The awesome feeling of designing and manufacturing your own design never gets old! Try Onshape, the feature-rich cloud based solution and support 247printing by clicking: Onshape.pro/247Printing
Wn =sqrt(k/m) Where Wn is the resonant frequency, k is the stiffness and m is the mass. In order to improve input shaping results, you have to aim for the highest possible frequency. That means you not only have to reduce mass, but also increase stiffness!
So skeletonized titanium, gotcha (Or arguably alloy steel (which one is debatable but a few come to mind), I assume it would since while it's denser most of that mass is gone and it's significantly stiffer, brass, nickel, and zinc would also work since they should all be stiffer than aluminium [at same dimensions]).
Very nice work! As you see with your natural frequencies and maximum recommended accelerations, what matters is the stiffness to mass ratio. Even with better stiffness to mass ratio of the CF, hyperbeam and ultrabeams in isolation, once you add the mass of the hot end, the complete performance may be worse since the alternatives have an absolute lower stiffness than the default. If you had absolute stiffness included in addition to the stiffness/weight I think it would be more apparent. It would be nice to know eventually what size of beam makes this crossover point viable for using the lightweight beam designs.
This is pretty much what I was going to comment too. The acceleration is lower with those lighter beams so they don't flex too much. It would be pretty interesting to see what would topology optimized beam look like.
yep too much stiffness reduction in comparison ... Another factor might be a reduction in damping of the beams themselves although I expect the impact to be smaller in comparison to the stiffness reduction.
Lightweight only helps if your motors have reached their maximum acceleration values due to inertia. His motors are clearly able to overcome the inertia of the heavier, stiffer beam. He will see practically zero improvement by trying to make lightweight parts. This is wasted effort in this instance.
@@UnitSe7en True, but as the printer size increases and the relative mass of the beam to the hot end gets larger, maybe this solution would be worthwhile (perhaps on a Voron 2.4). I guess that can be settled by future testing or at least a simulation if the overall stiffness/mass are known. From most of what I have seen, the melting and cooling rates of polymer will start to limit performance more than mechanical resonances in these smaller fast printers.
@@UnitSe7en exactly. This sort of research has already been done in the case of speaker drivers/larger magnets/stiffer-lighter weight cones. It makes no difference if your motor (its called that in speakers too) is already able to achieve the required acceleration to cope with the demands of the frequency being requested. It turns out to make no difference at all to make the cone lighter. Here, longer beams will make this problem worse. If its already of little yo no benefit on the v0, larger sizes will see more problems, as rigidity is lower already on the shorter beams, it will be more apparent on longer ones.
Awesome video, I love seeing the science and explanations behind these types of things. I ordered a FYSETC carbon beam for my 2.4, I'm excited to see if it makes a difference, but it at least looks cool.
I appreciate the video, but right at the start when you were comparing the percentage weight reductions, my brain was screaming, irrelevant without comparing the total mass of the moving parts. Of course you said this at the end.
You need to start the comparison somewhere - I did directly on the component. I also stated already in the video, that those beams don't make sense :-)
this! especially considering printers like the k3 work with no beam at all, it’s pretty obvious that this really doesn’t matter on such a small printer
Great video! "Because" definitely plays a part in many our customization's. This is my take as someone that isn't concerned with chasing maximum speeds. Its a relief to know that the other beams available for V0. Have no downsides to them in regard to performance. I would hate to once again buy something shiny because I wanted it, and find it cost me in other areas.
Had a feeling these were the results we would see based off the input shaper graphs people have shown with these beams on the Voron discord. Many are getting worse graphs on larger printers (350mm V2.4 for instance) with the skeletonized beams. Going to a bowden extruder setup is far more effective in removing weight as you've proven.
Re: Bowden V0; The problem there is that you've removed a bunch of weight *above* the toolhead CG/pivot point, and none below. It's still not ideal, but it does work for Stefan's stated goals, and I'm reasonably sure he's considered the increased rotational torsion on the X beam.
@viru52000 yea different printer but still x and y moves the same and a v2.4 350 x beam is the same as my x beam in length minus a few mm and with my trident switching to a cf x beam helped alot
@@Mikehatespigs the entire gantry and toolhead are designed differently on the VZbot and does have the same twisting loads as a Voron. Plus the Voron Discord being full of CF tubes delaminating from ABS printing temps, meaning it's a consumable at that point.
AFAIK, Input shaping is looking for it's first fundamental frequency to figure out how fast it can go. It isn't looking for torque on the motors, so weight is not part of the equation for the max acceleration it outputs. If you look at what makes up a fundamental frequency, a simplified equation would be fn = sqrt(k / m), where fn is the fundamental frequency (larger is better), k is the stiffness of the system, and m is the mass of the system. In a Voron, changing the x beam, you are lowering the stiffness of the best (lowering k) when you cut holes in it. The mass you need to think about is the mass of the whole moving system, not just the beam. So lightening the beam lowers the stiffness of the whole system, but only lowers the mass of part of it. My guess is going to a rectangular aluminum beam (like carbon fiber shape, but aluminum) would be an improvement. Or go to a thicker wall section carbon fiber. Of if you are really adventurous, move the linear rail to the front or back of the beam. It will then improve the area moment of inertia of the beam. That rail is quite stiff, if loaded in the correct way. Love to see people trying to push the limits! Great video
Based on my differential equations/control theory knowledge, the two factors in an undamped system are stiffness and mass. Given that most of the damping naturally comes from the friction in the motors and slider bearings and is likely the same for every x bar, I would take a closer look at how stiff each bar is and see if a stiffer material, such as titanium, gives better resonance testing results.
Cast CF would provide some damping also, wouldnt it? while maintaining stiffness. Or CF with titanium backer perhaps. But that might add some less predictable/linear resonance characteristics with the 2 materials interacting.
Stiffer materials helps, a little. But remeber that stiffness of a part is the CUBE of the depth. He would gain massive results if a 30 x 30 beam would fit in the space. I think the designers choose 15x15 just becuse the rest of the printer was 1515. Also I thing the flex on the beam is tortional. The center of mass of the extruder likely not level with the beam
I find it really useful for larger machines like my 500mm vcore 3. the weight savings there are huge compared to the weight savings on a tiny machine like the v0.
Have you measured this? Or is it subjective? Are the motors on the vcore not up to the job? There is nothing to suggest that it would make more difference on a longer beam unless the motors arent up to the job of moving the stock beam.
@@jeremyglover5541 boy wait until this guy hears about the concept of mass and how longer, thicker aluminum extrusions have considerably more mass than shorter ones
@@jeremyglover5541 Let's say the stock beam on a v0 weighs x grams. Since the vcore 3 500 is around 3x larger on that axis, the vcore 3 beam can be 3x grams. Let's say the ultralight gantry is half the weight of a normal gantry. In the case of the v0, the weight savings are x/2 grams, but in the case of the ratrig, the weight savings are 3x/2 grams. Since the motors still have the same amount of torque, the vcore 3 motors will have had a much larger reduction in mass to push, while the v0 motors would have a smaller reduction in mass to push.
@@microArc i know. This simple fact seems to have been missed by everyone using this argument. Its a symptom of just doing stuff, based on half baked theories designed to separate you from your money, without doing before/after measurements/experiments.
A triangangular beam would use a bit less mass than a square one. I would try to scrap off as much mass as possible from that linear rail. Then i would glue it to the beam and only use screws for application. Maybe the weight in the small v0.1 is not that relevant, but as you mentioned, it would be the correct way in terms of physics.
Mass is not the limiting factor. In fact, it just might be that a more massive beam is better. What matters here is total stiffness. Perhaps a 15 x 15 solid aluminum beam might be the best aluminum design. If you could afford titanium then perhaps a solid Ti beam? These CNC'd beams optimized the WRONG thing.
I would try a different take on the beam. The front and back faces are left mostly intact with the top an bottom faces more open. The front and back faces get more material removed towards the ends with next to no removal at the middle, like a beam of equal tension in mechanical engineering. The deflection in the middle is obviously the greatest and thus needs the highest amount of local stiffness. Maybe ill try that once or even a composite approach, combining aluminium and carbon fiber parts
This is true. The design of the box structure does not take into consideration the variable cantilever forces. If designed more appropriately, there is potential for improvement of stiffness:weight.
@@UnitSe7en or even focus more on stiffness than weight in that case as the x axis is already insignificant compared to the toolhead. I would wonder how a regular 15x15 tube with 2mm wall thickness would have compared, especially to the standard extrusion
No, the force on the beam is torsional. The heavy part is the extruder motor and it is not in line with the beam's center. The acceleration of the beam by the belt is about 2X higher than the acceleration by gravity. (gravity on Earth is "only" 10 meters per second squared) You can ignore the force of gravity here. You need a stiff beam, not a light beam. The mistake was to optimize the wrong thing. He tried to remove mass and mass does not matter here, total stiffness matters. I think a SOLID 15 x 15 beam would beat the 1515 v-slot beam
Super interesting! There are two ways to get higher resonance frequencies, less weight or more strength. I start to think that maybe it is worth trying to go into the opposite direction. Heavier but stronger, no printed parts, thicker extrusions, heavier linear rails, strong motors and maybe even ballscrews instead of belts. What do you think about it? Basically a CNC with faster motors, but a hotend instead of a milling motor. With exchangeable toolheads it could double as a CNC.
Thank you for that amazing comparison and great video! I can imagine how much work went in to that video and really appreciate your effort and time you took to make it!
I feel like even though the ratios show differently, the CF is the best. It offers a little off the weight but keeps most of the stiffness. Also, I'm guessing the input shaper graphs on the UltraBeam and HyperBeam (can I even type the name HyperBeam? Is it trademarked that much?) are absolutely horrible?
resonance frequency is based on weight and stiffness. Higher frequencies damp out faster. Low frequencies have larger displacements and longer damping times. To get a high frequency, it should be light weight, and stiff. I'm guessing that the reduction in stiffness with the lighter beams is actually making the resonance frequency lower as you are sacrificing more stiffness than you are getting back for in weight reduction. Especially since there are significant fixed weights on the head, having a very stiff x beam is probably more important than minimum weight.
Lower recommanded speed probably comes from the shifted center of mass. Because the beam doesnt weigh much, the linear rail acts like a pendulum. You would need to move the extruder head to balance the gantry
I am just now working on a design that can be manufactured on a slm machine and using topology optimization to get the geometry right. Will be part of my next V0.2 build since I don't want to take my V0.1 apart :)
@@247printing For me your videos about the 0.1 are what got me to build it in the first place and the same probably goes for other people. Great videos with lots of honest and upfront information about these machines. You found yourself a nice niche and I am looking forward to whatever you come up with.
Absolutely. To keep parameter variation low and not too complex, I used the default voltage at 24V and slightly elevated currents at 0.85A. --> Realistic scenario for max acceleration test.
@@247printing Yes, I agree that your scenario is realistic and showing the truth: we are not limited by motor power => not much sense to reduce moving mass. Higher voltages could have showed the difference, but these voltages would be unrealistic. Lower current also could have showed the differences, but also not realistic. Thanks for great experience and learning!
As someone who has built multiple Voron V0 kits, I must say that this is a LOT of work. My hat goes off to you! Such a great way to showcase the gantry arm differences, or lack thereof.
Thanks a lot! It was work, but it's very much fun and relaxing for me to do that. In addition: I am used to it - somehow all my videos go into the "lot of efforts category" :-)
I think those ultra lightweight beams are sort of snakeoil. Yes, reducing mass ist a good thing for high acceleration and the beams looking like a low hanging fruit, but you have still the linear rail, x and y carriages, the fastners and the hotend assembly with extruder. this adds up quite fast. That would be intresting, how heavy all together is and how much weigt can be saved by just using a bowden setup.
In the v0 that seems to be definitely the case, but in larger 300-500mm printers the normal X beam extrusion weighs a lot more. Also the v0 toolhead is actually quite heavy too so if he used a vzbot tool head or similar it would be lighter again.
The total weight of the xy carriage (bearing blocks, linear rail, extruder, hotend, brackets, housing and beam ) is what like 400g? And this just looks at a 80% reduction of 10% of the mass. Maybe a lightweight beam becomes relevant when you reduce the other components weight by 80% too… Regardless. I suspect the rotating mass, reluctance and inductance inside the stepper exceeds the entire mass inertia of the carriage. A coreless wide airgap long but small diameter high amp fiberglass servo would be interesting…
Here's a crazy idea, but what if you used alumina for your X-axis beam? There are a few approaches, but I think the most versatile and accessible method would be SLA 3D printing with sintering, followed by grinding to final tolerance. It would not be cheap in the first place, but complexity comes almost for free, so you could really squeeze a lot of performance out of it. It's almost twice as dense as 6061 aluminum, but it's leaps and bounds stiffer, and 3D printing opens the door to some serious topology optimization. You could probably shave off enough mass to come in under 10 grams while also achieving a stiffer beam with better vibration characteristics than anything tested here. And while it is a crunchy material, you can compensate for manufacturing errors or minor damage with some clever latticing and probably achieve even better performance.
My Mind is blown, I'm building a Voron V2.4 r2 350mm, a patreon and would love more of the "Red Pill:" AKA Klipper setup/tuning/everyday driver and perhaps Ludicrus speed Please!
I am curious why the beams are square shaped instead of triangle shaped. That should save even more weight. I also can't help but notice that those carriers y axis carriers on each end of the beam look quite... bulky and not at all optimized. Maybe there is yet more to save there
Square beams are way easier to make. But what happens if you remove the beam completely and use just the rail. Then you optimize the y axis carriers and belt idlers.
If the ringing test is essentially measuring resonance, then you should be able to go beyond that frequency to clear up the ringing again - As long as your accelerations are not a harmonic of the prime, it shouldn't happen... But I don't think the ringing test actually measures resonance - at least not as the primary measurement, because you're far too low frequency to be seeing resonance in the beam - only the part flexibility. In this case, it is of no surprise whatsoever that the lightweight parts are more flexible than a solid extrusion. Lightweight only helps if your motors have reached their maximum acceleration values due to inertia. Your motors are clearly able to overcome the inertia of the heavier, stiffer beam. You will see practically zero improvement by trying to make lightweight parts. This is wasted effort in this instance.
Maybe try testing using tension in your beams. Concrete beams in construction sometimes uses threaded bar inside to apply compression to the concrete and increase its stiffness. The term to look up is post tensioned concete. Either a thin threaded rod down the middle of the beam or steel wire, Kevlar thread, something with high tensile strength anchored at either end of the beam would maybe increase your stiffness & capability. Of course you would have to beef up your beam to take the compressive load & increase weight but would be interesting to see tested ✌️
Tensioning concrete is not to increase its stiffness, it's to prevent it from seeing internal tension loads, as it it not good in tension, unlike carbon fiber and metals.
The fact that the beams all get the same recommendation from input shaper surely would indicate that the limiting factor for improving accel here is stiffness, not mass - would be interested to see how a significantly stiffer beam (at the expense of more mass) would compare in this case (perhaps an aluminium square tube with unmilled walls?)
Great video and scientific comparisons as always 👍 very eye opening too. Someone like you is absolutely a must for us 3D printers makers. Also too I won’t go for sports rims unless the cost performance ratio is there. Maybe next we can see length comparisons on x-beam. Thanks bro, u r awesome.
I can tell you before watching the video it's useless because the linear rail and hotend/extruder will weigh much more in realtion, so the few gramms you save on the bar are useless.
Excellent video and even if your testing is home made it is still very relevant for hobbyist. Going fast is great but in the end how much time do you save and is the time saved worth your set-up time, effort and money. As a hobbyist I invest in things that don't always make sense just because I enjoy thinkering but since money rules the world..... How long does it take for someone to see a return on this investment? I know I know we don't do this for money as stated before but in the end should I buy one for the look and a few minutes in printing time? For commercial ventures maybe. Again, excellent work and I enjoyed the video.
I talked to Fail fast about lightweight X beams for a while. I ended up going with a Pultruded Carbon fiber one because it was much stiffer than the aluminum extrusion and the ultra lightweight one. Is your carbon fiber pultruded or woven, because woven preform much worse. I also tested a hand milled aluminum extrusion that I made that was 30 grams and I lost 25hz of resonances because of it. Can't wait to beat Baravel for the WR ;) I also didn't know about the titanium requirement could you elaborate further on it? Also you can buy carbon fiber x beams on ali for the 0.1 (that's where I got mine)
Hey, from what it looks like, the CF beam is woven - afaik you wouldn't see the fibers on pultruded ones. Looking forward to your WR - you can make it! Galvanic corrosion, google: "carbon fiber galvanic corrosion" ;-)
What is happening at 2:11 in the video? The scale starts counting up before the beam is even on it! Did you edit that with a different take so the LCD would be more readable?
Wow you have a great perception- while editing I was also wondering why it looks like this, but I bet it’s only because it was already slightly touching the scale from one side. Just one take here, no fancy editing
Thinking about generative design since I started with those beams in 2021, but I don't have operative knowledge on that. Maybe I'll revisit that topic some day.
Great set of tests. You could make it even lighter with a triangular truss design. Only the top part really needs to be flat for mounting the rail. The part of the triangular truss could be thickened a bit to better transmit loads along its length, but it would probably have even better stiffness to weight ratio than the HyperBeam.
It's a great video. Und I fully agree with normal mini ab or sb it makes no sense but looks good ;). But in combination with a light weight tool head it is possible to lose 50% of total mass or more.
Hmm. If i had to put forward a theory on why the Y acceleration performed worse. It would be that it involves the linear rail mounting. The Hyperbeam afterall lacks mounting holes for a good number of the rail's own holes, the CF Beam is a bit diy, bringing probably some tolerance errors and the Ultra-beam has that the holes diagonally connected to the rest of the frame, increasing effective length between hole and frame. These may cause minute motions within the rail that cause the software to take umbrage more rapidly. Even the best mount would suck if the linear rail isn't as rigidly mounted as possible. At least i would assume so.
The efficiency of the beam alone is academic, less relevant than efficiency measured by dividing the stiffness by the System weight (beam mass including the extruder mass). This would show even less difference. Great video.
Very nice investigation. I think maybe you underestimate the effect of the torsional force. The lateral deflection test technically would also be a torsional force since the the weights seem to be off-centered from the centre of the beam themselves. The from the lateral deflection and torsional defelection test seem to correlate with the input shaping results. Keep doing what you are doing! You have a great channel.
@@247printing I agree that torsional deflection is very important. Can you mount the accelerometer as far from the center of twist as possible and show the amount of vertical component realized. By adding weight to the print head to move the center of gravity, in line with the center of twist you may actually get higher acceleration recommendations from Klipper because that vertical component is reduced.
Love that you hot rod these machines! Is the beam being a constant section a constraint? On of the super cool things with using a product like carbon is it can be made to any shape, also the laminates alter the Performance tremendously. It looks like there is also a huge weight loss potential in machining pockets in the underside of the linear rail. Loosing weight out of structure is a fun part of my job setting up racing boats.
I was also thinking.. did your aluminum and CF beams clamp on the ends where the bought aluminum beam was threaded? If true, I wonder if that accounted for the difference in expected performance.
Wouldve loved to see how much stiffness is sacrificed if the carbon fiber tube is cnc machined for lightweight and if the deflection is comparable to aluminium from cutting the fibers on all 4 sides
One other thing to keep in mind when selecting material: fatigue. Aluminium is not great at that. An ultralight piece might need replacement every few months.
@@247printing (or trying to use the same thickness on a bigger machine) This little machine would probably work with the steel guide and two end pieces only. I bet some people will mistakenly use beams that are lighter than necessary and be surprised. My comment is just for a future "I told you so". 😉
Interesting how the beams behave with input shaper. Seems they are all different constructions and thus resonate different with different resonances. Fore sure something to look into. for a such a small beam its hard to see big gains. Maybe on a 350 voron or 500ratrig.
I'm curious... Could you perhaps mathematically model these tests and material properties to try and correlate your IRL findings? Then you could possibly control more of the variables you mentioned.
Great video Albert! It would have been nice also to see a test with the unsupported rail to see the input shaping and torsion difference. For the small size it shouldn't be an issue to use unsupported rails, I know is not the best thing to do for flatness and torsion but I saw some people doing it and seems to be a working solution.
@@247printing based on the deflection of a simple beam, when you double the length the deflection raises to the 3rd power so we should definitely see a much more conclusive result if light weight beams work or not!
Great video with in-depth analysis. Subscribed! PS: echo is quite noticeable in the audio and slightly annoying. A little bit of work with your sound capture setup would push the video realization to the top👌
Why do you get worse results with the lighter beams? It should be obvious... The mass of the beam is not the limiting factor. It is the stiffness of the beam that limits acceleration. You need to engineer a stiffer beam and don't worry so much about mass. Try maybe an aluminum box-shaped beam with very thick walls that weigh close to the same as the 1515 extrusion. Or even try a solid 15 x 15 I-beam. My bet is the thick-wall box is stiffer than the 1515 exstusion for the same mass. Try a box with thick enogh walls then you do not need nuts to hold the rail. Voron used the v-slot 1515 not because it is the best structural shape but because you can drop nuts into the slot and not have to drill and tap holes. It requires MUCH less skill and tools to build with V-slot because you don't need a drill press or mill. But if you are going to make an sell metal Voron Parts, you wouild have CNC equipment and
Have you looked into Aluminum/Titanium hardware? Bigger SHCS you can get a hole through hold although I don't think they make them for such small bolts. Can probably save a few grams here with minimal to no performance loss. Also looking at the vibrations of the frame in your slow mo. The frame is put together by blind through holes how much more ridge would it become with using the 80/20 90° elbows on everything but the X axis then your not adding weight but stiffness to the entire frame. On the X axis maybe play around with different materials instead of ABS. Something glassfilled or HIPS. Wondering if gaining a few grams say 5% weight vs gaining 20% in strength is beneficial on the brackets? How do the frequencies and vibrations play into effect. Engineering mind is running wild late at night here. Love your detailed videos though they keep me thinking!
Next, design a beam using topological optimization. Come up with something that will itself require metal (or PA12) 3d printing to build. I bet AutoCAD would sponsor you (if it works!). Also: The 1550 beam likely has far more torsional stiffness and out-of-plane shearing stiffness than the box beams. You probably need something like a "space truss" design. Going even further: once the beams are not a single solid material, you are going to get some cross terms between tensile, shear, bending, and torsion terms. The same thing comes up when you work out the response properties of a layered composite material. If I am remembering this correctly: you can end up with a material that for example, when loaded in bending will internally give rise to a torsional deflection, and vice versa.
I can imagine the weight of the stock beam and density of it causing a bit of a dampening effect when discussing acceleration. Usually when these beams are discussed they are only looked at from the perspective of deflection when these analyses are run. It doesn't take into account the dampening affect it would have when being thrown at high speeds in the printer. Think about it like how tuning forks work. The amount of material on a tuning fork will change how it sounds. This is a resonance vibration similar to what we are talking about with 3d printing. I bet each beam makes a different sound too...
so, I have a theory, the lighter beams have more flex but think of it as a vibration, that disrupts the input shaping vibration. kind of like wobbling around and causing a ruckus while its trying to smooth out the vibrations. there could possibly be a math equation written up to adjust for that extra vibration using your deflection measurements. i wouldn't know how to reprogram input shaping for this so its probably not feasible right now. I was just thinking what if you wrapped your light weight hyperbeam in carbon fiber?
Great video. I think it's not conclusive, though. Just the start of a thorough investigation. Maybe a hollow carbon fibre cilinder could improve stiffness, maybe I'm talking bullshit. All I know is that you're onto a very interesting subject, natürlich.
You should try a sort of minimilast CF composite beam. Like make a foam core with carbon plies over top and bottom, chop them and laminate, Because that's how carbon fiber is most effectively utilized.
This beam is only a small fraction of the gantry mass, yet needs to support the entire hotend mass in motion. So, calculating a Stiffness:Weight ratio using only the beam’s properties will artificially inflate its performance. A lighter beam may drop 60% weight and lose only 30% stiffness… but if it was only 10% of the gantry mass to begin with, the assembly as a whole exhibits a very different total: just 6% mass drop for 30% stiffness loss!
Amazing visual analogy, hyper optimized 'beams' when while still using PLA parts (of all things) in an assembly is just like driving those hoopties with oversized rims, all show, no go.
Very nice video. Thumbs up... I hope being not too much of a smart-ass but when explaining the low efficiency by the total mass safe to be only 8%, I need to say that it is even less, because lacking the mass of 2x mgn9h sliders plus the mass of the rotator in the nema14 motors ( rotation wise). I would expect a bigger impact on larger printers although these printers will not top the little voron on absolute speeds.
I Lookign at how the weight interacts witht he beam, I wonder how a hybrid lightweight beam would perform. Granted, the weight savign is on the most stressed and least heavy part, but it's a step. If a 2020 extrusion was cnc machined in only one direction, front ot back, with the other direction left alone then since there is no up/down movement the forward/back movement would retain the advantage of more material resisting flexion while possibly retaining enough stiffness to resist twisting. I have a background in competitive paintball from a manufacturer's perspective and I've heard players swear up and down that ounces and grams matter. While I remain skeptical in its application in shaving weight off a paintball gun, I can see it's real advantages here. I wonder how much weight could also be shaved off of the other mechanical parts to achieve a truly featherweight xy system.
Look at adding damping. Aluminum has very low damping properties, something like a constrained layer damping (search CLD) would add mass but limit the magnitude of vibration
It might be more cost effective to buy bigger stepper motors, than spending money on fancy beams. Algorithms fine tuning the accelerometers would help too. Mass is the cheapest vibration damper , ,so maybe the best solution would be to build a carbon fiber beam ,perhaps 75mm square, using bigger stepper motors. Then, set up a program to simulate a high speed back and forth motion. Then tune the beam to dampen any resonants, by adding mass to specific parts of the beam. (Resin painted on ,or shim stock glued on)
Can you add the Ender 7 x-axis beam setup into your next test video, just to see the difference since it's just a flat/thick piece of metal, currently have one running and trying to figure out what I should be looking at to upgrade current print quality and thought that something completely different than boxed metal?
I'm no machinist nor expert, but I have some machining experience. Mass might be your enemy for speed as it has to start and stop and that can cause overshooting aka ringing. Too much mass and you need to use bigger motors to move it and can't accelerate so quickly. At the same time too little mass will introduce harmonic vibration. Also ringing. So it is a double edged sword. There's a way to have your cake and eat it too, ramping your speeds accordingly.
My guess is mass is dampening. I seriously doubt input shaping is tuned for ultralight high performance. Kinda like a killer gpu with generic drivers…meh blah.
I am sorry. The aluminum beam weight is not much. Reduce it weight is meaning less. We should reduce the weight of the linear rail it self. Why it have to be so thick while the linear rail it self rely on how straight the beam is?
So, I'd say you're getting worse results with input shaper recommendations because the lighter mass is not able to absorb the vibrations produced as well as a heavier mass. This doesn't mean that the lighter gantries can't go faster, because they certainly can, but that's only half the problem. Think of it like two boats on rough seas. One is much larger and heavier and has an overall slower top speed, but because of its mass and larger profile, it can handle the rough waves much better, providing a smoother ride. The other boat, which is small and lightweight, has a higher top speed, but gets tossed around by the waves, limiting how fast it can really go without breaking against the waves. I realize this analogy isn't perfect, but I hope it gets my point across well enough. How to fix this, I'm not exactly sure. Perhaps progressively increasing the density of the lattice towards the middle of the beam to reduce flexing in the center might help?
Maximum acceleration is highly correlated to beam stiffness so a solid material will generally outperform hollow material with the same outer dimensions. If you want a lighter beam with the same stiffness you need to make sure the moment of inertia is the same, which for these beams, isn't the case. The carbon fiber beam and Ultrabeam likely have a similar wall thickness so they get around the same stiffness and maximum recommended acceleration. Your Hyperbeam has thinner walls that both the carbon fiber and Ultrabeam so the stiffness will be lower. This means you'll need to adjust both your outer dimensions and your inner dimensions as opposed to simply hollowing out the center.
Drag of the linear rail bearings is maybe also a flaw. Have you tried some special no slip “grease”? It’s used on mtb suspension. Have shop sized bins, can send samples for free. Greets from Nbg
Out of curiosity, could you ditch the beam and just get a bit larger or longer linear rail to bridge across the axis? Admittedly you'd probably end up needing a much higher quality linear rail that isn't bowed or anything of the sort, but, just my curiosity.
Really interesting! I've been diving down the rabbit hole of fast printers last few days, love seeing all the thought going into these printers! what about the linear rail? what would be the lightest option for that?
Even as janky as that test kinda was I think you still made your point fairly well. Lightweight beams are clearly just a gimmick that look cool & don't provide any meaningful performance gain.
The awesome feeling of designing and manufacturing your own design never gets old!
Try Onshape, the feature-rich cloud based solution and support 247printing by clicking: Onshape.pro/247Printing
As unnecessary as this is for most us, I greatly appreciate it. I learned a lot & it’s nice to see people pushing boundaries.
Thanks a lot, Daniel!
We are not going to act like this didn't make our day! We are much honored to be one part of your great X-BEAM, Albert! 🥳🥳
, m
Thx a lot for the collaboration, I bet we find ways to go on 👍
Wn =sqrt(k/m)
Where Wn is the resonant frequency, k is the stiffness and m is the mass. In order to improve input shaping results, you have to aim for the highest possible frequency. That means you not only have to reduce mass, but also increase stiffness!
TRUE!
So skeletonized titanium, gotcha (Or arguably alloy steel (which one is debatable but a few come to mind), I assume it would since while it's denser most of that mass is gone and it's significantly stiffer, brass, nickel, and zinc would also work since they should all be stiffer than aluminium [at same dimensions]).
Very nice work! As you see with your natural frequencies and maximum recommended accelerations, what matters is the stiffness to mass ratio. Even with better stiffness to mass ratio of the CF, hyperbeam and ultrabeams in isolation, once you add the mass of the hot end, the complete performance may be worse since the alternatives have an absolute lower stiffness than the default. If you had absolute stiffness included in addition to the stiffness/weight I think it would be more apparent. It would be nice to know eventually what size of beam makes this crossover point viable for using the lightweight beam designs.
This is pretty much what I was going to comment too. The acceleration is lower with those lighter beams so they don't flex too much. It would be pretty interesting to see what would topology optimized beam look like.
yep too much stiffness reduction in comparison ...
Another factor might be a reduction in damping of the beams themselves although I expect the impact to be smaller in comparison to the stiffness reduction.
Lightweight only helps if your motors have reached their maximum acceleration values due to inertia. His motors are clearly able to overcome the inertia of the heavier, stiffer beam. He will see practically zero improvement by trying to make lightweight parts. This is wasted effort in this instance.
@@UnitSe7en True, but as the printer size increases and the relative mass of the beam to the hot end gets larger, maybe this solution would be worthwhile (perhaps on a Voron 2.4). I guess that can be settled by future testing or at least a simulation if the overall stiffness/mass are known. From most of what I have seen, the melting and cooling rates of polymer will start to limit performance more than mechanical resonances in these smaller fast printers.
@@UnitSe7en exactly. This sort of research has already been done in the case of speaker drivers/larger magnets/stiffer-lighter weight cones. It makes no difference if your motor (its called that in speakers too) is already able to achieve the required acceleration to cope with the demands of the frequency being requested. It turns out to make no difference at all to make the cone lighter. Here, longer beams will make this problem worse. If its already of little yo no benefit on the v0, larger sizes will see more problems, as rigidity is lower already on the shorter beams, it will be more apparent on longer ones.
Awesome video, I love seeing the science and explanations behind these types of things. I ordered a FYSETC carbon beam for my 2.4, I'm excited to see if it makes a difference, but it at least looks cool.
I appreciate the video, but right at the start when you were comparing the percentage weight reductions, my brain was screaming, irrelevant without comparing the total mass of the moving parts. Of course you said this at the end.
You need to start the comparison somewhere - I did directly on the component. I also stated already in the video, that those beams don't make sense :-)
this! especially considering printers like the k3 work with no beam at all, it’s pretty obvious that this really doesn’t matter on such a small printer
Great video! "Because" definitely plays a part in many our customization's. This is my take as someone that isn't concerned with chasing maximum speeds. Its a relief to know that the other beams available for V0. Have no downsides to them in regard to performance. I would hate to once again buy something shiny because I wanted it, and find it cost me in other areas.
Had a feeling these were the results we would see based off the input shaper graphs people have shown with these beams on the Voron discord. Many are getting worse graphs on larger printers (350mm V2.4 for instance) with the skeletonized beams. Going to a bowden extruder setup is far more effective in removing weight as you've proven.
Re: Bowden V0; The problem there is that you've removed a bunch of weight *above* the toolhead CG/pivot point, and none below. It's still not ideal, but it does work for Stefan's stated goals, and I'm reasonably sure he's considered the increased rotational torsion on the X beam.
My vz bot ip improved with the carbon fiber beam and no difference when changed to the cnc
@@Mikehatespigs different printer design, they're not comparable in this aspect.
@viru52000 yea different printer but still x and y moves the same and a v2.4 350 x beam is the same as my x beam in length minus a few mm and with my trident switching to a cf x beam helped alot
@@Mikehatespigs the entire gantry and toolhead are designed differently on the VZbot and does have the same twisting loads as a Voron. Plus the Voron Discord being full of CF tubes delaminating from ABS printing temps, meaning it's a consumable at that point.
Albert, your videos are getting so much better... I anxiously look forward to each one. I love how you're pushing the whole hobby forward
Thanks a lot Jonathan! My plan is to get better with every video :-)
AFAIK, Input shaping is looking for it's first fundamental frequency to figure out how fast it can go. It isn't looking for torque on the motors, so weight is not part of the equation for the max acceleration it outputs. If you look at what makes up a fundamental frequency, a simplified equation would be fn = sqrt(k / m), where fn is the fundamental frequency (larger is better), k is the stiffness of the system, and m is the mass of the system. In a Voron, changing the x beam, you are lowering the stiffness of the best (lowering k) when you cut holes in it. The mass you need to think about is the mass of the whole moving system, not just the beam. So lightening the beam lowers the stiffness of the whole system, but only lowers the mass of part of it. My guess is going to a rectangular aluminum beam (like carbon fiber shape, but aluminum) would be an improvement. Or go to a thicker wall section carbon fiber. Of if you are really adventurous, move the linear rail to the front or back of the beam. It will then improve the area moment of inertia of the beam. That rail is quite stiff, if loaded in the correct way. Love to see people trying to push the limits! Great video
Based on my differential equations/control theory knowledge, the two factors in an undamped system are stiffness and mass. Given that most of the damping naturally comes from the friction in the motors and slider bearings and is likely the same for every x bar, I would take a closer look at how stiff each bar is and see if a stiffer material, such as titanium, gives better resonance testing results.
Cast CF would provide some damping also, wouldnt it? while maintaining stiffness. Or CF with titanium backer perhaps. But that might add some less predictable/linear resonance characteristics with the 2 materials interacting.
Stiffer materials helps, a little. But remeber that stiffness of a part is the CUBE of the depth. He would gain massive results if a 30 x 30 beam would fit in the space. I think the designers choose 15x15 just becuse the rest of the printer was 1515.
Also I thing the flex on the beam is tortional. The center of mass of the extruder likely not level with the beam
Another excellent dive into faster printing Albert! Love to see more videos as of late.
The cfk rod is made of quasiisotropic laminate. If you feel like it, you could use a pultruded rectangular profile. This is certainly much stiffer.
If I would ask the machinist at work to make me the HyperBeam he would tell me I'm mad and that I shall fu*k off lol
I find it really useful for larger machines like my 500mm vcore 3. the weight savings there are huge compared to the weight savings on a tiny machine like the v0.
500mm vcore 3s are beasts! I'm jealous over here with my 350mm v2.4
Have you measured this? Or is it subjective? Are the motors on the vcore not up to the job? There is nothing to suggest that it would make more difference on a longer beam unless the motors arent up to the job of moving the stock beam.
@@jeremyglover5541 boy wait until this guy hears about the concept of mass and how longer, thicker aluminum extrusions have considerably more mass than shorter ones
@@jeremyglover5541 Let's say the stock beam on a v0 weighs x grams. Since the vcore 3 500 is around 3x larger on that axis, the vcore 3 beam can be 3x grams. Let's say the ultralight gantry is half the weight of a normal gantry. In the case of the v0, the weight savings are x/2 grams, but in the case of the ratrig, the weight savings are 3x/2 grams. Since the motors still have the same amount of torque, the vcore 3 motors will have had a much larger reduction in mass to push, while the v0 motors would have a smaller reduction in mass to push.
@@microArc i know. This simple fact seems to have been missed by everyone using this argument. Its a symptom of just doing stuff, based on half baked theories designed to separate you from your money, without doing before/after measurements/experiments.
A triangangular beam would use a bit less mass than a square one.
I would try to scrap off as much mass as possible from that linear rail. Then i would glue it to the beam and only use screws for application.
Maybe the weight in the small v0.1 is not that relevant, but as you mentioned, it would be the correct way in terms of physics.
Mass is not the limiting factor. In fact, it just might be that a more massive beam is better. What matters here is total stiffness. Perhaps a 15 x 15 solid aluminum beam might be the best aluminum design. If you could afford titanium then perhaps a solid Ti beam? These CNC'd beams optimized the WRONG thing.
I would try a different take on the beam. The front and back faces are left mostly intact with the top an bottom faces more open. The front and back faces get more material removed towards the ends with next to no removal at the middle, like a beam of equal tension in mechanical engineering. The deflection in the middle is obviously the greatest and thus needs the highest amount of local stiffness. Maybe ill try that once or even a composite approach, combining aluminium and carbon fiber parts
This is true. The design of the box structure does not take into consideration the variable cantilever forces. If designed more appropriately, there is potential for improvement of stiffness:weight.
@@UnitSe7en or even focus more on stiffness than weight in that case as the x axis is already insignificant compared to the toolhead. I would wonder how a regular 15x15 tube with 2mm wall thickness would have compared, especially to the standard extrusion
No, the force on the beam is torsional. The heavy part is the extruder motor and it is not in line with the beam's center. The acceleration of the beam by the belt is about 2X higher than the acceleration by gravity. (gravity on Earth is "only" 10 meters per second squared) You can ignore the force of gravity here. You need a stiff beam, not a light beam. The mistake was to optimize the wrong thing. He tried to remove mass and mass does not matter here, total stiffness matters. I think a SOLID 15 x 15 beam would beat the 1515 v-slot beam
Super interesting! There are two ways to get higher resonance frequencies, less weight or more strength.
I start to think that maybe it is worth trying to go into the opposite direction. Heavier but stronger, no printed parts, thicker extrusions, heavier linear rails, strong motors and maybe even ballscrews instead of belts. What do you think about it?
Basically a CNC with faster motors, but a hotend instead of a milling motor. With exchangeable toolheads it could double as a CNC.
Maybe the lighter beams have too much flex, and that's why the measurement shows lower max acceleration for them?
Thank you for that amazing comparison and great video! I can imagine how much work went in to that video and really appreciate your effort and time you took to make it!
Thanks a lot mate. I enjoy your beam a lot and I'll use it for sure on one of my V0s for sure!
I feel like even though the ratios show differently, the CF is the best. It offers a little off the weight but keeps most of the stiffness.
Also, I'm guessing the input shaper graphs on the UltraBeam and HyperBeam (can I even type the name HyperBeam? Is it trademarked that much?) are absolutely horrible?
Most potential for sure by its "nature".The beams of all of them are quite comparable and not really bad, but also not really good.
resonance frequency is based on weight and stiffness. Higher frequencies damp out faster. Low frequencies have larger displacements and longer damping times. To get a high frequency, it should be light weight, and stiff. I'm guessing that the reduction in stiffness with the lighter beams is actually making the resonance frequency lower as you are sacrificing more stiffness than you are getting back for in weight reduction. Especially since there are significant fixed weights on the head, having a very stiff x beam is probably more important than minimum weight.
Lower recommanded speed probably comes from the shifted center of mass. Because the beam doesnt weigh much, the linear rail acts like a pendulum. You would need to move the extruder head to balance the gantry
Totally blowed my mind, looked like less weight not always raise the acce! Thanks for great video!
Thank you very much for supporting 247printing!
I am just now working on a design that can be manufactured on a slm machine and using topology optimization to get the geometry right.
Will be part of my next V0.2 build since I don't want to take my V0.1 apart :)
So cool that so many people work on that topic - it's a lot of fun!
@@247printing For me your videos about the 0.1 are what got me to build it in the first place and the same probably goes for other people.
Great videos with lots of honest and upfront information about these machines. You found yourself a nice niche and I am looking forward to whatever you come up with.
@@comatron231 Thank you so much!
Wait. There is a 0.2?? Nice!
Better update the STL collection and BOM if i want to build one this year, before i accidentally start building a 0.1
@@Foxhood the change is quite minimal but is a good QoL from what it seems
Also, to verify that you are not limited by motor power but rather voltage - you can test these beams at reduced motor current.
Absolutely. To keep parameter variation low and not too complex, I used the default voltage at 24V and slightly elevated currents at 0.85A.
--> Realistic scenario for max acceleration test.
@@247printing Yes, I agree that your scenario is realistic and showing the truth: we are not limited by motor power => not much sense to reduce moving mass.
Higher voltages could have showed the difference, but these voltages would be unrealistic. Lower current also could have showed the differences, but also not realistic.
Thanks for great experience and learning!
As someone who has built multiple Voron V0 kits, I must say that this is a LOT of work. My hat goes off to you! Such a great way to showcase the gantry arm differences, or lack thereof.
Thanks a lot! It was work, but it's very much fun and relaxing for me to do that. In addition: I am used to it - somehow all my videos go into the "lot of efforts category" :-)
I think those ultra lightweight beams are sort of snakeoil. Yes, reducing mass ist a good thing for high acceleration and the beams looking like a low hanging fruit, but you have still the linear rail, x and y carriages, the fastners and the hotend assembly with extruder. this adds up quite fast. That would be intresting, how heavy all together is and how much weigt can be saved by just using a bowden setup.
Bowden is key, yes, I'll come back to this. In the end the sum of weight reductions makes a difference.
In the v0 that seems to be definitely the case, but in larger 300-500mm printers the normal X beam extrusion weighs a lot more. Also the v0 toolhead is actually quite heavy too so if he used a vzbot tool head or similar it would be lighter again.
The total weight of the xy carriage (bearing blocks, linear rail, extruder, hotend, brackets, housing and beam ) is what like 400g? And this just looks at a 80% reduction of 10% of the mass. Maybe a lightweight beam becomes relevant when you reduce the other components weight by 80% too…
Regardless. I suspect the rotating mass, reluctance and inductance inside the stepper exceeds the entire mass inertia of the carriage. A coreless wide airgap long but small diameter high amp fiberglass servo would be interesting…
Yup, it's 385g, see the conclusion chapter
Here's a crazy idea, but what if you used alumina for your X-axis beam? There are a few approaches, but I think the most versatile and accessible method would be SLA 3D printing with sintering, followed by grinding to final tolerance. It would not be cheap in the first place, but complexity comes almost for free, so you could really squeeze a lot of performance out of it.
It's almost twice as dense as 6061 aluminum, but it's leaps and bounds stiffer, and 3D printing opens the door to some serious topology optimization. You could probably shave off enough mass to come in under 10 grams while also achieving a stiffer beam with better vibration characteristics than anything tested here. And while it is a crunchy material, you can compensate for manufacturing errors or minor damage with some clever latticing and probably achieve even better performance.
My Mind is blown, I'm building a Voron V2.4 r2 350mm, a patreon and would love more of the "Red Pill:" AKA Klipper setup/tuning/everyday driver and perhaps Ludicrus speed Please!
I am curious why the beams are square shaped instead of triangle shaped. That should save even more weight.
I also can't help but notice that those carriers y axis carriers on each end of the beam look quite... bulky and not at all optimized. Maybe there is yet more to save there
Triangular section has less torsional rigidity than a square for the same weight and wall thickness.
A circular section is actually the best.
Square beams are way easier to make. But what happens if you remove the beam completely and use just the rail. Then you optimize the y axis carriers and belt idlers.
If the ringing test is essentially measuring resonance, then you should be able to go beyond that frequency to clear up the ringing again - As long as your accelerations are not a harmonic of the prime, it shouldn't happen...
But I don't think the ringing test actually measures resonance - at least not as the primary measurement, because you're far too low frequency to be seeing resonance in the beam - only the part flexibility. In this case, it is of no surprise whatsoever that the lightweight parts are more flexible than a solid extrusion.
Lightweight only helps if your motors have reached their maximum acceleration values due to inertia. Your motors are clearly able to overcome the inertia of the heavier, stiffer beam. You will see practically zero improvement by trying to make lightweight parts. This is wasted effort in this instance.
Maybe try testing using tension in your beams. Concrete beams in construction sometimes uses threaded bar inside to apply compression to the concrete and increase its stiffness. The term to look up is post tensioned concete. Either a thin threaded rod down the middle of the beam or steel wire, Kevlar thread, something with high tensile strength anchored at either end of the beam would maybe increase your stiffness & capability. Of course you would have to beef up your beam to take the compressive load & increase weight but would be interesting to see tested ✌️
Tensioning concrete is not to increase its stiffness, it's to prevent it from seeing internal tension loads, as it it not good in tension, unlike carbon fiber and metals.
The fact that the beams all get the same recommendation from input shaper surely would indicate that the limiting factor for improving accel here is stiffness, not mass - would be interested to see how a significantly stiffer beam (at the expense of more mass) would compare in this case (perhaps an aluminium square tube with unmilled walls?)
My favorite part "Check GPT"
For printed parts I found a Color fabb tsells Lightweight ASA. I’m going to try that in my next toolhead
Wonder what a topology optimized beam would look and perform like...
Input shaping graph analysis would be awesome! There is too little coverage of this topic
I saw a big improvement in consistency on my Tronxy X5SA when I changed to a carbon fiber gantry. it was quite affordable too.
The X beam would have been very heavy for a large 440mm extrusion compared with a CF beam. I wish he did this test on a much larger printer.
Why isn't the X linear rail used as a structural element and ends before the corner mounts?
Your video was scientifically inaccurate, but still sufficient. Bam oida
Great video and scientific comparisons as always 👍 very eye opening too. Someone like you is absolutely a must for us 3D printers makers. Also too I won’t go for sports rims unless the cost performance ratio is there. Maybe next we can see length comparisons on x-beam. Thanks bro, u r awesome.
YOU are awesome, thank you!
If you want to improve the performance of the carbon fiber beam, you need to have the fibers oriented in the direction of the load.
I can tell you before watching the video it's useless because the linear rail and hotend/extruder will weigh much more in realtion, so the few gramms you save on the bar are useless.
What a waste of time - why didn't you tell me?
Excellent video and even if your testing is home made it is still very relevant for hobbyist. Going fast is great but in the end how much time do you save and is the time saved worth your set-up time, effort and money. As a hobbyist I invest in things that don't always make sense just because I enjoy thinkering but since money rules the world..... How long does it take for someone to see a return on this investment? I know I know we don't do this for money as stated before but in the end should I buy one for the look and a few minutes in printing time? For commercial ventures maybe.
Again, excellent work and I enjoyed the video.
Where can i get your hyperbeam or the carbon one with mellow?
I talked to Fail fast about lightweight X beams for a while. I ended up going with a Pultruded Carbon fiber one because it was much stiffer than the aluminum extrusion and the ultra lightweight one. Is your carbon fiber pultruded or woven, because woven preform much worse. I also tested a hand milled aluminum extrusion that I made that was 30 grams and I lost 25hz of resonances because of it. Can't wait to beat Baravel for the WR ;) I also didn't know about the titanium requirement could you elaborate further on it? Also you can buy carbon fiber x beams on ali for the 0.1
(that's where I got mine)
Hey, from what it looks like, the CF beam is woven - afaik you wouldn't see the fibers on pultruded ones.
Looking forward to your WR - you can make it!
Galvanic corrosion, google: "carbon fiber galvanic corrosion" ;-)
So .... lessons learned: the super lightweight stuff should theoretically perform better, but in practice performs similar to the regular 1515 piece?
👍
yes please, we want input shaping video!
What kind of aluminum did you use for your hyper beam? 7075 aluminum is something you might want to look into if you haven't already.
It's 6061. I think 7075 wasn't an option back then when I ordered
brb lemme watch this again at x0.5 speed
It was to fast? You always have the option to hit the pause button
Great video, really enjoyed it! Do you have a link to the tool shown to insert the nut carrier at 4:16? This looks like a useful tool.
Hi Greg, atm it’s only in my hard drive - not published yet, I am sorry
What is happening at 2:11 in the video? The scale starts counting up before the beam is even on it! Did you edit that with a different take so the LCD would be more readable?
Wow you have a great perception- while editing I was also wondering why it looks like this, but I bet it’s only because it was already slightly touching the scale from one side. Just one take here, no fancy editing
You don't have a V-Core3 500 by chance to repeat the same with the big 2020 gantry it has?
Im sure there is a bigger difference there.
In summer I'll build one and the kit should be on the way. I have some more carbon fiber (2020) here :-)
Here's an idea 3d print a generative designed beam in metal, could get lightweight and a geometry that minimizes vibration transfer
Thinking about generative design since I started with those beams in 2021, but I don't have operative knowledge on that.
Maybe I'll revisit that topic some day.
Great set of tests. You could make it even lighter with a triangular truss design. Only the top part really needs to be flat for mounting the rail. The part of the triangular truss could be thickened a bit to better transmit loads along its length, but it would probably have even better stiffness to weight ratio than the HyperBeam.
Just a thought experiment, would a titanium ultra beam theoretically perform any better than the aluminum one?
Might be!
It's a great video. Und I fully agree with normal mini ab or sb it makes no sense but looks good ;). But in combination with a light weight tool head it is possible to lose 50% of total mass or more.
Awesome video! Bought myself an lightweight aluminium beam for my customized Sapphire Pro/Panda Cube
Hmm. If i had to put forward a theory on why the Y acceleration performed worse. It would be that it involves the linear rail mounting.
The Hyperbeam afterall lacks mounting holes for a good number of the rail's own holes, the CF Beam is a bit diy, bringing probably some tolerance errors and the Ultra-beam has that the holes diagonally connected to the rest of the frame, increasing effective length between hole and frame. These may cause minute motions within the rail that cause the software to take umbrage more rapidly.
Even the best mount would suck if the linear rail isn't as rigidly mounted as possible. At least i would assume so.
The efficiency of the beam alone is academic, less relevant than efficiency measured by dividing the stiffness by the System weight (beam mass including the extruder mass). This would show even less difference. Great video.
Really enjoying your content came across it by chance image 3d printing rabbit hole you got some good stuff going here
Interesting video. Good job 👍
Very nice investigation. I think maybe you underestimate the effect of the torsional force. The lateral deflection test technically would also be a torsional force since the the weights seem to be off-centered from the centre of the beam themselves. The from the lateral deflection and torsional defelection test seem to correlate with the input shaping results.
Keep doing what you are doing! You have a great channel.
Thanks a lot and absolutely, I explained that exactly this is the reason why I did the torsional tests
@@247printing I agree that torsional deflection is very important. Can you mount the accelerometer as far from the center of twist as possible and show the amount of vertical component realized. By adding weight to the print head to move the center of gravity, in line with the center of twist you may actually get higher acceleration recommendations from Klipper because that vertical component is reduced.
Love that you hot rod these machines! Is the beam being a constant section a constraint? On of the super cool things with using a product like carbon is it can be made to any shape, also the laminates alter the Performance tremendously. It looks like there is also a huge weight loss potential in machining pockets in the underside of the linear rail.
Loosing weight out of structure is a fun part of my job setting up racing boats.
Think you got weight and mass mixed up.. pretty sure you measured weight.
Won't change anything thing since you were consistent.
totally true!
I was also thinking.. did your aluminum and CF beams clamp on the ends where the bought aluminum beam was threaded?
If true, I wonder if that accounted for the difference in expected performance.
Wouldve loved to see how much stiffness is sacrificed if the carbon fiber tube is cnc machined for lightweight and if the deflection is comparable to aluminium from cutting the fibers on all 4 sides
Me too! Maybe I'll do that...
@@247printing basically a "While you're at it" kind of adventure xD but great stuff
One other thing to keep in mind when selecting material: fatigue. Aluminium is not great at that. An ultralight piece might need replacement every few months.
Very, very good point (when going at crazy accelerations)
@@247printing (or trying to use the same thickness on a bigger machine)
This little machine would probably work with the steel guide and two end pieces only. I bet some people will mistakenly use beams that are lighter than necessary and be surprised.
My comment is just for a future "I told you so". 😉
Interesting how the beams behave with input shaper.
Seems they are all different constructions and thus resonate different with different resonances. Fore sure something to look into.
for a such a small beam its hard to see big gains. Maybe on a 350 voron or 500ratrig.
I'm curious... Could you perhaps mathematically model these tests and material properties to try and correlate your IRL findings? Then you could possibly control more of the variables you mentioned.
You mean FEA modelling? On the list :-)
Great video Albert! It would have been nice also to see a test with the unsupported rail to see the input shaping and torsion difference. For the small size it shouldn't be an issue to use unsupported rails, I know is not the best thing to do for flatness and torsion but I saw some people doing it and seems to be a working solution.
I definitely think we need to see this test again on your rat rig with 2020 extrusions.
We need this!
@@247printing based on the deflection of a simple beam, when you double the length the deflection raises to the 3rd power so we should definitely see a much more conclusive result if light weight beams work or not!
Great video with in-depth analysis. Subscribed!
PS: echo is quite noticeable in the audio and slightly annoying. A little bit of work with your sound capture setup would push the video realization to the top👌
Thanks a lot, also for your PS! Working on the reverb, I also find it annoying 👍
Why do you get worse results with the lighter beams? It should be obvious...
The mass of the beam is not the limiting factor. It is the stiffness of the beam that limits acceleration. You need to engineer a stiffer beam and don't worry so much about mass.
Try maybe an aluminum box-shaped beam with very thick walls that weigh close to the same as the 1515 extrusion. Or even try a solid 15 x 15 I-beam. My bet is the thick-wall box is stiffer than the 1515 exstusion for the same mass. Try a box with thick enogh walls then you do not need nuts to hold the rail.
Voron used the v-slot 1515 not because it is the best structural shape but because you can drop nuts into the slot and not have to drill and tap holes. It requires MUCH less skill and tools to build with V-slot because you don't need a drill press or mill.
But if you are going to make an sell metal Voron Parts, you wouild have CNC equipment and
Have you looked into Aluminum/Titanium hardware? Bigger SHCS you can get a hole through hold although I don't think they make them for such small bolts. Can probably save a few grams here with minimal to no performance loss. Also looking at the vibrations of the frame in your slow mo. The frame is put together by blind through holes how much more ridge would it become with using the 80/20 90° elbows on everything but the X axis then your not adding weight but stiffness to the entire frame. On the X axis maybe play around with different materials instead of ABS. Something glassfilled or HIPS. Wondering if gaining a few grams say 5% weight vs gaining 20% in strength is beneficial on the brackets? How do the frequencies and vibrations play into effect. Engineering mind is running wild late at night here. Love your detailed videos though they keep me thinking!
Next, design a beam using topological optimization. Come up with something that will itself require metal (or PA12) 3d printing to build. I bet AutoCAD would sponsor you (if it works!).
Also: The 1550 beam likely has far more torsional stiffness and out-of-plane shearing stiffness than the box beams. You probably need something like a "space truss" design.
Going even further: once the beams are not a single solid material, you are going to get some cross terms between tensile, shear, bending, and torsion terms. The same thing comes up when you work out the response properties of a layered composite material.
If I am remembering this correctly: you can end up with a material that for example, when loaded in bending will internally give rise to a torsional deflection, and vice versa.
I can imagine the weight of the stock beam and density of it causing a bit of a dampening effect when discussing acceleration. Usually when these beams are discussed they are only looked at from the perspective of deflection when these analyses are run. It doesn't take into account the dampening affect it would have when being thrown at high speeds in the printer. Think about it like how tuning forks work. The amount of material on a tuning fork will change how it sounds. This is a resonance vibration similar to what we are talking about with 3d printing. I bet each beam makes a different sound too...
Thank you for sharing your thoughts on that 🙏
so, I have a theory, the lighter beams have more flex but think of it as a vibration, that disrupts the input shaping vibration. kind of like wobbling around and causing a ruckus while its trying to smooth out the vibrations. there could possibly be a math equation written up to adjust for that extra vibration using your deflection measurements. i wouldn't know how to reprogram input shaping for this so its probably not feasible right now. I was just thinking what if you wrapped your light weight hyperbeam in carbon fiber?
Great video. I think it's not conclusive, though. Just the start of a thorough investigation. Maybe a hollow carbon fibre cilinder could improve stiffness, maybe I'm talking bullshit. All I know is that you're onto a very interesting subject, natürlich.
You should try a sort of minimilast CF composite beam. Like make a foam core with carbon plies over top and bottom, chop them and laminate, Because that's how carbon fiber is most effectively utilized.
This beam is only a small fraction of the gantry mass, yet needs to support the entire hotend mass in motion. So, calculating a Stiffness:Weight ratio using only the beam’s properties will artificially inflate its performance. A lighter beam may drop 60% weight and lose only 30% stiffness… but if it was only 10% of the gantry mass to begin with, the assembly as a whole exhibits a very different total: just 6% mass drop for 30% stiffness loss!
Amazing visual analogy, hyper optimized 'beams' when while still using PLA parts (of all things) in an assembly is just like driving those hoopties with oversized rims, all show, no go.
Very nice video. Thumbs up...
I hope being not too much of a smart-ass but when explaining the low efficiency by the total mass safe to be only 8%, I need to say that it is even less, because lacking the mass of 2x mgn9h sliders plus the mass of the rotator in the nema14 motors ( rotation wise).
I would expect a bigger impact on larger printers although these printers will not top the little voron on absolute speeds.
I Lookign at how the weight interacts witht he beam, I wonder how a hybrid lightweight beam would perform. Granted, the weight savign is on the most stressed and least heavy part, but it's a step. If a 2020 extrusion was cnc machined in only one direction, front ot back, with the other direction left alone then since there is no up/down movement the forward/back movement would retain the advantage of more material resisting flexion while possibly retaining enough stiffness to resist twisting. I have a background in competitive paintball from a manufacturer's perspective and I've heard players swear up and down that ounces and grams matter. While I remain skeptical in its application in shaving weight off a paintball gun, I can see it's real advantages here. I wonder how much weight could also be shaved off of the other mechanical parts to achieve a truly featherweight xy system.
Look at adding damping. Aluminum has very low damping properties, something like a constrained layer damping (search CLD) would add mass but limit the magnitude of vibration
It might be more cost effective to buy bigger stepper motors, than spending money on fancy beams. Algorithms fine tuning the accelerometers would help too.
Mass is the cheapest vibration damper , ,so maybe the best solution would be to build a carbon fiber beam ,perhaps 75mm square, using bigger stepper motors. Then, set up a program to simulate a high speed back and forth motion. Then tune the beam to dampen any resonants, by adding mass to specific parts of the beam. (Resin painted on ,or shim stock glued on)
Can you add the Ender 7 x-axis beam setup into your next test video, just to see the difference since it's just a flat/thick piece of metal, currently have one running and trying to figure out what I should be looking at to upgrade current print quality and thought that something completely different than boxed metal?
I'm no machinist nor expert, but I have some machining experience.
Mass might be your enemy for speed as it has to start and stop and that can cause overshooting aka ringing.
Too much mass and you need to use bigger motors to move it and can't accelerate so quickly.
At the same time too little mass will introduce harmonic vibration. Also ringing.
So it is a double edged sword.
There's a way to have your cake and eat it too, ramping your speeds accordingly.
My guess is mass is dampening. I seriously doubt input shaping is tuned for ultralight high performance. Kinda like a killer gpu with generic drivers…meh blah.
I am sorry. The aluminum beam weight is not much. Reduce it weight is meaning less. We should reduce the weight of the linear rail it self. Why it have to be so thick while the linear rail it self rely on how straight the beam is?
So, I'd say you're getting worse results with input shaper recommendations because the lighter mass is not able to absorb the vibrations produced as well as a heavier mass. This doesn't mean that the lighter gantries can't go faster, because they certainly can, but that's only half the problem.
Think of it like two boats on rough seas. One is much larger and heavier and has an overall slower top speed, but because of its mass and larger profile, it can handle the rough waves much better, providing a smoother ride. The other boat, which is small and lightweight, has a higher top speed, but gets tossed around by the waves, limiting how fast it can really go without breaking against the waves.
I realize this analogy isn't perfect, but I hope it gets my point across well enough. How to fix this, I'm not exactly sure. Perhaps progressively increasing the density of the lattice towards the middle of the beam to reduce flexing in the center might help?
Maximum acceleration is highly correlated to beam stiffness so a solid material will generally outperform hollow material with the same outer dimensions. If you want a lighter beam with the same stiffness you need to make sure the moment of inertia is the same, which for these beams, isn't the case. The carbon fiber beam and Ultrabeam likely have a similar wall thickness so they get around the same stiffness and maximum recommended acceleration. Your Hyperbeam has thinner walls that both the carbon fiber and Ultrabeam so the stiffness will be lower.
This means you'll need to adjust both your outer dimensions and your inner dimensions as opposed to simply hollowing out the center.
Drag of the linear rail bearings is maybe also a flaw. Have you tried some special no slip “grease”? It’s used on mtb suspension. Have shop sized bins, can send samples for free. Greets from Nbg
Out of curiosity, could you ditch the beam and just get a bit larger or longer linear rail to bridge across the axis? Admittedly you'd probably end up needing a much higher quality linear rail that isn't bowed or anything of the sort, but, just my curiosity.
Been watching you for a little while and only just realized what your channel name means
What does it mean? :-D
@247printing I always read it as 247 printing, but as I said, I just realized it means 24/7 printing
Really interesting! I've been diving down the rabbit hole of fast printers last few days, love seeing all the thought going into these printers! what about the linear rail? what would be the lightest option for that?
I love creators and companies collaborating to make awesome new products
Even as janky as that test kinda was I think you still made your point fairly well. Lightweight beams are clearly just a gimmick that look cool & don't provide any meaningful performance gain.