All about Lead Screws for 3D printers
Vložit
- čas přidán 28. 08. 2024
- This video tries to explain all about lead screws for 3D printers. Where to use them, where not to use them (and the reasons why), what size to use, the difference between lead and pitch, the optimum lead to use, as well as backlash and how to deal with it.
One of those rare longish CZcams videos that you watch wholly. No skipping is necessary or desirable.
Thanks! Subscribed! Please keep em coming!
Thank you for that feedback.
This is brilliant. Very informative and concise. Great explanations. Now I feel I have a good grasp of the fundamentals.
Thanks - glad it helped.
I am comforted to know....that despite all the Idiocy that is happening in the World....there is still Extreme Intelligence out there. Thank you. (I really needed to know the things in this video)
Thank you. In turn, I am pleased that you found the video helpful (although I prefer to think of myself as having a degree of common sense, rather than being extremely intelligent).
Thank you for the excellent technical explanation of lead screws for the layman. I was looking at backlash nuts and wobble limiters, chanced upon your video, giving me an understanding of why I might not need them.
I'm pleased that you found the video helpful and that it might have saved you some money.
What a lecture!!! Thank you so much for putting this together and explaining in such amazingly easy way. Respect.
Glad you found it useful.
@@deckingman very useful
As an Engineering educator, I have to say well presented and well described for the laymen to understand. Nicely Done
Thank you kind sir! As an engineer (retired) but not a educator, I'll take that compliment. Personally, I think I need to include more humour. There are certain facts that I can still remember being taught some 50 years or so in the past because they were "wrapped up" in a humorous anecdote. Yet I have long forgotten much that was presented as simple, dry facts.
I would like to encourage you to do more videos, so I did subscribe. I am starting to build my own printer from scratch, as I've ran out of things to upgrade on my current large volume printer. I always figured that belts were used on the X and Y for speed, and to some extent to keep per unit costs down. I knew that using lead screws on X and Y would require a more course thread, and all the bad things that go along with that, but never considered backlash as well. I would enjoy hearing you speak about the actual root causes of artifacts such as ringing, how it finds its way into our prints, and how we can reduce it as much as possible. Something tells me that your counterbalance system has something to do with that. Thanks again for the well presented, and valuable information.
Thanks for the encouragement. I now have over 1,000 subscribers but the viewing hours are still less than half what they need to be before I can start thinking about monetisation. It takes me quite a while to put a video together and I get nothing for my time and effort. But I'll keep putting videos up as and when I can find the time.
Ref ringing - you are partly right in that my machine does not suffer from it. But it never has done - even before I added the load balancing gantry. I do know the reasons why and for sure I can cover that in a future video. Part of the reason might surprise you and I expect I'll get a lot of flack because much of the cure goes against what seems to be the consensus view of things.
What a fantastic and detailed speech we have with you here, so, thank you so much for generously sharing your knowledge.
Thank you!
That's about all the info you'd ever need when dealing with lead screws, amazing job! Wanted to argue first about term "lead" being more important than "pitch", but few thoughts quickly turned it into understanding of my misunderstanding :)
About choosing screw with low lead: I don't see big of a problem to choose layer height with relation to full step, I always do that with my 8 mm lead (so 0.04 increments for layer height and z-hop). Real models wise, only if you need to make a thin film or model with precise height (say exactly 30.1 mm).
Once again, great job!
Thank you , it is the first time I can say I understand it .
This is so well explained. i knew most of this but it is always good to refresh some concepts.
Verry nice and mostly complete explanation.
Two points:
1) As you mostly drive the z-Motor with microstepping (simply because the movement will be more quite) you can not be sure the endstop triggers at a full step. So z(0) might be at a 0.75 step and z(0.2) also should be. So changing from 8mm lead to 2mm lead does not make the layers using fullstep positions. So while the therory you show is verry common, it actually not complete. The real advantage of using 2mm lead is the precision of positioning by theroticaly by a factor of 16. If for example the holding tourque of a 1/4 on 7mm lead step would be just reliable enough holding z, but the "target position" is on a 3/8 microstep and the z would stay on 1/4 due to not enough tourqe for the last 1/8, be incorrect by 8mm/200*1/8 = 0.005mm. Which at 0.1mm layerhight is an error of 5%. The theory about "magical layerhight" is not to get "fullstep-layers" but to make the error constant (if its 5% of in one layer and the next layer have the same microstep position, it most likely will be the same 5% off). So exept for first layer the error will hopefully be the same everywhere. You do not need 2mm for that - why using 0.1 mm layerhight and not 0.08 or 0.12? Why 0.15 and not 0.16? But off course 1/4 step reliable holding z does not mean 1/8 step never holds it. Might stop, might not. So the 5% error might sometimes take effect. Expect of course you set the motors to 1/4 microstepping (but the vibrations...)
The real advantage of 2mm lead: 4 times the tourque means if with 8mm lead a 1/4 step reliable holds z, now a 1/16 steps holds reliable (theoreticly. Due to more tourqe-loss by friction maybe 1/8 step). Also a (micro)step on 2mm lead means only 1/4 movement compared to 8mm lead. So theoretical 16-times more precision (in real life about 8-times). So the error gets so low you can not see it anymore.
This is esp. usefull using ABL or vasemode, there you do not have a pattern of z-positions but z changing depending on x/y coordinates.
1/4 step as reliable holding tourqe of course only is an example, it totaly depends of the printers mechanic, wich microstep might reliable hold z.
2.) While you are totaly right about anti-backslash nuts on z are not needed as long you do not install your printer on the ISS, most people using them do not install an anti-backslash nut to avoid backslash. As on T8 screws the friction is low, some printers (esp. early Ender 5) thends to bed falling when powwer of. An anti-backslash nut provides more friction, pressing to booth sides of the tread. So most people install the anti-backslash nut to solve bed-falling-on-power-off problem...
You raise a couple of interesting points. I'm no expert on stepper motors so the point about them holding accurately between two full steps is one that I have asked many people, many times. Invariably, the answer is something like "yes/no/maybe" and always with caveats. The consensus that I have gleaned (which might not be correct), is that they would tend to "settle" at a full step - especially when "idle hold" kicks in and reduces the current. A similar question is "would a move command of one or two micro steps actually translate to any physical movement of the motor?" Again the answer is invariably "yes/no/maybe" depending on any number of factors. But I think we can agree that a 2mm lead screw will give 4 times better resolution than an 8 mm lead screw.
Also, using a remote motor driving the lead screw via a belt, with a (say) 20 tooth pulley on the motor and an 80 tooth pulley on the screw would improve both the resolution and torque requirement for an 8mm lead screw.
Regarding the "auto rotation" on power loss due to the steep helix angle of an 8mm lead screw, that is something that I meant to talk about but forgot. I hadn't considered that anti-backlash nuts would help in that situation so thanks for pointing that out.
@@deckingman It´s even more complicated. For example: 1/16 microstep alone certainly will not make a move at all. But if the motion direction is up and the mass of axis already rotating (as changing one layer even on 8mm lead means 2.5 fullsteps at 0.1mm layerhight) might have enough momentum to cary up. And maybe momentum takes it up to the desired position - and fricton-torque of the 1/16 step will hold it there. But going down (for example after z-hop), one microstep above the next fullstep certanly will not be able to stop movement - it will end on the next fullstep.
So as Isaid - magical layerhight does not make the layerchange on a fullstep position (with 1/16 microstepping the chance is 1:16 endstop randomly trigger on a full step), but making the error mostly constant.
You misunderstood my comment about using 2mm lead instead 8mm increases the torque (English is a foraign language to me, so a bit hard to explain). What I ment: As with one rotation you only go up 2mm, the torque you need to move up is only 1/4 as you need for 8mm. So if you only change the leadscrew but keep current and stepper, all steps - including microsteps - will result in 4 times the force upwards; this is equaly to a stepper with 4 times the torque on a 8mm leadscrew. So without extra gentry, by changeing from 8mm to 2mm lead, at the same time you get 4 times the resulution, but also the torque results in 4- times the force upwards. Now the tourqe of microstep has better chance to produce the force hold z. So if before changeing the leadscrew a 1/4 position can hold z, now the torque of a 1/16 (theoreticly) will hold a z position. It is not exact, as 1/16 microstep does not have one quarter of the torque of 1/4 step, but a bit less. So the 4-times more resulition and the 4 times force by same (low) torque of a microstep sums up to an even better resolution / less error in positioning. So the result is not only 4 times as good, but 16-times (theroretical, in reality about 8 times).
I already changed most of my printers to 2mm lead. Nearly one of the first mods. Only the mk3(s) with integrated leadscrew I did not make a change, as I found no source for integreted 2mm lead leadscrews. Two weeks ago I first such motors for a reasanable price (about 16$ with 320mm leadscrew with lead of 2mm). So I ordered two, to try it on one of my mk3(s).
Yes, as I saw the first mods with ant-backslash on z, I thought "why? gravety is always down..." until I understood they actually only try to increase friction. As I use POM-nuts anyway (which have more friction, but are more silent), I took some time to understand it...
Holding torque of a stepper motor is measured with one full step of error. So if your error budget is a 1/16th micro-step, then you only get 1/16th of the holding torque to use to move the load. This is where the drag in an lead-screw can bite you and re-introduce "backlash" in how the printer follows commanded moves.
There's also many more types of anti-backlash nuts than just the cheap split-nut with a spring. Kerk in particular makes fancy nuts that take up backlash with a spring driven screw mechanism so that drag torque can be kept low even with high reversed loads.
Thank you. I discovered that the lead screws I have on my Ender 3 V2 are too short to be fully supported at the top and at the coupler, so I was looking at getting some new screws, but I wasn't sure whether a single start T8x2 lead screw was what I needed or not (I'm currently running the stock 4 start T8x8 lead screws that came with the printer and the dual Z upgrade), but given that it means I can theoretically achieve sub 0.1mm layer heights with full step accuracy on an FDM printer, I'd say that about wraps it up with a nice little bow.
Very good explanation on lead screws for 3D printers.
Thanks. Hope you found it useful or helpful in some way.
@@deckingman I did. I'm doing a second mod on my Ender 5s for dual Z axis. The first was front to back, this one will have the lead screws on the sides. I found a good design on Thingiverse and am making some changes. I went with the T8 x2. Which I received yesterday, Thanks for reinforcing that I made the correct choice.
Thank you for excellent explanation and sharing your knowledge and regards from Europe to England!
Great sir
Amazing video. I have understood everything from the beginning to the end.
Thank you. Glad you found it informative.
@@deckingman If I can have a question. Is it harder to keep the smaller pitch screws running smooth? Because of dust having bigger effect on them for example?
Not really. It's the thread form that is more important than the lead when it comes to dust or debris. If you look again at the graphic for trapezoidal threads, you'll see that because the peaks are removed, there is clearance between the tip of the thread form and the root (the valley). So any dust or debris will tend to collect in this space, rather than cause the thread to "bind". Whereas, "normal" threaded rod where the thread form comes to a sharp point does not have this space, so it is more susceptible to "binding" if dust or debris is present. But the lead or pitch is largely unimportant in that respect.
Great video, I’d love to see one about belts too!
Thank you!
never knew there so much to it. mind blown! thanks for the tutorial.
Glad you found it helpful/useful.
Thank you so much
Glad you found it helpful.
Great video thank you very much , hope you keep making them
Fantastic presentation, very much appreciated. I do feel like a have a much better command of the subject.
smart man
So since screw with 4 starting thread have Highest content Area so dones it mean it produce less backlash than of screw with single face thread ? Does the choice between these two matter? If I want to make milling cnc machine?
Well it's not quite that simple. If you look at the pitch, rather than the lead, then the contact area will be roughly the same. That is to say, an 8mm, 4 start thread will have a lead of 8mm but because it has 4 threads, the pitch (i.e the distance from one thread root to the next adjacent thread) will be 2mm. Whereas a 2mm lead, single start screw also has a pitch of 2mm (which happens to be the same as the lead). So all else being equal, the backlash is likely to be the same although the steeper helix angle of a 4 start thread might play a small role in that. For a CNC machine, you'll likely need higher speeds than if you were using screw driven X and Y axes on a 3D printer, which are easier to accomplish with a 4 start screw. Just be aware that you'll need bigger motors than you could use with a single start thread because of the gearing effect. It's a trade off between speed, torque and to some extent positional accuracy and I can't make that decision for you.
26:07 This excellent explanation makes me realize that z-height baby-stepping can actually trample all the positional accuracy even of a 2 mm lead screw. Because if, as a result of z-height baby steps calibration, your calibrated reference / starting height coincides with a micro-step position, then ALL your layers will be at micro-step positions, which is less than ideal. The solution to this would be to actually switch to using full-step mode while calibrating nozzle height, which means that the last micro-adjustments need to be done on the build plate, not on the z-axis itself. I'm wondering why I don't see anybody talking about this problem...
I never thought of this. It does make sense. But I love using babysteping. We can set the amount of babysteping in the firmware. We can set it to one full step.
Thanks :)
Thanks for a reall good explanation !
i always thought that if you want to remove an error from a cheep screw drive you just need to get a long enough nut to average the inperfections
and softer spring loaded pair of nuts could get rid of the backlash even on a cheapest threaded rod but is see it works a bit different as you put it
then again when building by wierd small 3d printer i decided to go with ..an m8 rod for the z because 1,25mm lead could work nice with 200 steps motor and i could get 0,1mm each 16 steps without microstepping the motor and 16 works great with binary conversions of the controller, 0,05 would be 8 steps, 0,025 4 and i doubt we need thinner layers
so i wouldn't worry about using a hardware screw as a lead screw,
what i find out on the internet people upgrading the m6 or m5 rods to proper lead screws only after the oryginal noot wears out and using the same non hardened metal for both kind of make it the bigges drawback actually
getting 2 brass nuts could have saved them from premature screw wear and nobody ever remember that with "first diy 3d printers" where budget is the main constraint , 2nd being avalability of parts
Pls, talking about DIY CNC machine. Would you advise 8mm lead or 2mm lead screws on the x and y axis ?
I hesitate to answer that because I have no experience of using a DIY CNC machine. My "gut feel" is that feed rates are likely to be lower than a 3D printer (at least for cutting moves) but that precision needs to be at least as good if not better than a 3D printer. So on that basis, I'd have thought that the finer, 2mm lead would be the better choice. But I would suggest that you pose the question on a CNC forum or at least ask someone with more experience than I have.
@@deckingman thankyou very much sir. Appreciate that.
This incredibly detailed. Thank you. One question though:
- If I want to replace the 8mm with a 2mm lead screw, does that mean I have to change adapt the firmware as well?
- if yes, would you happen to have a guide on doing this or a link to a video, please?
The only thing you would have to do is change the steps per mm for that axis. With an 8mm lead screw, one revolution would give 8mm of linear movement and using 1.8 degree stepper motors, one revolution would also take 200 full steps. So we know that (200/8=) 25 full steps would give 1mm movement and most firmwares assume 16x micro-stepping so your steps per mm are probably (25 x 16 =) 400. With a 2mm lead screw, one revolution would give 2mm of movement so in that case you would now have 100 full steps per mm which at 16X micro-stepping would be 1,600. You may have different motors or gearing to that example but you can see that whatever steps per mm you have with an 8mm lead, would need to use four times higher for a 2mm lead. You might also want to change the speed and acceleration settings for that axis but that that depends on many factors. So try it and see what works best for you on your machine. HTH
Thank you, it was a lot of useful information, I was thinking make an actuator with 200mm range, but able to pull or push up to 20 kN. I thinking use normal threads rod M12. Is it totally wrong?
So what about if you added spring pressures for you x and y to help with the back lash play ???
That's one way. Anti-backlash nuts often use that technique. If you've considered all the factors but still want to use leads screws for X and Y, then you'll certainly need to use some type of anti--backlash mechanism.
I found your video when searching for a design guide to select lead screws for a 3D Printer. Thanks for the tutorial. Learned quite a lot. But I am a bit confused when you say "mil" referring to mm (Video time 26:11 onwards)
Isn't mil = one-thousandth of an inch, or 0.001 inch? One mil also equals 0.0254 mm (millimeter).
Could you please clarify. Thanks.
It's just an abbreviated way of saying millimetres. Just like we say "a couple of secs" instead of "a couple of seconds". When referring to fractions of a millimetre we might say "point two mill" instead of '"nought point two millimeters". We often write "mm" as an abbreviation for "millimetres" so saying "mil" instead of "millimetres" is just a verbal abbreviation of the written expression. It's a sloppy use of language for which I apologise but it makes perfect sense to native English speakers and/or students of engineering.
@@deckingmanThank you for the clarification. I was confused because as an Electronics Engineer, I always interpreted "mil" as 0.001 inch. I did not know that "mill" in colloquial speech.= "millimetre" abbreviated. Learned something and will know in future to avoid confusion.
To confuse things even more, before the UK went metric, we used feet and inches. A thousanth of an inch was referred to as a "thou" (pronounced like "how"). If we worked to tight tolerances, we'd use "a tenth of a thou" meaning a tenth of a thousanth of an inch. Or a hundredth of an inch might be ten thou. But we never used "mil" as a unit of measurement.
Very interesting, one question, why not going on the 1mm lead so 1 full circle is 1 mm of move? It should have less back lash and very precise, speed would be slower then 2mm but torque applied considering the same motor should be higher. It would really like to have your consideration. Probably resolution would be too high considering print layer size.
I used to have 1mm lead screws but changed them for 2mm lead because the Z axis speed was limited by the maximum step pulse frequency that the Duet 3 firmware was capable of at that time. 1mm lead gives 3200 steps per mm at 16X micro-stepping which is a lot of steps per second that the firmware has to generate. Z axis speed is not normally a problem because we usually only move a fraction of a mm during printing. However, I can print objects that are around 800 mm tall so to home the printer after making such an object meant that the Z axis had to travel the full 800mm. When the speed was limited to 300mm/minute(5mm/sec), that meant that it would take 2.6 minutes( 160 seconds) just to home the Z axis. Changing to 2mm lead screws doubled the speed so halved that time while still retaining sufficient resolution. Also the Z motor is more than capable motor of providing the required torque. 1mm lead will screws will give a resolution of 0.05mm per full step but we don't need the resolution to be that high and can't make use of it. So there for me, there was no benefit to offset the reduction is speed.
How about if I want to use lead screws on a Delta printer, which lead screw should I try first?
I don't have any "hands on" experience of Deltas so I'd be reluctant to comment but I'd have thought that you would need zero backlash, which might be difficult to accomplish.
So, if i use a ball screw instead of lead screw in my printer- i don`t see any difference in print quality?
Assuming you mean for the Z axis, then I'd be very surprised if you did see any improvement in print quality by changing lead screws to ball screws.
Thanks for the video! Always glad to learn. Would you say that a ball screw has no job in a 3d printer then? I see lots of designs with 3 ballscrews for the z axis leveling and considering these are really expensive compared to the leadscrews it seems like premature optimization.
IMO, ball screws are designed for lower friction - even lower than lead screws. As you rightly say, they are very expensive because the "nuts" have ball races inside which makes them expensive to produce. The threads are often ground too - which is also a costly process. I wouldn't go as far as to say that they have no place in 3D printers, but personally I don't believe that the additional cost can be justified.
@@deckingman Thanks for the reply!
@@deckingman how about in a homemade cnc router machine ?
@@Robert-zx2df Well, that's not my area of expertise so I'd be hesitant to advise. But the same principles apply. Importantly speed and back lash but also size and loads. CNC machines tend to be larger than 3D printers so if you plan to build the machine larger than say 1,000 mm in X or Y then you might find lead screws are hard to come by, and you'll likely need to use a diameter greater than 8mm to prevent "whipping". So that might force you down the belt route. On the other hand, CNC heads are heavy but more importantly, the cutting forces are much higher than the (near zero) force required to move a nozzle through air. So belt stretch might be an issue. The denser the materials and the bigger the cut, then the higher the forces which might cause belt stretch. So I'd say the decision would also depend on the type of material and size of cutter you plan to use. Screw driven would be a better choice for large cutters and thick, dense material due to the loads which might lead to belt stretch. But belts would be a better choice for taking small cuts and/or for cutting low density material where speed is more important and loads/forces are low. But as I said, that's just my best guess - suggest you take advice from a CNC router forum or user group. If it were me, I'd probably use belts but much wider than the 6mm we use with 3D printers. HTH
@@deckingman thank you for that indepth reply. I have no hesitation listening to your advice for cnc even though as you say it's not your expertise.
My concern with belts were torque. A belt drive is more or less 1:1 ratio. If you're driving a light load then thats that's fine like you say. I like how there is a gearing effect with screws giving more torque. I don't mind if my machine cuts slow, nothing wrong with a slow cutting machine as long as it gets the job done. I think it will still be way faster than me doing it by hand.
The kit I found only comes with nema 17 motors. The torque on these things doesn't sound very high so I can't imagine them going fast in belt drive. If I didnt care how fast my machine cuts , could Ialso keep gearing it to achieve the torque required or does my logic not apply to stepper motors ?
I'm trying to build this for cheap just to learn and have something I use once in awhile. I was looking at using basic all threaded rod. You say there is more friction in it but I'm hoping that doesn't make the machine unusable?
I'd want to make the table as large as I can but will probably be limited to 3' or 4' max in x/y. Used to cut wood , no metal. As cheap as possible , doesn't have to be fast.
Thanks
You forgot to talk about positional accuracy and tolerances. Are the tolerances of lead screws typically better than that of "threaded rods"?
I touched on it when I said that both types (lead screw and threaded rod) are made the same way using the same machines - just with different shaped cutters.
Usually, it's very difficult to find specifications for threaded rod. There are "armchair engineers" (the type who have never cut a thread and have no idea how it's done) that you will come across on forums and who will trawl the internet and find specifications and tolerances for lead screws and then compare that with a "generic" specification for threaded rod. From that they will deduce that all threaded rod is inferior to leads screws (even though a lead screw is by definition a threaded rod).
The way that threads are formed on machines (either screw cut or by using a die) means that any variability would be in the depth of the root or the height of the peak. But the critical dimension is the the flank to flank distance and that will always be the same - even if a cutter wears. That's not strictly true but the probability of seeing any significant flank to flank variability along the length of any threaded rod is up there with getting struck by lightening while standing on one leg.
In a nutshell. it's perfectly possible to find threaded rod which is more dimensionally accurate than a lead screw and vice versa but either one will produce consistent linear movement of a nut for any angular movement of the screw (because the flank to flank distance will be the same due to the way the thread is formed).
deckingman That’s good to know :D I tried trawling the internet for specs and only found the tolerances for the diameters as you mentioned. I did the misstake of assuming this diameter changed across one rod and estimated the difference in height from there, but you are correct that these tolerances are for the whole production line so a single rod should be very accurate :D
Do you know if there’s anything to be gained from having the rod in tension instead of compression? I’ve never seen a printer use rods which are under tension, but I feel like it would be better, maybe the whipping will be reduced? :s You could also get away with thinner threaded rods that way, but maybe M1 rods are harder to get than M3 rods, so it doesn’t make much sense to go thinner.
@@beaconofwierd1883 To put screws under tension would mean that they would need to be constrained at both ends. That's generally not a good idea because it can cause problems in getting them perfectly aligned with the linear guides. Screws should not be used to guide the bed or carriage or to constrain it - that's the job of linear guides or rails. For that reason, some amount of "float" is usually desirable and the easiest way to achieve that is to only constrain one end of the screw.
If by "under tension" you mean something like a screw supported at the top with a build platform "hanging" on a nut, it really makes no difference. I've seen printers built that way with the stepper motor at the top. The main practical disadvantage is that the entire weight of the bed is then transferred to the stepper motor bearings, and they generally are not designed for that sort of loading. I prefer to support the weight of the bed by using thrust bearings at the base of the screws. which are designed to take that sort of load.
deckingman Couldn’t you use thrust bearings in the same way when hanging? Just support the coupler with a trust bearing and have the stepper take no load, only apply torque?
@@deckingman Also, isn't the bearing in most stepper motors (for example Nema 17) equally good (or rather bad) at axial loads in either direction? Or is there a thrust bearing at the bottom?