Magneto X linear motors: Deep dive and performance testing

[@VolkanTaninmis]

A new category unlocked for aliexpress seller: linear motors for 3d printers.

[@BeefIngot]

Would buy so fast. So obviously the easiest way to make a toolchanger.

[@lassikinnunen]

​@@BeefIngot you would drive it off the rails end to park or something?

[@BeefIngot]

@@lassikinnunen No, just have the rail extend to off of the bed and have the unused ones simply not move since the active part is on the thing that is moved and the magnets are stationary.

[@geauxracerx]

@@lassikinnunenso instead of changing the tool head on a single carriage, every tool head would have its own carriage? Idk seems like loading and unloading the entire carriage would be the weak point in repeatability/reliability. Where as once dialed in a single carriage and multiple tool heads could achieve the goal more easily. It’s already a proven technique in CNC machinery

[@heavyhemi2828]

@@BeefIngotoh crap! like all of them on a single rail, and the rail is just extra long on both ends, so the toolheads can just cycle between sides? Like as if each toolhead were the balls on an abacus? You would just Park the unused ones off to the sides

[@markbooth3066]

While the X/Y arrangement may seem counter intuitive, this is standard in robotics. The X motor is almost always further down the kinematic chain than the Y motor. When driving a linear motor with a stepper interface you will never see 'missed steps', you will either see a successful move, or the axis will trip. The controller will have a maximum allowable following error, the difference between expected and actual position for any servo cycle, and if the measured following error exceeds that value, the controller will emergency stop that axis, and by extension, the whole machine. If you have access to the following error, you will see it creep up as weight, speed and acceleration increase. As an example, if a move has a max following error of 10um, a successful move will never have been more than 10um away from the expected position at any point during the move. If the planned trajectory requires an axis to be at 1000um at time T, but it only gets to 989um because it isn't accelerating hard enough, the axis will trip. Similarly, if the planned trajectory requires an axis to be at 1000um at time T, but it's already at 1011um because wasn't able to decelerate hard enough, again the axis will trip. What won't happen is you get to the end of the test and the axis is in a different place to what you expect, it will either be within the following error limit of the desired position, or the axis will have tripped. This is one of the big advantages of a closed loop motor, you will never end up in the wrong position without knowing it, and errors will never accumulate over time. The fact that the fast move tripped at the end of the move, rather than the start, suggests to me that that the drive electronics can't cope with the back EMF generated by the motor for that combination of feed rate (speed), deceleration and mass. It can get up to that speed, with the given acceleration rate, but the amplifier can't dissipate enough energy to counter the inertia of the mass. I would be interested to know which axis tripped. I would expect it to be the Y axis, since it has to cope with the weight of the X axis as well as the weight of the print head. On one robot that I worked on, we ended up having to add dump resistors to the Y axis amplifiers (drivers) to improve the performance of those motors under deceleration, even once we'd tuned the Y axis motors more conservatively than the X axes. It was the difference between 2m/s & 2g for the Y-axis beams, verses 5m/s and 5g for the X-axis shuttles.

[@lack527]

I'd like to know more about the Bambu tensioning and centering method you mentioned in the VFA section.

[@PaulDominguez]

Me too. Can we get a link?

[@MrDehicka]

It was described on the babmu forum in the VFA thread. For me it had zero effect on the VFA.

[@strictchannel662]

Can anyone please share the link to this belt tensioning?

[@elektricmdolez2900]

pls share a link would be really helpful

[@TeachingTech]

Link added to the description under the VFAs chapter. I tried but couldn't replicate the results. It needs persistence and a very fine touch.

[@myself248]

Haha, thank you! That's exactly the madly-whipping-around footage I was hoping to see, and a thoroughly impressive showing, to boot. This is one of the most exciting developments in a long time.

[@capitalinventor4823]

Thanks for the follow-up video. It’s wonderful when you contact the developers to find the answers people have. You have access that ordinary people don’t have because of your channel. Thank you for all you have done for the community. When specifying a speed or acceleration greater than the machine is capable of handling it would be nice for an error code and/or message to be returned instead of letting the machine attempt something it’s not capable of doing and failing. I don’t know if this is an issue with the manufacturer implemented their linear motors, something in Klipper, or how the manufacturer configured Klipper.

[@arthurmoore9488]

While I don't have one and am not familiar with this particular setup, I can relatively confidently say it's not that easy. The most interesting thing about the approach taken is instead of modifying Klipper they developed their own controller board. As far as Klipper is concerned, these motors are being run by a cheap controller that only supports the bare minimum. I mean, instead of telling the actual controller boards "go to X position", Klipper calculates the difference between the current position and X in steps and sends that many pulses. Keep in mind that separate stepper motor controller board to talk to the drivers doesn't have to exist. It's just duplicating the functionality of the main controller board, but with a worse processor. The only thing the main board is doing is handling the Z axis, and heating the bed. The only reasons I can think of to make those decisions is if they either re-purposed CNC hardware and couldn't integrate it fully, or are planning to release a CNC.

[@anthonyalbanese1993]

​@@arthurmoore9488 Yeah these are good points - like, does klipper even support closed loop steppers / servos for the main movement axis in the first place? And if it does, does it "fully" support all the functions you can typically expect from controller that do? This machine is closer to a machining CNC with a printhead on it than a build from the ground up 3D printer with standard 3D printer parts; and I think that's a good thing. I'm actually curious why 1) people have such a hardon for Klipper support, and if Klipper was even the right thing for this machine. I understand klipper being open source, and the functionality it adds to something like an Ender 3 or the hundreds of bedslinger clones is great, but a machine with essentially completely different kinematics from anything Klipper was designed for just seems so odd to me. I suppose it saves them a ton of development time / overhead in probably having to make their own (or, say, converting Mach 3 - with licensing - to something usable by an average 3D printer user with 3D printer modules) - but I'm curious if this will end up inherently limiting the capability of the machine down the road. Or maybe it won't - Klipper is open source and someone could write their own proper linear motor modules to it that actually work with feedback. But that's a bet, not a guarantee.

[@calebharris292]

It'd be really cool to see the linear motors as a conversation kit for other 300x300 printers, especially if they show far better durability and stability over time.

[@MarinusMakesStuff]

Thank you for always being such a valuable contributor to the scene Michael. I've been following you for a very long time and you've always been so fair and complete about the subjects you discuss. I love your work! Beyond that, it would be really exciting to see kits coming out with which people can build their own linear motor based printers. I would totally use such a linear motor in one of my projects. This project needs a linear motor that's longer than 1 meter, but it looks like this technology is perfectly scalable so if they ever bring out such a kit, I would love to hear about it.

[@Bitfrogess]

Great video! I operate an Anca FX7 for a living and the thing is whisper quiet and easily accurate to a few tenths of thou. I can't wait for more refined versions of this 3D printer!

[@thethubbedone]

Anca machines are accurate to more like tenths of a tenth of a thou. Those are some of the best machine tools available at any price

[@woodwaker1]

I cancelled my pre-order in late November, had a tough choice to make finish paying for my Prusa XL or wait on the Peoply. I went with the Prusa XL and they Peoply refunded my $1499. Now waiting to see how the final production version turns out. My guess is May before they start shipping.

[@TheCreat]

Fantastic deep dive, thank you for the huge amount of work it must have been putting this together! I have a comment about "maintaining perfect positional accuracy" when you're talking about 30k+ acceleration at 1k+ mm/s² (15:30 or so). The closed loop system will always ensure this, as would a servo motor based system (belted/ballscrew), assuming tight enough tolerances and sufficiently beefy construction in the drive system. But the systems just get back to the perfect positional accuracy after possibly losing it while under heavy load, just not losing it enough to fully error out. So it doesn't "maintain" it, it just gets back to it. So the real measure would be how far and for how long the system was ever off it's target position. For high performance printing that's what matters, not that the head ends up in the correct x/y spot after printing for an hour, but that the filament is deposited exactly where and when it was supposed to. If the head lacks behind a bit, or overshoots the target position a bit after correcting for a previous error, that'll show up. It would requiring coupling the positional feedback of the X/Y with the extrusion system somehow, but I'd guess just the latency because of physics such a thing would incur would probably make it either impossible or at the very least impractical. Obviously given the sheer power of the motion system, the required precision in the control of the extrusion system needed to keep up with numbers like this is likely going to be the limiting factor. Also even measuring or quantifying what the actual error is would probably be a masters or doctoral thesis in itself. What even would be the unit, some sort of integral of the positional delta (target vs. actual) over time?

[@paulthetexan]

Nice to see a fellow fan of Jimmers!

[@3dpathfinder]

Im a fairly newer 3d printer and content creator and , I have to say your videos are beyond good. It would be a dream to have a Magneto X one day, but will be a long way off to afford the 2000 .

[@chicoxiba]

the future of XY movement on these machines.. for sure. Too many advantages.

[@tadhgd350]

I’d love to see the input shaper graphs not just the text. And an actual attempt at tuning IS

[@beermanuk]

Every time you post a video I notice you're usually wearing a t shirt of motorsport legend. No different this week with the Jimmy B t shirt :)

[@TeachingTech]

All of my shirts are either plain or motorsport/sports related :)

[@rob_over_9000]

This is such a strong first printer out of the gate. With a tool change or multi filament system, it’s hard to see any reason this wouldn’t become the gold standard going forward. I will say, the quoted specs for thermal safety on the electronics and motor should be revisited to ensure they can operate at higher temperatures since that doesn’t leave a whole lot of headroom but… very nice showing from these guys.

[@millpreetk1406]

Full Review when? I know it's an actively progressing printer, not the smoothest out of box experience yet, but I feel it has great potential based on my initial first few prints!

[@Thorhian]

I feel like a machine like this should be driven by LinuxCNC, something built for driving closed loop real time machine systems. Wouldn’t mind if Klipper was modified to have proper closed loop control and *real* real time control (stop means stop now, not when the command buffer runs out).

[@arthurmoore9488]

I mentioned it somewhere else, but I think that's what it's designed to be run with. So many of the decisions, like a 2nd micro-controller for commands and emulating stepper drivers, points to this originally being developed for LinuxCNC or similar with Klipper slapped on.

[@myself248]

@@arthurmoore9488Which is why I'm so intrigued by the idea of being able to buy the motors separately, I'm really glad he asked about that! Or perhaps it might be fun to just buy the whole printer and forklift-upgrade the controls.

[@Muz994]

That's what I think as well. I switched my printer from Linuxcnc to klipper to gain all the features of a printer specific controller, but it always feels like a "toy" in comparison. No support for external e-stop, no way to trigger e-stop if a limit switch is triggered during printing, no way to instantly pause a print etc... On larger and more powerful machines it can get dangerous without those things.

[@billverine765]

VFA has been driving me nuts on my Qidi XMax3. I think this is the future of 3D printers and so awesome to see VFA goes away at high speeds. You buy a CoreXY expecting to be able to use it at advertised speeds, but realistically you can only use about 1/3 Max speed for decent quality. This was a great review as always and thanks for your thoroughness.

[@redkingrauri3769]

I hope linear motors become an option in general for other printer kits as well. Peopoly said they designed theirs from the ground up to be cheaper and more available for the consumer, so I could see them as options later for kits like ratrigs. Be great if the v-core 4 or something got that.

[@mscir]

Brilliant analysis. That is the future of 3d printing. The prie will come down.

[@NickBR57]

Good instructive video, thanks Michael.

[@isaacnguyen6944]

First time I encountered linear motors was in the early 80s. I was to energize a big drawer to deliver medecine. A big flop at the time!

[@mdfyui8000]

With closed loop motors, the move and measure test intended for steppers isn't sufficient to determine if the system is useable, the closed loop system will always hide any problems. You don't know dynamically if the system is experiencing substantial overshoots if you only measure the static starting and return points - the only feedback about the dynamic condition is really the fault stop of the drive being triggered (so exceeding 2mm of slip or some other factor). But since the implementation of the linear motors seems to be opaque at the Klipper firmware level, information about this probably can't be collected there.

[@carlettoburacco9235]

I have a HEAVYLY modified Ender 5 Pro and an Ender 5 Plus. (linear rail, bigger hot bed 310x310 / 435x425 , beefer steppermotors, custom head, Volcano, Laser and Spindle fast attachment.....) I reached the speed limits of the Y especially on the Plus. (the X axis weighs a ton) These linear motors could be an alternative and better solution than putting a double stepper on the Y.

[@_..-.._..-.._]

You run spindles on an ender? 😂 how well does it do?

[@carlettoburacco9235]

@@_..-.._..-.._ Ender 5 Pro and Plus with linear rail: milled alu with 0.16 DOC, 3mm end mill, 16k rpm......slow, but better than nothing.

[@DeagleBingo]

@@carlettoburacco9235 oh see this is interesting bc i was going to comment that with the linear motors perhaps a spindle option could be had with a rail upgrade ? i don't know the first thing but i get the idea its more possible in this machine than others. sometime someones going to realize that being able to convert a machine easily from additive to subtractive for whatever someone needs to fabricate at the moment is the way to go.

[@The3DPrintingGrandad]

Great video. Sure linear drives will be to next big deal in 3D printing, until something better comes along. I see linear drives being a pretty complicated upgrade for the moment. I remember when everyone one said revolvers and potentiometers were good enough, who needs a fancy encoder thing-a-ma-bob!

[@XoNMan1]

It's interesting to see this thing driven by off the shelf electronics like an Octopus and USB to CAN board.

[@_..-.._..-.._]

I would assume the “direct drive” nature with the large magnets and coils would make these far less accurate than a geared down stepper motor with smaller steps.

[@kniefi]

Looks like an awesome machine! Very thourough test and review you did here! From what I have seen in this video - although the aluminum extrusions used on the printer have beefy wall thicknesses, for such a powerful machine, it still seems weak and flimsy, seeing how the whole machine shakes and wobbles around. I would LOVE to see this very kinematics being indtalled onto a machined/milled steel-frame chassis with a SOLID half inch netal basplate and start from there. with some square steel tubes - very overengineered but with probably 8x the stability of it's current state. It would not nearly shake as much and have greatly reduced deflection while moving very fast.... Great concept - This seems like a printer worth owning when you want dimensional accuracy and trouble-free printing of high-quality functional parts - especially considering it comes with a hardened steel hotend -> CF-PLA - CF ABS is now on the table!

[@BeefIngot]

I just want some hobbyist accessible closed loop absolute linear motors. So mamy possibilities and benefits. No belt stretching, always precise, no tensioning or belt care necessary, more space to do things like tool changers and if cost effective enough you can have one x rail per tool head.

[@Fjprints]

still very expensive unfortunately.

[@BeefIngot]

@@Fjprints even at 250 per rail id buy 6 to make a tool changer

[@mytuberforyou]

Personally, I think these are designed "backwards". It would be better to have the coils in the rail, and the permanent magnets in the carriage. This would reduce weight (mass) of the carriage, eliminate the motor wiring to the carriage, and allow the carriage to be hot-swapped. This would translate to simplification, speed increase, and convenience. In the case of something like a CO2 laser that uses mirrors and requires no power to the head, it eliminateds the need for a drag chain or ribbon cable.

[@DeagleBingo]

but does that work well enough from a physics perspective? i don't know. i see your point about the weight but i'm assuming they did what they did for a reason. to me... the issue is heat in an enclosed build chamber, i want a machine that will do functional stuff that has to be printed hot, as others have commented, and maybe 60c is good enough, i don't know till i have something and the opportunity to use more advanced materials more. i don't need that at 800mm/sec.. i probably would be super happy at 250mm/sec or less as long as it was reliable and cheap and repeatable and simple to set up.

[@BlaineRush-b8h]

I believe there are magnets that can withstand 120c, so maybe different (more expensive) magnets would be an option.

[@mytuberforyou]

@@DeagleBingo It would actually run cooler because the windings in thew rail would have the entire rail to dissipate heat and only be active where the carriage was, and the carriage would not have current flowing through it getting heated up. It would be a little more expensive to manufacture the rail since you'd need an array of windings in it insterad of just some permanent magnets, but at least to me, it makes all kinds of sense to have the head be the passive part. For all the reaons I listed above

[@esaheiskanen8131]

I thinks the biggest advantage is that printers can scale up in print size without loosing resolution unlike belt printers.

[@Livingdiceman]

Any chance for a link to more instructions on Patrick's VFA calibration method at 10:19?

[@TeachingTech]

Link now added to the description in the VFAs chapter.

[@Livingdiceman]

@@TeachingTech Thank you!

[@JoeStoffa]

I feel like we're not quite where we should be on the prosumer printers yet. I really want a psuedo open-source printer with klipper and standard nozzles that can reach 95+ C in the chamber (and ideally 450C on the hot end). I'm running a Makerbot Method-X (95C chamber and 315C extruder), and the difference between a 95C chamber and 60C chamber is night and day. I can print large pieces in polycarbonate (not PC blend and not fiber filled) at 95% infill and it's pancake flat. But, that printer cost me $4800, only has ONE nozzle option (0.4mm), and uses a proprietary (crap) slicer without support for anything fancy (e.g. input shaping). If you want higher T (e.g. Funmat HT) or open source then you're looking at a minimum of $5-$10k. There's awesome printers at the consumer level (e.g. P1S), but for serious engineering/lab applications there's a real dead space between $1k and $5k,...

[@DeagleBingo]

this. i'm no engineer... but to me the benefits of 3d printing are in the ability to print totally FUNCTIONAL parts, immediately, at high temp, without even looking at the machine printing them, just click and walk away back to whatever project, then come get the widget and install it in the project. so even if i can get away with printing something on my highly modified enders in some more advanced material... what i want is exactly what you are talking about. and yeah... can't justify more than 1.5-2 k on this kind of tool for the shop. but considering the state of the tech here... it seems highly doable and that really the industry should be pushing consumers into higher temp materials not acting like people only want to print pla and petg. i feel like this is how companies like langmuir systems got going so well... filling the gap you mention. there is room.

[@damon5455]

I'm not sure what you had the printer sitting on but it was wobbling like crazy. Perhaps the accelerometer just freaked out due to the unexpected movement when you tried to max out the speed?

[@mike-stirling]

Hey Michael! Awesome video as always! Could you refer me to a procedure you would recommend for Bambu belt tensioning? I saw the part of this video where you mention fine tuning belts and concerns with vfa's and wanted to ask! Thanks again

[@TeachingTech]

There is a link in the description that should help you.

[@mihpop9733]

I love the aplpication from the reliability standpoint alone. I don't want to buy tools to constantly tension and replace parts for it to work. Can't wait for Bambu to implement these haha

[@dankilmer9070]

LOVE THE CONTENT! QUESTION HOW TO I ALINE THE BELTS ON MY BAMBU PRINTER??? THANKS

[@alfredanil]

Can you test the holdings/static force for both the linear motors? How much does the smaller linear motor and extruder assembly weigh? Also how does the carriage of the larger linear motor weigh? I'm trying to understand how much dynamic force the moving mass produces and if it is equal to the holding/static force of the motor.

[@henninghoefer]

I would love to see closed loop setups come to other printers - I believe a lot of the advantages could be had even without linear motors.

[@paulhimle]

I feel like some cross bracing or corner gussets would do this printer some good. Maybe they should look at a different corner design with a connection style that would change how the slop from the screws are handled. Hard to explain without pictures.

[@oldman1944]

Great video, thanks for this.

[@trajesus]

Great video! By the way, I remember a video of yours where you showed a github project that made the terminal of the pi have the klipper, or octoprint logo when connecting via ssh, but can't seem to find it, could you help?

[@marclevitt8191]

Excellent testing. But why does every print tester on KZbin put their machine on a wobbly desk? I would put the machine on the floor, especially if you have slab-on-grade construction at your house. The acrylic sheets should also add a lot of stiffness to the frame, and that, combined with being on the floor, with perhaps some Sorbothane feet under the printer, seems like it would enable the printer to both obtain better results and higher speeds. The algorithms used for input shaping will have a hell of a time compensating for the extra swinging that comes from the printer wobbling the table when it moves. No?

[@songhan1586]

not really, the printer itself is in a enclosed system if you think about its as long as there no extra movement between the print bed and nozzle, i dont think the printer really cares much if the table moves or not. Someone did the test with a a1 held by a rope, still printed fine

[@scratchfisch6425]

The max acceleration is certainly impressive, but could you do tests to figure out the max acceleration it manages without ringing or other defects? The input shaper results you showed would make this an insanely slow printer (my Ender 3's recommended acceleration is twice of what your klipper console recommends), but this might be because you tested it on a wobbly surface.

[@zachary3777]

It looks like the frame on that machine isn't stiff enough for how much weight it's throwing around.

[@tonykyle2655]

Thank you for a very informative video.

[@Rob_65]

I am hoping that the 0.4 degree skew is a flaw of this pre-production model. A full 1 mm skew on a part that is 143 mm is a bit too much to my liking (my XL has only 0.07 deg, out of the box - no tuning from my side). The speed test shows some extreme shaking of the bed so it needs an enclosure to make the whole thing more rigid. I guess that needs more than just the acrylic case - too bad since that thing looks absolutely beautiful.

[@ipadize]

is it possible to have a graph plotting the following error?

[@elana4048]

It’s not actually the motors making the noise but the linear bearings on the rail. Higher quality bearings would reduce that noise.

[@rhotix403]

I dont trust the skew correction since your measurements will change with changing parameters like flow or line width but its interesting to see that the same pattern still exists while having no belts at all🤔

[@Tom--Ace]

I would like to know if the peopoly produces better results than say a p1s or x1c at extremely fine and small models and layer heights - eg 0.2mm nozzle, 0.02-0.06 mm layer heights and printing fine figurines. Does the peopoly's high repeatability produce a meaningful advantage in fine detail printing, and do the results approach those of a resin printer any more than say a bambu machine?

[@matsv201]

That is pretty much exactly how a transrapid maglev train work. Apart that its up side down. The motor is in the rail and the static magnet is on the carrage

[@GlennPierce]

I just want the encoder strips and reader + supports in common firmwares being able to recover from a nozzle nock on a blob of plastic is worth while.

[@Pappagar]

Looks like a good prebuilt for tinkerers

[@OldCurmudgeon3DP]

The Klipper speed/accel test is interesting, but doesn't tell the whole story. The final result passes, but how would the surface of the print fair assuming the closed-loop system is making corrections? i.e. correctness of the whole path between start/finish. The specs indicate it would look good, but specs don't define results.

[@arthurmoore9488]

Same thought. Closed loop means it could overshoot and auto correct, but we would still see it on the part. If Klipper has access to the raw encoder data, that could be measured.

[@mathewphillips4185]

Bamboo carbon x1 plus magneto X love child would be Supreme

[@klave8511]

If my Ender 3 is off and I turn the stepper motors they generate power so they contain magnets. Now I don’t heat the chamber but the motors do get quite hot at times during regular operation. I’ve never heard any complaints about the effect of temperature on the regular stepper motors. I’ve had stepper motors get too hot to touch in other positioners with no ill effects. Maybe the motors, linear or rotary, should have cooling added, a fan would be a minimum but with the slow rotation rate and direction reversal it should probably be externally powered.

[@arthurmoore9488]

Some motors do have cooling! It's just the rated continuous operating temperature is often around 100C. Remember, the hotter something is the faster it cools down to ambient. That's why the motors get to a certain temperature and stay there even though they aren't drawing less power. Engineering is about compromises, and active cooling both costs money and has other downsides.

[@MrDehicka]

X1C only have VFA on lower speeds too. Above certain speed I have no visible VFA. To justify the price of Magneto we need no VFA at low speeds.

[@bepstein111]

Where's the link to Lesic's video? I couldn't find it....

[@TeachingTech]

Fixed

[@bepstein111]

Thank you much@@TeachingTech !

[@lazyman1011]

Awesome testvideo

[@KuroSanArts]

I'm all for anything that gets rid of belts! I was wondering if anyone had done anything with hydraulic/pneumatic systems?

[@qwertyzxaszc6323]

If these are reliable, getting rid of the mess that belts are would be amazing!

[@CADautomaton2a]

anyone have a link to fine tuning the belts ?

[@1288832]

9.34 The plax is good, it pleases, but it is better to measure on iron

[@DeagleBingo]

ok... so bigger elephant... ratrig v core 4 or peopoly magneto x? same price... and the pros cons of those 2 kind of seem to balance each other out (i'm aware there are a lot of differences still but overall)

[@JoshuaLotion]

Holy crap I was so hoping to see a benchy at that max 1400 speed

[@kazolar]

The future of this machine is only as good as Peopoly's commitment to it, which based on their track record leaves something to be desired. I backed their first KS for the Moai, and it wasn't too long before they abandoned it without ever finishing the firmware optical distortion compensation features. When we're dealing with so much propriety tech, it's all about trust that in 2 years you're not sitting with a $2k paperweight. Everything inside the Bambu printers can be swapped for commodity part equivalents outside of the lidar scanner -- which is of questionable usefulness. Considering there are driver boards, feedback boards, tuning requirements, this machine is more closed source than purported closed source machines are. I paused the input shaping screenshot, and klipper recommended accelerations are ~2k -- that's far from impressive with advertised max accelerations being 10x that -- so it can go fast, it can accelerate fast, but the electronics are not tuned for it. As I suspected, this is a tech demo. Impressive one, but that's all it is.

[@deltacx1059]

These are a bit in the experimental phase for larger cnc machines that need high speed, high power and repeatability within .0005 inches, not much benefit for fdm machines especially when. Ballscrews are already able to get very high speeds and repeatability and belts seem to work well on even very fast printers.

[@marchingknight11]

How fast of a speedboat can you get with this printer?

[@peplegal8253]

Linear motors are just Brushless motors with extre steps. Yeah...they got rid of belts...but at the cost of complexity and reliability. Well...belts add variance and imprecision to the system...but those can be mathematically modeled into the control firmware. I would put my chips on Closed Loop Brushless for 3D Printers...but instead of using the ultra simplistic STEP/DIR signals...it would be better to bring the Loop control code inside the main control board.

[@NavySturmGewehr]

I really want to buy one now... I blame you!

[@thegps7197]

I would assume as the motors are brushless with only 3 wires they are not accurate enough on their own for position and these linear motors have to have an encoding to know where it is. Rather than a feedback loop and a stepper this printer using the encoder directly to position the toolhead

[@arthurmoore9488]

I think this video briefly mentioned it, but the X and Y axes have linear optical encoders, like you would find on a milling machine. That's how it's doing the position hold. The weird thing is that's all handled outside of Klipper's knowledge. I may be mistaken, but I believe Klipper may not have access to the actual position information.

[@thegps7197]

Very interesting. Definitely seems like a different way to drive both axis. Seems to work really well though. Looks like with these updates their coming out with it should get better and better

[@1288832]

7:25 The meaning of the current, if you ride all magnets?

[@joescalon541]

So this is not a corexy, but a linear motor Cartesian printer?

[@calebharris292]

I think it's called h-bot. Same as the ender 5

[@kazolar]

corexy is a belt path system, no belts, so obviously not corexy

[@97hondacivic]

​@@calebharris292the ender 5 is cartesian. this is also cartesian.

[@calebharris292]

@@97hondacivic double checked, my mistake

[@nauseouscustody1440]

Great👍

[@user-jp1qt8ut3s]

I use them for ultra high acceleration (>50 g) without additional mechanical loads, but I wonder if these motors are suitable for desktop CNC machines. Nema23 is so 20th century

[@arthurmoore9488]

I think that's their plan. Klipper can't run CNC machines, but both can run regular stepper drivers. Which is what they're emulating.

[@user-jp1qt8ut3s]

@@arthurmoore9488 I wonder how difficult the wide range of loads is for the controller

[@GeorgeGraves]

"Less down time" is the selling point? I've had all sorts of stuff fail - but a stepper motor isn't one of them. Never had a stepper go bad. I have had the wires wear out from motion, but that's a 10 mins fix.

[@nonchalanto]

The motors don’t seem as silent as they were made out to be, at least not at high speeds. The claim of less artifacts doesn’t seem to pan out either. Yes, they disappeared at high speeds, but that’s true of VFA’s on belted printers too. Also linear motors are extremely power hungry for the amount of torque they provide. They can only energize one side of the coil, whereas a circular rotor can be energized from both sides. That plus the huge, non-segmented bed heater (like with the Prusa XL) seems like it will come close to tripping North American breakers on 110V. Configuration and tuning of those motors seems like a nightmare, and god help you if one of the drivers or motors breaks-those are totally bespoke parts and assemblies. I’m not sure how I feel about this printer. It’s awesome that they’re trying something new, but it does seem like an impractical technology for 3D printing. Similarly to the Pantheon printers that use ballscrews-it looks cool, but it’s wildly impractical and you cannot actually reap the benefits they supposedly bring.

[@SkateSoup]

I think the early measurements have been between 7 and 9 amps so it will run by itself without tripping a 15 amp breaker, but if you have anything else that's power hungry on that same circuit you may have problems.

[@BlaineRush-b8h]

It’s 1000w, I don’t see how that would trip a breaker.

[@nonchalanto]

@@BlaineRush-b8h the bed by itself is 1000W. That, plus crazy-inefficient linear motors, hotend, mcu, etc could get close. And it’s unlikely the printer is the only thing plugged into that circuit. In a North American house, it could easily trip a breaker.

[@davidmorton8170]

Looks like a ESP32 on the driver board…. another of my hobbies…

[@user-jp1qt8ut3s]

Your way of explaining sounds a little bit like Aurora Tech

[@leggysoft]

I'd love to use these but the price is HORRIBLE right now.

[@dedogster]

There was a lot of shaking going on, one is that normal?

[@TeachingTech]

It's just physics. Any time you have mass moving that fast things are going to shake (in this case the whole table).

[@gilb6982]

Are these motor equivalent of a nema 17 or a nema 23 ?

[@IkarimTheCreature]

Nema 17 and 23 is just a form factor and doesnt say anything directly about power

[@stevesloan6775]

If there was a system set up to check machines worked together… this could s it!!! 🇦🇺🤜🏼🤛🏼🙏🏻🤙🧠🍀😂☮️💙

[@tobiasrausch1251]

Danke sehr interessant

[@henrymach]

It's basically a matter of cost

[@jonwebb9261]

The 3 micron repeatability is such bs. Some CMMs don't achieve that

[@Heiserton]

Am I the only one who doesn’t really care about ridiculous speeds, but more good speed and excellent capabilities such as max hotend temps/reliability/enclosure and repeatability?

[@floodo1]

Those input shaper results look really bad. My main question with this machine is "What are the input shaping results with all that moving mass?"

[@dhruvgulati1667]

Its not mm per second per second, acceleration is measured as mm per second².

[@powersv2]

Ofcourse you have a cinelifter!!

[@TeachingTech]

I do not. In fact I had to look up what it was.

[@dylanlasky2389]

I wonder how you could even combine these linear motors into a core xy. The issue is the magnets would have to move through the motor because the motors are stationary.

[@dexi9814]

You don’t want to have a corexy with this, it’s not even possible because corexy is specifically designed for crossing belts. What you can do is a croxy, something like the annex engineering k3. It would require a set of 4 linear motors and 6 linear rails, very expensive indeed.

[@dylanlasky2389]

@@dexi9814 is that similar to an all wheel drive corexy?

[@dexi9814]

@@dylanlasky2389 its actually the exact same as a Cartesian just with all the heavy parts moved out of the way. Therefore the belt length is about 4x shorter than an awd corexy and its less computationally heavy (if that matters at all). Croxy is the best way to build a high speed machine, downside is that its very expensive.

[@dylanlasky2389]

@@dexi9814 ok interesting do you have any helpful links on that?

[@KToMmi]

Is your T-shirt from Jardier? 😂

[@TeachingTech]

Jimmy Broadbent

[@KToMmi]

@@TeachingTech went close 😂

[@michaelgleason4791]

2:42 not if you rotate the machine 90 degrees

[@akierum]

The magneto x sample shown is out of focus, why? What are you hiding?

[@TS_Mind_Swept]

I definitely don't think it's a gimmick; would be nice if it was cheaper, but I see it as a perfectly valid method of movement (certainly more than a "core XZ" printer..)

[@hakunamatata324]

10:04: I had hoped they learned from BambuLab printers but they didn't. That belt tensioner is the worst design even, it's easy to strip the screw and the belt moves around due to the lack of guides. I released a video showing how to improve X1C belt tension improving printing quality when I had mine before going Voron. It's disappointing after releasing such a beautiful machine only to find out, they ignore one of the most important part: Belt ....... No comments.

[@TeachingTech]

That belt footage is from a Bambu Lab P1P. This printer has no belts.

[@tvrv9774]

Watching the printer shake around so much had me screaming lots of profanities. Unstable footing is never good. Get yourself a cement counter or granite or something, I use a 24"x24"x30mm drop of granite i got for 100 bucks on a 2x4 framework. Please post a video showing results from Klippain-shaketune results, if this is already klipper based. Those give you the BEST metric on what a printer can really run at without artifacts, imho. I'm printing ASA at the upper limit that my Rapido UHF can handle with a .5 mm bozzle nozzle on a v2 350mm build (48v AB)... faster than ever now with Shaketune. This machine's advertised speeds are higher than default settings, be advised, so don't be afraid to go above the default settings on the test commands. I can only print at a max of about 225mm/s at .2 mm height due to flow (about 25 mm^3/s) , which is why I question your logic on not changing the hot end. I can obviously go higher speeds and do when printing lower layer heights but I also adjust PA and everything else to match faster speeds. Hilarious running calibration tests at 400 mm/s btw. The klippain vibration test is AWESOME at choosing which speeds your machine can handle and where the harmonics ARE NOT. Then tune based on the chosen intersection of calculated linear travel print speed based on layer height and nozzle diabmeter with your max speeds under an acceptable vibration level, run klippain's input shaper (better, yes), then run PA at the speed selected in the vibration levels for outer wall. There is a neat USB nozzle mounted accelerometer that you can use for this.

[@YourName-ox4ec]

Pricy. But cool