Accuracy vs Precision. Before we get started, let's take a moment to look at the difference between accuracy and precision.
To steal a metaphor (from the image above), imagine targets at a shooting range.
Many Personal 3D Printers use threaded rod as a drive mechanism, usually just for the Z-axis, most of the rest use precision leadscrews. We've seen debate in multiple forums about which is better, and about how much of a difference it actually makes. We've even heard the argument that leadscrews are a waste of money, or that people using them in printer kits are just trying to extort more money from you. We'd like to contribute to this discussion and mention some of the reasons we feel leadscrews are a better choice (and worth the cost) for this application.
Both accuracy and precision matter in 3D Printing.
Another word that can be used for precision in this context is repeatability. This is the ability for the printer to reproduce the same movements again and again identically between layers or even between prints. The more repeatable the tool-path is in real space, the more precise it can be said to be.
Accuracy is not necessarily the ability to hit the same spot every time, but is rather the ability to get within an accepted level of precision of the place where you actually intend to be. Accuracy is the property that more closely determines accurate geometries and dimensions.
I'll apply these concepts below.
If you're interested in reading more about this distinction, please see this NOAA page on the topic. Though it is being applied to an entirely different field, the principles and terminology are essentially the same. This is also where I stole the above graphic from.
About Threaded Rod as a 3D Printer Drive MechanismThe biggest advantages threaded rod has over leadscrews are its price and its ubiquity/availability. You can walk into a hardware store and buy a sizeable length of threaded rod without breaking the bank. That's great, especially for a repstrap or a project that has an extremely tight budget or is time senesitive.
Let me stress that threaded rod DOES WORK for linear motion. There are many printers using it. In fact, the foundations of the Personal 3D Printer industry were built on printers that used threaded rod almost exclusively. Great results can be achieved with threaded rod. That said, it does have disadvantages.
Intended UseThreaded rod is designed for use in constructing things. It is meant to hold things together, to suspend things, or to run through things for internal strength. They are designed to have washers and nuts put on them and for those things to hold their place.
The goal of the design and manufacture of threaded rod is to create a product that you can easily thread a nut onto and get that nut into position and have it stay in position. In short, they are designed to bind when there is any (non-rotational) force applied to a nut (including in-axis force like the weight of a z-stage).
Tension vs CompressionThey are also designed to be used in tension (see the Wikipedia article). If a printer does use threaded rod for motion, the results will generally be better if the rods are suspended from the top of the printer (as in a common Prusa design), supporting weight in tension rather than if the motors are on the bottom (see the now outdated Mendel design) holding the rods up and supporting weight in compression. Keeping the rods in tension rather than compression will minimize bowing and bending, though a bent rod can still cause problems. While the deflection from compression may not be very visible, you can bet it's playing a role in your accuracy when your carriage is near the top. Many printer designers minimize the compression issue by using very large diameter smooth rods that are more resistant to bending, but a tensile arrangement with or without larger rods is a better solution.
Though not related to the linear motion, it's also worth noting that most RepRap designs that incorporate threaded rods utilize most of them in varying degrees of compression, rather than tension. This is a poor choice of material here, as these members can flex and bend; construction of this type will require much more constraint to make a solid frame, possibly leading to over-constraints that can cause issues. Metal extrusion (such as is used in the Mendel Max) is a much better choice for a framing element. Other newer designs, such as the Prusa i3 also avoid this problem by using a large plate of material instead of the rods.
StraightnessBecause of the usual applications of threaded rod, there are also no guarantees about the straightness of the rod to begin with. Poor handling in transportation and storage, or the manufacturing processes themselves may produce rods that have some natural bend to them. To some extent this is overcome by using solid, stable smooth rods or other precision linear motion mechanics to constrain x/y-axis motion and letting the un-driven end of the threaded rod float so it can only influence the z-axis motion, but the quality of that linear motion system (both components and design) will have a direct impact on how much or little of an impact the curvature of a threaded rod will have.
Overall QualityThe quality of threaded rod will also vary wildly, but in general won't be great as compared to leadscrews. For its standard applications, the tolerances of the rod only need to be good enough to reliably thread a nut of the proper size. There can sometimes be a fair degree of slop between the nuts and the rods. Rods of different sizes, or from different manufacturers will vary, and you can bet they're not optimizing their processes for consistency beyond what is sufficient for its intended application.
Smoothness and Surface FinishThe surface finish on a threaded rod is a combination of roughness from the treatment against corrosion, and burs that have only been removed so well as to make sure a nut will thread. Again, this comes back to the idea that they are designed to bind. Threaded rod should not be any smoother than is required to thread a nut; the roughness on the surface helps things to bind when non-rotational force is applied.
About Leadscrews as a 3D Printer Drive MechanismLeadscrews are designed for linear motion. They can certainly be costly, and as such may not be appropriate for every build, but they are a benefit to any printer you put them in.
Most of the points presented above about threaded rod are really raised in comparison to leadscrews. Leadscrews are made of strong enough materials to combat bowing and bending and can be used in tension or compression (within reasonable limits), are guaranteed to be straight (if handled appropriately and coming from a reputable supplier), are smooth, and overall are designed to keep from binding.
As an additional bonus, motion on leadscrews is also generally quieter than on threaded rod. It's worth noting that when employed for the z-axis motion backlash generally isn't an issue because of the force of gravity acting on the z-carriage and the fact that all movements are in one direction. This does benefit from gravity though, which this team undoubtedly took into consideration. In cases where you're performing fast z-movements (such as travel lifts, "dynamic z," or when using leadscrews to drive a delta bot), don't have the benefit of gravity, or when employing leadscrews for horizontal movements, an anti-backlash nut is a tremendous benefit, both to noise reduction and accuracy.
Lead, Pitch, and StartsIf you're designing a printer, or planning an upgrade for a printer you have and need to select a leadscrew setup, you need to be aware of the main properties you're going to run into.
The pitch of a leadscrew (or any screw) is the distance between adjacent threads. This might be expressed in distance, or by the inverse relationship of threads per unit distance. Some screws have multiple starts, which means they have multiple distinct threads side-by-side.
The lead is the distance that will be traveled in the course of one revolution of the leadscrew. In the case of a single-start screw, the lead is the same as the pitch.
10 threads per inch is a 1/10 inch pitch, which means one rotation takes you 1/10 inches
For a multi-start screw, the lead is the pitch multiplied by the number of starts.
For more information, see: Wikipedia: Screw Thread - Lead, Pitch, and Starts, and Wikipedia: Lead (Engineering).
In general, FFF/FDM printers use relatively infrequent, small, precise movements on the z-axis and consistent, fast movements on the x and y axes. A single start leadscrew with the tightest pitch possible (highest thread density, smallest pitch) is generally going to be your best bet for the z-axis, while you may or may not need something a little steeper to get the speeds you'd like from your x and y axes. While this may seem somewhat arbitrary given the precision of movement you can get from a stepper motor, an important factor to remember here is torque.
A more aggressive leadscrew will require more torque to drive. We have one kit printer we bought a couple years ago that has an overly aggressive multi-start leadscrew for the z-axis. The small motors included in the kit do not have the torque required to reliably start upward movement of the carriage, leaving it sitting there skipping steps until the carriage is given a little upward nudge to get it going (no, it's not a lubrication issue or a driver that needs turning up).
Application of Accuracy vs PrecisionBecause of the way they are made, and because of the requirements of their normal application, threaded rods should often be accurate enough for most applications. This is to say that the number of turns it takes to move a particular distance should be pretty close to correct. When people say decent accuracy is achievable using threaded rod, they're probably correct, generally speaking.
Threaded rod is not good, however, for precision. Because of the rough finish, the poor tolerances, potential slop between the rod and the nut, and other potential issues either in the rod itself or a printer design that utilizes the rods, you never know that each step on your motors is going to move your tool head by the precise distance you thought it would. The standard pitches on threaded rod are generally tight enough to make the accuracy high enough that the loss of precision may be acceptable, but for truly exceptional prints you'll likely find this to be a limitation.
A little extra movement in one step and a little less in the next average out to the right amount and give you overall accuracy over time, but you may wind up with subtle z-axis artifacts because layers are very slightly different heights. These artifacts will of course be much more pronounced at low layer heights (high resolutions). While you may be able to achieve successful prints at incredibly high resolutions with threaded rod, the surface finish at these resolutions will generally be much improved by the higher degree of precision achieved with a leadscrew.
Accuracy is important for getting the right overall shape and dimension of a print. Precision allows for smooth, straight edges, good tolerances on printed parts (particularly applicable when printing mating parts), and quality surface finish.
Other Options for Mechanical DrivesIt is possible to get your printer motion using a mechanism other than threaded rod or leadscrews. Some printers, especially delta bots, just use smooth rod with bearings (or rail systems), then drive motion with belts, wire, Syncromesh, or equivalent just like the X and Y axis on most 3D Printers are driven. An argument could be made that it's more difficult to get the level of precision and accuracy that you get from a leadscrew, but these options may be able to be comparable with excellent engineering and have an advantage in cost and speed.
April 30, 2014
Australian engineer Daniel Brown has been experimenting with overhangs, the bane of 3D printer operators worldwide. It looks like he’s managed to overcome them.
But first, what’s the deal with overhangs? They are geometric shapes in a 3D model that have no material underneath them, making layer-based 3D printing challenging. Most personal 3D printers must print “support” structures underneath to prevent the overhangs from immediately collapsing during printing.
Some overhangs are tolerable, though. You can often safely 3D print overhangs with as much as a 45 degree angle overhang without issue. More than that and you’re asking for trouble, which in this case means droopy filament strands.
Brown did an experiment to see how much overhang is really possible on a MakerBot Replicator. By progressively increasing the angle one could see where a failure occurs. However, it appears from the images above that he was able to tip the column to a mere ten degrees from horizontal.
The result is not quite perfect, as you can see some very slight blurbs on the bottom of the ten-degree column. But the result is very impressive nonetheless.
We suspect this extreme success has something to do with the overall geometry of the particular model being printed. Brown says:
What I think allowed this actually was the cross section of the part, that is a circular column, slanted therefore the maximum overhang only exists at the furthest point of the structure while it seemed to be held in place by the sides which comparatively we not overhanging at all.
Regardless of how it was accomplished, we’re definitely taking a much more aggressive stance on overhang angles from now on.