17
$\begingroup$

Whenever a 3D printer that uses linear rails is announced (case in point: the cetus), the Internet (well... at least that corner of it dealing with 3D printing) gets abuzz with excitement.

I researched a bit the topic myself, and while I understand that linear rails can be produced to a fantastic degree of precision for super-heavy machinery, it escapes my comprehension why they are considered superior "by default", relative to the classic linear bearings on a shaft.

3D printing is a lightweight application, and the motion of at least 2 axis does not happen against a solid surface (where you could bolt a linear rail every few cm) but suspended between the 2 ends of the axis. Furthermore, the internals of the bearings used on linear rails are substantially identical to those used on a shaft.

The cetus site says under the heading "Quality Linear Rails":

Self-lubricated | Maintenance Free | High Precision | Long Lifespan | Quiet

but this - in my experience - can be said of "Quality Linear Bearings on a Shaft" as well, and in some cases even bushing deliver to a high standard on 3D printers.

So, what am I missing?

$\endgroup$
  • $\begingroup$ Looks nice, easy to think about when constructing or assessing because linear bearings block all degrees of freedom except one. The linked one must be slow as heck but it's also in the ballpark for what people could spend kind of easily (~$400 with tax & shipping I guess. Could be okay too if it works okay and doesn't get all wobbly after some use or electronics fry or... there is definitely other things to a printer than the linear bearings :-) $\endgroup$ – Valmond Jan 8 '18 at 11:57
  • $\begingroup$ ps. what is it in this community, does nobody like ever upvote? $\endgroup$ – Valmond Jan 8 '18 at 11:58
  • $\begingroup$ It's probably down to the fact that they are expensive, and also the snobbery associated with over-engineered products. $\endgroup$ – Mick Jan 8 '18 at 16:09
  • $\begingroup$ @Valmond - I'm guilty of forgetting to upvote questions myself, a few days back I actually thanked in my reply the OP for a nicely formulated question and forgot to upvote it! I don't take it personally. My question was not about the cetus specifically, but yes... for the price it apparently is a fantastic machine according to various reviews. $\endgroup$ – mac Jan 8 '18 at 22:13
14
$\begingroup$

The following is a compilation of the input from a number of sources.

Linear rails in general are mechanical components that - when designing equipment - offer great flexibility.

The profile of the rail can be designed in nearly infinite ways. This in turn allows for:

  • Different levels of stiffness in different directions (for example you may have stresses only on a given plane, or you may actually want the rail to slightly flex in one plane but not in another one).
  • Placing the surfaces for the rollers strategically, for example in a location that is unlikely to get contaminated, or where the maximum force will be applied.
  • Curved paths, so that the carriage can move along a line that is not straight.

Because the contact surface between the rollers and the bearings is flat, cylinders can be used instead of spheres. This in turns diminishes the mechanical stresses, and the amount of play, increases longevity and allows for more bearing capacity, among others.

Linear rails can be anchored along their full length, rather that at their extremes, thus increasing the accuracy of their positioning, their stiffness and their bearing capacity.

Linear rails can be machined while pre-loaded, thus achieving maximum accuracy when in use, rather than when coming out of the factory.

The bearings on a linear rail only allow for one degree of movement. There need to be two rods with linear bearings/bushes to achieve the same result.

All that said, when it comes to the specific application of consumer-grade FDM 3D printers, it seems that none of the above is very relevant, nor confers any real advantage to the printer in terms of quality of the final print:

  • the mechanical stresses involved in 3D printing are very small,
  • the movements all happen along straight lines,
  • most of the axis cannot be anchored to a large, rigid body,
  • ...

On the other hand, the design with rods + linear bearings is cheap, effective, simpler and lightweight, all characteristics that are highly desirable in a 3D printer.

All in all, it seems that there is no good reason to prefer linear rails over rods in general.

Still, there may be specific designs that may benefit from their adoption. I postulate that the Cetus printer linked in the question may be such a design: the cantilever arrangement of its axis - for example - is well served by the fact that a single rail locks movement in all but one direction, and the orientation of the X rail offers maximum rigidity against the action of gravity.

| improve this answer | |
$\endgroup$
  • $\begingroup$ The dynamic stresses are not that small... $\endgroup$ – Sean Houlihane Jan 13 '18 at 10:09
  • 1
    $\begingroup$ They are if compared to other CNC machines that typically use linear rails (like CNC cutters and mills). Or even compared to the humble lathe. But sure, among all mechanical stresses a 3D printer has to withstand, those are probably the major ones and by a stretch! :) $\endgroup$ – mac Jan 13 '18 at 10:21
-1
$\begingroup$

Linear rails will always produce a high degree of accuracy and stability and more so than round rod with PTFE bushings and/or bearings.

One may argue the fact however even as a product developer and one who is involved with the mechanicals and development of machinery on a day to day basis comparing the two we see “significant” improvements using rails over rods and if properly utilized on a Z axis print bed you will have a bed free of leveling issues that can smoothly and accurately operate with one driver versus two.

I will further add that getting rods aligned perfectly is a difficult task for the average person and even a slight twist or angular position can affect final print quality. I’ve seen many linear rods that appear visually straight and when chucked into a lathe spindle with 0.0000 accuracy there will be 0.005 or more in runout. In fact I’ve yet to see one perfecting concentric motion that is longer than 6 inches. This tells me that they cannot plausibly be as accurate and that while they may function they will never function with a high degree of accuracy.

Do we need higher accuracy in 3D printer axis? Sure we can have quality control boards that compensate to some degree however the mechanics of the machine are utmost important before you choose the quality of board and software. Why install a \$300 motion control on a cheap linear rod printer if you’re not going to see the full benefits?

With technology further advancing into 64 bit and eventually 128 bit and higher degrees of precisional accuracy 3D printing is turning a page and will if not already be capable of micro resolutional accuracy and can only do so if all the components function properly together.

So sure, your rod guided printer works. However, it will never work as well as my linear rail guided printer with ballscrews and servos. You can have your layered fuzzy prints. I will keep my smooth finished injection molded looking parts that are made from materials a typical desktop cannot even print. So to argue it’s not needed is arguing that high quality isn’t accepted in a lower price point market.

One other addition here. Ask yourself how level and square is your printer? I’m not talking about using a carpenter's level for checking your machine they can be inaccurate up to a 1/4” per 10 feet. When you can dial your printers bed down to 0.00005 or less every direction and your structure is just as accurate than you know what a quality printer and print looks like.

I guarantee no printer priced \$300-\$1000 comes even close to that degree of accuracy. The average consumer is so drawn into the technology of a final print itself they overlook the precisional aspects involved and learn to settle for less. Then you wouldn’t expect your \$500 printer to compete with my \$10,000 printer either.

Bottom line you get what you pay for.

| improve this answer | |
$\endgroup$
  • 7
    $\begingroup$ would you mind to add a few paragraphs to add readability? $\endgroup$ – Trish Sep 12 '18 at 15:54
  • 2
    $\begingroup$ With technology further advancing into 64 bit and eventually 128 bit and higher degrees of precisinal accuracy -- what does 64 or 128 bit accuracy mean? How do "bits" play into this? One $2^{64}$th of the distance to the sun is 8 nanometers; 64 "bit" (let alone 128 "bit") accuracy isn't anywhere near achievable, how does it relate to the choice between linear rails and smooth rods? $\endgroup$ – Tom van der Zanden Sep 16 '18 at 12:12
  • 3
    $\begingroup$ When you can dial your printers bed down to .00005 or less every direction and your structure is just as accurate than you know what a quality printer and print looks like Nah... when you have dialled to .00005mm, you have most likely wasted time and money to something irrelevant, given that you will never print anything with that level of resolution. :) $\endgroup$ – mac Sep 23 '18 at 21:47
  • $\begingroup$ Bits = Steps = Accuracy .. 1st boards were 8 bit..now 16 is norm Stepper motors use micro steps 16 32....250 ( 16 is the norm on a -1000 printer Salmon skin, threads,fuzz, or problems are caused by tiny problems amplified due to repetition.. like how a 2$ seemingly innocuous plastic tube change can dramatically improve a print many fold(ptfe tube). Pardon grammar, my pad is moving and I am eating.. cheers $\endgroup$ – Lolipopssuck Jul 29 '19 at 12:46
  • $\begingroup$ -1 As this is unreadable $\endgroup$ – Greenonline Aug 31 '19 at 18:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.