outwards.
I've a Prusa Mk2, can you please tell me how to fix it?
Edit:I'm not quite sure, but maybe at around 5:00, you can see this on your mk3 print as well. Some layers are clearly more outwards than others.
thanks a lot dude! really objective and helpful!!
I have weighted (2.5kg) and shock mounted on a Yoga mat my 3D printer, will have to print this fast like 60mms.
I´ve some questions regarding 3D printers, and I hope someone in here could help me out
Can you print your own transformer or action figures with ANY 3D printer or do you need an especific type of printer?
What´s the best material to use if you want solid toys: ABS or PLA?
How solid is going to be the toy? Like Hasbro´s solid, like a Lego´s brick or is going to be like a really cheap plastic?
Do you need to sand the pieces or they´re already smooth as soon as they leave the print?
What´s the best software to use to create toys for 3D printing?
How durable is going to be the toy? Is going to break as soon as it hits the floor?
And what´s the best way to deal with tolerance´s issues?
Should I also make the table rigid (to the wall for instance) or let it be loose as some form of dampening?
increasing rigidity could avoid layer shifting ?
Dude, love your videos, but do you know what moment of inertia is?
I would call it resonace, but when I show our chief engineer a component we're trying to machine I tell him, Malc, I've got wobbly bits again!
Rigidity is key. Hobby grade 3d printers are basically low tolerance machines, so good positional accuracy and smooth surfaces are mostly controlled by either better quality components and, or speed of X/Y moves.
We get this in our milling machines once they are a few years old. So, effectively it's caused by wear in the lead screws and the CNC not being able to deal with it. Yes, slowing down the X or Y axes moves will usually solve it, but of course, time is money. It doesn't happen in Z to the best of my knowledge.
.
I hadn't read this before today.
Quote from Wiki.
-------------
The running speed for a leadscrew (or ball screw) is typically limited to, at most, 80% of the calculated critical speed. The critical speed is the speed that excites the natural frequency of the screw. For a steel leadscrew or steel ballscrew, the critical speed is approximately[17]
N = ( 4.76 × 10 6 ) d r C L 2 {\displaystyle N={(4.76\times 10^{6})d_{r}C \over L^{2}}} N={(4.76\times 10^{6})d_{r}C \over L^{2}}
where
N = critical speed in RPM
dr = smallest (root) diameter of the leadscrew in inches
L = length between bearing supports in inches
C = .36 for one end fixed, one end free
C = 1.00 for both ends simple
C = 1.47 for one end fixed, one end simple
C = 2.23 for both ends fixed
So if your machine, whatever it is, already had issues, ie: low grade parts or any flex in the chassis, any time you hit the critical speed of your lead screws, X or Y, all the other probs only make it worse.
No lead screws on X and Y on 99% of the FDM printers. It's 6mm GT2 belts.
i noticed some ghosting on a reprap model 3d printer.
which happens when you put support between prints (overhang)
this only happens on a few layers tough.
the layer where the support layers start and where they end.
the file is a really simple file with no text and less than 15 sides with 1 half circle.
I dont even have a 3D printer. I just watch your videos because they're interesting....
hi I did a print of it no Ghosting at my normal print speed, so I up the print speed till I started getting ghosting,eg 15mm/s to 25mm/s so now print at 20mm/s with no ghosting, THANKS MAKERS MUSE
Ghosts in the machine. Gotta print this on the MK2 and Sigma to compare. My printers are so competitive, that why I can't leave them on overnight ... they fight lol
Ghosting is bad, but scars are worse!! Thanks a lot for addressing ghosting, but next please let's fix these stupid scars of mine. Thumbs up Angus!
I downloaded this stl, sliced it in Ideamaker for my Flsun Q5. It is printing now but the print head is moving so slowly you have to touch it as to make sure it is printing. Is this test only for Cartesian style printers?
Should I disable the speed reduction for the short layers for that test?
Nice video! But it's just inertia, not a moment in my ominion! A moment means it has something to do with a rotation which is not the case because the head only move (rapidly) in the x and y axis.
Have you done anything to improve your Cetus, or do you just print slower for detailed prints?
Just print slower, though I'm working on a mod to constrain the gantry a little better (when I have the time!)
I "upgraded" my lower tray (the structural one below the heated one) on mine to a metal one and I guess it increased the weight because it ghosts horribly now. Need to change it back.
Nice, going to test this on my raise 3d n2
I would have mentioned belts also, their elasticity is a big contributor to ghosting.