Most 3D printers use lead screws on at least one shaft. These are simple devices, essentially a steel thread and a brass nut moving on them. However, for maximum accuracy, you need to use a ball screw. These are usually very expensive, but have many advantages over lead screws. [MirageC] A cheaper ball screw was discovered, but it has certain limitations due to its low price. He designed a simple device to improve the performance of these cheap ball screws.

On the surface, the ball screw looks like a screw with strange threads. However, nuts are very different. The inside of the nut is a ball bearing, which can be installed in the groove to make the nut rotate with less friction. A special path collects the ball bearings and recirculates them to the other side of the nut. Generally speaking, ball screws are very durable, can withstand higher loads and higher speeds, and require less maintenance. Unlike lead screws, they are more expensive and are usually very rigid. However, they are also a bit noisy.

The rated accuracy of the ball screw is C0 to C10, of which C10 has the lowest accuracy, and the price will increase with accuracy-all the way up. [MirageC] shows how to replace precision grinding with a cheaper ball screw. These screws are cheaper and harder, but have more runout than precision screws.

This kind of jitter will cause oscillation during the 3D printing process, which is obvious on the print. Using the mechanic's dial indicator, [MirageC] found that the screw was not straight at all, and even the relatively poor C7 ball screw would be more accurate.

solution? Clever arrangement of 3D printed parts. Ball bearings and magnets. This device allows the nut to move laterally without having to transport it to the print bed. This is a clever design and seems to work well.

If you want to learn more about ball screws, we can help. Although the accompanying video is dead, Sanmei also has some very good information to share.

The ball screw is a cool solution for ultra-low loads, although I feel that for such a low load, a simple belt or cable drive can do the same thing at a lower cost.

Belts and cables have a series of problems of their own, even a cheap lead screw rarely solves them.

This is because the main movement—typically, steeper operations with 1.8° resolution—is reduced by the pitch of the lead screw, allowing (for example) 0.002" accuracy and 10tpi thread.

I think the point here is that the effective accuracy is limited. I don't know how to compare the slope between the belt and the lead screw, but I suspect that the slope of a cheap lead screw will not be less than 2 mils!

Usually the printer uses some kind of lead screw on the vertical axis, which allows you to use gravity as an anti-backlash measure to absorb the slope and provide more precise movement on the layer thickness axis (you need more). You might use ball screws more to reduce friction (so you can move faster) and get a _a bit_ less extreme gearing instead of gaining precision or accuracy.

I am also considering the problem of backlash, b/c I installed a ball screw on a new CNC milling machine in the basement. It is much better than a trapezoidal screw for the purpose: a lot of back and forth movement.

But on the Z axis of 3DP, you will only really rise, and rebound is almost irrelevant, so OP focuses on reliability and maintenance.

It is interesting to see that z-wobble still affects the printer. That was the cupcake curse (and Darwin!) back then, because it was completely fixed to the screw. Then there are generations of Mendelian and derivative products. People hope that the screws will not move and restrict them, leading to the "tug-of-war" mentioned in the video...

I like this solution with magnets and balls, and with additional magnets for torque. I am a little surprised that mechanical engineers have not yet optimized the solution for this problem: translation on two axes and constrained on the other four axes.

If there is any start, you can still get the same anti-clearance effect from the gravity on the ball screw. The nut on the ball screw in the video looks like an anti-backlash nut. His solution introduces a direction-dependent axial nut rotation, that is, backlash, which does not seem to be corrected by gravity.

The best Thomson screw has an accuracy between C7 and C8 when it is new, with a linearity of about 76 microns/300 mm. Their standard lead screw is 250um/300mm, or worse than C10. Another advantage of ball screws is that they can better maintain their initial accuracy. Despite this, there are still many old Monarch lathes that still manufacture precise parts on the old screw.

The reality is that we are discussing a 3D printer made of aluminum extrusions whose linear guides are not even properly bolted to the frame.

> I am a little surprised that mechanical engineers have not yet optimized the solution for this problem: translation on two axes, constrained on the other four axes.

They did it, the guy showed a version of it, but decided not to build it, but chose magnets and balls. This is a variant of Oldham coupling.

Facts have proved that belts can provide movement with a smaller slope/backlash compared to ball screws. Anyone who has studied building a CNC or 3D printer from scratch knows this. A properly tensioned belt can greatly exceed the accuracy of a ball screw. The problem with belt drives is that they usually increase the complexity of the drive system. Therefore, they are usually avoided when possible. Especially in DIY design.

You also confuse resolution with precision...

The pitch on the screw or ball screw is exactly the same as using two different-sized pulleys on the belt drive. In fact, many designs use this method to drive ball screws at different ratios to increase torque or improve resolution. But I want to emphasize that this has nothing to do with accuracy. I can get a resolution of 0.002 inches on a ball screw or belt drive. However, this does not mean that I have an accuracy of 0.002 inches. When you consider rebound, or when this hack tries to fix the "swing", your accuracy will be greatly reduced.

My (previous) Ender 3 Pro uses a belt-driven Z axis, and I can actually print at a finer resolution...more reliable than using a cheap lead screw. Only need gear reduction and step/mm adjustment.

The reason I went that way was because the printer was sucked by the lead screw. No matter what I do, I always have bands or offsets on the Z axis. I changed the lead screw (twice), I switched to Delrin nut, brass nut, delrin anti-backlash nut, brass anti-backlash nut, new bracket, everything.

The point is: In some applications, belts are far superior to cheap lead screws.

That is a very informative video! There is also an interesting project. Thanks for posting!

"This kind of jitter causes shaking during the 3D printing process, which is obvious on the print."

The only way this can happen is if he uses a transportation-grade ball screw (such as the one used for mobile home jacks) for motion control, no matter how much you pay, this is a bad idea. To be visible like this, the deviation must be a significant fraction of one millimeter, which is several orders of magnitude higher than the specifications of all motion control ball screw grades.

Looking at his method of measuring "run-out", he put the dial indicator on the part of the nut that may not be the reference surface-IE is not suitable for measuring anything. In addition, measuring the ball screw horizontally like this will only measure sagging, and moving the DTI along the nut like this will only measure the surface of the nut, not the screw. In order to measure the axial runout, he must support the screw vertically and rotate it, measuring the distance from the nut to the parallel surface.

I admire his efforts in reprocessing, but I disagree that the result is due to what he said... It is more likely to be the backlash of the transmission system or the problem of the drive motor, or even the position deviation caused by the warped screw installation. The question of volume.

After using them to build, I know that C7 rolled screws from excellent suppliers will far exceed the precision and accuracy of most acme screws, many of which are used to produce perfect 3D printing.

Although you are right, his measurement technique may bring errors to the measurement as shown in the picture, but the test he did shows that these very cheap ball screws come with orders of magnitude, which greatly exceeds his measurement. Overflow level in the fixture. Its image rotates vertically and oscillates vigorously (ie, it bends).

My biggest complaint is that if you want to use such magnets and don’t need maintenance, then it doesn’t make sense to stick them inside-the glue will fail one day, but in those hard-to-see places-his 3D The printing really should have a layer under the magnet (even a chamfered magnet), so the magnet will never loosen. I also think it might be the worst way to "fix" this bad swing I can imagine, the bending system is better to stay stiff in all the right directions.

I have seen a lot of lead screws that have a lot of beating, or bend in their center... It would be great if there is an adapter like this to connect with ordinary T8 screws.

There, I fixed it for you with cheap ACME screws. https://i.imgur.com/ZnipGdg.jpg

However, the solution is problematic. @ 17:24 I can clearly see that Z recoil has been introduced. The swinging wing reduces the fluctuation of the nut rotation, but it cannot completely stop it. You can see that the nut rotates upward in one direction, while rotating it downward the opposite. This means that during the transition from one direction to another, the lead screw is rotating, but the nut does not move up or down, but keeps the same Z position while rotating. The vertical movement starts only after the rotation of the nut stops. For many 3D printers, it may be acceptable in Z because layers are usually printed in only one direction. It will be interesting to know if the nut keeps rotating while the entire layer is printed, or if it drifts back to zero because of the magnet (which will cause each layer to grow over time). This may be no problem because the magnet can bear the weight of the bed and parts. But I will not use it on X or Y. A better solution is plates with radial ball grooves, because they allow movement in a plane while ensuring angular constraints, but who can resist 3D printing solutions that include magnets?

A better solution, because he has a dial indicator, put the screws in the brackets of the two wooden V-shaped blocks, and then use wooden mallets or rubber mallets to carefully tap them straight. I would never recommend this to chase one-tenth with expensive screws, but when you use one-thousandth (obviously!) cheap screws, you really have nothing to lose.

I also want to know why his linear guide does not restrict movement. In my opinion, the screws may be a bit big for the size of the rail, but they are not terrible. The guide rail is installed on an extruded aluminum channel steel, which has a poor flatness tolerance, but should provide sufficient rigidity for the guide rail. But I noticed that he bolted the rail into the channel every four holes. This is a major problem, especially when the screw is large and eccentric. The rail may be bent to conform to the screw. Adding more than 300% of the bolts will significantly increase its stiffness.

Some other observations: Please do not test the stiffness of the ball screw by hand like that. Precision screws from reputable manufacturers are usually checked for straightness and corrected as necessary. If you try, you will be able to bend them beyond the tolerance range. Besides, when he dropped one ball screw on the other, I cringed. You don't want your screws to clink.

I disagree with his pro/no evaluation that the ball screw is more rigid than the lead screw. He is actually saying, "I compared a thin screw with a thick ball screw. The rigidity of the ball screw is higher, so all ball screws have rigidity problems." For a given material, tempering and Diameter, they will be very close. Similarly, he should use EP2 grease instead of oil, the noise will be greatly reduced. (But, what do I expect from a video that includes WD40 as an example lubricant?). In addition, the ball screw does require maintenance. They need to inject a little grease into the nipple on the nut from time to time. The excess part will seep out, removing the tiny steel particles worn from the passages in the balls and nuts.

Compared with the ball screw, the only advantage of the screw is the cost and the greatly reduced reverse drive in vertical applications; the second is meaningless-brakes and weight compensation (counterweight or gas spring) are cheap there s solution. This actually boils down to accuracy and price, and the cost of maintaining these accuracy.

There are also errors in the return path animation. I don't know where it came from, but it is incorrect for that ball screw. The animation shows a single return of a single channel through multiple circuits. A simple visual inspection showed that the metal did not have enough depth to achieve this, and his nut contained 3 plastic plugs instead of two. Most single-channel nuts use an outer tube to return the ball. Processing the inner tube as shown in the animation is not easy. His nut contains 3 independent circuits, each of which rotates around the screw once before returning through a plug. The hole into which the plug is inserted is drilled so that the ball jumps out of the thread and returns to the other side of the thread as the screw rotates.

It also has an advantage-preloading. When using a single-channel nut, the balls can only be loaded on one side of the nut channel-they are pressed by the screw. When the screws reverse direction, they move to press to the opposite side. Since the ball must move, the channel must be slightly larger than the ball, and this space will cause a rebound. Double nuts (literally two nuts back-to-back) push each nut outward by compression, so that each of them will preload one side of the ball, thereby eliminating the gap. Although his nut has been preloaded-the thread in the intermediate circuit has a slight axial displacement relative to the other two threads. It is actually a triple nut. This is one of the reasons why the introduced nut rotates so badly. He started with a zero vertical rebound, and now has more.

It looks like I don’t like this video, but it’s great on many levels. The production is really clever, using animations, making video clips, etc., and explaining what a ball screw and a lead screw are, even if I object to some of his pro/op investigation results. Obviously, a lot of effort was put into making it. The fact that the video is a dubious solution to the wrong problem due to a lack of basic understanding (and bolts) does not detract from this.

In the past, rolled C5 ball screws could be purchased from the British company IBL. They were absorbed by Thomson. When I wanted to replace the worn-out D16mm 5tpi L18.5" ball screw with a nut, my quotation was close to US$2,400-excluding shipping, customs or taxes from Germany. The accuracy is expensive. Because Thomson is not in their The C5 ball screw is listed in the product list, so the replacement may be outdated. Needless to say, I swallowed, cleaned and reassembled the existing screw, and then purchased another one from eBay.

Some general rules of thumb: Assume that the rail is wet noodles until it is sufficiently bolted to a hard surface. Use all bolt holes. The extrusion will never be flat or straight. If you install a (nominal straight and flat) 12 mm thick rail on a bent 25 mm thick aluminum extrusion, will the rail stay straight? You may create miracles by smoothing the extruded aluminum mounting surface with several grades of dry and wet sandpaper on the granite countertop. Yes, if you do anodizing, it may not look so pretty, but hey. No amount of magnetic trickery can solve bad fundamentals. First, let your structure become rigid and square, and the installation surface is flat. Reduce at least one size of your ball screw. If you have a 20mm guide rail, assuming you need to handle cutting forces, then a 15/16mm ball screw is fine. Since 3D printing does not generate cutting forces, you do not need strong screws. Try using 12mm screws and 20mm rails. If you have a 12mm rail, try using 8mm screws. Make the ball screw consistent with the guide rail, not the other way around.

Some ball screw refurbishment companies will regrind them and then use the largest ball that fits. I have seen people claim that they can grind rolled screws and improve their accuracy to match the ground screws. The total cost of rolled screws plus grinding is lower than that of fully ground screws.

Some ball screws become smaller every other ball. This "spacer" ball is used to reduce friction, but it also reduces accuracy and load capacity. Replacing the spacer ball with a full-size spacer ball will increase load capacity and accuracy. The number of balls must usually be reduced by 1 or 2 to accommodate balls of all sizes.

Random participant: "These ball screws are very weak, let's strengthen them a bit!"

Depending on whether you are participating in a hackathon or a BDSM party, you may have completely different reactions.

I saw this a few weeks ago and wondered why he took this route to solve it.

Considering that he actually has a linear guide to constrain any movement that is not on the Z axis, why not let the ball screw be unconstrained at one end. Due to the bounce, the end can swing without applying any significant lateral force to the linear guide-except for vibrations that can be controlled by speed.

Do a trick done with a long screw. Install the screw under a certain tension to prevent it from sagging when it is level. This can also straighten slight bends. Tension can also prevent bending or sagging from high-speed swing.

In order to tighten the vertical ball screw with an aluminum extrusion frame, I will use some sturdy extrusions, such as 40×40 mm, and then machine a steel rocker to install on the top. Place the ball screw bearing seat on one side and a threaded rod on the other side, which is connected down to the base to apply tension. The goal is to apply enough force to straighten the ball screw without causing the aluminum extrusion support tower to bend.

It is radial runout-the change is in the direction of the screw radius.

Good timing and overview of article reviews. Again, why I am still hesitant to disassemble the printer carrier and belt drive for the 8mm SS pole. The current idea may stick to the original plan and try to use it in a multi-functional design...maybe linear bearings can be added to the shaft, whether it is ultimately a sled/carrier/slider/gantry. Will remember the design of the future, but larger drives load more.

Please be kind and respectful to help make the comment section great. (Comment Policy)

This website uses Akismet to reduce spam. Learn how to handle your comment data.

By using our website and services, you explicitly agree to the placement of our performance, functionality and advertising cookies. learn more