Minuteman Dryruns on World Record pace. Is a sub 1 min 3DBenchy possible? (Episode 7)

[@Splarkszter]

Sub 2 minute benchy: A highly modified but still normal looking Ender3. Sub 1 min benchy: A death machine

[@ZURAD]

Halving is very difficult

[@Splarkszter]

@ZURAD Yeah. Diminishing returns territory.

[@kevok3714]

You mentioned shaving weight down off the build plate, have you considered moving your air bearing from the build plate to the surface? Something similar to an air hockey table, lots of tiny holes providing lift against the whole buildplate surface, rather than a single large centralized hole. That would eliminate the ducting of the build plate, allowing for a simpler, lighter piece. No idea if it would help, and it would be a lot of work to drill a bunch of .5mm holes in the surface. All the best, I’m enjoying this series, and I’m learning a lot from it!

[@Vez3D]

Awesome man! Lots of research and engineering. May i suggest one thing? Maybe you did it already but just in case... Optimizing the slicing layer by layer. There is tons of things at each layer you can set differently. Use modifier and tune each layer individually :) That is where you will gain the max time i think once you have reached the max from the hardware

[@wavyseahill]

@@Vez3D hello Vez! we were discussing the kilomillimeters in coments. You use that as well. Can you tell us the reason?

[@Vez3D]

@@wavyseahill im not sure I understand the question.. sorry.

[@wavyseahill]

​​@@Vez3DJan and You as well quite often write acceleration in several K of mm/s^2, so basically kilo-millimeters. Why do a lot of people do such a thing instead of just converting to m/s^2. Writing 500k mm/s^2 is basically (1000/1000)*X. Question is - why?

[@Roetz40]

Yeah i know thats a thing. But I also got a regular job and social life :D maybe ill invest the time and analyse the motion vectors of each layer and apply a filter on top. Setting parameters layer by layer is just a pain...

[@Roetz40]

standards... we started with milimeters and continue to do so for easy comparison sake.

[@SarahKchannel]

Neodymium magnets are very sensitive to temperature. A motor with magnets on the stator, where all the heat of windings accumulates would most likely be more sensitive to heat than any other motor. Cooling the motor housing/windings will make most likely a noticeable difference. Its the same reason that speakers use much less often Neodymium magnets then regular ceramic magnets.

[@rocketslizardsandall]

Hi there! I'd like to point out that you're 100% correct on the heat issues with neodymium magnets, they don't hold their magnetism as well as say ferrite magnets at temperature, but that's not why they're not used in speakers! In fact many of the highest power handling (ie heat) speakers use neodymium magnets, it's simply a matter of cost. The magnet mass in large subwoofers even for neodymium can approach the kilogram mark, which is a very large amount of cost. As such ferrite is simply more economical. To clarify, I'm a speaker designer, so I have a very thorough background in the field.

[@satibel]

@@rocketslizardsandall btw why don't we see more dual coil speakers? (as in speakers with 2 coils in opposite directions as opposed to a permanent magnet)

[@TrippyRiddimKid]

During your simulated weight tests I noticed that the steppers at that speed are actually flexing the mounting brackets causing the stepper to tilt downwards at the shaft end. This MIGHT be where some of your inconsistencies are coming from, as that slight movement would put extra tension on the lines and cause an artificial increase in weight which changes based on the angle of the stepper. Maybe redesign the mounting to include a clamp that holds the steppers down at the back as well?

[@peterdavis9403]

I noticed this too at 30:00 in the video. It may add some small Coriolis? force to the mix which clamping the motor at both ends should solve.

[@dscottwilliamson]

Have you considered dual motion platforms? A core XY gantry with extruder over the moving bed, each moving in complementary directions. The relative acceleration’s and top speeds would far exceeded the capabilities of either approach separately, potentially doubling them.

[@regun2434]

I had idea about dual delta where one awd delta moves printhead and another moves buildplate, but I'm lazy

[@lumotroph]

Nothing to contribute to the build, but about the videos: I don’t think anybody minds them being long! This sort of content is what I watch KZbin for - if it’s longer it just means I watch more of your videos rather than changing between a few shorter ones. Keep it up!

[@dihler55]

I love how the test sounds like pneumatic actuators firing instead of steppers :D

[@BlackHeartScyther]

With that massive delta between just the pulley and the weighted pulley, I'd love to see a few of these motors linked together shaft-to-shaft to see how high acceleration you can get with that. Ideally, 2x motors at 4a each makes an 8a motor, it would be very interesting to see if the motors working together would increase overall acceleration or just torque. If just torque, it looks like that would make the larger pulleys work better to possibly achieve higher acceleration through the gear ratios at play.

[@Brainstormer_Industires]

The amount of headroom between the "motor only' curve and the "motor with load" curve indicate there could be significant gains from more motors.

[@rocketeer8719]

I agree with this insight and I think it’ll boost the possible maximums, I just don’t know what new issues will arise from combining two or more motors together

[@satibel]

something interesting you can do with doubled motors is offset them by a half step to effectively double the resolution. however with that you may need custom drivers that can work in lockstep. at this point you might even wanna run an fpga driver with closed loop control.

[@spanny52]

I build wire rope assemblies for my day job. I would like to offer some help/advice. You can buy pre-stretched wire rope in addition to internally lubricated constructions that will help with performance and longevity. Maybe you know all this already but any adjustments you can refine to take out constructional stretch and application stretch, it’ll make a big difference. You call also proof load your assembly up to 60%of the nominal breaking strength take out almost all the stretch. All of those wire rope clips and splices add up as you know. I would use a bolt with a hole drilled through and a ball shank to tension your bolts. I can send more details if you like.

[@Roetz40]

Thank you, these information are valuable. Are you interested to join my discord? Link is in the video describtion. Thats easier to communicate :)

[@polarbear2272]

Two quick questions: 1. Could the drop in performance at certain accelerations be related to harmonics? It seems like a significant amount of energy might be transferring into the cable, judging by the noise. Perhaps a solution could be to add a sub-plate and mount the motors closer to the build plate. 2. Have you considered biasing the motor orientation along the length of the boat? This might allow for greater acceleration on the longer travels, making the power use more efficient.

[@802Garage]

Wow this really makes the 2 minute Benchy even more impressive. Awesome work look forward to the next episode as always!

[@misterbean29]

I much enjoy your journey and I especially appreciate that you gave Monika the recognition and mention she deserves. Her Ender3 project was great and she certainly did set a valuable milestone. Did you ever try to find the max accel of your motor without any load? I can not remember you ever did.

[@michaelbraaten]

I’m so glad I clicked on this video! I love this stuff! I started school as an engineering student but went a different way in life (live and learn), but I am absolutely nerding out on 3D printing and modding. I have built many custom watercooled PCs in my time, so 3D printer modding and engineering is right up my alley 😁 I’m now subscribed and I have to go back and watch your previous vids on this! Great stuff!!!

[@eviethekiwi7178]

so many commenters thinking they know better after watching a few minutes of video, not considering the hours and hours of development and work put into this setup.

[@eepydragonloaf]

Well yes but communication is also key when optimising something. Especially if you work on something for a very long time you can get locked in on a solution and fail to see a more optimal solution that may be apparent to someone who has not spent much time on it. Don't think that's the case for most people in this comment section though i might very well be wrong, however allowing communication whilst it may lead to a lot of terrible ideas might actually lead to something good. Just in general. Communication is essential. Speaking from experience.

[@jankington216]

This applies to every youtube video

[@charlesstaton8104]

@@eviethekiwi7178 in the video, he asked for feedback. In the comments, people are criticized for giving it?

[@orphax1925]

when you are in a project mixing mechanics, electronics and software you have to accept that some people know better than you in specific areas

[@satibel]

it's almost as if people who are familiar with a subject have a high tendency to talk about it, for example I have a degree in yapping...

[@ianroberts8371]

What about using carbon tube push rods instead of cable, similar to a delta machine but flat? This project is crazy cool!!

[@tfxWarrior]

Hey super nice video series! Just had a weird shower thought kinda idea: could you make the extruder/hot end move as well? ofc not as fast as the build plate, but maybe a solid amount of movement opposite to the buildplate could increase the effectvive acceleration you can achieve.

[@vinny142]

A) Have you seen the size of the hot end? :-) B) If the hot end moves then it has to make it's move in the same time that it takes the bed to move, there can be no lag, they have to be at the same position at the same time. If the hot end moves then the bed doesn't have to travel as far, but if the bed doesn't have to travel as far then it can get to it's next position even faster and so the hot end has to move faster too. It's a circle.

[@TakeApartLab]

also, then the cooling will have to move with the hot end, and that cooling is gonna have to be massive.

[@leonnundel2443]

Mega cooles projekt was du da machst. Bin seit Tag 1 dabei und das prinzip von dem Mechanismus ist richtig geil

[@awesomelyslowplastic]

Such a cool project! Really looking forward to seeing your next tests! More tension on strings or a stiffer material maybe to prevent the string from "flopping" around?

[@Roetz40]

Yes I have to adress that. Next iteration will have the strings more supported on the longer sections

[@sammiller5509]

wuld it be better to tension the cable indirectly by removing the tension springs in line with the cable and instead use a pully which can be moved outward to tension, like a belt tension mechanism with a thumb screw?

[@Sh4de2k]

@@sammiller5509the springs are there because the lines that connect the motors diagonally, have to be able to lengthen when there is movement in the other axis.

[@phizc]

​@@Sh4de2kwhy not compensate for that with the other steppers? If the build plate moves away from the center in the Y direction, both X steppers should release tension on their cables. Or would custom G-code violate the speed benchy rules?

[@awesomelyslowplastic]

@@phizc very uneducated guess here, but I think that would be insanity trying to get timing and distances so accurate. Especially running steppers, maybe with closed loop servos

[@alexcrouse]

You should ask the Slowmo Guys to visit and help you diagnose motion controls at these speeds!

[@deckname5794]

Sure, since they are basically neighbors, they could just come over and help out.

[@cxob2134]

Awesome project, cant wait to see this thing Print!

[@Bu5H84]

Hahaha the crazy part even though the actual benchy wasn't printed i could visualize what part was being printer just based on the movement of the hardware, iv been watching way too many benchy being printed :)

[@B0A2]

Loving every episode of this!

[@SkansgardCNC]

IMHO a lot of people underestimate steppers and go for closed-loop without actually understanding why it worked. FOr something like a CNC, closed loop is mostly useful for stopping the machine _if_ it looses position/steps. An adequately sized stepper with the correct parameters will perform the same job as a closed-loop stepper, except you will not get the reliability. When ppl go from an open-loop nema17 to a closed-loop nema23, it is the size/power that is the biggest improvement, not necessarily the closed-loopedness

[@link7417]

closed loop at least when it comes to servos can also provide active correction (which is the real utility whit it whit closed loop servo though I don't know about stepper but I cant see why that should not be possible whit a proper setup) not just aborting if something goes wrong, for when the motor losses steps it will not have travelled far enough so the closed loop system can correct that and make it travel further, this also works if it overshoots the target and make it back to the right position though that function is only useful whit travel between points whit no other operation

[@SkansgardCNC]

@link7417 in a cnc-mill you don't really want that in many cases, as you will just get inaccurate results. failing to reach desired width of cut means next round it will have to work twice as hard, and is more likely to fail. If for some reason it does not fail, your part will not be accurate

[@SkansgardCNC]

On a 3d-printer where anything within +/-0.1mm is good, this does not matter much, but in metal you can feel details down to 0.01mm or something, and you get visual imperfections even smaller than this 😅

[@link7417]

@@SkansgardCNC um no, it does not have to work twice as hard on the next pass, it does not compensate on the next pass whit the tool, if you tell a CNC Mill to travel 100 mm on the x axis but it only travel 99 mm it will compensate for that right away because it will not blindly trust that the motor will reach the desired position, yes ideally you don't wanna push a CNC mill to where it could lose steps in the first place

[@SkansgardCNC]

@@link7417 If you just use closed-loop steppers as a drop-in replacement, the cnc-controller has no idea that the desired position were not reached. It is all internal in the stepper-driver. I am sure that professional cnc's have no problem dealing with it, and that it CAN be dealt with, but grbl etc does not give a shit 😅

[@markp5726]

This looks quite challenging! One thing to keep in mind is that you won't be able to push bed acceleration all that much, since the object being printed will contribute a changing amount of mass to the system. Or do you plan to compensate for that as you go and reduce acceleration based on amount of filament extruded / mass added? I also wonder how much acceleration is possible without breaking the bond between bed and object.

[@weirdsciencetv4999]

Parallel cable machine. Pretty damn brilliant! Subscribed!

[@kuglepen64]

Brackets are flexing, not along the string axis but still. Is the string you use too elastic at these speeds?

[@Roetz40]

i guess not. Everything resonates, I found a mode at this printing speed. I'll constraint the movement next iteration

[@wktodd]

How about.. move both nozzle and part ? Halves the require distance for each axis doubling the acceleration.

[@Roetz40]

I've considered doing this. Problem is that my hotend weights about 20 fold what the moving bed weights... Thats going to cause a lot of problems

[@wktodd]

@@Roetz40 long heated nozzle pivoting to reduce moment , perhaps? Or just move the nozzle for shortest moves (infill chimney etc. ) . Missing out on the possible gain from relative motion seems wasteful

[@KnowArt]

awesome stuff!

[@MarcelRobitaille]

Is the weight really a valid substitute for the rope? The weight resists accelerating in either direction, but the rope only resists when pulling.

[@Roetz40]

Youre right. I have to investigate if we can run faster/slower values in any of these directions!

[@tyranbrack2637]

Not sure what the exact rules / regulations are but a possible design could be mounting the bed up higher and allow the connecting belts to run underneath it. Laying the bed across the belts with the corners pointed towards the motors, giving it the most surface area for attachment / support / stabilization. But you could then have the opposing motors work in conjunction puling the belt (string or whatever) instead of opposing each other. This would also shorten the belt / string, and could make it a large loop so its pulling against itself? Removing slack / pulleys. There would be way more inertia and what have you, all things that would need to be worked out. "Design" just randomly came to my mind so thought I would suggest it.

[@anderslagerqvist2642]

Just a thought...A piston on the extruder to control retraction would be a cool test. Normal retraction feels like it will have some limitations that will be hard to handle. The piston could handle fast changes in the flow. This way the extruder can focus on pushing forward mostly. Controling the extruder would be much more complicated though...

[@thomaskamp9365]

@Jan Du hast bei 15:55 min ein Problem: Wenn die Neodym-Magnete erst einmal heiß geworden sind, dann verlieren die auch Ihr magnetisches Feld (Magnetkraft). Du musst also von Anfang an aktiv kühlen.

[@Roetz40]

Ich bin hier im Bereich wo neodym noch keine Probleme machen sollte. Mein Flaschenhals ist der Kunststoff der weich wird und dann verändern sich alle Abmaße im System. Aber ansonsten hast du recht ;)

[@user-vy5hc9ud6l]

Why did you say that only open loop steppers are allowed? I thought it only said "The machine must be powered by stepper motors". So maybe closed loop to be able to push closer to the limit.

[@Roetz40]

I asked this into the annex discord and got this answer. As soon as i hit real limits i will ditch the open loop aproach

[@charlesstaton8104]

​​​@@Roetz40We need a new speedboat competition. The rules of annex one are stifling innovation, not promoting it. Servos are no longer something that only companies and wealthy hobbyists have access to. They totally make sense to put on a 3D printer now. I can make a servo out of an RC hobby BLDC motor and a FOC driver that costs less than high end NEMA17 steppers. Also my personal peeve, is the focus it puts on *_one_* benchy. I asked recently on their discord server if it is OK to print, for example 10 benchies in 20 minutes, and call them 2 minute benchies. They said no. It must be a single benchy. There is no way to cool that much plastic in that amount of time. It is a material limit. Physics. You will only ever make poops at a sub-3min pace and they would rather see one pile of poop at 2 minute pace, than 10 actual boats at 2 minute pace. I'm not about to invest that time in making poops. BTW i made it clear that I meant one toolhead, one nozzle; not 20x nozzles printing parallel like you did before. Still NO. this speedboat thing has run its course and Annex are not willing to evolve it to the next level. I see no reason we should maintain interest in internet credits minted in 2016.

[@merkatorix]

At 20:35 : What is the "pfioui" sound after the ratteling? It seems to occur after the steppers stop.

[@davidreinhart373]

I may be wrong but I think that the way you simulated the weight of the printing platform using flywheels is not accurate since the motor will have to accelerate AND decelerate the flywheel all by itself, but when printing it only accelerates the mass by itself decelerating is done by the other motor since you can't push on the string. I don't know how much it matters but I thought to point it out.

[@Roetz40]

Yes you're right. But it gets me very close :)

[@davidreinhart373]

@@Roetz40 You could take two steppers tie the pulleys together with string and then use the flywheel again just to see how big the difference in testing methodology is.

[@davidreinhart373]

Thinking about it more this method wouldn't be accurate either, best would be hanging lead weights on the string.

[@Roetz40]

@@davidreinhart373 everything with hanging weights needs to have a few stages of mechanical advantage. Otherwise were limited to 1g of acceleration. We want to go 100g+ :)

[@davidreinhart373]

@@Roetz40 I meant two steppers connected with a string and weight simulating the weight of the print bed between. To get rid of the "pushing on a string" problem

[@Infinion]

I looked at the stepper motors you were comparing and Lin Engineering's 76mm long NEMA23 part X5718-077-50 "Extreme Torque Hybrid Stepper Motor" beats the nanotec torque at 325rpm by 15% (2.95 Nm) and inertia that is only 30% at 144 g-cm inertia. If you want to add a gearbox, they add a 5:1 to that same motor with part number X5718-077-50-Y05-A0-A12F and your inertia is doubled (still less than nanotec), holding torque goes from 3.3 Nm to 14.85 Nm, and torque at 325 rpm is 5.83 Nm at your elevated voltage.

[@elvendragonhammer5433]

Technically a sub 1 min benchy has already been done; twice. However neither of those methods use an actual "3D printing mechanism." Both use a gel to stabilize the formation of the object in 3D-Space. One is pure liquid deposition, where it injects a liquid into a reusable gel, & is the more versatile of the two, as you can create objects with plastics, rubber, foam, resin, ceramics etc, all combined into one object. They used a full size automobile seat as a "print in place" original design & test case. The second is far more expensive, requires much more physical setup space, technical know-how & skill, is much less versatile, (at least for the time being), but is the fastest in the world. It, like the prior method uses a gel, but it was much firmer- like ballistics gel instead of the more "goopy" nature of the first one. It also wasn't stated if this kind of gel was reusable. Unlike the other one, it has a UV curable resin fully encapsulated in the cube of gel prior to the start of the process. It then uses a femto-second laser split into 3 beams & each beam is projected through a polarizing screen creating a hologram projected from 3 axis into the cube. The UV light essentially "flash freezes" the hologram into outer shell of the object & then the laser is turned off & full curing starts from there. The one I saw was a Mardi-Gras style candy skull fully finished in just under 13 seconds, with immense detail, (easily as good if not better than any 8-K resin printer) but with even more strength, because there are no actual deposition layers. (There still are with conventional resin printers, they're usually just too small for the naked eye to see, but they still weaken the part compared to a mold-cast cured part) Regardless I think this is a cool project, & that you & whoever else is venturing into this area will likely benefit many more people, as their use case & cost, though more expensive for the avg user, it's still more likely to eventually improve printing infrastructure in the long run rather than the niche use cases I mentioned above. From a technological perspective, I'm most concerned about the vibration at the speeds you want, & the accuracy level. It seems to me layer shift, adhesion & extrusion speeds are going to take a ton of work if you ever intend to make it be able to produce a viable product using this method, unless your only goal is to see if you can just break that record with a more conventional style printer. Either way, I hope you succeed though.

[@techman8817]

That hologram system sounds very cool.

[@nippelfrost]

Great video and thanks for the time and effort you are putting in! Have you considered modifying the Motors? Machining off the end for closed-loop application and reducing diameter and length on the other end is what I would consider.

[@excitedbox5705]

I think the slack comes from slipping of the loop on your cable, since missed steps would move the stopping point of the build plate. A second possibility is the cable slipping on the pulley and moving slack/tension from behind the pulley to the platform side. The issue with the larger pullies comes from the rotating mass of the heavy weight so far from the center. This issue will be resolved by lowering the weight of the pulley. The more mass the pulley has the more energy is used to change direction.

[@99seaweed]

This has probably already been discussed, but can fishing line be used instead of the thicker cable/line that you are using? There is a big variety of strong/light/stiff fishing line.

[@RegularOldDan]

I remember not long after the Prusa MK3 came out, I was talking with someone who was ABSOLUTELY convinced that the quoted 200mm/s max speed would never be reached as he felt the accelerations would shake it apart or at least move it across the table. (Silly, I know.) If only I could show him this now. WOW. (To be clear, I understand how older printers often wouldn't reach their theoretical top speed due to lower accelerations. With a Benchy, there's no way the MK3 would reach 200mm/s, but not because it would shake itself apart, but because the necessary accelerations just wouldn't be able to be achieved with the stock hardware.)

[@TinSVM]

I've never clicked play so fast.

[@timha4102]

Did you do it in less than a minute though?

[@mvadu]

44:14 that accelerometer is going have one heck of a shock when you set up the input shaping!

[@derekrussell9925]

You should definitely try clearpath servo motors! From Teknik torque will go up dependability will go up as well! You can even set them up to follow step and direction signals, among other things

[@Adrellias]

Really awesome build 😁 as an idea why not attach an imu to the build plate, just for measurement and calibration puproses?

[@matthiaspenzlin6465]

time to go to archery bow strings.

[@mjodr]

I like how the video compression codec really struggles to show the fast movements.

[@bluereaper6.472]

Have u considered a thin metal wire vs rope to reduce mass, resistance and stretch. Also have u considered 2 idlers or one larger one at the corners to reduce drag and stress by the high cornering angle.

[@DavidMoscoeUni]

would be a software nightmare to implement but if you also had the extruder moving with a separate set of steppers you could squeeze some more acceleration out of the system. Feeding the heating element from Boden tubes too could get the practical weight of what's actually being moved much lower

[@justparanoid]

Novanta IMS makes Nema23 stepper motors that on paper looks great. Might be worth taking a look.

[@tangent_theta]

Bro this is fantastic. I wanna see that with the extruder on. How might you go about measuring positional accuracy?

[@JoeMalovich]

Can a linear motor be built that operates in 2 axises? Not compound, but a plane of coils and magnets.

[@810mb3Rg]

Do you measure the current in the coils directly, or just state those numbers based on the current setting of the drivers? If so, how are you measureing the current? Im also trying to drive a stepper very quickly and trying to optimize the drivers tuning 😁

[@CarloVaccariPlus]

Input shaping can help reduce cable end vibrations.

[@kreglamirand2637]

replace the cable ends/adjusters you have with spliced amsteel or UHMWP cables. you could do a splice that also works as the adjustment mechanism but ideally that can be done on the back end to remove all excess weight.

[@roboman2444]

So.... what about two steppers per pulley? One on either side of the pulley. Just wire them in parallel and run them at 4 amps (2amps per stepper each).

[@user-xe2ek1td1x]

I know this is pretty far in your build to change up now, but what about for your next build, for your moving platform, instead of cables you change your drive to an air piston system? You seem to have access to the a compressor, a 3d printer, and machining equipment. You can have air solenoids connected with pistons situated in each axis. As long as you create a feedback loop, inertia shouldn't be as much of an issue. Robots already use air motors and pistons for the exact challenges you're running into. As for cooling, a liquid cooling system and reverse heater core could keep your systems ice cold. It would be a beast power wise, but you could easily surpass your bench marks. Good luck!

[@loiblanche5863]

I'm not really familiar with the scene but why don't you move the bed and the hade at the same time, I get that it would be harder with certain head sizes but it would still be faster right?

[@grougrouhh1727]

really nice idea for the air plate

[@InductorMan]

Gosh, I think input shaping could help a whole lot. Your data seems to show that the inertial contributions of the build platform and the stepper rotor/pulley are comparable. Then they're connected by the significant compliance of that Dyneema or whatever you've got there. And we can see the slack side of the cable system flopping around during certain parts of the video when the platform motion aliases with the camera shutter. It seems like input shaping or something similar could do a whole lot to prevent the build platform from jerking the slack cable as the tension is removed, and knocking the rotor out of the magnetic detent. You're controlling three masses connected by very significant compliances by applying force to only one of them, and treating the system as rigid. It seems like the oscillatory motions of the system could easily be producing for spikes equal to the applied forces: in the worst case maybe doubling the propensity of the system to skip steps. Obviously, you have managed to tune it through cable tension adjustments and gotten performance much closer to the rigid body ideal than this, since you're getting clearly quite close to your rigid inertial mass test data. But I question whether this motion control system is actually useful when tuned this way. I would guess that you're being forced to allow excessive cable slack and excessive build platform wiggle in order to avoid spending your torque on the resonant motions that would occur between the two rotors and build platform if the cable were properly taut the whole time. I think you're going to need more than inverse kinematics here. You're going to need a separate feed-forward/input shaping algorithm for both the driving motor and the slack motor during any given motion, to smoothly tension the cable between the masses on the driving side, smoothly relax the cable on the paying-out side, and then smoothly return to normal tension as the motion segment terminates. Maybe it could be the same code, but I have a feeling you're going to need to treat the Dyneema elasticity as nonlinear on the paying-out side. Point 2: You're talking about how much trouble you have dealing with the cable tension: how about fixing that directly? One thing which comes to mind is that it might actually be helpful to have extra compliance in this system to set that tension, in parallel with a very stiff viscous damping element or even a simple locking clamp. The idea would be to have a soft compliance/spring that sets the resting torque of the motor, and a stiff damper (or clamp) in parallel with it that keeps the system rigid during motion. For instance, you could dispense with the cable tensioners, and one of the motors of each pair could be mounted on a sliding mount. The mount could then be provided with a long, compliant spring that applied the static bias force to the cable segment between opposing motors, setting their resting torque. Then, you need the motor platform to remain locked in place during printing. Ideally you'd have a parallel damper or mass/damper element to keep it pinned in place dynamically: a very stiff shock absorber from the platform to the motor mount. Perhaps more simple, although not as ideal, you could just lock down the stepper mount with screws in a slot after allowing the spring to properly tension it. This isn't as nice because it doesn't deal with nonlinearities in the cable tension that would start showing up as the platform begins to move. The damper would automatically deal with that, and keep the average motor torque fixed to the desired value. But then again, setting the tension and then locking it down might be more predictable.

[@3D_Printing]

The Creality K2 is said to use Forced Feedback Servo's this is said to be very quiet and may be push speeds

[@Zilli_341]

It looks like open loop control is reaching its limits. What about mechanically coupling two steppers toghether, do you think it would help? It would of course double the number of motors and drivers, but it shouldn't increase the complexity too much. I don't even think you need 8 of the same motors for testing, the ones you already have should be enough to test the acceleration of one axis. People have been using 2 motors per belt on CoreXY machines for a while and it seems to work well (but I'm not sure how hard it would be to sync two motors that are rigidly coupled togheter, as usually there is a somewhat stretchy belt in between them)

[@stanciuandrei7932]

At this point the accel numbers could be measured in Km/s² or mach(345m/s). The nanotech motor with the 60mm pulley basically hits mach 5.5/s acceleration (mach 5.539/s or 1900m/s²)

[@lumotroph]

That’s a change of measurement units I can get behind! Mach 5 🎉🎉😂

[@scottshipley570]

Should build this on a air hockey table to make the bed frictionless

[@imviper]

Looks very interesting. I had a thought which may or may not make sense. Have you considered instead of syncing motors using 2 motors with a constant rate spring and maybe linear motors.

[@thomaskletzl6493]

if i understand it correctly wouldnt you reduce the gap between just pully and simulated weight if you use 8 motos. So you only have half of the inertia per motor?

[@Roetz40]

Yes. I still think its smarter to reduce the weight of everything in the first place instead of making it more complex :D

[@Sewad_J]

Why not get the encoders for the closed loop bit? wouldn't that give you all the data you need about the motors as well as a reliable way to control them?

[@SW-qr8qe]

Great work!

[@David-gk2ml]

Can you have a two stage hot end? two temp zones, preheat then melt. I was thinking about less melted-volume that potentially drips with poor control...

[@renevile]

it looked like there was a lot of rope side wobble, do you think that has an effect on your acc limits. i think that it might have a little effect perhaps there is 10k to gain by mitigating the wobble. you know how rollercoasters brake using permanent magnets and copper bars, simply using eddie currents to brake. what if you made a fairly light 'ring' where the middle of the rope can pass through, so there is no touching when That rope is doing the pulling. but when the rope is wobbeling sideways because the other axis is moving then the ring is also getting moved side to side. now if there are some small magnets on that ring and a copper strip above and below, then the ring shuld resist 'fast' oscillations. the question is if there is something to be gained from a system like that, and if the added resistance is worse than the potential gains. the reason i thought of this is high voltage power lines actually have dampners on the wires at the point where they hang from the towers, this is to stop them swinging in the wind. now i probably just wasted 5 mins of your time if you read this, but we will never know if it works unless someone who knows the machine actually thinks it through, i can easily sit here on this side of the screen and think something will work, but you are the only person that can decide if there might be something to gain.

[@tomekpeki]

Bro, you probably know that, but I'll leave it here just in case. Changes in the current settings don't affect max acceleration above a certain level, because real current is not rising in the motor anymore. It's all because of the motor's coils' inductance. There is simply not enough time for current to start to flow through the coils. To get higher accelerations/max speeds, you need higher PSU voltage for the motor drivers, so the current "is pumped" faster into the motor.

[@michaelbraaten]

One question: on the printing parameters of the sub 2 minute Benchy dry run, what does the term “square corner velocity” refer to? I’ve never heard that term before. Thanks!

[@Roetz40]

It's the speed in mm/s the printer will maintain when it's doing a 90° corner (instead of doing a full stop at the corner point)

[@michaelbraaten]

@ thank you! I saw in another video in the series you called it jerk which I have actually heard of lol

[@Charliewebbwins]

someone needs to get this guy a high speed camera

[@stevenbacon3878]

Do the long electrical cables between the motor and the stepper drivers have any negative impact?

[@gv100_blitz]

Shouldn’t the rotational mass test be off center from the motor?

[@anthonykelly8480]

Instead of tying off the cables at the base can you return it through the pulleys to the spring? Like a double cable looped at the base and fixed back at the spring. Might fix the tension issue.

[@guestc142]

The minute this thing starts printing, it will have finished it.

[@slevinshafel9395]

15:08 i like your wokrshop. i mean is relax and enginering not fancy RGB led and shiny things. I like it.

[@NuggetsInclusive]

have you conciderd an ice hoky table type mechanism to reduce friction?

[@Roetz40]

This is basically the same thing here... :)

[@handdancin]

doesnt the distribution of the weight relative to the axis of rotation matter with those weights you are using to simulate stage inertia? i.e. if you made them smaller diameter but longer vs larger diameter shorter, etc etc

[@MrKRSO]

Escap / Portescap disc magnet stepper motors have very low inertia and should be ideal for rapid acceleration like this.

[@pipboy370]

Hailo erstmal, ich habe jetzt das Video fast komplett durch und frage mich die ganze Zeit, was gegen ein reduzieren der Schwungmassen der Antriebsräder spricht. Auch sehe ich noch das Potenzial sämtlichen Umlenkrollen Kugellager zu verpassen. Die maximale Geschwindigkeit wird erreicht, wenn die dynamischen Massen Null erreichen. Aber im großen und ganzen ein sehr interessantes Projekt allein dafür eine Messplatte als Basis zu verwenden 😃👍

[@luddebosse1439]

It kinda makes sense that you will get better acceleration from a smaller pulley thanks to the mechanical advantage you gain. you get more torque but less speed/ distance per rpm

[@Roetz40]

its always about finding the sweet spot...

[@chuckrouse2451]

Would duet3d parts be of advatage? 🤷

[@vinny142]

I'm new to the series but still wondering: If one motor isn't enough then why not two?

[@andreasp.7246]

might be a dumb idea, but why dont you just use 2 motors pulling on each side? double the torque? and if you connect the shafts together, than you have bearings on two sides of the pullys

[@martin09091989]

More Power!!! 💪 You should first find a drive solution that can handel high acceleration on its own. Those steppers are near there limit what they can do with just there own inertia anyway! Large motors with very large pulleys! Angular acceleration will be the same, but the bigger motors can handle the higher load!

[@JoeMalovich]

Can you just run a rigid arm to the cable, or a pushrod instead of cable? Kinematics would change for sure.

[@IngoHerges]

What about splicing the dyneema cords to get rid of the plastic parts? There’s even a special splice for tensioning from the hammock community

[@IngoHerges]

It’s called Whoopi sling.

[@luddebosse1439]

this is insane speeds! i think eventually you will run into the spring action being limited, similar to valve float in a car XD

[@Muffinfreak2]

Was sind das denn für Lesezeichen in deinem Browser? 😂

[@Roetz40]

haha :D

[@rsquared9703]

Other commenters mention heat and the magnets. I would invest in motor cooling methods. Heat sinks and fans. It will help motor life if your doing tests back to back.

[@NathanBuildsRobots]

29:13 that thing sounds like a machinegun

[@martin-it4jb]

is this going to be in the 3Dprinting news someday?

[@clantz]

Any specific reason you insist on using steppers?

[@jkr9594]

You've gotten somewhat close to half the speed needed, by your own calculation, so why not simply move the hotend as well. I know it would be harder, and not as fast, but even if it goes only a 10th² as fast, that'd be a basically free bit of speed, right? I am sure there is a reason, so i am asking.

[@johnwiltson3164]

👍i know you will push it faster some how grate work

[@Louis-te7th]

And then there is me sitting in front of my P1S and still being completely blown away how fast something can move. Meanwhile this beast: 😴

[@SergeiPetrov]

But is it possible to use a pulley block and increase the torque even higher? The maximum linear speed is still not achieved with this type of printing.

[@Roetz40]

@@SergeiPetrov what is a pulley block?

[@SergeiPetrov]

@@Roetz40 block and tackle

[@Roetz40]

@@SergeiPetrov ah i didnt knew this word. So i think that wont help. At the end of the day youre want to find the sweetspot between speed an torque. Changing pulley sizes is the exact same thing as using block and tackle (but with less friction in the system) Also maximum speed makes a very small difference in printing times. Acc. is everything

[@802Garage]

NeedItMakeIt Mike would like to collab with you BTW! Maybe pop into his Disk-chord if you have time? :)