A good way to significantly reduce the inertia of a lead screw, at high accelerations, is to switch from a "fixed nut, rotating screw" to a "fixed screw, rotating nut" configuration. Another advantage of this is that it completely eliminates any whipping of the screw due to it not being perfectly straight, thus reducing vibration in the system. The only disadvantage (apart from the relatively higher build complexity) is the extra slop/backlash in the nut holding bearings, which could be mitigated by using bigger, preloaded radial bearings or smaller, more expensive angular ones.
Currently designing a 3D printer that uses multi-start leadscrews. Just a few comments: 1 - As others mentioned, those couplers will introduce backlash, and are not intended to be mounted that way. The screw should be retained axially, separately, via bearing block, with the coupler acting to absorb any shaft misalignment. Using those couplers as shown may actually cause or increase resonance at higher speeds. I'd agree in this application that solid couplers would be better. 2 - It should be fairly easy to copy the design of an actual anti-backlash leadnut to 3D print, and source a spring with proper tension. They actually could be had on eBay for relatively cheap. 3 - A leadscrew with anti-backlash leadnut is about 25-70% efficient depending on lead. A 4-start (.5" lead) about 40% and 8-start (1" lead) about 70%. While this may not sound too efficient, and larger steppers may need to be used, this also can act as a damper of sorts, since you wouldn't want a majority of the stepper's torque overcoming its own detent torque. 4 - I built a wood-framed router about 8 years ago or so. It doesn't really expand nor contract much, since wood does not expand much at all along the direction of grain. Plus, as long as the driven end of the screw is retained axially, you'll only see a change in accuracy as the screw rises in temperature. 5 - Repeatability (precision) will be very good because of the anti-backlash leadnut. Accuracy, you'll be at the mercy of the screw accuracy. But even an imported ACME screw is rated at .003"-.007"/ft (because it's rolled-forged), and that's worst-case scenario over that length. For the typical part made on a printer that can be adjusted out for the most part with the steps/unit setup. Screws from known suppliers like Kerk, Nook, or Roton have specs for their screws, and really don't cost much more than the no-name stuff. I should note you should be looking for "precision ACME" screws not standard ACME - they're slightly different. 6 - Aside from flex in belts, they do also have backlash with the teeth and how they mesh with the pinion. Maybe it's good enough for 3D printers, but unless I got the belts and pinions from Gates themselves, I'd have no guarantee of the precision of the belts from the "unknown" suppliers. 7 - If you're going through the trouble of using leadscrews, why not use supported linear rail? They can be had cheaply enough, and would actually add stiffness and strength to the plywood frame. 8 - There's a bit of discussion about steppers and heat. In actuality steppers are designed to run relatively hot, when you get the max power out of them. Yes with the cheaper steppers we normally use there are precautions, but the industrial-grade steppers have insulators rated to temperatures at least twice as much as the garden-variety import., and thus you can push them harder. Also, steppers are also designed with a flange and pilot which (1) absorb torque and (2) mate with a stepper plate or machine made of metal for transference of heat. Since most all 3D printer designs have steppers mounted to plastic, the heat is not dissipated well. But even some cheap CPU heatsinks would help greatly if there were heat concerns.
excellent Peter. This is by far the best tutorial on this topic. Luckily the small servo motors are dropping in price so that they can replace the stepper motor based losses and in combination with ball screws you can come to an optimal configuration. Surely not for everyone, but accounting time on printing, material spoils for bad prints and frustration, i totally vote for going to the best you can afford to build a printer. It will benefit in the long term. Greetings from Herri in China
This is a very interesting video. it directly relates to my new printer design. I will have 450mm x 450mm x 4mm aluminium build plate and was thinking of using acme threaded rods instead of belts. because i think with the inertia of the plate, i don't think it would take much of the bed movement for the belt to move without the stepper moving, which would make the bed move further than intended. Which is why i was going to use acme threaded rods. But the 2 main reasons i think printers usually use belts is: 1. It is more complicated for most people to tune a acme thread than a belt. lots of math and fine tuning for precises movement. 2. The cost of a belt is roughly half that of a acme thread per axis. But if you have a counter-point i would like to hear it.
very knowledgeable knows what he's talking about, could tell was nervous af for the camera (lots of little mistakes and studders), his video are great super helpful and very personal feeling, like talking to a human being unlike some of these more presentation/demonstration like videos
Thank you for your video, there is additional information I would like to add to your coverage. • using the aluminium spring couplers on X and Y will introduce backlash, as they are springs, and unless there is a thrust bearing preventing motion of the leadscrew along the axis, unacceptable loss of motion control would occur • as you suggest, there are limits that are driven by the hot end capability, which especially limit us when printing eg 0.4 mm layer heights to 250mm/sec or less without running out of filament. If however layer height is reduced, to 0.1 mm for example, the same hot end could potentially extrude up to 1000mm/sec. the benefit being that overall print time could be less affected for thinner layer heights. • TrinityLabs aluminatus A1 was sold with triple stack nema 17 25mm 10 start linear stages. These were approx $800 each based on conversations I had with Ezra at the time he was building these printer kits. Yes they are very expensive. • I have video on my CZcams channel of these linear stages running at 720mm/sec without additional weight ( eg extruder or build platten) these are really amazing linear stages, and are key to the high accuracy and speed the (properly built) aluminatus is capable of. • You have mentioned belts and belt ringing. These artifacts disappear with linear stages. • In practise, the repeatability and speed of these stages is definitely a worthwhile upgrade, and the improved accuracy of head placement makes prints at 0.3 mm layer heights look as nice as belt driven 0.1mm layer height results. • A perfectly acceptable print result is possible with belts and steppers, my i3 build manual on reprap.org features my Průša i3, built with Josef's aluminium frame parts, genuine J-head, metal gear system extruder and so forth. I will post some links as a follow up comment to this post, although they will likely be marked as spam for approval. I hope this helps support the conclusion of your video that lead screws can perform very well in 3D printing applications. Thanks for your video mate
Great video. I'm in the process of designing my first CNC machine and tried a hybrid of pulleys and lead screws. The screws seem much better for my needs, even if it's a little overkill. I have a belt driven 3d printer that works just fine, but CNC seems like it would benefit from the lead screws. I'll have to build a little printer with all the belts and gears I no longer need. Cheers
That was very educational, thanks! A toothed belt seems to win over a screw, for a cheaper, faster yet precise XYZ Cartesian Palletizing robot - very very small loads, for a school project really. So much to study :/
Thanks for a great video. You answered all my questions. I'm going for lead screws, 8mm/rev at X and Y axis, but I'll try XY on the head, and Z on the build plate. I rarely go beyond 80mm/sec, so I guess my 1,8deg/step motors will manage the speed.
Nice documentary. It came out right when was just about to make a plotter.I was thinking of going to a milling shop to get a custom lead screw made for myself with the finest thread pitch,I could get a nut for.
In terms of torque the screw wins most of the times, the only issue is how much slack your nuts have and a good leadscrew is much more expensive than gt2 . The gt2 is consistent and cheap, but on the other hand most printers are poorly designed in terms of lateral force on the motor shafts and pulls on the bearing on one spot. Ballscrew will not have that issue as long as it is aligned . It is more complex to build but more sturdy. Leadscrews is also used in many ink printers and work fine with a heavier setup than a bowden tube does.
While the application here is 3d printing, the same exact thing applies to CNC machines. The mechanics of both are almost the same, Its only the parameters you require that are different. In fact, even laser printers use the same basic setup. The important thing with a cnc is something known as chip load. Chip load is basically the ideal amount of material to be removed. Its a combination of spindle/router speed and the feed rate. Typically for wood and a typical router you must be able to get anywhere from 100 ipm to 200 ipm feed rates to hit that sweet spot. Granted, you can cut far slower but it often burns up bits. That is the critical issue to address with a CNC which usually means a far faster X and Y travel rates is required than a 3d printer and you will also be moving far more mass around On my home brew I use 1/2 10 5 start acme rods with high precision anti-backlash nuts driven by NEMA 23 450 oz/in steppers using divide by 2 microsteping that results in .00125 in resolution and up to 300 IPM rates. And I still dont hit the mid band resonance point that all steppers have. The divide by 2 microstepping reduces the torque by roughly 1/3. But there's still enough power to snap 1/4 bits clean into. There is NO WAY I could get that balance of resolution, speed, and accuracy with a belt drive.. Overall an excellent description the the math behind the positioning issue. The only issue that I will pick on is a bigger gear will always give more travel per step than a smaller one. The torque will also decrease. Easy to show this effect by drawing 2 circles on the same center, one at 2 inches in diameter and another at 1 inch diameter. Now draw 2 lines at some angle through the center point. Look at the distance between where the 2 lines on the small circle cross and the same points on the larger circle. The distance will be greater on the large circle. A bit of trig will give you the actual rotational distance per step but is easier to simply measure it. Mark a point on the belt and some reference point that wont move then take say 100 steps then mark a point right at the reference point as accuracte as you can. Now divide that by the steps taken ( 100 in this case ) and you will have the actual movement per step. Note the more steps used the more accurate you are. Any motion control program will have some means in it to compensate for reality and what math shows. If youre close with the math, then just fine tune the controller parameters to get things dead on. One of these days I just might try a scratch built 3d printer but my medium of choice is wood. Plastics just dont trip my trigger, I spent too many years doing small shop custom injection molding, and short run stuff. :) Once in a while I whittle something out of wood where a bit of 3d printed plastic would be acceptable.
Good vid. I enjoyed the presentation however there were a couple arguable points. For sake of that I'll go back and watch and get specific. They don't take anything away from your efforts and good advice. Thx
A very interesting vid. I have been thinking of changing my 20t pulleys for 16t then adjusting my steps to compensate which in turn will slightly increase my motor's rpm's. My 20t pulleys are showing signs of wear so need replacing soon anyhow. This vid makes me think going to the smaller pulleys is worth doing.
I think you've missed a thing, sorry if i overheard it; but basically another issue you have with lead screws is that any potential excentricity of the screw on the motor coupling will introduce a varying offset along the screw depending on the rotational angle that the screw is currently at. I think this is the reason why lead screw drive is not established for very precision sensitive XY in desktop 3D printing. Also i don't believe in anti backlash nuts. I mean they will remove backlash statically, but in movement, they will still be loaded to one side or the other against the spring and introduce a velocity dependent offset.
The eccentricity is a possibility but if the bed / gantry etc is securely mounted on good bearings then there still should not be any wobble. LEAD screws are established in many situations were precision is required especially at higher accelerations, BELTS have a tendency to stretch and oscillate too if your not using good ones, one of the primary reasons belts are used over screws would be 1. acceleration (Moment of inertia) and Cost. as with everything it is about picking the right parts for the right job and compromising were your able to get the best you can vs not breaking the bank
@@TheBreadboardca You're right, belts fundamentally behave akin to anti backlash nuts, also being a spring of sorts. At least the issues are velocity or acceleration dependent rather than periodic. I'm finding that centering even just the Z rods on flex couplers is a bit of challenge, but it can be done. Perhaps Prusa has been right though to use motors with integrated lead screws turned down to become motor's internal shaft.
Great tutorial and is very informative. I am building my first 3D printer which is a Prusa i3 clone. The X and Y axis is run by GT2 20 toothed pulley and belt. The Z axis is on 5mm threaded rod which means the Z axis steppers will be running like mad especially if Z hopping is active. I think the best bet is to get this running as it is but prioritise making printed parts to accommodate lead screw or better still ball screw rods and do away with the 5mm threaded rods and maybe change the GT2 to 16 tooth for better accuracy to the X and Y axis. It is quite clear why this guy has gone for lead screw on all axis as there is a good balance of torque and accuracy for it's purpose.
Good luck with it... one of the primary reasons lead screws or threaded rods are used for Z is so the X does not just come crashing down when power is removed from the steppers For this build (Also a prusa I3 mkII clone), I may yet change to belts for x and y but I want to go through the process to show my viewers what will happen and how good (Or Not) it is.
Thank you for your quick reply. I will be following your progress as this is really interesting. Just a quick question if I may, in your opinion is it worth increasing the ratio by using 16 tooth drive pulleys or will the end results be negligible ?.
I've used lead screws to move a laser to draw light images in a wall. It had to traverse fast enough to make it look like a line, not a dot roaming around. It not only moved fast enough, but had the precision to continue to draw the image repeatedly without it wobbling. Belts moving that fast can bounce and cause cumulative inaccuracy. A 3d printer isn't going nearly as fast, so a belt is viable, but I like my lead screws. I'm trying to figure out a printer using an arm so there's only the one vertical item to work around. I'm hoping to be able to print at angles and even upside down in one shot. Like Tony Stark's robot arm that's always carrying a fire extinguisher. Yeah... someday. lol.
We were changing the angle of a laser diode on a stick just a few degrees. It hardly weighed anything, so it wasn't an issue. Adding mirrors would just make it more complicated and harder to get done by the end of the semester. If we'd done it like scan lines a mirror or prism would have been the way to go. I found six printer/scanners in the trash at work. I'm thinking I might use all of the lasers and mirrors to get one consolidated output. See what I can cut and mount it on an old 3D printer... or the front of my car! :)
I'm doing a printer build that is mainly for research and learning to build a commercial printer to aid in fixturing for my CNC mill and to produce durable printed parts. It'll be my 3rd 3d printer but first custom build. It'll use linear components from PBC Linear. Goal is to make a precise decent speed printer with high temperature capabilities, meaning hot end capable of 450C and a heated chamber capable of around 160-180C. Because of the high temperatures I'm looking into not using belts. Reliability is also a goal. Discovered non captive lead screw stepper motors and am considering them. Extruder assembly will be using a remote direct drive extruder like a Zesty Nimble and liquid cooled Slice Engineering's 450C Mosquito hot end so extruder assembly on X axis will be fairly lightweight. Because of the weight savings I'm interested in using a non captive Nema14 lead screw stepper motor from PBC Linear fixed to extruder assembly on a linear rail. Since non captive Nema14's internal drive is what spins and the leadscrew doesn't spin I'm thinking this setup can achieve decent speeds and useful for large footprint designs since lead screw doesn't spin. I'm interested in what others think about non captive lead screw stepper motors for X and Y on a large footprint 3d printer.
Thats quite the project The fact that your "Spinning" the inside of the stepper rather than the lead screw does remove the moment of inertia of the lead screw but at the expense of the moment of inertia of the internal nut of the stepper + the mass of the actual stepper motor for two directions, this could be quite extensive. and the internals of the stepper could be subject to the very high temps of the build chamber. Another thought may be to stay with traditional ball screw / lead screw and have the steppers outside of the build chamber (Small holes would do it and not effect the temp stability of the chamber much. to compensate for the mass of the lead screws, simply use Nema 24's, you have much more space for them in this configuration. the rest of your build seems pretty well thought out, I look forward to hearing more about it.
@@TheBreadboardca Thanks for the response and input 👍 Regardless of how the extruder assembly is moved I'll have flat insulating accordion bellow covers to isolate motion system above heated chamber and more than likely liquid cooling stepper motors with same water loop that the Mosquito hot end is cooled with. After some more research I'm back to considering belts. But still think the non captive lead screw stepper motor is an interesting idea. More than likely will build this large printer I'm working on with high end belts but after that will make a 200mmx200mm or smaller proof of concept build trying out non captive lead screw stepper technology. For a large gantry style printer that has a print area of 2'-4' on X and 4'-8' on Y non captive lead screw(maybe ball screw if they exist) stepper(or servo) I'd think would have a lot lower moment of inertia than a 4'-8'+ spinning lead or ballscrew, not to mention the fact this avoids others issues involved with long fast spinning lead or ball screws. Because my extruder assembly is so lightweight I figured for X axis a Nema14 stepper motor would be the minimum needed to move itself, extruder assembly and overcome friction from leadscrew coupling/anti backlash nut. Y axis would have either a Nema17 or Nema23 non captive stepper for each Y axis linear rail, so 2 of each. Only reason I'm reconsidering belt drive in probably a coreXY arrangement(I like hbot arrangements simplicity but not gantry racking, additional rigidity and more precision assembly requirements)is it'll be a lot more cost effective, lower moment of inertia and only need 2 stepper motors for XY that can be located on outside of printer. If I was to build a large printer with non captive stepper for XY all 3 motors would be fixed to a beefy gantry which rides on precision linear rail assemblies. I'm avoiding spinning lead or ball screws for everything except Z(also considering non captive steppers for Z axis) because printer sizes I'm interested in are kind of long for a fast spinning lead or ball screw. As things progress a little farther I'm going to upload the build to CZcams. I'm also interested in carbon or glass fiber and G10 like composites and how they could be used to make a strong, stiff and lightweight X axis. I'll be using my 30"x40" milling area welded steel construction industrial grade cnc router capable of milling aluminum for the builds metal or composite parts.
Most of the belt stretch is actually in the elasticity of the teeth on the belt, not the belt backbone. also placing too much torque on the belt can strip teeth which can introduce intermittent faults that can be a bugger to find
I know that the cheap rubber GT2's will stretch pretty quickly initially. Kind of like a guitar string, which you have tune, stretch, tune, stretch, etc... Polyethylene steel re-enforced belts are said to not stretch as much...
I built a CNC mill and needed to use screws for the axis because I didn’t think belts were up to the job. I ended up having to drop a decent amount of money on some C5 rated ball screws because the nut on the leadscrew wears out too quickly and causes lots of issues.
Great video which go in details. I am also considering making a 3D printer with leadscrew and no belt. There are also other aspects which have to be considered with leadscrew: noise and power. I have heard that leadscrew require more powerfull motors than belt, meaning more prowerfull stepper driver and motors...
yeah due to friction, leadscrews are a lot less efficient (about 1/2 to 1/3 of belts), and you need to grease them from time to time. though if you have a bit more money (around 30-100€ depending on the dimensions of the printer) consider using ballscrews instead as they have virtually no backlash, are more efficient and may be more silent than leadscrews. an other technology that exists but is used mostly in bigger machines is the rack and pinion which has the same backlash problems as leadscrew, but is a lot more efficient.
I disagree with the comment at about 28:49. Higher voltage to the motor helps to counteract back EMF allowing higher speeds to be obtained. BEMF is typically what limits the maximum speed and is why higher voltages are utilized. To determine the maximum voltage a stepper motor can take use 32 * √L = max V where L is the motor inductance. The driver may ultimately determine the max V though.
Hello, I need your help to make some key decisions. I am making a large size 3d printer CoreXY type, bed will not move, size - 2300*1500*1500mm (X Y Z). Due to the large size of x & y, I have two options, one is to use large width belt 10mm, another solution is a ball screw. We don't have to compromise in speed and quality, so please guide according to your experience. In the view of the ball screw, I don't have experience of use of a ball screw, so there is doubt on the backlash problem in the ball screw. So please guide that, Have you observed backlash effect in ball screw, if yes than please guide on how to fix this problem. Please discuss in the view of ball screw SFU1610 for X & Y axis. Also guide, ball screw SFU1605 is better or lead screw for Z axis. In lead screws, we eliminate backlash effects by using anti backlash spring nuts. Same way is there any way to properly eliminate backlash in Ball Screw. You can also suggest your ideas for a large size printer. Please reply
You did not mention efficiency, which is also a factor. Belt drives are about 90% efficient. Lead screws are only roughly 50% efficient from what I have read. That will result in loss of power, more wear and probably a lot more noise (deal breaker). I debated going for lead screws in my own build, but ended up going the conventional way with GT2 belts for now. Thanks for the video. I hope you follow up and show us your build once it is finished.
Yes, lead screws are I believe nearer 70 but very dependant on the quality, a ball screw though can be as good if not better than a belt for efficiency, but at significant cost increase. now the losses are easy to compensate for by using higher drive voltage (Increasing low end torque) and also more powerful stepper motors. I am using GT3 belts as for 1, I already have the belts and 2 there bigger and less stretch, but less resolution in the drive per step. One thing I will also be trying are IGUS bearings, self lubricating and should be more efficient due to low friction
I think a lot depends on quality, certainly threaded rod will be terrible as it is not even designed for this application. For most 3D printers of a reasonable size though even if less than 50% should be easy to compensate for using bigger motors, nema17 style probably already have more than enough for this. My smaller CEL Robox only uses Nema 11 or smaller I think
Well i have a question if we use 2x microstepping will it make a 200 step/rev motor 400steps/rev? I don't think you can physically do this, you can make steps between step for smoother performance but can't create real step angle.
You supposedly "can" do this with widely debated results for simple 2x MS (many parties claim minimal loss of accuracy and precision, others claim none or significant; again with 4x), but beyond that the overwhelming opinion is that using MS positionally introduces massive loses of accuracy and precision. Then again, I've watched one video that demonstrated otherwise so I don't know anymore.
Wow, yes you are correct, I had not noticed, the ones I used in the with nuts on though were 4 or 1 start (Threaded rod) and the math still holds up, pitch * number of starts = lead so if this were a 5 start with 2mm pitch, it would have a 10mm lead. (10mm linear move per revolution), good catch,
GT3? Who uses GT3 on a 3D printer? GT2 is used on most printers. I don't see a follow-up to this conceptual design. It would be interesting to see the finished product.
The answer should be what is your skill level? Beginners should use belts. I agree screws are superior and built my 20x18x18 with screws. Add 3000rpm motors and wallah you have high resolution and a very fast printer. At some point printing at 1000ipm has other issues like getting the material out of the nozzle.
I.. just want to prevent my steppers from missing a step and screwing a 5+ hour print. (i tried everything to fix it) what if you gear the leadscrew to move at the speed a belt and either use an 84oz NEMA 17 or a NEMA 23? closed looped steppers are too expensive.. belt systems are easy to over or under tension. a minuscule amount of backlash is hardly a problem when a belt system can ruin your print and day.. as long as i get a part any surface imperfections can be ignored or sanded out.
the biggest thing I found to avoiding skipping is to have the right current dialed in on the stepper controllers, ignore the fact they get noisier, and they will run warmer, this is all normal and expected. Belt tension can be almost like a bow string, where if you "Twang it" it will make a note or almost make a note. What kind of printer are you running?
Belts have too much slop. Where as a rod would be more precise. Its like using belt drive turntables vs direct drive, the direct drive is more reliable and gives you that fine tuning compared to belts. On my 3d printer, I have noticed I have to adjust my belts after so long as they stretch/weaken this results in a sloppy print. If you replace that with a more reliable drive you cut down on a few things
Actually, I would not agree with that example. A turntable can benefit greatly from a belt drive, it helps to isolate motor noise from the audio, it can be equally as precice if not more so as the same control of the motor is available but often with a significant gear ratio allowing for much finer control of the turn table. The MASS of the turntable is what also provides significant stability and due to the gearing used with a belt system, the motor torque if "amplified". I also know audo systems quality can be a very subjective topic and I dont want to get into a phyisophical debate as to which is better, this is simply my current view point,
Recently I dissasembly a photocopier that uses a sort of 1mm fishing cord to move the scannig head. Do you think could be a good system for 3d printers?
My probably not answer was based on the hardware from a scanner, can you use spectra wire for 3d printers, I am sure you can, I dont know enough of the parameters of the larger form of this product or how well it will grip the pulleys (We would not want slippage etc)
One turn on the pulley is 60mm, while one turn on the lead screw is 8mm. So you would have to have a gear reduction of 7.5 to 1 on the belt to have the same resolution of the leadscrew. Leadscrew always wins hands down for accuracy. There I just did in 10 seconds what took you 35 minutes! P.S. use a 10 to1 planetary gearbox on the stepper motor, and now your belt has higher resolution/accuracy than the leadscrew. Now I got even more in depth than you in 15 seconds. lmao Summary, if you know how to do math you can use any form of mechanical means to obtain linear motion, and achieve the (same) desired results. Rack and pinion with active backlash compensation beats all others in speed, accuracy, reduced mass acceleration, strength, and simplicity every time. (20 seconds elapsed time)
I appreciate your video, thank you. I have a question regarding the video, however. In the Video you kept using 8mm as the travel distance for the lead screw relative to your revolutions. Wouldn't that number be equal to the pitch? (2mm) not 8mm even if you have a 4 start lead? one revolution should be just 2mm should it not? Which in turn would put all your math off by a factor of 4. Would it not? Perhaps I'm misunderstanding this? Also, if you wanted greater accuracy (Less speed/more torque) on something like a Core XY could you not go to a smaller number of teeth on your pulley?
nope, I am correct, with a lead screw with multiple starts, the way it is described is pitch=distance between threads, if there is only one thread then it also happens to equal the lead (Distance traveled for one rev), when there are 4 starts (Any thing other than 1 really), the distance traveled is the pitch (2mm in this case) * the number of leads (4 in this case) and that gives us 8mm. (I also physically verified this before responding), Weird I know but that's the way it works
Peter Oakes thanks Peter for the explanation. Seems somehow counterintuitive to me. As in how would the distance travelled change based on the number of times the threads on the lead go around the lead screw. I have a printer with z axis lead screws (4 start I think) I'm gonna test that when I get home tonight. Maybe I'm thinking about it incorrectly. Thank you for your response and insight.
Peter Oakes yeah, I'm getting exactly how you're explaining it... And that's the part that is counterintuitive to me. I'm almost home I'll definitely be checking my prusa style machine. I am completely curious to see how far it travels on one full rotation of the z axis
Isn't wood a very very very bad material to use for the frame? :S It distorts with humidity and well just time, and more importantly, it's auto ignition temperature is between 250 - 300 C. Meaning, if your heater element fails in some way (either bed, or worse the extruder heater falls out) your printer is going to go up in flames.
First, the wood, I explain I am using Marine Ply, this is very dimensionally stable so over the period of a few days will not change dimension's in any significant way, also most locations were the unit would be used is also pretty stable regarding humidity. also it is extremely immune to distortion (It would need to be especially in marine applications. If the extruder heater element falls out then it will or should land on the metal / glass bed if it even lands on anything, the cables attached should not be long enough for it to reach the frame, the heated bed should never get close to the temperature required to ignite wood, if it does then you have way worse issues than the wood to worry about. Most bed coverings wont survive much above 100 - 150C without disintegrating and producing toxic fumes or worse. No one should be running a printer without thermal runaway monitoring turned on these days either, In summary there are a very large set of things needed to go wrong way before I would be worrying about the wooden frame.
theBreadboard Maybe it’s not that unsafe then. I’m just thinking back to when I was tinkering with my clogged nozzle on my acrylic Anet A8 and heated it up and the heated element fell out. The head was close to one side and landed very close to the edge of the bed, had the head been moving it would most likely been pushed off the bed down onto the acrylic frame. But the cables on the Anet extruder is maybe a bit longer than they have to be. Do you know how good wood is at holding screws over time btw? I’m designing a new budget printer and I’m currently going all metal and 3d printed parts because I was worried about woods dimensional accuracy. Though now I’ve decided to chase micrometer accuracy so then wood might be out the window anyway.
@@beaconofwierd1883 If you have the budget then using OpenBuilds V slot aluminum extrusions would be a way to go, there not too expensive and are easy to assemble and re-assemble. This is what most 3D printers use these days as well, the only issue I find with most current printers is there using 20x20mm parts, I would use 20x60 or even 40x60 C beams for the main structure, this would ensure a very rigid frame. The wood will be good for a few "Ins and Outs" of a screw but I would go with nuts and Bolts for securing parts to the wood, drill the holes right through, slightly tight for the screw to ensure accuracy (Assuming the holes are drilled accurately of course.
theBreadboard It’s way outside my budget goal and way outside my accuracy goal. My plan is to buy cheap steel square tubing and lap two sides flat to within a couple of microns, then work from there :) So I can hopefully achieve micron accuracy for less money than the budget printers. Currently it looks like the frame, guides and screws will go for less than 30 bucks :D I don’t know if it will work, but it’s not a lot of money to waste to test it :)
@@beaconofwierd1883 Most importantly is a solid frame, square tubing securely fastened together could be stronger than the extrusions anyway, followed by near zero play in the moving parts. If you have the skills then I say go for it.
formula at ~28:20 looks wrong; According to that your MAX RPM decreases as you increase current. How can that be when you are increasing the power? Would make much more sense if it was Volts * Current divided by (Inductance * 2 * Steps). But indeed according to sources that is the correct formula. So the previous step is holding back from getting to the next step due to delay of magnetic field collapsing? Delay increases by the function of current applied? That would make sense why people are so adamant on setting "correct" Vref for drivers. However, it is a motor and a motor is motor. Volts * Current == Power AND Power equates the amount of work that can be done (sans losses). So i have typically just cranked drivers always to the max, which has always yielded the best results as long as steppers remain below 70C. If they do not, i have typically added heatsinks. Each time i do this print results are better, and you can drive higher speed. So what piece i am missing here, as that formula would suggest that setting your stepper drivers to 0.00A would give the best speed, which is impossible at 0W. www.allaboutcircuits.com/tools/stepper-motor-calculator/ In other words, this formula makes 0 sense. Stepper motors are essentially brushless DC motors (at their most basic)... So how would the laws of physics be reversed? I think you used wrong voltages, as per above link: APPLIED VOLTAGE. So output of the stepper driver, which will be a tiny fraction of the input 12V. Now that would give the correct results, as the current increases so does applied voltage. So specs: uk.rs-online.com/web/p/stepper-motors/5350489/ Resistance per phase is 1.65ohm, so to apply 12V means 7.27A == 87.24W (total ~174W), where as applying 2.8V would yield roughly 1.68A == 4.7W. In essence the power rating does not affect maximum torque max rpm rating at all. That is not the max speed off the driver, simply where it still has it's maximum torque. Max torque RPM rating tho is a bit misleading, it is not the maximum speed it can turn at all, typical electric motor goes steadily from Max torque @ 0RPM to no torque at their maximum RPM. To calculate actual torque curves. etc. back emf etc stuff needs to be figured out. to calculate the Kv rating. For example if Kv == 300RPM it means 0 load max rpm of 300rpm with 1volt applied. If the 100% torque rating is 7.8RPM (as with your motor) it means from 7.9 to 300rpm torque decreases linearly, and your maximum load will determine where is maximum real speed. You can increase this by voltage, which means also increasing current hand in hand; But does not change the basic characteristics of the motor (full torque max rpm known). There probably also is a formula to calculate this from the spec sheets, which i am not aware. Basicly, higher the rated voltage for the motor == higher Kv rating if torque is the same. Higher the voltage == higher speed it can go. The Kv ratings are extremely low in stepper motors, here is more information about it: learningrc.com/motor-kv/ Basicly you can measure this by using another motor to spin the stepper motor. We all have seen how LCD screens light up when you move the bed by hand :) Here is also some interesting discussion about motors: www.reddit.com/r/CNC/comments/6iai0v/low_voltage_vs_high_voltage_stepper_motors/ Even more interesting: electronics.stackexchange.com/questions/200324/how-to-interpret-the-stepper-motor-voltage-requirement Some MFGs release the Kv rating as Volts per 1000 RPM. RS does not, but they release some curves at docs-emea.rs-online.com/webdocs/13f7/0900766b813f73f3.pdf I think you can use the minimum time per step to calculate 0 torque max rpm, for Peter's motor at rated voltage it is 15 625rpm at 2.8V or 5580rpm/1v (Kv). Which is surprisingly high. So now the equations makes perfect sense, wrong values and assumptions were used. But it's no where near this simple. I am in process of upgrading one of my printers to 36V TB6600 drivers.
Love the analysis, are you a quad copter flyer by chance ? there is one thing you did not factor in your post... INDUCTANCE. for a motor it is a gross over simplification to simply say power is V*I, this only considers resistance (Not even impedance). While it is true that if you apply a DC voltage to a coil and leave it there, the current will eventually fall to simply V/R but this is not how a stepper / controller works. the controller monitors the current (The reason for the Vref) and when it exceeds Vref setting it will discontinue applying the voltage. it is not actually limiting the current to that, the instantaneous current is a function of the inductance and applied voltage, it is simply watching it and waiting for it to reach its max set point as indicated by a function of Vref The current in an inductor does not instantly go to max the instant you apply the voltage to it, an inductor is the opposite to a capacitor in that it resists change in current and therefor seems like an open circuit to a change in voltage (Capacitor resists change in voltage and therefor seems like a short to a change in voltage) What a stepper wants in order to achieve max torque is to get to max magnetic flux (But not over saturate). A higher voltage applied will allow the current to increase quicker to the set point max and therefor achieve max torque quicker. Due to the inductance etc, with all other things being equal, increasing the MAX current will simply delay the circuit from reaching this MAX current and the time curve is the same for a single step, the controller tries to maintain this MAX Current by switching on and off the applied voltage (Remember the current does not instantly go to max but neither does it go to zero when the applied voltage is disconnected and now the magnetic field is collapsing and trying to maintain that current flow (Inductor resists change in current), once the controller detects this current fall bellow a set value it will turn on the output again and re energize the coil one last point, it is a common miss understanding that a stepper (Brushless DC Motor) speed is a function of voltage, it is not. Torque is a function of voltage the running speed of a stepper is determined by the controller cycling through the phases (Coils) MAX torque is only reached by applying a voltage long enough for the magnetic flux to build to max, if you switch to fast you will never reach it which is why you loose torque at higher speeds (Yes you can go faster but at reduced torque). a lower inductance stepper motor will permit higher speeds or applying higher voltage will do the same This is why in industry on CNC machines etc, they will often run with 48 or even 80V DC supplies, the inductance may be similar, the max current is more but the speed and MAX torque is still achieved by having a higher applied voltage (NOT CURRENT). but you need a much better stepper controller to work at these voltages I know this is a whole bunch of explanation on different things but i hope it helps you see how it works, I smell a video in the works :)
I used to play around with RC cars, until i could afford a fast big one :) I was one of the early users of brushless DC motors with BIG lipo packs. Indeed; Inductance. That is why i said it's never that simple; And that is somewhere i need to do some additional research. Good explanation. And you are correct; Magnetic fields take their time to ramp up or down; i simply erased that bit from my earlier reply. And as for the speed being function of voltage -> Both are correct, if controller does not command for faster RPM, no faster RPM will happen, and that is why it is called a speed controller with RC. however, what i was referring to was the intrinsic motor max RPM per voltage. Higher voltage is higher potential (think as RPM); Where as higher current is higher "force" (think that as of torque) to make things happen (power) indeed. I am sorry if my explanation of the mistake in the calculations on your video made believe otherwise. it was a lot train of thought and poorly edited.
@@skaltura not a problem. Healthy discussion is always a good thing and I'm not perfect and therfor subject to the odd error here or there so I could have made an error. Oh and that formula is only for speed at max torque. Not other chacturistics
If it works with screws for X axis on the bed. Why couldn't it do that for Y? It is the same distance that both must travel depending on the object. A circle is a circle how much one twists it. If it handles a line with a screw in the X axis, it should also go on the Y axis. Right now I'm building a printer for myself, not just because I can't buy one but for the pleasure. Someone with experience can tell me if it has been tried. Otherwise I will try. Z moves so slowly up or down so there is no question.
Screws should work for all axis without issues, as with belts, Acme Screws or Ball screws its all about minimizing slop / play in the axis. Z is typically taken care of due to gravity, X and Y may need help. Belts have minimal play but they can stretch during acceleration etc, hence the ringing when changing directions or stopping at a corner (The belt is a bit like a bad spring). Screws (Ball or Acme) may have a little play (Backlash) depending on quality and if anti backlash measures are taken but they wont suffer from the same springy affect. But the motors may need to be more powerful due to increase in moment of inertia.
@@TheBreadboardca An update on what has happened. I have now built a 3D printer with 8mm screws, 2mm pitch with 4 starts, on al axis. Went down to 8 microsteps for more torque. After some fine tuning, it works GREAT! One problem I got is that the dual motors on the Z-axis do not get enough Vref voltage from the A4988 stepper driver. Sometimes skipping steps at startup. Only a restart of the printout. Normaly i do prints in 60mm/s, fast or slow? i dont got any to compare with.
@@STRB909 60mm/s is a reasonable speed so should not cause any problems, your Z axis should be able to move even with using an A4988 driver and two parallel motors, this is a very common configuration on many printers, I would check for any binding in the screws etc., an other option would be to use a separate driver for each motor, depending on your controller you may even have a spare driver position, often labeled for use with a second head.
@@TheBreadboardca r8 my bad...i dint notice the website which had all the calculations, I was referring to the doc file where u typed everything. PS thanks
theBreadboard can you please give me your contact info, I’m going to design a revolutionary 3D printer, I’m making my own motors, it’s not even comparable to servo or stepper in the way it works and it had all the pros of both and none of the cons. I’m going to need a specialized board to control it and new software to work with that. As in my magnetic coils can change their orientation, and no I’m not talking about switching the positive and negative, I mean as in XYZ orientation. , but I have some rod questions, and an idea for you to replace the spring type using magnetics
Dear Sirs Shen-Han Technology is established in 2010 and supplies ball screw and ball nuts which factory is sited in Bao-An, Shenshen, China. We are the high-tech ball screw manufacturer and meet the international standard of DIN 69051, GB/T17587 and JIS C5, etc. The factory is certified by ISO9001 & UKAS and we are professional in the industry and have the confidence to produce both of the global standard and the customized specification, and the quality is compatible with the international branded products. We are approved by Chinese high-tech enterprise certification and the plant has the equipment included CNC lathes, CNC 4 axis machining centers, CNC 5 axis machining centers, and CNC internal thread grinding machines. We have R&D team to develop the new designs of ball screw and has the specific patents including the new design and new appearance. We are also capable to develop OEM or ODM designed ball screws to satisfy with customized needs. The sketch of CAD/CAM customized transmission component design is available both for equipment and high lead ball screws in specific standards. Capabilities of Shen-Han Technology - . Ball screws and screw nuts in customized specific standards . High lead ball screws and screw nuts (1616, 1632, 2020, 2040, 2525, 2550, etc) . Module in precision positioning, XY tables (Single -axle or multiple-axle stages) Wei Tai Transmission Limited Company is the head office in Taipei and acts for all of international trade activities to handle the shipment supplied from Shen-Han Shenzhen factory. We are looking for the new customers in your regions and it’s thankful if you can give us a chance to work together with. Please feel free to contact the following: Wei Tai Transmission Limited Company Email: 02weitai.ltd@gmail.com
Belt pitch has absolutely NOTHING to do with linear distance on a belt drive. Linear distance is dependent on the pulley circumference and nothing else. You compared a 2pitch 20 tooth pulley (~13mm dia.) to a 3pitch 20 tooth pulley (~19mm dia.) a 13mm pulley will have exactly the same linear translation if it has 20 teeth or 200 teeth, which is 13*pi = 40.8407mm per rev. This is about 5 times faster than an 8mm lead lead screw, with about 1/5 the resolution and torque (ignoring frictional losses). On the subject of friction, a 2mm pitch screw will have the same (approximately) friction regardless of number of starts with no load. With linear load applied more starts has less friction. There are differences in thread design, a lead screw will use some variant of an Acme thread, while threaded rod will use some variant of a standard 60 degree profile. Acme threads are intended for linear motion, while 60degree threads are intended for strength. As you increase the number of starts for a given pitch on a thread, you increase the pressure being applied radially , and reduce the pressure applied linearly to the nut. In other words, adding starts, reduces available force to move your part. Imagine your 2pitch thread with 50 starts, the rod will look more like a spur gear than a thread, and will apply nearly all of the pressure radially (like a gear), and apply nearly no pressure linearly. What this means is that the 4start 2mm lead screw has about 1/4 the applied linear pressure as a 1start 2mm lead screw. Comparing this to a belt gets a bit messy, but it should be clear that a 4 start lead screw will not apply massively more pressure that a belt.
I disagree with "It has nothing to do with it", it is simply a simpler way of looking at it. If a belt has a 2mm pitch and a pulley has the matching pitch and also has 20 teeth around then it is really simple math to figure out how far it will travel in a single turn (2*20), (3*20 in the case of a 3mm pitch). No 2PiR etc required and no need to accuratly measure the diameter of the pulley, compensating for the teeth and flat spots. The math you present is perfectly correct and if you were using a smooth belt it would be the only way to figure out the distance traveled for a single turn, with a toothed belt, my way is simpler but no more or less accurate. Thanks for the great info on the friction effects with different screws atc, very informative.
@@ldruningman Everybody is actually correct, but to some "C=2πr " (2*3.14159* ??, as the radius is not given) is more complex than 20*2 (Teeth * Pitch), we could figure out R by reversing the process ie. 40/2 * 3.14159 giving R=6.366 so now we can do 2*3.14159*6.366 = 39.998 but you have to agree, that is too much unnecessary work right !!. most parts like this are advertised as a 20 tooth GT2 pully or simply a GT2 belt. the 2 meaning 2mm pitch
I am sure they do, but with the range of 3D printers being sold, most of them use Acme threads and for the purpose they work well enough. I know when i complete my test bed, there will be issues with backlash on the x and y but how much depends on how I mitigate it. if I can get a sponsor to support me with Ball Screws then I will be only too happy to create a video all about them too, along with using linear rails rather than ground rods. Are proper ball screws way better in oh so many ways, absolutely, but that comes with a considerable increase in cost. If you need that accuracy and repeatability and stability then it is worth the additional cost
All your mathematics is a waste. What really matters is how smooth are the lead screw and the belts. The errors of douse are much bigger than step of the movements. So way are you showing this?
Actually, you would be surprised how many people just dont know the details, they put it together or buy one built and just accept whatever it is without knowing (And sometimes that's ok) but when you want the extra bit of performance, it often helps if someone takes the time to explain, yes sometimes it sounds obvious but I have come across similar issues oh so many times
Can stand unprepared videos, if your going to do this, have your stuff ready to go and know what it is and what your going to say. Try creating a script. Moving on.
A good way to significantly reduce the inertia of a lead screw, at high accelerations, is to switch from a "fixed nut, rotating screw" to a "fixed screw, rotating nut" configuration. Another advantage of this is that it completely eliminates any whipping of the screw due to it not being perfectly straight, thus reducing vibration in the system. The only disadvantage (apart from the relatively higher build complexity) is the extra slop/backlash in the nut holding bearings, which could be mitigated by using bigger, preloaded radial bearings or smaller, more expensive angular ones.
Currently designing a 3D printer that uses multi-start leadscrews. Just a few comments:
1 - As others mentioned, those couplers will introduce backlash, and are not intended to be mounted that way. The screw should be retained axially, separately, via bearing block, with the coupler acting to absorb any shaft misalignment. Using those couplers as shown may actually cause or increase resonance at higher speeds. I'd agree in this application that solid couplers would be better.
2 - It should be fairly easy to copy the design of an actual anti-backlash leadnut to 3D print, and source a spring with proper tension. They actually could be had on eBay for relatively cheap.
3 - A leadscrew with anti-backlash leadnut is about 25-70% efficient depending on lead. A 4-start (.5" lead) about 40% and 8-start (1" lead) about 70%. While this may not sound too efficient, and larger steppers may need to be used, this also can act as a damper of sorts, since you wouldn't want a majority of the stepper's torque overcoming its own detent torque.
4 - I built a wood-framed router about 8 years ago or so. It doesn't really expand nor contract much, since wood does not expand much at all along the direction of grain. Plus, as long as the driven end of the screw is retained axially, you'll only see a change in accuracy as the screw rises in temperature.
5 - Repeatability (precision) will be very good because of the anti-backlash leadnut. Accuracy, you'll be at the mercy of the screw accuracy. But even an imported ACME screw is rated at .003"-.007"/ft (because it's rolled-forged), and that's worst-case scenario over that length. For the typical part made on a printer that can be adjusted out for the most part with the steps/unit setup. Screws from known suppliers like Kerk, Nook, or Roton have specs for their screws, and really don't cost much more than the no-name stuff. I should note you should be looking for "precision ACME" screws not standard ACME - they're slightly different.
6 - Aside from flex in belts, they do also have backlash with the teeth and how they mesh with the pinion. Maybe it's good enough for 3D printers, but unless I got the belts and pinions from Gates themselves, I'd have no guarantee of the precision of the belts from the "unknown" suppliers.
7 - If you're going through the trouble of using leadscrews, why not use supported linear rail? They can be had cheaply enough, and would actually add stiffness and strength to the plywood frame.
8 - There's a bit of discussion about steppers and heat. In actuality steppers are designed to run relatively hot, when you get the max power out of them. Yes with the cheaper steppers we normally use there are precautions, but the industrial-grade steppers have insulators rated to temperatures at least twice as much as the garden-variety import., and thus you can push them harder. Also, steppers are also designed with a flange and pilot which (1) absorb torque and (2) mate with a stepper plate or machine made of metal for transference of heat. Since most all 3D printer designs have steppers mounted to plastic, the heat is not dissipated well. But even some cheap CPU heatsinks would help greatly if there were heat concerns.
excellent Peter. This is by far the best tutorial on this topic. Luckily the small servo motors are dropping in price so that they can replace the stepper motor based losses and in combination with ball screws you can come to an optimal configuration. Surely not for everyone, but accounting time on printing, material spoils for bad prints and frustration, i totally vote for going to the best you can afford to build a printer. It will benefit in the long term. Greetings from Herri in China
Thanks! I've been trying to learn the language of CNC for a future project and it hasn't been easy! This helps.
Amazing content, thank you for the in depth theory and direct coupling to its effect in a 3d printer!
Glad it helped!
Looking forward to the rest of this series. Particularly the technical details and calculations that go into your design decisions.
Thank you a lot from Brazil. That explanation was awesome and have cleared my doubts.
This is a very interesting video. it directly relates to my new printer design. I will have 450mm x 450mm x 4mm aluminium build plate and was thinking of using acme threaded rods instead of belts. because i think with the inertia of the plate, i don't think it would take much of the bed movement for the belt to move without the stepper moving, which would make the bed move further than intended. Which is why i was going to use acme threaded rods.
But the 2 main reasons i think printers usually use belts is:
1. It is more complicated for most people to tune a acme thread than a belt. lots of math and fine tuning for precises movement.
2. The cost of a belt is roughly half that of a acme thread per axis.
But if you have a counter-point i would like to hear it.
very knowledgeable knows what he's talking about, could tell was nervous af for the camera (lots of little mistakes and studders), his video are great super helpful and very personal feeling, like talking to a human being unlike some of these more presentation/demonstration like videos
Thank you for your video, there is additional information I would like to add to your coverage.
• using the aluminium spring couplers on X and Y will introduce backlash, as they are springs, and unless there is a thrust bearing preventing motion of the leadscrew along the axis, unacceptable loss of motion control would occur
• as you suggest, there are limits that are driven by the hot end capability, which especially limit us when printing eg 0.4 mm layer heights to 250mm/sec or less without running out of filament. If however layer height is reduced, to 0.1 mm for example, the same hot end could potentially extrude up to 1000mm/sec. the benefit being that overall print time could be less affected for thinner layer heights.
• TrinityLabs aluminatus A1 was sold with triple stack nema 17 25mm 10 start linear stages. These were approx $800 each based on conversations I had with Ezra at the time he was building these printer kits. Yes they are very expensive.
• I have video on my CZcams channel of these linear stages running at 720mm/sec without additional weight ( eg extruder or build platten) these are really amazing linear stages, and are key to the high accuracy and speed the (properly built) aluminatus is capable of.
• You have mentioned belts and belt ringing. These artifacts disappear with linear stages.
• In practise, the repeatability and speed of these stages is definitely a worthwhile upgrade, and the improved accuracy of head placement makes prints at 0.3 mm layer heights look as nice as belt driven 0.1mm layer height results.
• A perfectly acceptable print result is possible with belts and steppers, my i3 build manual on reprap.org features my Průša i3, built with Josef's aluminium frame parts, genuine J-head, metal gear system extruder and so forth. I will post some links as a follow up comment to this post, although they will likely be marked as spam for approval. I hope this helps support the conclusion of your video that lead screws can perform very well in 3D printing applications. Thanks for your video mate
czcams.com/video/24T4EiHhXD4/video.html
Photographs and videos of my Aluminatus rebuild with smoothieboard controller flickr.com/photos/87180506@N08/sets/72157647547282865
Nice build there.. :)
Thanks John! A video of it printing here now also czcams.com/video/ZcmgkvvjljY/video.html
Thanks for the excellent explanations. Great inspiration for me to continue with my lead screw delta build.
maybe a bit slower than a regular delta......but with an amazing resolution. That....and magnetic joints......and up you go my friend.
I’ve been thinking about the same thing. Putting some anti backlash on lead screw and Delta will be more stable holding it position.
Great video. I'm in the process of designing my first CNC machine and tried a hybrid of pulleys and lead screws. The screws seem much better for my needs, even if it's a little overkill. I have a belt driven 3d printer that works just fine, but CNC seems like it would benefit from the lead screws. I'll have to build a little printer with all the belts and gears I no longer need.
Cheers
glad you found it useful, share your results if possible, it will be nice to see
That was very educational, thanks!
A toothed belt seems to win over a screw, for a cheaper, faster yet precise XYZ Cartesian Palletizing robot - very very small loads, for a school project really. So much to study :/
Feel free to ask questions as needed, I am always happy to help with education when I can
Thanks for a great video. You answered all my questions. I'm going for lead screws, 8mm/rev at X and Y axis, but I'll try XY on the head, and Z on the build plate.
I rarely go beyond 80mm/sec, so I guess my 1,8deg/step motors will manage the speed.
Thank you so much for your time!
I solved my problem by watching this video and got too many ideas in mind now. Excellent!
Good video and what you presented was very well done. You presented a lot of engineering concepts that I was presently unaware,
Glad you liked it, more to come in the future stay tuned
Is there a video on the final outcome of this build? I am quite interested in making a lead screw based printer.
Google Workhorse 3D printer.
Me too
Nice documentary. It came out right when was just about to make a plotter.I was thinking of going to a milling shop to get a custom lead screw made for myself with the finest thread pitch,I could get a nut for.
In terms of torque the screw wins most of the times, the only issue is how much slack your nuts have and a good leadscrew is much more expensive than gt2 . The gt2 is consistent and cheap, but on the other hand most printers are poorly designed in terms of lateral force on the motor shafts and pulls on the bearing on one spot. Ballscrew will not have that issue as long as it is aligned . It is more complex to build but more sturdy. Leadscrews is also used in many ink printers and work fine with a heavier setup than a bowden tube does.
While the application here is 3d printing, the same exact thing applies to CNC machines. The mechanics of both are almost the same, Its only the parameters you require that are different. In fact, even laser printers use the same basic setup. The important thing with a cnc is something known as chip load. Chip load is basically the ideal amount of material to be removed. Its a combination of spindle/router speed and the feed rate. Typically for wood and a typical router you must be able to get anywhere from 100 ipm to 200 ipm feed rates to hit that sweet spot. Granted, you can cut far slower but it often burns up bits. That is the critical issue to address with a CNC which usually means a far faster X and Y travel rates is required than a 3d printer and you will also be moving far more mass around
On my home brew I use 1/2 10 5 start acme rods with high precision anti-backlash nuts driven by NEMA 23 450 oz/in steppers using divide by 2 microsteping that results in .00125 in resolution and up to 300 IPM rates. And I still dont hit the mid band resonance point that all steppers have. The divide by 2 microstepping reduces the torque by roughly 1/3. But there's still enough power to snap 1/4 bits clean into. There is NO WAY I could get that balance of resolution, speed, and accuracy with a belt drive..
Overall an excellent description the the math behind the positioning issue. The only issue that I will pick on is a bigger gear will always give more travel per step than a smaller one. The torque will also decrease.
Easy to show this effect by drawing 2 circles on the same center, one at 2 inches in diameter and another at 1 inch diameter. Now draw 2 lines at some angle through the center point. Look at the distance between where the 2 lines on the small circle cross and the same points on the larger circle. The distance will be greater on the large circle.
A bit of trig will give you the actual rotational distance per step but is easier to simply measure it. Mark a point on the belt and some reference point that wont move then take say 100 steps then mark a point right at the reference point as accuracte as you can. Now divide that by the steps taken ( 100 in this case ) and you will have the actual movement per step. Note the more steps used the more accurate you are.
Any motion control program will have some means in it to compensate for reality and what math shows. If youre close with the math, then just fine tune the controller parameters to get things dead on.
One of these days I just might try a scratch built 3d printer but my medium of choice is wood. Plastics just dont trip my trigger, I spent too many years doing small shop custom injection molding, and short run stuff. :) Once in a while I whittle something out of wood where a bit of 3d printed plastic would be acceptable.
Good vid. I enjoyed the presentation however there were a couple arguable points. For sake of that I'll go back and watch and get specific. They don't take anything away from your efforts and good advice. Thx
A very interesting vid. I have been thinking of changing my 20t pulleys for 16t then adjusting my steps to compensate which in turn will slightly increase my motor's rpm's. My 20t pulleys are showing signs of wear so need replacing soon anyhow. This vid makes me think going to the smaller pulleys is worth doing.
Very interesting looking forward to joining you on your build journey. Regards RobUK
Thanks for a great video
I think you've missed a thing, sorry if i overheard it; but basically another issue you have with lead screws is that any potential excentricity of the screw on the motor coupling will introduce a varying offset along the screw depending on the rotational angle that the screw is currently at. I think this is the reason why lead screw drive is not established for very precision sensitive XY in desktop 3D printing.
Also i don't believe in anti backlash nuts. I mean they will remove backlash statically, but in movement, they will still be loaded to one side or the other against the spring and introduce a velocity dependent offset.
The eccentricity is a possibility but if the bed / gantry etc is securely mounted on good bearings then there still should not be any wobble.
LEAD screws are established in many situations were precision is required especially at higher accelerations, BELTS have a tendency to stretch and oscillate too if your not using good ones, one of the primary reasons belts are used over screws would be 1. acceleration (Moment of inertia) and Cost. as with everything it is about picking the right parts for the right job and compromising were your able to get the best you can vs not breaking the bank
@@TheBreadboardca You're right, belts fundamentally behave akin to anti backlash nuts, also being a spring of sorts. At least the issues are velocity or acceleration dependent rather than periodic.
I'm finding that centering even just the Z rods on flex couplers is a bit of challenge, but it can be done. Perhaps Prusa has been right though to use motors with integrated lead screws turned down to become motor's internal shaft.
Great tutorial and is very informative. I am building my first 3D printer which is a Prusa i3 clone. The X and Y axis is run by GT2 20 toothed pulley and belt. The Z axis is on 5mm threaded rod which means the Z axis steppers will be running like mad especially if Z hopping is active.
I think the best bet is to get this running as it is but prioritise making printed parts to accommodate lead screw or better still ball screw rods and do away with the 5mm threaded rods and maybe change the GT2 to 16 tooth for better accuracy to the X and Y axis.
It is quite clear why this guy has gone for lead screw on all axis as there is a good balance of torque and accuracy for it's purpose.
Good luck with it... one of the primary reasons lead screws or threaded rods are used for Z is so the X does not just come crashing down when power is removed from the steppers
For this build (Also a prusa I3 mkII clone), I may yet change to belts for x and y but I want to go through the process to show my viewers what will happen and how good (Or Not) it is.
Thank you for your quick reply. I will be following your progress as this is really interesting. Just a quick question if I may, in your opinion is it worth increasing the ratio by using 16 tooth drive pulleys or will the end results be negligible ?.
Andy P I think it can only help. It will not be detrimental
Friction can be simply reduced with grease or maybe a Teflon coating of sorts
GREAT VIDEO :) THANKS FOR TAKING THE TIME TO MAKE IT...
Now I understand why expensive 3D printers are expensive. Why cheap printers are cheap.. great video.
late, but great topic, thanks 👍
I've used lead screws to move a laser to draw light images in a wall. It had to traverse fast enough to make it look like a line, not a dot roaming around. It not only moved fast enough, but had the precision to continue to draw the image repeatedly without it wobbling. Belts moving that fast can bounce and cause cumulative inaccuracy. A 3d printer isn't going nearly as fast, so a belt is viable, but I like my lead screws. I'm trying to figure out a printer using an arm so there's only the one vertical item to work around. I'm hoping to be able to print at angles and even upside down in one shot. Like Tony Stark's robot arm that's always carrying a fire extinguisher. Yeah... someday. lol.
though rotating 2 mirrors instead of moving a laser is a lot faster for what you wanted to do ^^
We were changing the angle of a laser diode on a stick just a few degrees. It hardly weighed anything, so it wasn't an issue. Adding mirrors would just make it more complicated and harder to get done by the end of the semester. If we'd done it like scan lines a mirror or prism would have been the way to go.
I found six printer/scanners in the trash at work. I'm thinking I might use all of the lasers and mirrors to get one consolidated output. See what I can cut and mount it on an old 3D printer... or the front of my car! :)
Your accent is very unique. It's like English, American, and South African mixed together.
English and Canadian, but close
I'm doing a printer build that is mainly for research and learning to build a commercial printer to aid in fixturing for my CNC mill and to produce durable printed parts.
It'll be my 3rd 3d printer but first custom build. It'll use linear components from PBC Linear. Goal is to make a precise decent speed printer with high temperature capabilities, meaning hot end capable of 450C and a heated chamber capable of around 160-180C. Because of the high temperatures I'm looking into not using belts. Reliability is also a goal.
Discovered non captive lead screw stepper motors and am considering them. Extruder assembly will be using a remote direct drive extruder like a Zesty Nimble and liquid cooled Slice Engineering's 450C Mosquito hot end so extruder assembly on X axis will be fairly lightweight. Because of the weight savings I'm interested in using a non captive Nema14 lead screw stepper motor from PBC Linear fixed to extruder assembly on a linear rail.
Since non captive Nema14's internal drive is what spins and the leadscrew doesn't spin I'm thinking this setup can achieve decent speeds and useful for large footprint designs since lead screw doesn't spin.
I'm interested in what others think about non captive lead screw stepper motors for X and Y on a large footprint 3d printer.
Thats quite the project
The fact that your "Spinning" the inside of the stepper rather than the lead screw does remove the moment of inertia of the lead screw but at the expense of the moment of inertia of the internal nut of the stepper + the mass of the actual stepper motor for two directions, this could be quite extensive. and the internals of the stepper could be subject to the very high temps of the build chamber. Another thought may be to stay with traditional ball screw / lead screw and have the steppers outside of the build chamber (Small holes would do it and not effect the temp stability of the chamber much. to compensate for the mass of the lead screws, simply use Nema 24's, you have much more space for them in this configuration. the rest of your build seems pretty well thought out, I look forward to hearing more about it.
@@TheBreadboardca Thanks for the response and input 👍
Regardless of how the extruder assembly is moved I'll have flat insulating accordion bellow covers to isolate motion system above heated chamber and more than likely liquid cooling stepper motors with same water loop that the Mosquito hot end is cooled with.
After some more research I'm back to considering belts. But still think the non captive lead screw stepper motor is an interesting idea. More than likely will build this large printer I'm working on with high end belts but after that will make a 200mmx200mm or smaller proof of concept build trying out non captive lead screw stepper technology.
For a large gantry style printer that has a print area of 2'-4' on X and 4'-8' on Y non captive lead screw(maybe ball screw if they exist) stepper(or servo) I'd think would have a lot lower moment of inertia than a 4'-8'+ spinning lead or ballscrew, not to mention the fact this avoids others issues involved with long fast spinning lead or ball screws.
Because my extruder assembly is so lightweight I figured for X axis a Nema14 stepper motor would be the minimum needed to move itself, extruder assembly and overcome friction from leadscrew coupling/anti backlash nut.
Y axis would have either a Nema17 or Nema23 non captive stepper for each Y axis linear rail, so 2 of each.
Only reason I'm reconsidering belt drive in probably a coreXY arrangement(I like hbot arrangements simplicity but not gantry racking, additional rigidity and more precision assembly requirements)is it'll be a lot more cost effective, lower moment of inertia and only need 2 stepper motors for XY that can be located on outside of printer.
If I was to build a large printer with non captive stepper for XY all 3 motors would be fixed to a beefy gantry which rides on precision linear rail assemblies.
I'm avoiding spinning lead or ball screws for everything except Z(also considering non captive steppers for Z axis) because printer sizes I'm interested in are kind of long for a fast spinning lead or ball screw.
As things progress a little farther I'm going to upload the build to CZcams.
I'm also interested in carbon or glass fiber and G10 like composites and how they could be used to make a strong, stiff and lightweight X axis.
I'll be using my 30"x40" milling area welded steel construction industrial grade cnc router capable of milling aluminum for the builds metal or composite parts.
Most of the belt stretch is actually in the elasticity of the teeth on the belt, not the belt backbone. also placing too much torque on the belt can strip teeth which can introduce intermittent faults that can be a bugger to find
I can confirm that from experience: the rubber teeth will bend / rip long before any stretching in the tension element...
I know that the cheap rubber GT2's will stretch pretty quickly initially. Kind of like a guitar string, which you have tune, stretch, tune, stretch, etc...
Polyethylene steel re-enforced belts are said to not stretch as much...
Thank you for this professional video
I built a CNC mill and needed to use screws for the axis because I didn’t think belts were up to the job. I ended up having to drop a decent amount of money on some C5 rated ball screws because the nut on the leadscrew wears out too quickly and causes lots of issues.
Many thanks for the thorough guide 🌹
Great video which go in details. I am also considering making a 3D printer with leadscrew and no belt. There are also other aspects which have to be considered with leadscrew: noise and power. I have heard that leadscrew require more powerfull motors than belt, meaning more prowerfull stepper driver and motors...
yeah due to friction, leadscrews are a lot less efficient (about 1/2 to 1/3 of belts), and you need to grease them from time to time.
though if you have a bit more money (around 30-100€ depending on the dimensions of the printer) consider using ballscrews instead as they have virtually no backlash, are more efficient and may be more silent than leadscrews.
an other technology that exists but is used mostly in bigger machines is the rack and pinion which has the same backlash problems as leadscrew, but is a lot more efficient.
One of the strangest mixed accents I've heard for a while (as an Australian). British/American or even some Aussie in there as well? Good video.
British, Canadian actually, but close, glad you liked the video
Thanks for a very informative video.
Thank you for a great video.
there's been some people that used chain and sprocket, if it turns out a decent system, i'll use it.
I disagree with the comment at about 28:49. Higher voltage to the motor helps to counteract back EMF allowing higher speeds to be obtained. BEMF is typically what limits the maximum speed and is why higher voltages are utilized. To determine the maximum voltage a stepper motor can take use 32 * √L = max V where L is the motor inductance. The driver may ultimately determine the max V though.
Great video!
What would be the problem if you design a corexy printer with X Y and Z axis has a ballscrew or leadscrew? all axis?
please tell me which is best belt or screw?
Hello,
I need your help to make some key decisions. I am making a large size 3d printer CoreXY type, bed will not move, size - 2300*1500*1500mm (X Y Z).
Due to the large size of x & y, I have two options, one is to use large width belt 10mm, another solution is a ball screw.
We don't have to compromise in speed and quality, so please guide according to your experience.
In the view of the ball screw, I don't have experience of use of a ball screw, so there is doubt on the backlash problem in the ball screw. So please guide that, Have you observed backlash effect in ball screw, if yes than please guide on how to fix this problem.
Please discuss in the view of ball screw SFU1610 for X & Y axis.
Also guide, ball screw SFU1605 is better or lead screw for Z axis.
In lead screws, we eliminate backlash effects by using anti backlash spring nuts. Same way is there any way to properly eliminate backlash in Ball Screw.
You can also suggest your ideas for a large size printer.
Please reply
very helpful info thanks !
You did not mention efficiency, which is also a factor. Belt drives are about 90% efficient. Lead screws are only roughly 50% efficient from what I have read. That will result in loss of power, more wear and probably a lot more noise (deal breaker). I debated going for lead screws in my own build, but ended up going the conventional way with GT2 belts for now. Thanks for the video. I hope you follow up and show us your build once it is finished.
Yes, lead screws are I believe nearer 70 but very dependant on the quality, a ball screw though can be as good if not better than a belt for efficiency, but at significant cost increase. now the losses are easy to compensate for by using higher drive voltage (Increasing low end torque) and also more powerful stepper motors.
I am using GT3 belts as for 1, I already have the belts and 2 there bigger and less stretch, but less resolution in the drive per step.
One thing I will also be trying are IGUS bearings, self lubricating and should be more efficient due to low friction
I've read 30% for acme/trapezoidal. 20% for plain 60 degree thread, and 80% for ballscrews.
I think a lot depends on quality, certainly threaded rod will be terrible as it is not even designed for this application. For most 3D printers of a reasonable size though even if less than 50% should be easy to compensate for using bigger motors, nema17 style probably already have more than enough for this. My smaller CEL Robox only uses Nema 11 or smaller I think
@@MattOGormanSmith ehat is also why trapezoidal often is used for vertical movement so it will have locking if power is lost
Well i have a question if we use 2x microstepping will it make a 200 step/rev motor 400steps/rev?
I don't think you can physically do this, you can make steps between step for smoother performance but can't create real step angle.
You supposedly "can" do this with widely debated results for simple 2x MS (many parties claim minimal loss of accuracy and precision, others claim none or significant; again with 4x), but beyond that the overwhelming opinion is that using MS positionally introduces massive loses of accuracy and precision. Then again, I've watched one video that demonstrated otherwise so I don't know anymore.
Rack&Pinion can be a good alternative to belts and screws
what's you're take on double z lead screw with a closed belt loop solution?
Bytheway the leadsrew what on the rigth in the doc, that is a 5 start one, not a 4 start. Just a little notice.
Wow, yes you are correct, I had not noticed, the ones I used in the with nuts on though were 4 or 1 start (Threaded rod) and the math still holds up, pitch * number of starts = lead so if this were a 5 start with 2mm pitch, it would have a 10mm lead. (10mm linear move per revolution), good catch,
I know the math I'm an engineering student. Just noticed it ;)
Nice 👍
GT3? Who uses GT3 on a 3D printer? GT2 is used on most printers. I don't see a follow-up to this conceptual design. It would be interesting to see the finished product.
Thank you sir.
Welcome!
Ia it good to use m8 threaded rod instead of lead screw
In that picture of the pitch and stuff of the lead screws they should have colored each rail or whatever you would call it
You have changed my mind - I am now going to use leadscrews for my Prusa 3D!!! Now all I have to do is sell the belting I bought!!!
So how did you end up doing it? I want to do a Fisher style with 12mm smooth rods, 1000 by 500 mm frame.
The problem is not speed it is backslash...
what happened to the ball screws project (3d printer diy)
The answer should be what is your skill level? Beginners should use belts. I agree screws are superior and built my 20x18x18 with screws. Add 3000rpm motors and wallah you have high resolution and a very fast printer. At some point printing at 1000ipm has other issues like getting the material out of the nozzle.
I.. just want to prevent my steppers from missing a step and screwing a 5+ hour print. (i tried everything to fix it)
what if you gear the leadscrew to move at the speed a belt and either use an 84oz NEMA 17 or a NEMA 23?
closed looped steppers are too expensive.. belt systems are easy to over or under tension.
a minuscule amount of backlash is hardly a problem when a belt system can ruin your print and day..
as long as i get a part any surface imperfections can be ignored or sanded out.
the biggest thing I found to avoiding skipping is to have the right current dialed in on the stepper controllers, ignore the fact they get noisier, and they will run warmer, this is all normal and expected. Belt tension can be almost like a bow string, where if you "Twang it" it will make a note or almost make a note. What kind of printer are you running?
Backlash on lead screws?
Belts have too much slop. Where as a rod would be more precise. Its like using belt drive turntables vs direct drive, the direct drive is more reliable and gives you that fine tuning compared to belts. On my 3d printer, I have noticed I have to adjust my belts after so long as they stretch/weaken this results in a sloppy print. If you replace that with a more reliable drive you cut down on a few things
Actually, I would not agree with that example. A turntable can benefit greatly from a belt drive, it helps to isolate motor noise from the audio, it can be equally as precice if not more so as the same control of the motor is available but often with a significant gear ratio allowing for much finer control of the turn table. The MASS of the turntable is what also provides significant stability and due to the gearing used with a belt system, the motor torque if "amplified". I also know audo systems quality can be a very subjective topic and I dont want to get into a phyisophical debate as to which is better, this is simply my current view point,
@@TheBreadboardca i have used both types of turntables to dj on and saw the opposite 🤷♂️.
@@TheBreadboardca this time it has lead screw T8 (16mm lead) it can rotation speed close to the belt i try test print it very good on speed 100mm/s
Recently I dissasembly a photocopier that uses a sort of 1mm fishing cord to move the scannig head. Do you think could be a good system for 3d printers?
Probably not, the scan head is extremely light compared to a 3D printer and therefor i expect you will experience issues
That system is what I say, look possible
My probably not answer was based on the hardware from a scanner, can you use spectra wire for 3d printers, I am sure you can, I dont know enough of the parameters of the larger form of this product or how well it will grip the pulleys (We would not want slippage etc)
cool
if we change to lead screw drive..do we need to tweak the software..?
not the software but the configuration, you will need to change the steps per mm to match the screw vs what you had for the belt
How about the ball screw?
One turn on the pulley is 60mm, while one turn on the lead screw is 8mm. So you would have to have a gear reduction of 7.5 to 1 on the belt to have the same resolution of the leadscrew. Leadscrew always wins hands down for accuracy. There I just did in 10 seconds what took you 35 minutes! P.S. use a 10 to1 planetary gearbox on the stepper motor, and now your belt has higher resolution/accuracy than the leadscrew. Now I got even more in depth than you in 15 seconds. lmao Summary, if you know how to do math you can use any form of mechanical means to obtain linear motion, and achieve the (same) desired results. Rack and pinion with active backlash compensation beats all others in speed, accuracy, reduced mass acceleration, strength, and simplicity every time. (20 seconds elapsed time)
I can understand how a y axis would be done with rack and pinion but can you give examples of how the x and z axis would be done? Cheers
I appreciate your video, thank you. I have a question regarding the video, however. In the Video you kept using 8mm as the travel distance for the lead screw relative to your revolutions. Wouldn't that number be equal to the pitch? (2mm) not 8mm even if you have a 4 start lead? one revolution should be just 2mm should it not? Which in turn would put all your math off by a factor of 4. Would it not? Perhaps I'm misunderstanding this? Also, if you wanted greater accuracy (Less speed/more torque) on something like a Core XY could you not go to a smaller number of teeth on your pulley?
nope, I am correct, with a lead screw with multiple starts, the way it is described is pitch=distance between threads, if there is only one thread then it also happens to equal the lead (Distance traveled for one rev), when there are 4 starts (Any thing other than 1 really), the distance traveled is the pitch (2mm in this case) * the number of leads (4 in this case) and that gives us 8mm. (I also physically verified this before responding), Weird I know but that's the way it works
Peter Oakes thanks Peter for the explanation. Seems somehow counterintuitive to me. As in how would the distance travelled change based on the number of times the threads on the lead go around the lead screw. I have a printer with z axis lead screws (4 start I think) I'm gonna test that when I get home tonight.
Maybe I'm thinking about it incorrectly. Thank you for your response and insight.
even though it is described as 2mm pitch, 4 start etc, think of it as 4 threads with 8mm pitch
Peter Oakes yeah, I'm getting exactly how you're explaining it... And that's the part that is counterintuitive to me. I'm almost home I'll definitely be checking my prusa style machine. I am completely curious to see how far it travels on one full rotation of the z axis
Let me know, you got me curious now too :)
always thought the lead screw would be better..!
Can you do a video on "Lead screws vs ball screws" It would be interesting to see what differences there are for 3d Printing
Ballscrews are superior and a video a waste of time. Especially considering the cost of today’s c7 screws it’s a no brainer.
Isn't wood a very very very bad material to use for the frame? :S It distorts with humidity and well just time, and more importantly, it's auto ignition temperature is between 250 - 300 C. Meaning, if your heater element fails in some way (either bed, or worse the extruder heater falls out) your printer is going to go up in flames.
First, the wood, I explain I am using Marine Ply, this is very dimensionally stable so over the period of a few days will not change dimension's in any significant way, also most locations were the unit would be used is also pretty stable regarding humidity. also it is extremely immune to distortion (It would need to be especially in marine applications.
If the extruder heater element falls out then it will or should land on the metal / glass bed if it even lands on anything, the cables attached should not be long enough for it to reach the frame, the heated bed should never get close to the temperature required to ignite wood, if it does then you have way worse issues than the wood to worry about. Most bed coverings wont survive much above 100 - 150C without disintegrating and producing toxic fumes or worse.
No one should be running a printer without thermal runaway monitoring turned on these days either,
In summary there are a very large set of things needed to go wrong way before I would be worrying about the wooden frame.
theBreadboard Maybe it’s not that unsafe then. I’m just thinking back to when I was tinkering with my clogged nozzle on my acrylic Anet A8 and heated it up and the heated element fell out. The head was close to one side and landed very close to the edge of the bed, had the head been moving it would most likely been pushed off the bed down onto the acrylic frame. But the cables on the Anet extruder is maybe a bit longer than they have to be.
Do you know how good wood is at holding screws over time btw? I’m designing a new budget printer and I’m currently going all metal and 3d printed parts because I was worried about woods dimensional accuracy. Though now I’ve decided to chase micrometer accuracy so then wood might be out the window anyway.
@@beaconofwierd1883 If you have the budget then using OpenBuilds V slot aluminum extrusions would be a way to go, there not too expensive and are easy to assemble and re-assemble. This is what most 3D printers use these days as well, the only issue I find with most current printers is there using 20x20mm parts, I would use 20x60 or even 40x60 C beams for the main structure, this would ensure a very rigid frame. The wood will be good for a few "Ins and Outs" of a screw but I would go with nuts and Bolts for securing parts to the wood, drill the holes right through, slightly tight for the screw to ensure accuracy (Assuming the holes are drilled accurately of course.
theBreadboard It’s way outside my budget goal and way outside my accuracy goal. My plan is to buy cheap steel square tubing and lap two sides flat to within a couple of microns, then work from there :) So I can hopefully achieve micron accuracy for less money than the budget printers. Currently it looks like the frame, guides and screws will go for less than 30 bucks :D I don’t know if it will work, but it’s not a lot of money to waste to test it :)
@@beaconofwierd1883 Most importantly is a solid frame, square tubing securely fastened together could be stronger than the extrusions anyway, followed by near zero play in the moving parts. If you have the skills then I say go for it.
formula at ~28:20 looks wrong; According to that your MAX RPM decreases as you increase current. How can that be when you are increasing the power?
Would make much more sense if it was Volts * Current divided by (Inductance * 2 * Steps).
But indeed according to sources that is the correct formula.
So the previous step is holding back from getting to the next step due to delay of magnetic field collapsing? Delay increases by the function of current applied?
That would make sense why people are so adamant on setting "correct" Vref for drivers.
However, it is a motor and a motor is motor. Volts * Current == Power AND Power equates the amount of work that can be done (sans losses). So i have typically just cranked drivers always to the max, which has always yielded the best results as long as steppers remain below 70C. If they do not, i have typically added heatsinks. Each time i do this print results are better, and you can drive higher speed.
So what piece i am missing here, as that formula would suggest that setting your stepper drivers to 0.00A would give the best speed, which is impossible at 0W.
www.allaboutcircuits.com/tools/stepper-motor-calculator/
In other words, this formula makes 0 sense. Stepper motors are essentially brushless DC motors (at their most basic)... So how would the laws of physics be reversed?
I think you used wrong voltages, as per above link: APPLIED VOLTAGE. So output of the stepper driver, which will be a tiny fraction of the input 12V.
Now that would give the correct results, as the current increases so does applied voltage.
So specs: uk.rs-online.com/web/p/stepper-motors/5350489/
Resistance per phase is 1.65ohm, so to apply 12V means 7.27A == 87.24W (total ~174W), where as applying 2.8V would yield roughly 1.68A == 4.7W.
In essence the power rating does not affect maximum torque max rpm rating at all. That is not the max speed off the driver, simply where it still has it's maximum torque.
Max torque RPM rating tho is a bit misleading, it is not the maximum speed it can turn at all, typical electric motor goes steadily from Max torque @ 0RPM to no torque at their maximum RPM.
To calculate actual torque curves. etc. back emf etc stuff needs to be figured out. to calculate the Kv rating.
For example if Kv == 300RPM it means 0 load max rpm of 300rpm with 1volt applied. If the 100% torque rating is 7.8RPM (as with your motor) it means from 7.9 to 300rpm torque decreases linearly, and your maximum load will determine where is maximum real speed. You can increase this by voltage, which means also increasing current hand in hand; But does not change the basic characteristics of the motor (full torque max rpm known).
There probably also is a formula to calculate this from the spec sheets, which i am not aware. Basicly, higher the rated voltage for the motor == higher Kv rating if torque is the same. Higher the voltage == higher speed it can go.
The Kv ratings are extremely low in stepper motors, here is more information about it: learningrc.com/motor-kv/
Basicly you can measure this by using another motor to spin the stepper motor. We all have seen how LCD screens light up when you move the bed by hand :)
Here is also some interesting discussion about motors: www.reddit.com/r/CNC/comments/6iai0v/low_voltage_vs_high_voltage_stepper_motors/
Even more interesting: electronics.stackexchange.com/questions/200324/how-to-interpret-the-stepper-motor-voltage-requirement
Some MFGs release the Kv rating as Volts per 1000 RPM.
RS does not, but they release some curves at docs-emea.rs-online.com/webdocs/13f7/0900766b813f73f3.pdf
I think you can use the minimum time per step to calculate 0 torque max rpm, for Peter's motor at rated voltage it is 15 625rpm at 2.8V or 5580rpm/1v (Kv). Which is surprisingly high.
So now the equations makes perfect sense, wrong values and assumptions were used.
But it's no where near this simple. I am in process of upgrading one of my printers to 36V TB6600 drivers.
Love the analysis, are you a quad copter flyer by chance ?
there is one thing you did not factor in your post... INDUCTANCE. for a motor it is a gross over simplification to simply say power is V*I, this only considers resistance (Not even impedance). While it is true that if you apply a DC voltage to a coil and leave it there, the current will eventually fall to simply V/R but this is not how a stepper / controller works.
the controller monitors the current (The reason for the Vref) and when it exceeds Vref setting it will discontinue applying the voltage. it is not actually limiting the current to that, the instantaneous current is a function of the inductance and applied voltage, it is simply watching it and waiting for it to reach its max set point as indicated by a function of Vref
The current in an inductor does not instantly go to max the instant you apply the voltage to it, an inductor is the opposite to a capacitor in that it resists change in current and therefor seems like an open circuit to a change in voltage (Capacitor resists change in voltage and therefor seems like a short to a change in voltage)
What a stepper wants in order to achieve max torque is to get to max magnetic flux (But not over saturate). A higher voltage applied will allow the current to increase quicker to the set point max and therefor achieve max torque quicker.
Due to the inductance etc, with all other things being equal, increasing the MAX current will simply delay the circuit from reaching this MAX current and the time curve is the same
for a single step, the controller tries to maintain this MAX Current by switching on and off the applied voltage (Remember the current does not instantly go to max but neither does it go to zero when the applied voltage is disconnected and now the magnetic field is collapsing and trying to maintain that current flow (Inductor resists change in current), once the controller detects this current fall bellow a set value it will turn on the output again and re energize the coil
one last point, it is a common miss understanding that a stepper (Brushless DC Motor) speed is a function of voltage, it is not. Torque is a function of voltage
the running speed of a stepper is determined by the controller cycling through the phases (Coils)
MAX torque is only reached by applying a voltage long enough for the magnetic flux to build to max, if you switch to fast you will never reach it which is why you loose torque at higher speeds (Yes you can go faster but at reduced torque). a lower inductance stepper motor will permit higher speeds or applying higher voltage will do the same
This is why in industry on CNC machines etc, they will often run with 48 or even 80V DC supplies, the inductance may be similar, the max current is more but the speed and MAX torque is still achieved by having a higher applied voltage (NOT CURRENT). but you need a much better stepper controller to work at these voltages
I know this is a whole bunch of explanation on different things but i hope it helps you see how it works, I smell a video in the works :)
I used to play around with RC cars, until i could afford a fast big one :) I was one of the early users of brushless DC motors with BIG lipo packs.
Indeed; Inductance. That is why i said it's never that simple; And that is somewhere i need to do some additional research. Good explanation.
And you are correct; Magnetic fields take their time to ramp up or down; i simply erased that bit from my earlier reply.
And as for the speed being function of voltage -> Both are correct, if controller does not command for faster RPM, no faster RPM will happen, and that is why it is called a speed controller with RC. however, what i was referring to was the intrinsic motor max RPM per voltage.
Higher voltage is higher potential (think as RPM); Where as higher current is higher "force" (think that as of torque) to make things happen (power) indeed. I am sorry if my explanation of the mistake in the calculations on your video made believe otherwise. it was a lot train of thought and poorly edited.
@@skaltura not a problem. Healthy discussion is always a good thing and I'm not perfect and therfor subject to the odd error here or there so I could have made an error. Oh and that formula is only for speed at max torque. Not other chacturistics
If it works with screws for X axis on the bed. Why couldn't it do that for Y? It is the same distance that both must travel depending on the object. A circle is a circle how much one twists it. If it handles a line with a screw in the X axis, it should also go on the Y axis. Right now I'm building a printer for myself, not just because I can't buy one but for the pleasure. Someone with experience can tell me if it has been tried. Otherwise I will try. Z moves so slowly up or down so there is no question.
Screws should work for all axis without issues, as with belts, Acme Screws or Ball screws its all about minimizing slop / play in the axis. Z is typically taken care of due to gravity, X and Y may need help. Belts have minimal play but they can stretch during acceleration etc, hence the ringing when changing directions or stopping at a corner (The belt is a bit like a bad spring). Screws (Ball or Acme) may have a little play (Backlash) depending on quality and if anti backlash measures are taken but they wont suffer from the same springy affect. But the motors may need to be more powerful due to increase in moment of inertia.
@@TheBreadboardca Thanks for a great answer. Screws it is, the things you explain regarding belts is what i experiense on my laser engraver.
@@TheBreadboardca An update on what has happened. I have now built a 3D printer with 8mm screws, 2mm pitch with 4 starts, on al axis. Went down to 8 microsteps for more torque. After some fine tuning, it works GREAT! One problem I got is that the dual motors on the Z-axis do not get enough Vref voltage from the A4988 stepper driver. Sometimes skipping steps at startup. Only a restart of the printout. Normaly i do prints in 60mm/s, fast or slow? i dont got any to compare with.
@@STRB909 60mm/s is a reasonable speed so should not cause any problems, your Z axis should be able to move even with using an A4988 driver and two parallel motors, this is a very common configuration on many printers, I would check for any binding in the screws etc., an other option would be to use a separate driver for each motor, depending on your controller you may even have a spare driver position, often labeled for use with a second head.
Ball screw it is and what I run on all 3 axises.
So, whatever happened to this project?
CAN U MAKE THE DOC FILE AVAILABLE FOR US
Sorry, which DOC file are you referring to, many are already linked in the description.
@@TheBreadboardca r8 my bad...i dint notice the website which had all the calculations, I was referring to the doc file where u typed everything. PS thanks
@@adityams1659 ill see if i can find it, it was a while ago :)
theBreadboard can you please give me your contact info, I’m going to design a revolutionary 3D printer, I’m making my own motors, it’s not even comparable to servo or stepper in the way it works and it had all the pros of both and none of the cons. I’m going to need a specialized board to control it and new software to work with that. As in my magnetic coils can change their orientation, and no I’m not talking about switching the positive and negative, I mean as in XYZ orientation. , but I have some rod questions, and an idea for you to replace the spring type using magnetics
Backlash is a B
Dear Sirs
Shen-Han Technology is established in 2010 and supplies ball screw and ball nuts which factory is sited in Bao-An, Shenshen, China. We are the high-tech ball screw manufacturer and meet the international standard of DIN 69051, GB/T17587 and JIS C5, etc. The factory is certified by ISO9001 & UKAS and we are professional in the industry and have the confidence to produce both of the global standard and the customized specification, and the quality is compatible with the international branded products.
We are approved by Chinese high-tech enterprise certification and the plant has the equipment included CNC lathes, CNC 4 axis machining centers, CNC 5 axis machining centers, and CNC internal thread grinding machines. We have R&D team to develop the new designs of ball screw and has the specific patents including the new design and new appearance. We are also capable to develop OEM or ODM designed ball screws to satisfy with customized needs. The sketch of CAD/CAM customized transmission component design is available both for equipment and high lead ball screws in specific standards.
Capabilities of Shen-Han Technology -
. Ball screws and screw nuts in customized specific standards
. High lead ball screws and screw nuts (1616, 1632, 2020, 2040, 2525, 2550, etc)
. Module in precision positioning, XY tables (Single -axle or multiple-axle stages)
Wei Tai Transmission Limited Company is the head office in Taipei and acts for all of international trade activities to handle the shipment supplied from Shen-Han Shenzhen factory. We are looking for the new customers in your regions and it’s thankful if you can give us a chance to work together with. Please feel free to contact the following:
Wei Tai Transmission Limited Company
Email: 02weitai.ltd@gmail.com
Belt pitch has absolutely NOTHING to do with linear distance on a belt drive.
Linear distance is dependent on the pulley circumference and nothing else.
You compared a 2pitch 20 tooth pulley (~13mm dia.) to a 3pitch 20 tooth pulley (~19mm dia.)
a 13mm pulley will have exactly the same linear translation if it has 20 teeth or 200 teeth, which is 13*pi = 40.8407mm per rev. This is about 5 times faster than an 8mm lead lead screw, with about 1/5 the resolution and torque (ignoring frictional losses).
On the subject of friction, a 2mm pitch screw will have the same (approximately) friction regardless of number of starts with no load. With linear load applied more starts has less friction. There are differences in thread design, a lead screw will use some variant of an Acme thread, while threaded rod will use some variant of a standard 60 degree profile. Acme threads are intended for linear motion, while 60degree threads are intended for strength.
As you increase the number of starts for a given pitch on a thread, you increase the pressure being applied radially , and reduce the pressure applied linearly to the nut. In other words, adding starts, reduces available force to move your part. Imagine your 2pitch thread with 50 starts, the rod will look more like a spur gear than a thread, and will apply nearly all of the pressure radially (like a gear), and apply nearly no pressure linearly.
What this means is that the 4start 2mm lead screw has about 1/4 the applied linear pressure as a 1start 2mm lead screw. Comparing this to a belt gets a bit messy, but it should be clear that a 4 start lead screw will not apply massively more pressure that a belt.
I disagree with "It has nothing to do with it", it is simply a simpler way of looking at it. If a belt has a 2mm pitch and a pulley has the matching pitch and also has 20 teeth around then it is really simple math to figure out how far it will travel in a single turn (2*20), (3*20 in the case of a 3mm pitch). No 2PiR etc required and no need to accuratly measure the diameter of the pulley, compensating for the teeth and flat spots. The math you present is perfectly correct and if you were using a smooth belt it would be the only way to figure out the distance traveled for a single turn, with a toothed belt, my way is simpler but no more or less accurate.
Thanks for the great info on the friction effects with different screws atc, very informative.
I understand the confusion. That was my first thought but you're correct Peter.
@@TheBreadboardca I agree with Joshua43214. Distance traveled per revolution is not related to belt pitch, but rather circumference of the pulley.
@@ldruningman Everybody is actually correct, but to some "C=2πr " (2*3.14159* ??, as the radius is not given) is more complex than 20*2 (Teeth * Pitch), we could figure out R by reversing the process ie. 40/2 * 3.14159 giving R=6.366 so now we can do 2*3.14159*6.366 = 39.998 but you have to agree, that is too much unnecessary work right !!. most parts like this are advertised as a 20 tooth GT2 pully or simply a GT2 belt. the 2 meaning 2mm pitch
aged like a milk
Direct drive for the pro.
Rack and pinion
Dont want to watch the whole video
Can somebody tell me what is the best choice for a laser engraver belt or screw?
As far as I know, there is no such thing as RPMs (revolutions per minuteS). It is just RPM or revolutions per minute.
Nobody here seems to know what a real ballscrew is. Acme threads are for amateurs.
I am sure they do, but with the range of 3D printers being sold, most of them use Acme threads and for the purpose they work well enough. I know when i complete my test bed, there will be issues with backlash on the x and y but how much depends on how I mitigate it. if I can get a sponsor to support me with Ball Screws then I will be only too happy to create a video all about them too, along with using linear rails rather than ground rods.
Are proper ball screws way better in oh so many ways, absolutely, but that comes with a considerable increase in cost. If you need that accuracy and repeatability and stability then it is worth the additional cost
All your mathematics is a waste. What really matters is how smooth are the lead screw and the belts. The errors of douse are much bigger than step of the movements. So way are you showing this?
If you know so much... why are you searching for videos on the subject?
Too much high school level maths. Any one who tries to make a 3d printer can do this easily?
Actually, you would be surprised how many people just dont know the details, they put it together or buy one built and just accept whatever it is without knowing (And sometimes that's ok) but when you want the extra bit of performance, it often helps if someone takes the time to explain, yes sometimes it sounds obvious but I have come across similar issues oh so many times
Ой, мужик, какой же ты нудный. Все по теме. Но какой же ты нууууудный
Can stand unprepared videos, if your going to do this, have your stuff ready to go and know what it is and what your going to say. Try creating a script. Moving on.
Yea, I hate when somebody goes out of their way to explain things for free and they dont get it perfect. Even $400 textbooks have errors, man.