@@theomoons As they say... If you want help, don't ask a question, simply suggest doing it a certain way, and someone will surely explain a better one. ;)
I still feel like the sand in the frame might marginally helps, but adding the "sand containers" to all of the moving parts really just serves to add weight and increase the problems.
Although sand in the frame really is a poor-mans version of just using better aluminum extrusion profiles - i.e. the heavy-duty type instead of light-weight.
Not documenting "failures" is a huge problem in the medical industry b/c peeps don't want to embarrass themselves. Props to you for posting this. It helps inspire new hypotheses to come and still makes good content 👌
There's an amazing technique called "Motion Amplification" which uses special cameras and DSP to visually exaggerate imperceptible movements in video. It's used largely for identifying sources of resonance in structures and industrial machinery. I'd be fascinated to see it used on a 3D printer. I predict the main source of resonance is the belts.
Wow that sounds so cool. i remember a similar tech where a camera was used to observe a human face and detect imperceptible changes in color of the skin due to blood being pumped and amplify it to detect the heart beat.
@@BerraLilltroll I saw that video (or a similar one) recently and was impressed with his technique and results. As I recall he was balancing a rotating mass with weights, but the technique could be used to identify resonate frequencies and test various damping solutions without wasting filament or waiting for test prints to finish.
In the machine tool world, "epoxy granite" is used to fill machine frames to increase rigidity and dampening ability. It's basically just epoxy resin and sand. What the epoxy adds is it connects the mass of the sand to the structure, making it one. The aluminum extrusion with sand in it will ring nearly as much as without the sand as the sand is able to just move and allow the aluminum to move (ring).
True, however cast iron also has significantly different characteristics to aluminum extrusions. Plus, the lighter the machine then the slower / shallower you need to go. Even a mini lathe weights 40+kg.
@@arthurmoore9488 So you add mass to non moving parts only. If you're fairly certain you're not going to fiddle about with your design, you could even fill in the outer moldings.
The sand on a rapid-moving part was totally a mistake on my opinion. Getting down the rabit hole you could try with some mineral oil, like the ones that are inside car/bike suspensions...?
FYI: I put a FDM printer on a $20K active dampening vibration table we use for laser alignment and it made no difference. I think the issue is amplitude and frequency is not consistent. What about hanging the printer from a pivot point, with a mass dampener at the other end of the pivot point? A sort of Foucalt's pendulum with a mass dampener.
I think this is the comment closest to identifying the fundamental challenge. There must be multiple resonances in the printer structure and there are certainly a multitude of stimuli when you consider acceleration and vector forces. For a Cartesian system, it is probably a problem constrained to two planes; XY and ZX. I think it’s a lot harder than it looks. The analysis would be fairly involved, I think.
Aren't vibration damping tables mostly about preventing vibration from the environment from reaching equipment? Or is it designed for vibrating equipment as well? Ultimately, I think the problem is that most of the vibration is coming from the carriage itself, which is only loosely coupled to the rest of the system, and by definition needs degrees of freedom.
Hmm, makes sense that you shouldn't put weight on the moving parts. So I'd try the same sand setup but excluding the print bed since it's moving quickly and weighting it gives more inertia so a higher probability of skips.
I’ve put my printer on a square of *memory foam* . Using tick square of plywood on top so the printer doesn’t just sink into the memory foam. It works great & definitely makes the printer quieter.
Tom, this is such a great video! No meaningful results are still results, just as you said. Only you would build a Suspensorium Love it! I'm chasing resonance reduction as well and would be interested in how different belts affect things. You are both thorough and entertaining and it is always a pleasure watching your videos.
I wouldn't describe the results as non-meaningful. They were not 'positive' in the sense that there was no improvement in the printing artifacts due to 'ringing.' But it provided some evidence that some of the 'simple' fixes do not appear to work as miracle cures, and in some cases actually made things worse.
tuned mass damper are also commonly used in skyscrapers to absorb vibrations from swaying in the wind. there's one in Taiwan where the architects famously decided to make it a tourist attraction - instead of tucking a concrete block away in some utility room they made the main damper a large shiny decorated sphere hanging in the middle of a big atrium.
I would recommend removing the sand from the print bed. Then, run the tests again: With Feet, without feet, and then suspended. This series is great, and your willingness to be honest is appreciated.
@@brianfhunter actually the suspended test could be coupled with and earthquake reader and it can plot the sine wave of the vibrations. Then with the total mass of the 3D printer known, plus it's rought moment of inertia, you can calculate how much damping it needs
Lol, you seemed so tired & disappointed by the end of the video, but you make a good point that even the failures are useful information! This has taught me that if I want to deal with vibration, I'm better off putting my time into brushing up on my old engineering courses rather than on messy, half-baked quick-fixes. Thanks for that!
Love your honesty one of the reasons why we come to you you always keep it honest so we all don't go out to the store and buy lumber and try to make what you made lol I can only imagine how many people be trying to pull out their hair cuz they couldn't get what you have thanks again for keeping it honest and real
Yes definitely do a video on the belts please ...... Since I starting playing around with 3d printing and my velleman k8200 many years ago you have taught me so much, and are normally the go to videos when I have a problem or just want to improve something. Keep them coming and hope to see you around for many more years to come THANK YOU Tom
So I am not the only one that got hurt playing with the k8200. Still got mine. But I don't use it since I got my Ender3 original. I since replaced to pcb on the Ender. I could maybe use the old pcb into the K8200 and try to have it run better.
@@pierremartel3552 My 8200 was a good printer even thou I have done a lot of upgrades, it was working fine up to a few months ago when I parted it out to build a bigger n better printer from scratch. It was the 8400 that I scraped after a few prints when the MB froze up putting the hot end into a thermal run away ..... after a hard reset it was still showing the hot end at almost 400c
Interesting ideas! It might work better to put the 3d printer on legs that stick into a bucket of sand, so that there is lots of sand to absorb vibrations in the base. I am curious to see the result of a test like that!
I was going to comment with the exact same thing - this feels intuitively like a good move. If you had a large, square container filled with sand and just sat the printer in it, I think you might see a difference. (Probably too slight for it to be worth it, but at least it would be cheap to try.)
I think the best solution would be to have a counter weight on the same belt as the X or Y. The top of the belt moves the opposite direction as the bottom of the belt.. Let's say you have a weight on the X that weighs the same as the head.. They would move opposite directions, and hence when they stop, should vibrate the machine. Just a thought.
Or you could try Input Shaper, I sure hope that's where this is going, try a bunch of physical solutions, show they don't really work, then show the absolute magic of Input Shaper tuned with an ADXL.
Less initial vibrations for Input Shaper to negate would lead to sharper prints as it'll apply less smoothing. So removing as much resonance before using Input Shaper is the way to go. It's a shame this test wasn't shown without the sand containers on the bed axis, as that's somewhere you should be trying to remove weight on this style of printer.
Need stiffness + weight everywhere in the frame to shift the frequencies higher. That being said most ringing is due to how stepper motors work. Throw an oscilloscope on the stepper motor, then run Input shaper and compare.
I just bought some motor bike gel pads, cut them to roughly 5x5x2cm and put them under the printer. They have the advantage of being able to move in all directions and completely removed all ringing. The printers still stand on massive, 5cm thick sandstone slabs and those are on rubber mats, which results in an de-coupled, non resonating table and a still free moving printer While moving fast, you can see the gel visibly shaking and shifting the printer as it absorbs the energy, but the printer never moves on the gel itself. this might be worth a try if you have ringing and/or noise issues printer: prusa i3 mk3s
Have had a similar plans. Was thinking of using a section of 1-2cm thick exercise mat under a large concrete paver with gel on top. Do the feet of the printer dig into the 2cm gel pads? Thomas should test this.
@@Trevellian I removed the feat for this exact reason, didn't even try it out. One thing I found out is that the size of the gelpad has a direct influence on the movement of the printer, the smaller the pad the more it moved. I sliced them down from 5x10cm to 5x5cm and the change was noticeable when pushing the printer to one direction it is important that you enforce the table you put the stone on since it weighs a ton and it will bend your (possibly cheap) table. I enforced mine by placing a solid 3cm thick wooden board on top (no screws required)
Another great video Tom. This is just a collection of thoughts. Sand alone isn't great at dampening. It's coarse and rough and irritating and it gets everywhere. Solution Epoxy "Granit" (a with Epoxy damped mixture of sand that is filled and compresses in forms or extrusions to minimize resonances and vibrations on Professional Machines like Kern cnc mills) Weight alone helps a bit but Weight and a Dampener works great. I would love to see the printer mounted to the paver and than placed on a 2 to 5 inch think piece of upholstery foam. Oh and i think stifer belts will make a noticabal difrence (steel cored belts, heard that there are some out there in the wild) Keep up the great work and i cant wait to see the next episode in this series and the escalation that it could bring with it.
Stop hidiing Anakiin, we all know it's you. Although you bring a fair pont about the resin but it defeats the purpose of dampening. What makes sand such a great dampener is the fact that it's loose particles that have thousands of points of friction for the energy to dissipate into heat, having it all merge with the resin just make's one solid mass that, albeit having greater inertia and therefore harder to get moving, has a single natural frequencey that it can resonate with.
@@coolbionicle I disagree unfortunately. Having a Mass that not solid makes it very hard to predict the result and I cant find a mechanical loss coefficient for loose sand. Besides that, the “granit” part of the epoxy graint its mostly a filler too minimize the amount of epoxy needed (epoxy = expensive). The epoxy provides the dampening, the Aluminum frame the needed stiffens for the machine and the Sand is just a filler. If you don't trust me go over to Adam bender her on you tube and watch his How to Build Epoxy Granite Machine Base video. He goes in more detail with Loss Coefficient vs. Young's Modulus and how to find the right ratios of sand to epoxy.
I think you may get some interesting results if you modified the springs on the tesseract. What it is now is a vibration isolator not a damper. All of the vibrations made by the printer stay within the system. If you think about vibration dampers on cars there are two parts, the spring and the actual damper or shock absorber. The tesseract set up is only spring and does not have the best damping qualities.
Thank you for going through everything that didn't work. I've been really curious about sand on smaller machines. I'd love to see a video on the different belts. If you could please have one cheaper belt case using an old piece belt that would be great. I honestly wonder about my belts every time I see ringing.
Very interested in the belt comparison video. I found early on in continuous belt platforms (CoreXY) that there are parametric resonances that crosstalk in abrupt vector changes, leading to ringing artifacts. I used accelerometers and a 35670a dynamic signal analyzer to capture the response of the gantry vs the head. Data is on e3d's forum. I came to believe, but did not follow up on the idea of tuned dashpots that the belts can ride on... (shock absorbing idler pulleys)... might be a trick to try. Thanks for the hard work, Tom!
Very cool video. Not vibration or resonance related, but one thing i have been curious about for a long time is how flipping the printer upside down would affect overhangs and bridges. Since any sagging will be pushed back up by the next pass of the nozzle, I would expect it to improve the results. And now you have the perfect "printer enclosure" to test it 😄 Also seeing the effect of different belts would be very interesting.
@@robertbyrnes7075 I don't think so. The molten plastic is probably viscous and sticky enough to stay in place wherever it is fully supported by the previous layer. And in areas along slanted edges where it is not fully supported it will sag towards the nozzle. Which means that on the pass for the next layer the nozzle will lift the sagging but now much cooler plastic up again. As opposed to pushing it further down when printing with the nozzle pointing downwards. I find it hard to predict the end result of this interaction, but I think it might make it possible to print steeper overhangs with better quality.
Back in the late 1980s, some of the scientific equipment vendors, such Edmund Scientific, sold rubber ball kits for physics demonstrations. One ball would be a standard rubber ball and would bounce as such. The other ball would sometimes be called a dead ball as it wouldn't bounce. The idea was to show how energy could transferred in one direction (bounce) or dispersed in multiple simultaneous directions (dead). At that time, the dead balls were of particular interest in the amateur holography scene as lasers weren't very powerful and the holographic process was extremely sensitive to vibrations. I don't know if the dead balls or material are still available but if still available, that could be an interesting test.
I have observed quite a bit of NVH testing. The bottom rails likely didnt harm much, but the vertical supports having sand added was the main issue making prints worse. Adding mass to cantilevered components will just shift the frequency of the resonance. More mass generally = more resonance at lower frequencies (greater deflection distance). Adding reinforcement to the gantry along with soft mounts that absorb energy will help most. A support rod run from the top of each vertical support, to lower frame would mitigate the motion of the gantry. Soft materials used under the machine would help absorb vibrations from print head motion.
You could try small dampers from RC trucks. They are sprung and oil filled. This will actually adjust the frequencies instead of just adjusting the amplitudes. You need to absorb some energy, elastic straps provide almost no damping effect. I dont understand the fascination with optimizing single variables. It would be far more effective to increase the stiffness of the frame by putting braces on it or adding gussets at the connection. Damping a wet noodle is pointless. Brace it until it is rigid the dampen it. Otherwise you are finding ways to dampen a frequency that is irrelevant and ineffective. At least the oil filled damper would effect a far wider spectrum of frequencies and it is tunable using different weight oils.
Small dampers for belt tensioners, shouldnt help with skips but maybe will with resonanses. For skips you decrease moving mass or increase motor power.
@@Gebsfrom404 Wasnt thinking for belts. The exercise in the video is about dampening the vibration of the frame. I was thinking if they wanted to put the frames on an elastic base then could mount the dampers to the table. This way the elastics or flexible bases allow movement but the oil filled dampers absorb vibration. As for belt tensioners I ended up using a small coil spring from a brake drum hardware kit (had a lot of them)... it is used as a retention spring to hold the brake shoes to the backing plate... around 19mm diameter and 25mm tall compression spring... pretty stiff. Then I preloaded it with a threaded rod running through it to a yoke holding the belt pulley. Any compression spring would work as long as it is stiff enough to maintain control of the belt.
Should have used lead shot as is used in speaker pedestals - more mass and less leaky than sand. Ultimately the problem is with elasticity within the printer frame and mechanism. The lab robots I work on use large stiff belts that don't stretch to achieve precision. If I could be bothered to do it , I would reproduce the printer frame in welded steel RHS. That would eliminate a lot of the flex. Then the only thing to deal with is the flex in the mechanical drives. You could try wider belts. Most modern motorcycles have vibration dampers in the handlebar ends and often on other parts of the frame
Just a piece of advice for measuring squares, you can do the triangle method. 3:4:5, is the ratio your sides should be at and the distance between them. Its a commonly used carpentry trick for getting perfect 90 degree angles.
Since the idea is to apply dampening to counter the moving parts, and you already have the suspension rig, try loading two printers one facing upright and the upside down one facing the opposite direction... you want to load the same program into both and start them running (the bottom one does not need to be actually printing). The bottom printer will be moving in the opposite direction to the top one and should mostly cancel out the movements of the top one. Might be a relatively easy experiment to try out.
Any experiment that contributes to knowledge is always welcome. Whether the output is positive or negative, it is still knowkedge that can used in future endevours. Please continue with the belt swaps to see if there is any improvement to be gained on the prints at high speeds/accelerations. Thank you Tom.
Something I did to my printer, and I know this isn’t on the scope of which you made this video cover, was I replaced the x carriage. It originally was a heavy thick three part steel. Now composed a singular piece of aircraft grade aluminum of the same thickness. The shock went way down as the bed now shifts. Knowing this and then applying the issue of resonance, wouldn’t a solution of determining the weakest material causing the slop (artifacts) be best to replace? IE blue loctite all frame bolts. Reduce slack in bearings and such? Stiffer springs on heat bed?
I did the sand earlier this year. You need to tamp the tube until the sand settles. You can add a significantly more amount of sand when you do. All it takes is tapping the sides of the extrusion to make it vibrate.
What if we put equivalent weight on the belts..... So for example for the bed.... At the belt we can add the same weight as the bed just bellow the center of the bed on the belt.... So now when the bed moves in one direction.... Weights move in the opposite and kind of cancel out the effects.... Same can be done for all the belts..... That will kind of act like a balancing system for reciprocating components and maybe change the frequencies.....
Might be worth noting that springs do not dampen shocks or movement, they only absorb it, and if not dampened by a shock absorber, they can make the effect more pronounced. Try driving a car with blown shock absorbers. The springs still work, they just keep bouncing after you've hit the bump, and your car will be all over the road. The shock absorber is there to slow and halt the movement of the spring, after the bump has past.
Sand filled feet are a good, cheap and decently working solution. There are many different sand filled balls you can use with a printed holder to keep them in place below the printer
H! Great work - being systematic as presented makes the most sense in such a research. Different belts? - Hell Yes! I've switched to genuine Gates after factory Creality went bad - I've noticed less ringing - yet that could be biased. God tool to help would be Klipper Resonance Compensation, and it's measurements with plots. I've observed changes while belt tension being alternated, or X axle configuration changed. Best regards!
Had a good time watching this, thanks. I found vibration dampening a bit annoying too, especially when you have a wide range of excitation. Nvmd, white steel belts on heavy beds vs cheap or kevlar would be interesting indeed.
The sand will add more mass to the printer, making it still vibrate, but because of the mass, the amplitude will be much less. You should add lots of weight to the top crossbar of the printer. This should greatly decrease the amplitude of vibrations. Also, locate the filament spool somewhere else. If the spool rocks on the printer that could translate to the print quality. I've got a delta printer and this video has got me thinking I should just load up the top of the printer with something to keep it from moving around as much.
With that many skipping steps, I'd start looking into bumping up the stepper motor voltage via the driver pots. I'm curious whether fixing those skipped steps make the 'sandtainer' approach work at all, if only because stuffing them wherever they fit just kinda looks cool. But also, resonance tuning in software is way easier and very effective.
Keep this series going please. Super enternaining and gets everybody's gears going. What about non-symetric dampening setup? It could help with reducing resonation of the dampening system itself. For instance - your tesseract cube with some cables tensioned way more? Or usual dampening rubber feet but with one odd, softer/stiffer foot?
A damping mechanism can reduce high frequency vibrations, but it also enhance low frequency ones. Most of the vibrations coming from the moving nozzle are low frequency ones, so the dampener could make things worse.
TH-cam in parts seems to be more honest than most the research journals - also presenting what did not work, or what did not make the greatest WOW !!! effect. thanks so much
I am an electrical engineer so I haven't studied this topic, but I would first attach vibration sensors all over the frame and heating bed and do frequency analysis and then designed springs with resonant frequencies matching the strongest frequencies from the analysis.
Tom, dumb question as I haven't checked through all of your other experiments, yet - what have you done with the bed springs? Stock Ender 3 springs are woeful, I use the regular yellow upgrade springs on all of my printers, with a LOT of compression to eliminate some print bed wobble, these springs are better finished than ones fitted to cars - the ends are ground flat - [N.B. car springs break BECAUSE the ends are not finished properly so it transfers all the stress to the end of the first coil, saves the maker money, costs us for replacements] The other thing I do is fix the M4 countersunk [flat head] screws to the print surface from below, using M4 nylon nuts - this fixes the screws to the print bed, don't use metal nuts without nylon washers - you will damage the insulation surface or worse, short the bed to the frame, not unlike what happened with the early CR6SE's. Finally, after fixing the screws to the bed, putting the bed back onto the frame becomes difficult because the clearances on the screws are tight, so I enlarge three of the four holes with a 5mm drill.
On the UH60 BlackHawk Helicopter, we have tuned masses on springs that we test and tune to the aircraft to dampen unwanted vibrations, on the new model, there are counter rotating electromechanical vibration dampers that actively cancel unwanted frequencies.
FWIW, I put my Prusa i3 mk2 on a paving stone and under the paving stone I put a rectangular inflatable chair cushion (i.e. anti bedsore inflatable cushion for a wheel chair). Primarily I did this to reduce the sound, since the homemade wooden workbench I set my printer on was amplifying the vibrations of the printer. I can pretty much only hear the loud stock fans now.
Sand was common for audiophiles for a short time to deaden platforms and it did work but then things like soundproofing sheets ended that. One thing I am testing is turntable spikes/isolators. They are little 2 part metal pucks that either have a spike that fits In a cup foot or use a steel ball between 2 metal plates. They work well and so far are stopping a lot of transfer from the printer to the table.
This was fantastic. And thank you so much for testing things out. I was very curious as to the outcome, especially the sand damped tests. It also makes sense that BMW used such a complicated series of dampers. But in the industrial and aerospace worlds, there is a simpler solution - avoid the areas of operation that induce harmful resonances! In a light aircraft, this could be as simple as a plaque that says "Do not run propeller at RPM between 2500 and 2550 for more than 5 minutes". Or for a larger aircraft, a redesign of the vibration causing components, to shift their resonances rather than compensate in other areas. Where the moving masses are controlled by stepper motors, that's the obvious place to control vibrations. Match the speeds and accelerations to avoid areas of resonance. That's not cheating or a hack! That's proper design. Love your channel and the work that you do...not intended as a slight at all. That I learned that a printer will work upside down is worth more than the price of admission. To me that means it's just a matter of time before there's a 3d printer in space.
I love your Sciency thinking. Please keep valuing it as this sets you greatly apart from typical TH-cam content we all got used to! - You briefly mentioned the additional mass on the printbed might harm the print quality. I totally agree. I would have tried again withouout those printbed Santainers, maybe even adding them elsewhere… - I highly value the additional mass on the printer. The bigger the mass difference between bed and printer body, the more recoil you can convert into printbed acceleration. - A tip from back in the days (experience with laser interferometry in physics lab): For effective (nearly critical) dampening of alu extrusion profiles instead of pure frequency shifting fill the profiles with silicone (not silicon - Fugensilikon, nicht Silizium ;-) This also greatly assists survivability of your mechanical components when compared to sand… I would love to know the frequency range of problematic ringing. This would hugely narrow down the root of the evil (profile bending, profile ringing, belt stiffness etc.) okok, I’m getting to scientific. Looking forward to the engineering way of life: try it out! :-D
potential different question-problem pertaining to ringing in model layer toolpath with SHARP corners; it might even not be ringing of the frame or extruder itself in motion to make those effect once the large one is solved and you seek for more. on sharp corners, the extruded line is basically circle shape with offsets that gets squished by its own material in the toolpath; won't it be the extruded material amount itself trying to equalize within the toolpath? on that end, I simply model things with sharp corners filleted/rounded, so not only ringing is solved, but tolerancing is simpler and more effective
The last time i do something similar i made a weight, attached with some belts/rope to every axle so that it have some "Vorspannung". This improves the quality a lot. Of course with pulleys so that gravity can work in X and Y Axis
If you're chasing after resonances, tuning the structure will get you into diminishing returns fairly quickly. What you want to do is ramp all of your steppers up and down so that resonance never has time to build. Remember, one movement of the printer is easy to dampen out, it becomes a problem when you have many movements happening rhythmically that case resonance to build to the point where they cause significant movement. The other thing to do is to decrease the mass of your moving elements. The ender 3 is a poor platform to perform that modification on, however, something like a voron build gives ample opportunity to drill out 'speed holes' while maintaining rigidity.
I went through the trouble of getting Klipper working on my Ender 3 V2 for resonance tuning using an accelerometer (and Klipper's implementation of pressure advance works with the stepper drivers hardwired into legacy mode on a stock V2). It worked incredibly well and the Ender's frame and motion system is surprisingly stiff, allowing the use of the better dampening algorithms that provide compensation without excessive smoothing of corners.
Considering the small amount of sand you used, you could've just bought play sand, ran it though a sieve and blender. That would give you sand as fine as flour.
There is another option I think you should try. Make a 3D printer that has a coreXY gantry for both the printhead and heated bed that move in equal but opposite directions either by connecting both to one motor with a gear or a separate set of motors (assuming the mass of both is roughly equal.) For the same printspeed the nozzle (and bed) only half to move half as far (and half as fast) relative to the motors and frame. Their momentums should also cancel out and reduce vibrations. So in theory less vibrations and lower velocity required to achieve the same print speed.
I'm wondering whether replacing rigid containers with soft sand bags would help. They may allow for more movement and thus improve damping. A more professional solution would be to find out which frequencies affect ghosting the most. Then with a FEM model you can find out what dampers to use (e.g. oil dampers, tuned mass dampers or combination of those) to almost completely negate vibrations. This is a very complex task though, and it would probably fit your printer only since every machine is slightly different in terms of rigidity.
I got a Creality CR-6 SE and I haven’t ever gotten it to work, and I found out that both the extruder and the printing bed are completely different temperatures than what it says on the screen so I’ve been searching for videos to try to help me and I click on this video thinking I might find something, only to find you filling your 3-D printer with sand and hanging it upside down from bungee cords 😂 although your video didn’t help me with my printer, it did lift my spirit, so thank you.👍
This may require a redesign of the entire printer but since everything is belt driven, why not hook up weights on the other sizes of the belts so when the bed moves to one side a weight moves in the opposite direction. Similar to how elevators use weights and pulleys to make lifting easier. Use weights and even pulleys to cancel out as much inertia created by the printers movement on every axis. The weights would have to be highly calibrated. The bed would be the hardest as that gets heavier as the print goes on but, not by too much. This idea even works similarly to how sound cancelling headphones work with air vibrations (sound).
One thing that I found to minimize vibration and stepper sounds is to add a bearing to the other side of the motor. I used a v slot wheel that just happened to fit my set up.
just bolting the machine to a stiff and heavy bench helps degrees of magnitude more than adding sand. If anything, try adding a resin slurry/putty mix. In machining, they use a mix of gravel and resin, here probably metal powder and sand to create a slurry which is heavy and solid. but again, not much mass inside those flimsy aluminium beams
Thick silicone pads are the best. They absorb just the kind of vibrations that 3D printers create. You can DIY them from dried up silicone caulk tube. Just be very careful when slicing, the silicone sausage is difficult to handle. They also hold their shape very well under a load, just let them cure fully if you DIY, a week after slicing so they are properly hardened. Even a dried up silicone caulk can be soft in places..
Tom! You are literally adding weight (sand) to the end of a spring (bungee cord) . That is the analogy used to describe resonance . How do you figure this will dampen? Have you considered RC car shocks with silicone oil. maybe get enough of them to match the weight of an RC car
The energy vs stiffness vs weight debate! By far the best solution to ringing I have found is the resonance dampening of klipper. Everything is made of rubber, and flexes. With clever steering ability (awareness of the controller of the resonance frequencies of the system) I'm currently printing at 180mm/s with 10000mm/s2 acceleration with clean/way better results as with the old style on 60mm/s with 2500mm/s2 acceleration . I'm using tubes, half filled with sand, as reinforcement in my speaker cabinets. Sand works miracles in dampening, especially packed loose.
Best place to add damping is directly to the motion components, in this case the stepper (moving mass), the belt (spring component) or the head (moving mass). Easiest is probably to the belt, at the tensioner or with another idler wheel that has dome resistive (to motion) mounting. Sorbothane, memory foam, felt clutch - like a brake in a car or some other friction mount. Large buildings use dynamic systems, moving a mass in the opposite direction, even water that gets pumped around. This is a fun topic, am enjoying it.
Adding weights to the bed was just a bad idea, it's not been engineered to expect them. Also the sand is loose inside so will create some movement. Would be interesting to add metal weights instead of the sand ones, and not to the bed. To see the impact. Or to workout the rough frequency a 3D printer generates and try some of those car weights.
If you know your speed... measure the distance between the ripples and you can get a frequency. Do you get different ringing frequencies on the X vs Y axes? How about near the edge of the platform vs in the middle?
The sand trick works really well for desktop cnc machines which suffer from high frequency vibration from the cutter. I suspect the outcome her is likely the space being filled with sand isn't large enough, and the fairly low frequency vibrations are not easy to absorb with the sand. I can't really explain why the sand makes it worse
Cheers for experimenting so we don't have to. Curious if/how different dampening materials/fluids attached just to static frame components would help, or not. Detaching power supply and fan cooled controller board. Decouple spool. Basically isolate everything possible.
How about damping grease? Maybe put it on the stepper motor mounts and on any frame connections? I'm not sure if it will be be enough for the lower frequencies of a 3d printer but some put it on the motor mounts of their quadcopters.
Echoing what others have said, take the excess weight off the print bed and other moving parts. Might be an easy thing to test to see if it helps, but as of right now, it looks to me like you may be partially cancelling out any gains you otherwise would have made. That said, my printer is bolted to a heavy MDF slab in my shop with zero dampening, and while ringing does occur, it's barely noticeable. I think mechanical tuning will only get you so far in this case, especially if your feeds and speeds are maxed out. It's definitely an art, and I love it. Thanks for the vid, and your "3D" tesseract did look pretty awesome!
I set my printer on a heavy paving stone that is sitting on top of a thick piece of packing foam. I did it for acoustics but I think it makes a difference in resonance artifacts too. I guess it makes sense that they'd be connected.
Try kitchen cabinet gas struts: these should (sort of?) work like a dampner, either in a base (z) and/or pushing against each other in the same axis (x & y). They are also rated at different forces so you can play with that.
Steel belts, ditch the weights on the bed, and then build flextures that constrain in one axis and allow the printer to oscillate in the other axis. Then test both X and Y axes to see which is better for ringing. Also seeing some models available that have 2 axis suspension feet on top of squash and tennis balls. You really do have the perfect filament to test this stuff with.
With that suspended setup, you're making it work like a washing machine. I think you can try to make it smaller and add some suspension to slow down the movement, like in a washing machine: there are springs to absorb vibrations and small pistons to add friction and reduce the resonance effects. Nice project.
I think you should try strapping the highest point of your printer to anchors on the table. One strap across the X and two straps for the minimum and maximum Y. Get them as tight as you can without buckling the aluminum.
12:28 - 12:32 aren't vibration dampeners. They are CV (constant velocity) joints. Like Universal Joints but more linear. They allow rotation across an angle. The rubber bits on the rear differential are dampeners though, sure. I would really like to see the experiment again but with the sand weights removed from the Y (and X?) axis, just leaving the sand in the frame. Also comparison at a higher print speed perhaps.
To really understand what is going on you would need to do a full NVH survey using triaxial accelerometers to measure the resonance of the frame etc. Using this data you can tune the required masses for the appropriate areas of the printer.
I worked in the computer chip industry and this is what they do to print computer chips at 0.25 microns and less. XYZ stage is on granite slab. That slab is on air bladders and sits on another granite slabs. All cables are mounted with vibration dampening hardware. All motors are mounted with vibration damping hardware. To check for vibration we put accelerometer on each axis ( use one accelerometer 3 times). My touch was so tuned in after 16 years I could put my hand on the stage close my eyes and tell you where the vibration was coming from. Most 3D printers don’t isolate the motors or cables. I recommend silicon dampeners for the motors and cables and for 3D use a tile or concrete tile on silicone dampeners on top of another tile. Air bladder maybe overkill
Model railroading we glue are sand(track ballast) with Elmer’s white glue thin down with water and a drop or two of dish soap to allow the glue to penetrate.
You should just put an accelerometer on the hotend which has built in dead reckoning feature, log all displacements and run a Fourier analysis on it to see the major oscillating frequencies. Then you can try to dampen one or two of them and at least see if you're heading in the right direction. It shouldn't be too time consuming or difficult.
You should try sitting the printer on a paver that has a foam floor mat on it. The paver stopped my table from resonating the printer sound. And the foam mat between the printer and the concrete helped dampen the ringing.
I think you need to put a drone flight controller with blackbox on the frame in various places to get simple gyro recordings. Figure out the frequencies, then engineer dampeners. You can probably buy specific feet that absorb specific frequencies already engineered for you
We have a really weird setup we stumbled across by accident. Our Ender 5 Pro is sitting on an IKEA LACK table with rolls. Not exactly IKEA "RILL" rolls, but very similar. The intention was, to be able to easily move the printer around, but we forgot rolls with stoppers. So, while printing, the whole table started rolling around slightly. To stop it, we put the rolling table on a thick rug, similar to a doormat, which increases friction and prevents it from rolling around. Unintentionally it dampened the whole system so much, that it became a lot more silent and shows a lot less signs of vibration.
Hey there. Great Video! If you look for perfect dampening you could just put all of that sand in a bigger box and put the printer right on the sand. Would be interesting to see if that makes any difference. But I doubt it, I think the problem is not the vibration of the printer as a whole but the vibration of the printer parts (esp. print head and bed) relative to each other.
I don't know what Thomas thought when adding Sand weight to fast moving parts. Naturally, that worsens the performance. It would have been interesting to see if proper addition of sand would have helped.
Those sand containers are a great solution. Did you try adding the sand only to the static portions of the frame?
Sandtainers*
I was going to ask the same question. Leave those heated bead sand containers out.
Surprised he didn't try it without, first, to begin with...
Same question from me too. I'd love to see it only attached to the static frame parts
The ones on the bottom. Of the bed can't be a good idea
Try removing the sand blocks from the moving parts. Keep them on the non-moving parts.
He has to generate comments. That helps getting the clip prompted all over the interwebs.
I was wondering that as well
@@theomoons As they say... If you want help, don't ask a question, simply suggest doing it a certain way, and someone will surely explain a better one. ;)
I still feel like the sand in the frame might marginally helps, but adding the "sand containers" to all of the moving parts really just serves to add weight and increase the problems.
Not sand. It should be ie. lead to even have marginal difference.
Yes. I would have liked to see what would happen if sand was added to just the non moving parts.
Although sand in the frame really is a poor-mans version of just using better aluminum extrusion profiles - i.e. the heavy-duty type instead of light-weight.
@@alexbaeza4370 mhmm mm
@@graealex its not the same… one adds strengths and the other absorbs vibration energyyy
Not documenting "failures" is a huge problem in the medical industry b/c peeps don't want to embarrass themselves. Props to you for posting this. It helps inspire new hypotheses to come and still makes good content 👌
I think we all need to be more supportive of each other making mistakes.
There's an amazing technique called "Motion Amplification" which uses special cameras and DSP to visually exaggerate imperceptible movements in video. It's used largely for identifying sources of resonance in structures and industrial machinery. I'd be fascinated to see it used on a 3D printer. I predict the main source of resonance is the belts.
Wow that sounds so cool. i remember a similar tech where a camera was used to observe a human face and detect imperceptible changes in color of the skin due to blood being pumped and amplify it to detect the heart beat.
@@wizardOfRobots it's the same, it's called Eulerian Video Magnification
Ot watch Mathias Wandel checking vibrations with a loudspeaker and oscilloscope. Wandel is the master.
U,u
@@BerraLilltroll I saw that video (or a similar one) recently and was impressed with his technique and results. As I recall he was balancing a rotating mass with weights, but the technique could be used to identify resonate frequencies and test various damping solutions without wasting filament or waiting for test prints to finish.
In the machine tool world, "epoxy granite" is used to fill machine frames to increase rigidity and dampening ability. It's basically just epoxy resin and sand. What the epoxy adds is it connects the mass of the sand to the structure, making it one. The aluminum extrusion with sand in it will ring nearly as much as without the sand as the sand is able to just move and allow the aluminum to move (ring).
True, however cast iron also has significantly different characteristics to aluminum extrusions. Plus, the lighter the machine then the slower / shallower you need to go. Even a mini lathe weights 40+kg.
@@arthurmoore9488 So you add mass to non moving parts only. If you're fairly certain you're not going to fiddle about with your design, you could even fill in the outer moldings.
The sand on a rapid-moving part was totally a mistake on my opinion. Getting down the rabit hole you could try with some mineral oil, like the ones that are inside car/bike suspensions...?
Those dampeners only work at high amplitude, low frequency vibration.
The oil in dampers only works because it gets pushed through a valve. How would you install such a valve in a 3D printer.
Yes the increase weight on the bed destroy the test.
Moving parts = as light as possible
Frame = as heavy as possible
@@VanDerPol watercooling tubing...?
Filled with oil it would still act like a solid.
FYI: I put a FDM printer on a $20K active dampening vibration table we use for laser alignment and it made no difference. I think the issue is amplitude and frequency is not consistent.
What about hanging the printer from a pivot point, with a mass dampener at the other end of the pivot point? A sort of Foucalt's pendulum with a mass dampener.
What do you mean by a pivot point in printer?
I think this is the comment closest to identifying the fundamental challenge. There must be multiple resonances in the printer structure and there are certainly a multitude of stimuli when you consider acceleration and vector forces. For a Cartesian system, it is probably a problem constrained to two planes; XY and ZX. I think it’s a lot harder than it looks. The analysis would be fairly involved, I think.
Shouldn't input shaper be able to cancel out all frequencies?
Aren't vibration damping tables mostly about preventing vibration from the environment from reaching equipment? Or is it designed for vibrating equipment as well?
Ultimately, I think the problem is that most of the vibration is coming from the carriage itself, which is only loosely coupled to the rest of the system, and by definition needs degrees of freedom.
The stepper and belt is the culprit.
And this time alot driver help eliminate example trinamic.
Hmm, makes sense that you shouldn't put weight on the moving parts. So I'd try the same sand setup but excluding the print bed since it's moving quickly and weighting it gives more inertia so a higher probability of skips.
This for sure. I think that will help alot.
I had the same thought. Direct drive units are intentionally light weight to avoid inducing ringing.
Also makes sense that you should put weight on the moving parts to generate a lot of comments and viewer engagement for the algorithm.
I’ve put my printer on a square of *memory foam* . Using tick square of plywood on top so the printer doesn’t just sink into the memory foam. It works great & definitely makes the printer quieter.
I do the same, except I use a concrete tile instead of plywood and some cheap mattress foam from Amazon underneath
Tom, this is such a great video! No meaningful results are still results, just as you said. Only you would build a Suspensorium Love it! I'm chasing resonance reduction as well and would be interested in how different belts affect things. You are both thorough and entertaining and it is always a pleasure watching your videos.
How about input shaper?
"No meaningful results are still results" now i know its a bad idea to fill my printer with sand.
I wouldn't describe the results as non-meaningful. They were not 'positive' in the sense that there was no improvement in the printing artifacts due to 'ringing.' But it provided some evidence that some of the 'simple' fixes do not appear to work as miracle cures, and in some cases actually made things worse.
tuned mass damper are also commonly used in skyscrapers to absorb vibrations from swaying in the wind. there's one in Taiwan where the architects famously decided to make it a tourist attraction - instead of tucking a concrete block away in some utility room they made the main damper a large shiny decorated sphere hanging in the middle of a big atrium.
I would recommend removing the sand from the print bed. Then, run the tests again: With Feet, without feet, and then suspended.
This series is great, and your willingness to be honest is appreciated.
Agree, but the suspended one is just for fun, there is no logic on continuing that other than entertainment.
@@brianfhunter actually the suspended test could be coupled with and earthquake reader and it can plot the sine wave of the vibrations.
Then with the total mass of the 3D printer known, plus it's rought moment of inertia, you can calculate how much damping it needs
Lol, you seemed so tired & disappointed by the end of the video, but you make a good point that even the failures are useful information! This has taught me that if I want to deal with vibration, I'm better off putting my time into brushing up on my old engineering courses rather than on messy, half-baked quick-fixes. Thanks for that!
Love your honesty one of the reasons why we come to you you always keep it honest so we all don't go out to the store and buy lumber and try to make what you made lol I can only imagine how many people be trying to pull out their hair cuz they couldn't get what you have thanks again for keeping it honest and real
Yes definitely do a video on the belts please ...... Since I starting playing around with 3d printing and my velleman k8200 many years ago you have taught me so much, and are normally the go to videos when I have a problem or just want to improve something. Keep them coming and hope to see you around for many more years to come
THANK YOU Tom
So I am not the only one that got hurt playing with the k8200. Still got mine. But I don't use it since I got my Ender3 original. I since replaced to pcb on the Ender. I could maybe use the old pcb into the K8200 and try to have it run better.
@@pierremartel3552 My 8200 was a good printer even thou I have done a lot of upgrades, it was working fine up to a few months ago when I parted it out to build a bigger n better printer from scratch.
It was the 8400 that I scraped after a few prints when the MB froze up putting the hot end into a thermal run away ..... after a hard reset it was still showing the hot end at almost 400c
Interesting ideas! It might work better to put the 3d printer on legs that stick into a bucket of sand, so that there is lots of sand to absorb vibrations in the base. I am curious to see the result of a test like that!
I was going to suggest the same thing. Just place the printer in a sand box and see what happens. Hopefully it won't dig itself in. :-)
I was going to comment with the exact same thing - this feels intuitively like a good move. If you had a large, square container filled with sand and just sat the printer in it, I think you might see a difference.
(Probably too slight for it to be worth it, but at least it would be cheap to try.)
I think the best solution would be to have a counter weight on the same belt as the X or Y.
The top of the belt moves the opposite direction as the bottom of the belt.. Let's say you have a weight on the X that weighs the same as the head.. They would move opposite directions, and hence when they stop, should vibrate the machine.
Just a thought.
Or you could try Input Shaper, I sure hope that's where this is going, try a bunch of physical solutions, show they don't really work, then show the absolute magic of Input Shaper tuned with an ADXL.
Less initial vibrations for Input Shaper to negate would lead to sharper prints as it'll apply less smoothing. So removing as much resonance before using Input Shaper is the way to go.
It's a shame this test wasn't shown without the sand containers on the bed axis, as that's somewhere you should be trying to remove weight on this style of printer.
@@howaboutbecause686 Exactly, the GiGo principal still applies no matter how good your software compensation, garbage in garbage out.
@@m3chanist How dare you insult the best 3D printer to ever see the light?
Need stiffness + weight everywhere in the frame to shift the frequencies higher. That being said most ringing is due to how stepper motors work. Throw an oscilloscope on the stepper motor, then run Input shaper and compare.
More weight shifts frequencies to lower
I find most of the ringing comes from motor, belts, pulleys. Just lubricating and stabilising makes a huge difference.
I just bought some motor bike gel pads, cut them to roughly 5x5x2cm and put them under the printer. They have the advantage of being able to move in all directions and completely removed all ringing.
The printers still stand on massive, 5cm thick sandstone slabs and those are on rubber mats, which results in an de-coupled, non resonating table and a still free moving printer
While moving fast, you can see the gel visibly shaking and shifting the printer as it absorbs the energy, but the printer never moves on the gel itself.
this might be worth a try if you have ringing and/or noise issues
printer: prusa i3 mk3s
Have had a similar plans. Was thinking of using a section of 1-2cm thick exercise mat under a large concrete paver with gel on top. Do the feet of the printer dig into the 2cm gel pads?
Thomas should test this.
@@Trevellian I removed the feat for this exact reason, didn't even try it out.
One thing I found out is that the size of the gelpad has a direct influence on the movement of the printer, the smaller the pad the more it moved. I sliced them down from 5x10cm to 5x5cm and the change was noticeable when pushing the printer to one direction
it is important that you enforce the table you put the stone on since it weighs a ton and it will bend your (possibly cheap) table.
I enforced mine by placing a solid 3cm thick wooden board on top (no screws required)
Another great video Tom.
This is just a collection of thoughts.
Sand alone isn't great at dampening. It's coarse and rough and irritating and it gets everywhere. Solution Epoxy "Granit" (a with Epoxy damped mixture of sand that is filled and compresses in forms or extrusions to minimize resonances and vibrations on Professional Machines like Kern cnc mills)
Weight alone helps a bit but Weight and a Dampener works great. I would love to see the printer mounted to the paver and than placed on a 2 to 5 inch think piece of upholstery foam.
Oh and i think stifer belts will make a noticabal difrence (steel cored belts, heard that there are some out there in the wild)
Keep up the great work and i cant wait to see the next episode in this series and the escalation that it could bring with it.
Stop hidiing Anakiin, we all know it's you. Although you bring a fair pont about the resin but it defeats the purpose of dampening. What makes sand such a great dampener is the fact that it's loose particles that have thousands of points of friction for the energy to dissipate into heat, having it all merge with the resin just make's one solid mass that, albeit having greater inertia and therefore harder to get moving, has a single natural frequencey that it can resonate with.
@@coolbionicle I disagree unfortunately. Having a Mass that not solid makes it very hard to predict the result and I cant find a mechanical loss coefficient for loose sand.
Besides that, the “granit” part of the epoxy graint its mostly a filler too minimize the amount of epoxy needed (epoxy = expensive).
The epoxy provides the dampening, the Aluminum frame the needed stiffens for the machine and the Sand is just a filler.
If you don't trust me go over to Adam bender her on you tube and watch his How to Build Epoxy Granite Machine Base video. He goes in more detail with Loss Coefficient vs. Young's Modulus and how to find the right ratios of sand to epoxy.
I think you may get some interesting results if you modified the springs on the tesseract. What it is now is a vibration isolator not a damper. All of the vibrations made by the printer stay within the system. If you think about vibration dampers on cars there are two parts, the spring and the actual damper or shock absorber. The tesseract set up is only spring and does not have the best damping qualities.
Thank you for going through everything that didn't work. I've been really curious about sand on smaller machines. I'd love to see a video on the different belts. If you could please have one cheaper belt case using an old piece belt that would be great. I honestly wonder about my belts every time I see ringing.
Would absolutely love to see the belt comparisons
Great video as always, always learning from your content. Keep it up
I agree. I would love to see the belt comparisons too.
@@AntiVaganza I'm fine with feeding the algorithm to help a channel that is good
Very interested in the belt comparison video. I found early on in continuous belt platforms (CoreXY) that there are parametric resonances that crosstalk in abrupt vector changes, leading to ringing artifacts. I used accelerometers and a 35670a dynamic signal analyzer to capture the response of the gantry vs the head. Data is on e3d's forum. I came to believe, but did not follow up on the idea of tuned dashpots that the belts can ride on... (shock absorbing idler pulleys)... might be a trick to try. Thanks for the hard work, Tom!
Very cool video. Not vibration or resonance related, but one thing i have been curious about for a long time is how flipping the printer upside down would affect overhangs and bridges. Since any sagging will be pushed back up by the next pass of the nozzle, I would expect it to improve the results. And now you have the perfect "printer enclosure" to test it 😄 Also seeing the effect of different belts would be very interesting.
Wouldn't you literally just have the overhang problem on every layer?
@@robertbyrnes7075 I don't think so. The molten plastic is probably viscous and sticky enough to stay in place wherever it is fully supported by the previous layer. And in areas along slanted edges where it is not fully supported it will sag towards the nozzle. Which means that on the pass for the next layer the nozzle will lift the sagging but now much cooler plastic up again. As opposed to pushing it further down when printing with the nozzle pointing downwards. I find it hard to predict the end result of this interaction, but I think it might make it possible to print steeper overhangs with better quality.
Need a way to measure resonances and their vector/directions. Then matching those with damping would be clearer.
Back in the late 1980s, some of the scientific equipment vendors, such Edmund Scientific, sold rubber ball kits for physics demonstrations. One ball would be a standard rubber ball and would bounce as such. The other ball would sometimes be called a dead ball as it wouldn't bounce. The idea was to show how energy could transferred in one direction (bounce) or dispersed in multiple simultaneous directions (dead). At that time, the dead balls were of particular interest in the amateur holography scene as lasers weren't very powerful and the holographic process was extremely sensitive to vibrations. I don't know if the dead balls or material are still available but if still available, that could be an interesting test.
They are called squash balls and are available from most sports shops
I have observed quite a bit of NVH testing. The bottom rails likely didnt harm much, but the vertical supports having sand added was the main issue making prints worse. Adding mass to cantilevered components will just shift the frequency of the resonance. More mass generally = more resonance at lower frequencies (greater deflection distance). Adding reinforcement to the gantry along with soft mounts that absorb energy will help most. A support rod run from the top of each vertical support, to lower frame would mitigate the motion of the gantry. Soft materials used under the machine would help absorb vibrations from print head motion.
You could try small dampers from RC trucks. They are sprung and oil filled. This will actually adjust the frequencies instead of just adjusting the amplitudes. You need to absorb some energy, elastic straps provide almost no damping effect.
I dont understand the fascination with optimizing single variables. It would be far more effective to increase the stiffness of the frame by putting braces on it or adding gussets at the connection. Damping a wet noodle is pointless. Brace it until it is rigid the dampen it. Otherwise you are finding ways to dampen a frequency that is irrelevant and ineffective.
At least the oil filled damper would effect a far wider spectrum of frequencies and it is tunable using different weight oils.
Small dampers for belt tensioners, shouldnt help with skips but maybe will with resonanses. For skips you decrease moving mass or increase motor power.
@@Gebsfrom404 Wasnt thinking for belts. The exercise in the video is about dampening the vibration of the frame. I was thinking if they wanted to put the frames on an elastic base then could mount the dampers to the table. This way the elastics or flexible bases allow movement but the oil filled dampers absorb vibration. As for belt tensioners I ended up using a small coil spring from a brake drum hardware kit (had a lot of them)... it is used as a retention spring to hold the brake shoes to the backing plate... around 19mm diameter and 25mm tall compression spring... pretty stiff. Then I preloaded it with a threaded rod running through it to a yoke holding the belt pulley. Any compression spring would work as long as it is stiff enough to maintain control of the belt.
Should have used lead shot as is used in speaker pedestals - more mass and less leaky than sand.
Ultimately the problem is with elasticity within the printer frame and mechanism.
The lab robots I work on use large stiff belts that don't stretch to achieve precision.
If I could be bothered to do it , I would reproduce the printer frame in welded steel RHS. That would eliminate a lot of the flex. Then the only thing to deal with is the flex in the mechanical drives. You could try wider belts.
Most modern motorcycles have vibration dampers in the handlebar ends and often on other parts of the frame
Great ideas, Thomas. The super hero prison made me laugh out loud, very accurate description of how it looks lol
Just a piece of advice for measuring squares, you can do the triangle method. 3:4:5, is the ratio your sides should be at and the distance between them. Its a commonly used carpentry trick for getting perfect 90 degree angles.
Anakin hates this video so much
Since the idea is to apply dampening to counter the moving parts, and you already have the suspension rig, try loading two printers one facing upright and the upside down one facing the opposite direction... you want to load the same program into both and start them running (the bottom one does not need to be actually printing). The bottom printer will be moving in the opposite direction to the top one and should mostly cancel out the movements of the top one. Might be a relatively easy experiment to try out.
Why did you add weight to the moving parts?
Obviously that wouldn't work.
The question was if sand in the *non*-moving parts would make a difference.
Any experiment that contributes to knowledge is always welcome. Whether the output is positive or negative, it is still knowkedge that can used in future endevours. Please continue with the belt swaps to see if there is any improvement to be gained on the prints at high speeds/accelerations. Thank you Tom.
Something I did to my printer, and I know this isn’t on the scope of which you made this video cover, was I replaced the x carriage. It originally was a heavy thick three part steel. Now composed a singular piece of aircraft grade aluminum of the same thickness. The shock went way down as the bed now shifts. Knowing this and then applying the issue of resonance, wouldn’t a solution of determining the weakest material causing the slop (artifacts) be best to replace? IE blue loctite all frame bolts. Reduce slack in bearings and such? Stiffer springs on heat bed?
I did the sand earlier this year. You need to tamp the tube until the sand settles. You can add a significantly more amount of sand when you do. All it takes is tapping the sides of the extrusion to make it vibrate.
What if we put equivalent weight on the belts..... So for example for the bed.... At the belt we can add the same weight as the bed just bellow the center of the bed on the belt.... So now when the bed moves in one direction.... Weights move in the opposite and kind of cancel out the effects.... Same can be done for all the belts..... That will kind of act like a balancing system for reciprocating components and maybe change the frequencies.....
Might be worth noting that springs do not dampen shocks or movement, they only absorb it, and if not dampened by a shock absorber, they can make the effect more pronounced.
Try driving a car with blown shock absorbers. The springs still work, they just keep bouncing after you've hit the bump, and your car will be all over the road. The shock absorber is there to slow and halt the movement of the spring, after the bump has past.
Sand filled feet are a good, cheap and decently working solution. There are many different sand filled balls you can use with a printed holder to keep them in place below the printer
H! Great work - being systematic as presented makes the most sense in such a research. Different belts? - Hell Yes! I've switched to genuine Gates after factory Creality went bad - I've noticed less ringing - yet that could be biased. God tool to help would be Klipper Resonance Compensation, and it's measurements with plots. I've observed changes while belt tension being alternated, or X axle configuration changed.
Best regards!
I cut the molded foam and box my E3V3KE came in and placed it in that. No problems so far. Not sure even if it helps but it's peace of mind.
Had a good time watching this, thanks. I found vibration dampening a bit annoying too, especially when you have a wide range of excitation.
Nvmd, white steel belts on heavy beds vs cheap or kevlar would be interesting indeed.
The sand will add more mass to the printer, making it still vibrate, but because of the mass, the amplitude will be much less. You should add lots of weight to the top crossbar of the printer. This should greatly decrease the amplitude of vibrations.
Also, locate the filament spool somewhere else. If the spool rocks on the printer that could translate to the print quality.
I've got a delta printer and this video has got me thinking I should just load up the top of the printer with something to keep it from moving around as much.
With that many skipping steps, I'd start looking into bumping up the stepper motor voltage via the driver pots. I'm curious whether fixing those skipped steps make the 'sandtainer' approach work at all, if only because stuffing them wherever they fit just kinda looks cool.
But also, resonance tuning in software is way easier and very effective.
Keep this series going please. Super enternaining and gets everybody's gears going. What about non-symetric dampening setup? It could help with reducing resonation of the dampening system itself. For instance - your tesseract cube with some cables tensioned way more? Or usual dampening rubber feet but with one odd, softer/stiffer foot?
A damping mechanism can reduce high frequency vibrations, but it also enhance low frequency ones. Most of the vibrations coming from the moving nozzle are low frequency ones, so the dampener could make things worse.
Wow!
I really thank you soooo much for showing all this and still be so sympathetic! Cheers a lot!
TH-cam in parts seems to be more honest than most the research journals - also presenting what did not work, or what did not make the greatest WOW !!! effect. thanks so much
I am an electrical engineer so I haven't studied this topic, but I would first attach vibration sensors all over the frame and heating bed and do frequency analysis and then designed springs with resonant frequencies matching the strongest frequencies from the analysis.
Tom, dumb question as I haven't checked through all of your other experiments, yet - what have you done with the bed springs?
Stock Ender 3 springs are woeful, I use the regular yellow upgrade springs on all of my printers, with a LOT of compression to eliminate some print bed wobble, these springs are better finished than ones fitted to cars - the ends are ground flat - [N.B. car springs break BECAUSE the ends are not finished properly so it transfers all the stress to the end of the first coil, saves the maker money, costs us for replacements]
The other thing I do is fix the M4 countersunk [flat head] screws to the print surface from below, using M4 nylon nuts - this fixes the screws to the print bed, don't use metal nuts without nylon washers - you will damage the insulation surface or worse, short the bed to the frame, not unlike what happened with the early CR6SE's.
Finally, after fixing the screws to the bed, putting the bed back onto the frame becomes difficult because the clearances on the screws are tight, so I enlarge three of the four holes with a 5mm drill.
On the UH60 BlackHawk Helicopter, we have tuned masses on springs that we test and tune to the aircraft to dampen unwanted vibrations, on the new model, there are counter rotating electromechanical vibration dampers that actively cancel unwanted frequencies.
FWIW, I put my Prusa i3 mk2 on a paving stone and under the paving stone I put a rectangular inflatable chair cushion (i.e. anti bedsore inflatable cushion for a wheel chair). Primarily I did this to reduce the sound, since the homemade wooden workbench I set my printer on was amplifying the vibrations of the printer. I can pretty much only hear the loud stock fans now.
Sand was common for audiophiles for a short time to deaden platforms and it did work but then things like soundproofing sheets ended that.
One thing I am testing is turntable spikes/isolators. They are little 2 part metal pucks that either have a spike that fits In a cup foot or use a steel ball between 2 metal plates. They work well and so far are stopping a lot of transfer from the printer to the table.
This was fantastic. And thank you so much for testing things out. I was very curious as to the outcome, especially the sand damped tests. It also makes sense that BMW used such a complicated series of dampers. But in the industrial and aerospace worlds, there is a simpler solution - avoid the areas of operation that induce harmful resonances! In a light aircraft, this could be as simple as a plaque that says "Do not run propeller at RPM between 2500 and 2550 for more than 5 minutes". Or for a larger aircraft, a redesign of the vibration causing components, to shift their resonances rather than compensate in other areas. Where the moving masses are controlled by stepper motors, that's the obvious place to control vibrations. Match the speeds and accelerations to avoid areas of resonance. That's not cheating or a hack! That's proper design. Love your channel and the work that you do...not intended as a slight at all. That I learned that a printer will work upside down is worth more than the price of admission. To me that means it's just a matter of time before there's a 3d printer in space.
I love your Sciency thinking. Please keep valuing it as this sets you greatly apart from typical TH-cam content we all got used to!
- You briefly mentioned the additional mass on the printbed might harm the print quality. I totally agree. I would have tried again withouout those printbed Santainers, maybe even adding them elsewhere…
- I highly value the additional mass on the printer. The bigger the mass difference between bed and printer body, the more recoil you can convert into printbed acceleration.
- A tip from back in the days (experience with laser interferometry in physics lab): For effective (nearly critical) dampening of alu extrusion profiles instead of pure frequency shifting fill the profiles with silicone (not silicon - Fugensilikon, nicht Silizium ;-) This also greatly assists survivability of your mechanical components when compared to sand…
I would love to know the frequency range of problematic ringing. This would hugely narrow down the root of the evil (profile bending, profile ringing, belt stiffness etc.) okok, I’m getting to scientific. Looking forward to the engineering way of life: try it out! :-D
Thanks for the attempt Thomas! I know I appreciate all the effort you put in these videos!
potential different question-problem pertaining to ringing in model layer toolpath with SHARP corners; it might even not be ringing of the frame or extruder itself in motion to make those effect once the large one is solved and you seek for more. on sharp corners, the extruded line is basically circle shape with offsets that gets squished by its own material in the toolpath; won't it be the extruded material amount itself trying to equalize within the toolpath?
on that end, I simply model things with sharp corners filleted/rounded, so not only ringing is solved, but tolerancing is simpler and more effective
The last time i do something similar i made a weight, attached with some belts/rope to every axle so that it have some "Vorspannung". This improves the quality a lot.
Of course with pulleys so that gravity can work in X and Y Axis
If you're chasing after resonances, tuning the structure will get you into diminishing returns fairly quickly. What you want to do is ramp all of your steppers up and down so that resonance never has time to build. Remember, one movement of the printer is easy to dampen out, it becomes a problem when you have many movements happening rhythmically that case resonance to build to the point where they cause significant movement.
The other thing to do is to decrease the mass of your moving elements. The ender 3 is a poor platform to perform that modification on, however, something like a voron build gives ample opportunity to drill out 'speed holes' while maintaining rigidity.
I went through the trouble of getting Klipper working on my Ender 3 V2 for resonance tuning using an accelerometer (and Klipper's implementation of pressure advance works with the stepper drivers hardwired into legacy mode on a stock V2). It worked incredibly well and the Ender's frame and motion system is surprisingly stiff, allowing the use of the better dampening algorithms that provide compensation without excessive smoothing of corners.
Considering the small amount of sand you used, you could've just bought play sand, ran it though a sieve and blender. That would give you sand as fine as flour.
There is another option I think you should try. Make a 3D printer that has a coreXY gantry for both the printhead and heated bed that move in equal but opposite directions either by connecting both to one motor with a gear or a separate set of motors (assuming the mass of both is roughly equal.) For the same printspeed the nozzle (and bed) only half to move half as far (and half as fast) relative to the motors and frame. Their momentums should also cancel out and reduce vibrations.
So in theory less vibrations and lower velocity required to achieve the same print speed.
I'm wondering whether replacing rigid containers with soft sand bags would help. They may allow for more movement and thus improve damping.
A more professional solution would be to find out which frequencies affect ghosting the most. Then with a FEM model you can find out what dampers to use (e.g. oil dampers, tuned mass dampers or combination of those) to almost completely negate vibrations. This is a very complex task though, and it would probably fit your printer only since every machine is slightly different in terms of rigidity.
I got a Creality CR-6 SE and I haven’t ever gotten it to work, and I found out that both the extruder and the printing bed are completely different temperatures than what it says on the screen so I’ve been searching for videos to try to help me and I click on this video thinking I might find something, only to find you filling your 3-D printer with sand and hanging it upside down from bungee cords 😂 although your video didn’t help me with my printer, it did lift my spirit, so thank you.👍
Maybe a bad thermistor or not fitted right. Sometimes it also helps to add thermal paste.
Thanks for the help I’ll try that
This may require a redesign of the entire printer but since everything is belt driven, why not hook up weights on the other sizes of the belts so when the bed moves to one side a weight moves in the opposite direction. Similar to how elevators use weights and pulleys to make lifting easier. Use weights and even pulleys to cancel out as much inertia created by the printers movement on every axis. The weights would have to be highly calibrated. The bed would be the hardest as that gets heavier as the print goes on but, not by too much. This idea even works similarly to how sound cancelling headphones work with air vibrations (sound).
One thing that I found to minimize vibration and stepper sounds is to add a bearing to the other side of the motor. I used a v slot wheel that just happened to fit my set up.
just bolting the machine to a stiff and heavy bench helps degrees of magnitude more than adding sand.
If anything, try adding a resin slurry/putty mix. In machining, they use a mix of gravel and resin, here probably metal powder and sand to create a slurry which is heavy and solid. but again, not much mass inside those flimsy aluminium beams
That look of concern at 3:37 XD
Thick silicone pads are the best. They absorb just the kind of vibrations that 3D printers create. You can DIY them from dried up silicone caulk tube. Just be very careful when slicing, the silicone sausage is difficult to handle. They also hold their shape very well under a load, just let them cure fully if you DIY, a week after slicing so they are properly hardened. Even a dried up silicone caulk can be soft in places..
Tom! You are literally adding weight (sand) to the end of a spring (bungee cord) . That is the analogy used to describe resonance . How do you figure this will dampen? Have you considered RC car shocks with silicone oil. maybe get enough of them to match the weight of an RC car
The energy vs stiffness vs weight debate!
By far the best solution to ringing I have found is the resonance dampening of klipper. Everything is made of rubber, and flexes. With clever steering ability (awareness of the controller of the resonance frequencies of the system) I'm currently printing at 180mm/s with 10000mm/s2 acceleration with clean/way better results as with the old style on 60mm/s with 2500mm/s2 acceleration .
I'm using tubes, half filled with sand, as reinforcement in my speaker cabinets. Sand works miracles in dampening, especially packed loose.
Best place to add damping is directly to the motion components, in this case the stepper (moving mass), the belt (spring component) or the head (moving mass). Easiest is probably to the belt, at the tensioner or with another idler wheel that has dome resistive (to motion) mounting. Sorbothane, memory foam, felt clutch - like a brake in a car or some other friction mount. Large buildings use dynamic systems, moving a mass in the opposite direction, even water that gets pumped around. This is a fun topic, am enjoying it.
Adding weights to the bed was just a bad idea, it's not been engineered to expect them. Also the sand is loose inside so will create some movement.
Would be interesting to add metal weights instead of the sand ones, and not to the bed. To see the impact.
Or to workout the rough frequency a 3D printer generates and try some of those car weights.
If you know your speed... measure the distance between the ripples and you can get a frequency.
Do you get different ringing frequencies on the X vs Y axes? How about near the edge of the platform vs in the middle?
The sand trick works really well for desktop cnc machines which suffer from high frequency vibration from the cutter. I suspect the outcome her is likely the space being filled with sand isn't large enough, and the fairly low frequency vibrations are not easy to absorb with the sand. I can't really explain why the sand makes it worse
Cheers for experimenting so we don't have to. Curious if/how different dampening materials/fluids attached just to static frame components would help, or not. Detaching power supply and fan cooled controller board. Decouple spool. Basically isolate everything possible.
You can take a osci with a piezo and messure the frequency and than buy a right piece
How about damping grease? Maybe put it on the stepper motor mounts and on any frame connections? I'm not sure if it will be be enough for the lower frequencies of a 3d printer but some put it on the motor mounts of their quadcopters.
Echoing what others have said, take the excess weight off the print bed and other moving parts. Might be an easy thing to test to see if it helps, but as of right now, it looks to me like you may be partially cancelling out any gains you otherwise would have made.
That said, my printer is bolted to a heavy MDF slab in my shop with zero dampening, and while ringing does occur, it's barely noticeable. I think mechanical tuning will only get you so far in this case, especially if your feeds and speeds are maxed out. It's definitely an art, and I love it. Thanks for the vid, and your "3D" tesseract did look pretty awesome!
I set my printer on a heavy paving stone that is sitting on top of a thick piece of packing foam. I did it for acoustics but I think it makes a difference in resonance artifacts too. I guess it makes sense that they'd be connected.
Try kitchen cabinet gas struts: these should (sort of?) work like a dampner, either in a base (z) and/or pushing against each other in the same axis (x & y). They are also rated at different forces so you can play with that.
Steel belts, ditch the weights on the bed, and then build flextures that constrain in one axis and allow the printer to oscillate in the other axis. Then test both X and Y axes to see which is better for ringing. Also seeing some models available that have 2 axis suspension feet on top of squash and tennis balls. You really do have the perfect filament to test this stuff with.
With that suspended setup, you're making it work like a washing machine.
I think you can try to make it smaller and add some suspension to slow down the movement, like in a washing machine: there are springs to absorb vibrations and small pistons to add friction and reduce the resonance effects.
Nice project.
I think you should try strapping the highest point of your printer to anchors on the table. One strap across the X and two straps for the minimum and maximum Y. Get them as tight as you can without buckling the aluminum.
12:28 - 12:32 aren't vibration dampeners. They are CV (constant velocity) joints. Like Universal Joints but more linear. They allow rotation across an angle.
The rubber bits on the rear differential are dampeners though, sure.
I would really like to see the experiment again but with the sand weights removed from the Y (and X?) axis, just leaving the sand in the frame. Also comparison at a higher print speed perhaps.
You got lots of points! Sometimes involving plastics or other absorbent material. Thanks for making this video.
Hot glue is great for what you did there. A few drops of IPA on it when / if you want to remove it will pop it right out.
To really understand what is going on you would need to do a full NVH survey using triaxial accelerometers to measure the resonance of the frame etc. Using this data you can tune the required masses for the appropriate areas of the printer.
I worked in the computer chip industry and this is what they do to print computer chips at 0.25 microns and less. XYZ stage is on granite slab. That slab is on air bladders and sits on another granite slabs. All cables are mounted with vibration dampening hardware. All motors are mounted with vibration damping hardware. To check for vibration we put accelerometer on each axis ( use one accelerometer 3 times). My touch was so tuned in after 16 years I could put my hand on the stage close my eyes and tell you where the vibration was coming from. Most 3D printers don’t isolate the motors or cables. I recommend silicon dampeners for the motors and cables and for 3D use a tile or concrete tile on silicone dampeners on top of another tile. Air bladder maybe overkill
Model railroading we glue are sand(track ballast) with Elmer’s white glue thin down with water and a drop or two of dish soap to allow the glue to penetrate.
You should just put an accelerometer on the hotend which has built in dead reckoning feature, log all displacements and run a Fourier analysis on it to see the major oscillating frequencies. Then you can try to dampen one or two of them and at least see if you're heading in the right direction. It shouldn't be too time consuming or difficult.
It's just input shaping, it is well known
You should try sitting the printer on a paver that has a foam floor mat on it. The paver stopped my table from resonating the printer sound. And the foam mat between the printer and the concrete helped dampen the ringing.
I think you need to put a drone flight controller with blackbox on the frame in various places to get simple gyro recordings. Figure out the frequencies, then engineer dampeners. You can probably buy specific feet that absorb specific frequencies already engineered for you
We have a really weird setup we stumbled across by accident.
Our Ender 5 Pro is sitting on an IKEA LACK table with rolls. Not exactly IKEA "RILL" rolls, but very similar.
The intention was, to be able to easily move the printer around, but we forgot rolls with stoppers. So, while printing, the whole table started rolling around slightly.
To stop it, we put the rolling table on a thick rug, similar to a doormat, which increases friction and prevents it from rolling around. Unintentionally it dampened the whole system so much, that it became a lot more silent and shows a lot less signs of vibration.
Hey there. Great Video! If you look for perfect dampening you could just put all of that sand in a bigger box and put the printer right on the sand. Would be interesting to see if that makes any difference. But I doubt it, I think the problem is not the vibration of the printer as a whole but the vibration of the printer parts (esp. print head and bed) relative to each other.
I don't know what Thomas thought when adding Sand weight to fast moving parts. Naturally, that worsens the performance. It would have been interesting to see if proper addition of sand would have helped.