MinuteMan Motor Showdown. Is a sub 1 min 3DBenchy possible? (Episode 6)

[@timonix2]

What you are looking for is a FOC controller. Those can absolutely get the most possible torque out of your stepper motors. And you can't really lose steps either

[@guillaumequevy418]

Yes! VESC is the gold standard in the space

[@JamesBaber]

One thing you don't need to worry about: Higher stepper motor driver supply voltages won't cause increased motor heat. All it should do is allow faster current changes within the motor's (inductive) windings and therefore allow them to maintain torque at higher RPMs. Additionally, in my own testing, I have seen a 30% reduction in supply power demand to the driver by changing from a 12V to 48V supply (44W -> 34W) with a constant motor current. If those are the motors I assume (23HE45-4204S), you should really be able to run them closer to their 4.2A rating without causing damage. What surface temperatures are they reaching at 2.5A?

[@peterbonham5540]

Yes he doesn't mention stepper inductance. Am doing nothing like this, but the biggest change I got was low inductance steppers, dumb power supply with a big transformer, bridge rect and a lot of capacitance and the highest voltage stepper drivers you can find.

[@Rubacava_]

@@peterbonham5540 An external high/side low side drivers and bootstrap circuit can be added to an existing stepper driver for higher voltages. For high speed, it is needed because the back emf of the motor dominates the supply.

[@peterbonham5540]

@@Rubacava_ Wouldn't that limit the ability of the stepper driver to control/measure the current (assuming they actually do)

[@chrisparussin5359]

@@peterbonham5540thats a good idea, for testing purposes a good lipo battery is best cause it can deliver tons of amps without any problem

[@Dennzo1337]

In the process of testing various motors for my printer, i discovered that the driver setting and step table configurations in klipper will make a huge difference in performance. Espacially with nema 23 motors on 5160 drivers, it took me a long time and finally the decision of going back to nema 17 for the maximum acceleration. Also i discovered that there is a humongus surge for energy espacially at those accelerations with such huge motors wich causes step loss at the beginning of the asked for move, i would try to compensate for that with either a overpowered psu or a bus bar of capacitors because even those big drivers with their caps that you have there where originally made for nema 17 motors. Im quite sure with the right amount of effort in electrical engineering you could get a lot more out of those motors.

[@imst4722]

For the annex speed boat test, they need to be stepper motors for it to qualify. Not sure if you plan on following their guidelines

[@Roetz40]

Thats why were doing all of this... :)

[@lunaumbra5179]

@@Roetz40 As a new viewer I am surprised you are concerned with a silly organizations rules to follow. I thought getting a sub 1 was the only important thing. Why do you care if the "annex speed boat test" will approve?

[@youtubehandlesux]

​@@lunaumbra5179Least condescending youtube comment

[@youtubehandlesux]

​@@lunaumbra5179least condescending youtube comment

[@youtubehandlesux]

​@@lunaumbra5179 l‮tsae‬ c‮gnidnecsedno‬ y‮ebutuo‬ co‮tnemm‬

[@JulienBaut]

The first Sponsor I really appreciate! 😅

[@tebla2074]

Another great video! I do think it might be worth exploring the idea of using a moving tool head with your moving bed setup. It seems like any acceleration the tool head could do would just be added to what your bed can do. max speed would be the sum too, and as an added bonus all the movement distances would be effectively halved.

[@Roetz40]

I know...I dont want to do that if there is any other way around... I got respect for this kind of setup :P

[@psbpsbp]

​@@Roetz40Instead of moving the tool head,you can move the base with stacked 2 stacked motors

[@4991Ares]

If you look closely at the slow motion shots at 19:54, you can see the pulley deforming plastically within the amplitude of the movements it makes. You have plastic backlash in your system! That would certainly explain a non-linear increase from system power to acceleration: something else is eating away at your excess power... If I am correct, that means some stiffer pulleys might be able to solve some of your current issues.

[@TheHanutaXD]

Could this possibly also be a "glich" from the rolling shutter of the camera, making it look like it is deforming ? (maybe I am looking at the wrong part of the pully)

[@DERTOBL]

Thats more like the whole Mounting Bracket deforming from Torque. It's convenient to print them but some proper Brackets made from Metal would do a lot apparently

[@maerten9517]

I saw that too.

[@TheNamelessOne12357]

If you need more RPM out of steppers, increace voltage, not current. There are 110VDC/80VAC drivers from Leadshine/StepperOnline, like DM860, try them first. Servos are controlled by PID, so they can deviate from set trajectory. It's OK for robots, because there are reducers with really high reduction rates, but not for directly driven 3D printers.

[@citymonkey5929]

Yep. I also say more voltage, not more amps. In my Company we used 24 V steppers with 72 v for short but quick movements.

[@Chad.The.Flornadian]

It's interesting going along on this journey with you. It's also a great lesson for people out there struggling with their own ideas or issues. Solve one problem at a time. That way you don't get discouraged when you hit roadblocks, which you will. And as you saw at the end of this video...don't be afraid, or too proud, to reach out for help. Keep up the great work Jan!

[@gamerpaddy]

you could try modifying those nema17 motors by stacking them. remove the back plate and bearing, trim and connect the shafts together etc. would be interessting if ldo tries to make low intertia rotors. they are usually a solid chunk of laminated steel with magnets inbetween. theres a lot of weight that can be saved.

[@Roetz40]

Haha interesting idea :D

[@TeslaNick2]

I wonder what could be done with BLDC motors and closed loop FOC drivers. I definitely think lightening the stepper rotors will help. There's got to be a sweet spot that optimises for acceleration of mass vs flux. Liquid cooled BLDC motors.

[@mermerico2]

I think if I were in your position I would want to test things with decreased complexity and increased points of measurement. The first thing I would do is print an inertia simulator that clamps onto the shaft of a motor and see I'm able to achieve the required accelerations. Then I would try two motors connected by wire. One thing I wonder- is it possible the timing between the motors are slightly off and they are fighting each other? I think that would get you roughly the same acceleration no matter how powerful the motors are.

[@Roetz40]

I do fear that it is a problem of this sort... Very hard to identify and almost impossible to get rid off...

[@dRepby]

​@@Roetz40 I agree! You definitely should test one motor, looking at it's shaft with high speed camera, to confirm that it can indeed perform at acceleration you need.

[@802Garage]

Something weird happened at 19:29 where you can see slack in the rope. Should never happen ideally I would think. Just thought I would mention that. Changing the pulley diameter has a very small effect on inertia, but a very large effect on movement per rotation, so I think that may be a key. Look forward to the next installment!

[@Person1873]

That probably has more to do with spring rates than steppers. The damper springs get stretched but the line goes slack before the spring has a chance to retract.

[@802Garage]

@@Person1873 Yes I agree. I'm thinking either a rope still slipped or something went weird with the spring. My thoughts on the pulley diameter were meant to be separate haha.

[@vadimm6432]

Came here to say this as well, that shouldnt happen, it means there is some uncontrolled (unaccounted) movement. Do the opposite motors perform "breaking"? For each other? Is this possible to do, to cancel out the inertia? Might help with stopping and changing dorection

[@46449771]

I think an even larger motor pulley will help. You definitely have enough torque for it. Larger pulley will give you more speed of course. I think maybe more acceleration too? As long as you have the torque which you do.

[@thomaskamp9365]

Hallo Jan, Dippel hier aus dem Discord. Ich kann Dir sagen, warum Du Schritte verlierst ... lach. Springe im Video mal nach 19:25 min, stelle das Video auf maximale Zeitlupe und achte nur auf die Seilspannung hinten links am Stepper. Es sieht im ersten Augenblick so aus, als wenn das Seil sich selber aufschwingen würde. Ich habe davor im Video aber mehrfach gehört, dass die Stepper Schritte verlieren, obwohl die Seile ruhig und glatt waren wie ein Kinderpopo. Das ist kein Fehler in der Programmierung und wahrscheinlich auch kein Aufschwingen irgendwelcher Resonanzen. Das ist ein verklemmtes Seil zwischen den doppelten Führungsrollen vor dem Pully. Dur hasst zwar die kreisförmige Aussparung zwischen den Rollen, wo das Seil durchlaufen sollte. Leider sind solche Seile aus Dynema, kleine Arschlöcher und halten sich nicht immer daran! Die ziehen sich dann auch gerne einmal zwischen die engen Stellen und verklemmen sich dann kurzfristig. Das ist nicht lange, denn die Fäden im Dynema gehen ja spiralförmig um die Längsache des Seiles, bedauerlicherweise aber lange genug, um mal eben ein Schritt zu verlieren ohne offensichtlichen Grund. Fehlerbehebung: Mache am besten vor und hinter die doppelten Führungsrollen 4 Zylinder aus Rundmaterial, die das Seil so führen, dass es nicht oben oder unten in den Spalt zwischen den Führungsrollen gelangen kann. Leider sind Seile keine Festkörper und drücken sich nicht in das Loch, sondern nur eine Ansammlung an Fäden, die gerne mal Sonderwege gehen.

[@jessebarker3602]

Sharing my opinion on what could potentially be going on. Just a theory from my little knowledge, i have been learning heaps from this series. Your nema 17s got performance boost when you increased current. Because they lacked power for your demand. These nema 23 steppers arent reacting the same to more power. Pretend they are completely unloaded. They have a max exceleration unloaded. No matter current input if you have already hit that limit it wont increase. I believe bigger diameter pulleys to really load these motors up will put them back into a range where more current = more performance. Right now they are potentially unloaded and just spinning at max potential without full current with heaps torque to spare. I would swap pulleys to ones much bigger like 100mm diameter just to see how these motors react.

[@Roetz40]

Yes. I need to do open load testing with different pulleys...

[@martin-it4jb]

What about using big subwoofer coils for the movement , the range would be really small but the speed and power could be there 🤔

[@CarloVaccariPlus]

You'd need position feedback for such a linear motor.

[@thomaskletzl6493]

speakers just pull or push nothing in between

[@Supamic1]

❤ I love how you removed your singing over the intro reel, well done! 😂🎉

[@Roetz40]

Purely by accident. Will never happen again! ;D

[@heyrpe]

Long Shot idea: How many steps per second does the MCU have to send to the drivers at the speeds you want to achieve? Can your MCU send that many impulses per second (or are pulses that fast even arriving over the signal lines properly)? If not, you'll need a faster board / better firmware.

[@Person1873]

You're also probably getting into radio frequency oscillation rates, so increased shielding and beefier grounding may help. It's also possible that you're approaching the maximum switching rate of the stepper drivers.

[@SetKat-Alex]

I get your idea, but that would just influence the maximum speed of the motors, not the accels.

[@Schradermusic]

@@Person1873 My bet is on the drivers. Just a feeling, though.

[@frederikdubois7945]

i think one of the problem is that in linear move (only one axis) the 2 others stepper needs to turn slightly also to account for the longer rope length required when off-axis, we can already see it at 18:50, the two unused stepper are wiggling, that means that even with higher current, either the rope slips on the pulley which it can't as you screwed it to the pulley or the two active motor needs to pull against another stepper to make it loose step or at least if a step is not lost, it need to pull against the holding current. It will not help with higher current as the holding current of the non-running stepper will also be stronger cancelling the higher power of the active stepper. The springs position (after the motor) is related to this issue being placed after the stepper loop...

[@CarloVaccariPlus]

Stepper drivers are a *very* deep rabbit hole. Many of the standard ones are not actually suitable for extreme currents or for extreme speeds, and you may need to tune your chopper parameters.

[@kaiwheeler64]

X Y insight might be found in plotter mode, fitting a Sharpie to your hot end and getting it to draw intricate geometry, layer one.

[@MarcStollmeyer]

If you have a lot of torque but are maxing out acceleration of the rotors, put a larger pulley on it.

[@SwissLEN]

I would suggest to use steppers with less steps per rotation or larger pulleys. You can't change the magnetic field in the motor infinitely quick and stepper motors need many field changes per revolution (1 per step). Servos need way less field changes per revolution (2-6) and can spin therefore much faster than steppers.

[@kylewellman402]

I would love to see a test using Linear Induction Motors. Same concept the peopoly magneto x 3d printer uses. It would also make a great video on how to make them:D I am having trouble finding any documentation about the actual motor drivers. I know the magneto X uses regular stepper drivers, but if memory serves there is a power supplier between the stepper driver and linear motor. The stepper is only used for the step and direction signals, but a different power board actually supplies the current to the motors. The concept is awesome because standard boards used for 3d printers will work so as far as modeling, slicing, and printing there isn't anything different (aside from some step values and some other parameters)

[@JoeMalovich]

You have the space to mount them face to face, so two motors per pulley connected to form a common shaft.

[@SnobbishGoose]

Besides lightening the motor rotor would counter rotating rotors work too ?

[@OZtwo]

I can not help here other then to say: DO NOT give up! This project isn't about how to get a 1 min Benchy on the current system but to find the system that will allow you to print at that speed.

[@gamerpaddy]

you might look into 5 phase steppers, hard to come by, expensive drivers but apparently more agile than bi/unipolar steppers.

[@t3chn0w1z4rd]

usually when you scale up the motor, you scale down the pulley, otherwise you're just trading torque for speed. The bigger motors can handle the higher torque, but your motion system cannot, so if you scale down your pulley, you'll have more accuracy.

[@WilliDurpel]

try closed loop systems, i think they will get u closer to your goal

[@Mattvbro]

What about making a bed that rotates then have the print head just move back and forth while the bed spins? this would alleviate the issues of the whole lot having to suddenly change direction

[@zeveris913]

I think you need some mechanical advantage to the motors. Turning the rotary motion into linear motion with a fulcrum and pivot point. Now increasing the lever size will magnify the movement the stepper is doing. You need to find a way to make the steppers get more movement in a linear motion at lower speeds, at that point more amps and different lever sizes should equal more speed. Like one step gets twice the distance now or three times the distance

[@SetKat-Alex]

No need for fulcrum or pivot point. Just increase the pulley diameter.

[@zeveris913]

@@SetKat-Alex in his test i saw the string coming off the pulley, a bigger pulley is perfect but i feel like we need a rigid linkage. Something about the the string and spring tied to the other motors doesn’t seem right and you get losses with bends it’s taking

[@zeveris913]

@@SetKat-Alex a slow motion camera would help so much

[@SetKat-Alex]

@@zeveris913 okay I understand what you're saying! If think you're onto something, by using lightweight linkages (CF maybe), like what is used on delta printers or picker robots, he can achieve much better rigidity.

[@a-a-ron5911]

Is the problem the acceleration of the motors or the inertia of the system? Like could you use much larger pulleys with those theoretically stronger motors? And achieve much faster real acceleration while turning and accelerating the motors slower? Or is the issue the inertia of the system causing cable stretch and snapback at higher speeds stalling the motors?

[@brydenquirk1176]

Allso abandoning micro stepping can rule out processing related bottlenecks

[@alexscarbro796]

Could linear motors be used to drive the pulley cables?

[@cylosgarage]

Spectacular engineering, amazing project. Hope you can find some quicker motors! This is still crazy to watch as is

[@simontillson482]

To achieve high torque and high acceleration at the same time, please just drop the stepper motors, that’s a real weakness for them. Only a brushless DC motor can do that. The problem with steppers is that to increase torque, you need more metal in the rotor, so high torque motors have really big, heavy rotors. A BLDC on the other hand, has a smaller diameter rotor and much larger saturation flux, which increases the torque massively while keeping rotor mass (and therefore inertia) quite low - massively lower than an equivalent torque stepper. BLDCs are also far more efficient, so drive power and overheating will be much less of an issue.

[@jdogsherman4790]

Would the higher torque of the bigger motors allow for a higher spring ratio on the pulleys? I wonder if the springs are introducing an interference at higher accelerations due to them not returning to their initial length “fast” enough.

[@mariusj8542]

Interesting topic. So if you have calculated that you need acceleration of 500.000mm/s^2 to get a one minute benchy and since acceleration =net torque/ total inertia none of the nema17 or nema23 motors are capable of reaching 500.000mm/s^2 due to the angular acceleration, which if you think about it’s almost 50g. You are in the range of railguns or coilguns the military is experimenting on. Perhaps some extremely tuned servo motor could do it, but I highly doubt it. Suggested solution: linear motors - what about taking a Voron 0.2 with y axis in neodymium magnets on each sides with an iron backplate to direct the magnetic field. Just epoxy in the magnet in a 3d printed form, or perhaps do a small milled form in carbonfiber or aluminium. The problem is the xaxis gantry, here you need to really optimise weight, spacing of the magnets and the windings on the “coil forcer” on the printing head to balance inertia/mass. Linear motors are capable of reaching 100g, so it could perhaps,maybe, no guarantee work 😁

[@ellabun]

Please put the 23s away :) Enjoying following this 👍

[@maerten9517]

I'm not an engineer or even close, but is there any way to ditch the strings and use rods instead? Sort of like stupidly fast servo arms?

[@ericcmcgraw]

I also came to the same theoretical result which showed that maximum acceleration can be achieved when the radius is chosen to match rotor inertia to load inertia. Is there any reason you chose to only 2x the radius when the rotor inertia is 4x? Is it just to preserve step resolution?

[@Roetz40]

Resolution is not a problem. I can still hold 64/128 which is plenty

[@ericcmcgraw]

@@Roetz40 will you do further testing with a larger diameter pulley to match the inertia? I guess for that you’ll need to design mounts to lift the motors up off the surface.

[@LachlanMcDermott-u7j]

You should look into the moving stages used in EUV machines.

[@WoLpH]

If you really want to optimize your motors you probably need to find some high performance accelerometers, something like a A0220HF would probably give you much better results already but I'm not sure what kind of G forced it needs to be able to handle, that unit maxes out at 50g but it can read it at 15kHz. A 621C40 is a much better option, up to 30kHz and 500g, but also quite a bit pricier.

[@dungadumbs428]

had an interesting thought.... appears there is adequate power to run larger pulleys but then step resolution is an issue. So... how about ganging 2 motors connected to a pulley - 2 motors in to 1 output. One motor for gross/fast motion with a huge pulley and a second(smaller) motor for fine resolution(trim)? Run the fine motor at a fast but reliable speed and the gross motor at near failure speed? No idea if klipper could deal with this scenario but seems plausible. love this project because it's nuts!!!

[@mrvolucris1863]

If the motors have enough torque regardless, why not increase pulley-size even further to require lower rotational speeds? With an even larger pulley-diameter you could even up the amperage further to provide the needed torque?

[@user-vy5hc9ud6l]

Just smack a flywheel / a block of metal on a singel motor that has eqivilant innertia to your system. Test it with that to minimise variables to better compare motors.

[@somebody566]

Perhaps the induction of the motor windings is limiting the current flow? In other words is the current setpoint actually reached? Would be interesting to measure this.

[@snapo1750]

wouldnt it be better (from a acceleration point) to have per axis 4 nema 17 in parallel so 16 in total motors? the rotor of the nema 17 is much smaller therefore lighter to accelerate. Just my 2 sats.... 🙂

[@bdykes7316]

Servo motor Yaskawa SGM7A-04A*A with a 30mm pulley, a 90g load (1:1 Inertia ratio), 25mm trapezoidal move profile with a max speed of 2m/s should hit 1400m/s^2 of acceleration and 2000000m/s^3 jerk according to the Yaskawa sizing software. The next step is longer stroke voice coil actuators. Potentially suitable example: Brand "The Modal Shop" Model: "K2075E075 Kit", which lists 38G peak acceleration with a 450g load.

[@sleepib]

I think you need to find the peak power point of the motors, and tune pulley size around that. slippage should be solvable by adding another loop around the pulley.

[@coledavidson5630]

Someone in the comments who knows about cameras, get this man a good slowmo cam lol

[@BenjaminEverest]

yep, try the shorter NEMA 23s (the increase will still be marginal, but they will perform better than the 17s). The extra rotor weight cancels out the torque increase on acceleration. Also because the bed is so light, getting a larger and larger stepper motor wont help. You would have to make the bed significantly heavier (through mechanical advantage) to see any real improvements with the larger stepper motor. (Basically to the point of stalling the 17s, it's very unintuitive and you will lose a lot of resolution)

[@DmytroPochapskyi]

On those kind of speeds for steppers I would look into another possible sources of problems - if you say, that on nema17 and nema23 you can achieve almost the same accelerations and other people in 3D printing community have same max accelerations it looks obvious to me that this is not a coincidence. stm32, which is sending step signals to drivers, is fast, but it may be not enough for this application. try to make small test with raspberry on its own sending a step signals to drivers and make a small python code where you will be able to change accelerations and speeds by yourself. maybe this will show you some usefull data.

[@beautifulsmall]

I do not have the hardware to do what your asking. But, Ive been trying the ICM-20649 30g accelerometer ~ 1500Ksps for a project, its aduino compatible and as good as ive seen. For low rotor inertia wth high torque a coreless motor EG Maxon, may be a good fit but its DC so needs an encoder and different driver. It will be extreemley instructive to see the stepper limits. Fascinating project. Beyond coreless it might be linear motors or using clutches with two continuously spinning motors but the presision would be a nightmare.

[@Roetz40]

Interested to joining the discord? :) would like to know more about the accelerometer!

[@henninghoefer]

To me it looks like the rope and pulley setup might be the limiting factor … How about going for solid rods pushing the bed around? Or leadscrews with a really tall pitch?

[@sammiller5509]

Würde es helfen aus zwei nema 17 Motoren einen längeren zu bauen? Bzw die Achsen von zwei Motoren verbinden um höheres Drehmoment mit niedriger rotierender Masse zu erzeugen. Ich denke der Radius von dem Rotor ist auch entscheidend(kleiner= weniger träge)

[@algos-ryansheehy]

Why not use a retro-reflective small ball on the center platform and a fixed camera on the nozzle to pick it up as your feedback mechanism, instead of encoders, which can lose steps? The camera might not be fast enough to pick up the movement accurately.

[@Bl4ckSupra]

You should rotate both motors on the right side for 180° to avoid the accidental swastika design :) They will not interfere with the pulleys and wires.

[@Saminvent05]

That German engineering...

[@evren.builds]

Btw does it have to be steppers? Was that a restriction of the benchy challenge Would small servo motors be applicable? Also I'm curious if closed loop steppers would result in better performance at the higher speeds as well.

[@jwrm22]

Afaik steppers were a requirement for the speed boat challenge, but more to stay true to the normal 3D printing constraints.

[@evren.builds]

@@jwrm22 Yep, just look through the rules again. It says: -The machine must be powered by stepper motors Still doesn't cancel the closed loop steppers

[@Roetz40]

Unfortunately it does according to the officials :P

[@anderslagerqvist2642]

Once you got the motors at insane speed... Will the material stick to the plate or will g-forces pull it off?

[@Roetz40]

It will stick :))

[@Ravedave5]

Any reason you have 4 in the mix to start? Why bot do a linear 2 motor test for limit testing? No build plate etc go to low number of variables.

[@anonym9323]

Well I do not see in general any area in the friction equation it is Umschlingungswinkel and friction coefficient for F1 to F2 or am I complete wrong

[@mathiaschaves7604]

Did you saw the variable diameter nozzle from sculpman? I saw an earlyer video were you used a lot of hotends. But if you make a setup inspired by him you could print way faster because you don't move the head.

[@florianschneider2256]

I am also in the process of building an ultra high-speed printer. In my design I came across the same problem of maximum acceleration. Somewhere around 300m/s² are the limits of stepper motors due to the comparatively high rotor inertia. My approach is to overcome this limitation with brushless motors in a servo loop. These motors have about half the rotor inertia at the same torque. In my project I can use them without any problems because my goal is not the fastest speed benchi but the fastest practical printing speed (extreme printing speed in a format that is practical for everyday use with a print quality comparable to a Prußa or Bambula printer) via excessive use of feedback loops. I currently design my printer for an acceleration of 500 m/s² and thus reach the practical limits of normal hobby BLDC motors. Axial flux motors would be more ideal. With these motors, the torque in the cube increases with the diameter while the moment of inertia only increases in the square -> linearly higher dynamics with higher diameter. However, I have not found any such motors for my size and price range.

[@Roetz40]

Thats very interesting! Would you like to join my discord and show some of your findings? Link is in the video desc. If I really found the limits of these Steppers, I would like to go a different route. :)

[@nathaniellangston5130]

You should definitely try out low rotor inertia drone style motors with Odrive over either their shaft mounted encoders or field oriented control. Also I worked at an industrial 3d printer manufacturing company and we used heavy duty Dyneema fishing line for our CoreXY machines we built for very low weight while essentially no elasticity. Using Dyneema on a pulley and having it wound 4 or 5 times around creates enough friction to avoid slipping without the need to screw the cable down. Personally while I am certainly less intelligent than you and I know you've probably thought of most the stuff I've said. I feel like lowering the elasticity/slippage between the motors to bed and 3D printing a vacuum preloaded air bearing bed while more complicated may help with high accelerations. I'm not sure if that is already built in and I just missed it. Kinematically I'm mildly concerned about the angle of attachment to the bed being so close together rather than say 90 degrees apart around the bed but I think I potentially my largest concern with the current angle of attachment for myself if I was doing it would be that I don't understand software well enough to write software to compensate. Great project!! I love seeing these videos come out! Personally as someone who has 4-5 Odrive S1's and 2-3 Odrive 3.6's just laying around and many more currently in use I feel like they are off the shelf enough of items that I don't see them as that big of a difference "off the shelf" wise from "off the shelf" steppers / servos. I also think you'd have a much easier time hitting your goal using the Odrive as you can essentially build whatever the perfect servo no matter your application is with regards rotational inertia, acceleration and speed. Given that you have a huge variety of hobby/RC motors to pick from rather than prebuilt servos that are largely all the same.

[@ShelleyD1]

The LDO Nema17 2504's take 56v from my PSU and I can set them at 2450mah, but over 150k it just loses the accuracy, i get best results and lowest noise much lower. I assume you've done the math on if those springs are even needed at these speeds. Seems like the rebound time is too short for the wire tension. Having the wire sync across motors is nice, can you link them through the center for a shorter 'belt' maybe to reverse backlash a little better?

[@vadimm6432]

Would it be possible to add more winding on your rope pulley? And make the channels a bit narrower to hold the rope tighter. I think you are on the right track with larger pulleys, as long as that doesnt reduce the accuracy of step tracking, and providing motors have enough power (and power supply) to drive them. Like the other comment said, there was visible slack in the ropw during high speed test, look into that please, that shouldnt really happen on the inside loop (rope inside of the motors)

[@gingerflow5594]

what u need is a dude that can tune your step sticks properly , so much to achive in terms of power . The stepper can do is the step stick cant

[@josephgauthier5018]

maybe extra long nema 17 motors? that way you can get more torque, but keep that inertia down, especially since its a squared function of the radius

[@Roetz40]

Well you were on a binge watching marathon:D I think nema 17's in generally don't develop the necessary torque. Nema 23 is the way to go ...

[@josephgauthier5018]

​@@Roetz40 lol, i definitely was. watched it all at 1.75x speed, like i do with most videos i watch... so much content so little time XD well, i hope you can find a way to lighten the cores of those motors, like you discussed in the video. out of curiosity, have you ever considered playing around with material science related stuff for you videos? subscribed, and looking forward to more content!!

[@jangrewe]

Switch from steppers to FOC? VESC will be your friend! ;-)

[@SnobbishGoose]

I feel like this project is very ingenious and it reminds me of @StuffMadeHere and their cable moving basket ball net If Shane were to be tackling this project he would probably have slapped on some tracking balls on the build plate and made it closed loop with a motion tracking camera

[@kingnikman4108]

LDO steppers are so good because they have very very very low inductance. This massively affects speed vs torque. Those nema 23 could have a lot of inductance and bad speed torque response. From looking at steppers in general, 2x longer stepper has 2x more inductance roughly.

[@chrismayer8990]

I think you are extremely limited with the cable system and the springs. Possibly a toothed belt system and gas spring could solve a few problems. A provisional toothed belt system with suspended counterweights would be quick to implement and you could quickly test the effects this has on the system by changing the counterweights.

[@brydenquirk1176]

Test the motors uncoupled from the mechanism they will never accelerate faster than that. Personally I think you need a lot of very small motors but that's easy to test by just getting one and testing the acceleration uncoupled.

[@Fabian3331234333]

Dang. What a job. Kind of a little bit eny. Let me know if you need a german machine learning and computer vision expert haha

[@joeldriver381]

Servo time!

[@ViniciusMiguel1988]

Steppers are not really the choice for those kinds of dynamics, I hope you are running those motors are 48V, they perform dramatically better at higher voltages.

[@OwlingBishop]

I don't know why/how/who convinced you to use the two anchor points layout on the moving plate / air baring but it feels / looks very wrong to me from a dynamic point of view and we can see the plate flying off the "ground" in this video because of the two points layout .. I suspect it would really be better to get back to the four anchor points.

[@electricalychalanged4911]

Not an expert here but you are forcing huge and super fast power fluctuation on your power supply. This might lead to instances were one motor for a ms just does not have the power to move as alle the caps are draind and you loose steps while the ps is regulating. Would be interesting to see the voltage profile of the supply during a run. Just adding a big as capacitor to the Inlet of each motor driver might be a quick and dirty test to see if that is the limit.

[@nosenseofhumor1]

i also suspect a power sag is in play

[@youtubehandlesux]

This is actually what limits my voron 0's max acceleration. Why wouldn't they sell small form factor large power supplies :(

[@wingunder]

Won't you be better served with some fast electromagnetic linear actuators? By using motors in your setup, you have to translate a rotary movement to a linear movement. By using linear electromagnetic actuators, you'll cut out this conversion, making the whole thing less complex.

[@Roetz40]

For the time being, I try to stick to the speedbenchy rules which only allow open loop steppers. But maybe in a future iteration ;)

[@MPenzlin]

Ich würde das rückwärts rechnen: Welche lineare Beschleunigung brauche ich? Dann würde ich ausprobieren, welche Rotationbeschleunigung der Motor ohne irgendetwas schafft? Keine pullies, keine Kabel, nichts. Wenn der Motor das dann schafft, wann würde ich ausrechnen, welchen Durchmesser der pully haben muss, damit ich aus der rotationbeschleunigung die lineare Beschleunigung erzeugt. Dann ausprobieren, ob es der Motor+ diesem pully immer noch die rotationsbeschleunigung schafft. u.s.w. Außerdem würde nich nicht die Dehnbarkeit der Kabel unterschätzen. Bei 400g wird aus 100g halt 40kg. Da dehnt sich ein Kabel sicherlich schon wo sehr, dass es weit mehr als ein Schritt der Motoren ist. Diese Dehnbarkeit ergibt also auch noch mal die maximale Beschleunigung die man erreichen kann. Da könnte man also ausrechnen, wie dick die Kabel mindestens sein müssen, damit man die Beschleunigung schafft. u.s.w.

[@Iamtherodlight1453.]

So awesome good sir.

[@robinc.5077]

Based on absolutely nothing, my guess is increase the voltage, a lot.

[@Roetz40]

Already at max for non industrial solutions

[@philipweser8774]

Do you still need the rope behind the steppers? If I see this correctly the screws hold the rope now, so everything behind that should be unneeded and therefor dead weight.

[@jaykop2611]

Sei vorsichtig mit der Motorlänge jan! Ein sehr DEUTSCHES Design😂

[@SimplyElectronicsOfficial]

I think gearing might be a solution.

[@Person1873]

Backlash

[@madmatrac]

shorter motors, bigger pulleys

[@Fabian3331234333]

Have you took a look at the Kraken Steppers?

[@lambda7652]

More motors instead of bigger motors

[@monela2696]

I know how to do this but im too lazy to explain, sorry. Maybe if you try for 4-5 more months then i will find some time to help you. But atleast now you know its possible

[@Glocktopus42]

CAN I come work for you? I'm a mechanical engineer:)

[@elmstrong]

Try 67 amps...😊

[@jasontang6725]

Move the bed and print head - double speed and acceleration.

[@chronokoks]

Should have used shorter motors and yes you need amps dude.

[@ToniThePerkele]

Voice coils

[@ToniThePerkele]

In closed loop

[@PeeDee3D]

Have you looked into Devil Design-LDO Krakens?

[@zewarmdur9195]

Hey, I’ve been trying to watch your videos, but I found them a bit too long-around 20 minutes is ideal for me. Also, there’s a lot of face time and chatting, which makes it feel more like a vlog. No offense, but I’m more interested in seeing the end result, the process, and how you solve problems rather than just watching you talk. It would be much better if you used more visuals or clips to show what you’re explaining, rather than just showing your face. Also, the video titles could be more descriptive. Just some constructive feedback. You know better, but I hope this helps!

[@24Tigger24]

I would use a better computer than the Bigtreetech Pi, e.g. a Raspberry or an old mini PC with Intel or AMD. Instead of Nema23 motors, I would first try the Kraken stepper motors from Omranello, which were specially designed for fast 3D printers

[@Roetz40]

I already send him an email... ;)