Many 3D printers employ a GT2 band for the y-axis plate and the printhead in the x-axis carriage.

However, there seems to be the obvious disadvantages of:

  • stretching (and need for adjustment) and;
  • possibly slight inaccuracy, or "wobble" (for want of a better word (as wobble is often used in describing z-axis deviations)), due to the elasticity of the band.

Is there any particular reason why GT2 is used over a straight forward rack and pinion system?

Rack and pinion

In particular, for the y-axis plate, as a rack would appear to me, be:

  • easy to install;
  • more accurate, and;
  • require a lot less adjustment1.

If the rack is made from aluminium, surely weight can not be a major factor... or can it?

Is cost a factor? A pinion seems to cost around \$10-\$15 (650 B\$, here in BKK, Thailand), which is obviously more than a reel of GT2 and a couple of GT2 pulleys.

Would the rigidity of the rack be less forgiving of an imperfectly aligned axis, something which the GT2 band and pulleys combination would not be so affected by?


Which of these factors cause designers to use GT2 en lieu of a rack?

  • Accuracy
  • Weight
  • Cost
  • Maintenance
  • Tolerence of non-square axes, manufacturing errors.

1 Maybe constant adjustment is not required on an everyday basis, but the tension would still need to be checked now and again, whereas a pinion would not have this requirement.


With a belt system, the belt engages roughly half the pulley. This, and the tension in the belt, ensures the belt always engages the pulley tightly. A belt and pulley system is thus relatively forgiving.

With a rack and pinion system, only a few teeth engage at any given time. To avoid backlash and get the same kind of "tight" engagement, both the gear and the rack need to be made with very high precision. The carriage also needs to be very well constrained, because any wobble of the rack relative to the gear introduces backlash (or binding). Moreover, you also need to keep the rack and pinion well lubricated lest they wear out prematurely.

Given that belt and pulley work well enough, I don't see why you'd need to move to rack and pinion. The main advantage of rack and pinion is that the rack isn't elastic. The maximum length of a belt system is limited by its elasticity, but given we aren't building meter-long 3D printers anyway, that advantage of rack and pinion does not apply.

  • 1
    $\begingroup$ Nice explanation, thanks. So it is similar situation to my query about the z-axis threaded rods being constrained at the top - the more rigid the driving mechanism the more backlash comes into play. Which leads me to wonder whether the use of a GT2 band, instead of a threaded rod, would reduce wobble and banding in the z-axis... $\endgroup$
    – Greenonline
    Feb 5 '17 at 14:13
  • $\begingroup$ I know this is old, but it is also factually wrong. There is NOTHING that prohibits a pinion based system to use multiple gears to connect. And there are PLENTY of examples (as in: commercial products) that demonstrate this system with high precision, including commercial 3d printers. Also, it all depends on "buy quality components, maybe have the whole setup machines well", which is not "not possible". And finally, it has the advantage of a fixed carriage weight and fixed known compression, while this is a variable for a classical belt system where longer printers mean a WAY longer belt. $\endgroup$
    – TomTom
    Jul 21 '20 at 18:32
  • $\begingroup$ @TomTom I never said it was "not possible", which is something you put in quotation marks as if I said it. The answer doesn't even imply that; it is simply stating some reasons why the vast majority of printers use belts (and not racks). It would be interesting if you provided some examples of commercial 3D printers using rack and pinions so we can see how they do it. $\endgroup$ Jul 22 '20 at 6:30
  • $\begingroup$ Thre is no need for this as I would assume anyone interested is smart enough to use google. And check CNC machines - QUITE common. MORE concerning is that you miss the obvious alternative - multi belt systems, either as double belt or as belt drive. Those combine pretty much both advantages, including WAY lower stretch on the belt (as the receriving belt on the machine CAN be a rack, OR can be GLUED to the machine over the whole length. The moving belt part then is ABSURDLY small and CONSTANT (i.e. 15cm, regardless of length of movement). $\endgroup$
    – TomTom
    Jul 22 '20 at 7:53
  • 1
    $\begingroup$ @Svartalf The question is why "many 3D printers employ a GT2 band [...]". Very large 3D printers were and still are very uncommon, and so are 3D printers that use rack-and-pinion in place of belts. Moreover, the answer doesn't claim that belts are "superior", but that it is more difficult to make them (and thus more expensive) for relatively little advantage. This explains why we see belts on many 3D printers (considering that most 3D printers are $300 consumer-grade ones). Your example of a (presumably) very expensive industrial machine only confirms this. $\endgroup$ Aug 16 at 8:30

The answer so others can understand is that it's cheap and simple to run with a belt. Much of the 3D printers these days either stem from the sintering systems that were developed years ago and things that rose out of the RepRap community where a Lead Screw that had the precision needed to do proper 3D printing at decent speeds was out of reach for most of the community.

It's a bit of, "we've always used belts," and then people come up with all sorts of stories to validate the dubious take to begin with.

At one point in time (to be specific...around a couple of years BEFORE the question was asked here...) the costs of a lead/belt-screw system for anything other than a Z axis solution for a Cartesian or a Core-XY was prohibitive for a hobbyist or a system bought off shelf or build by them.

It was too hard and expensive to get straight enough parts with precision to actually make anything other than small toy systems with those. We wouldn't get into racks as they were even MORE expensive. With the precisions we were previously working on and with in the community at the time this question was originally posed, it was something that only made sense for select solutions.

As you scale larger (Hey, we're now professionally making HUGE parts with this stuff) it makes much, much more sense. For certain classes of manufacturing...it actually makes sense with certain filaments to make short run parts for almost ANYTHING, including things in the automotive space. At that juncture, prototyping, fixing "impossible" to repair situations on older cars, etc. makes some sense to get more precision or vastly larger size. Desireable becomes the ability to make printers with volumes up to 2 meters in size. Professional space? Maybe. But telling people that a pinion isn't making any sense? Heh...hardly. Same goes for lead/ball-screws. Several have made incredible printers fairly cheaply that can print seemingly impossible print runs (Hung at a 45 degree angle in mid air and print...) with some of this stuff. Is it cheaper? No. Does it make sense? Possibly. Especially if one wants to DIY a massive print volume printer (waves hand). I want a meter and a half cubed on a side print volume that I can enclose to crank out ASA prints of Smart body panels, for example... You're not getting that with belts- too stretchy, even with PU steel core belts.

Frame it in from a relative cost perspective (It's only going to be about 25% more costly on a DIY or commercial product with a speed upside and precision that might make it WORTH that...but understand that an Ender 3 designed that way will run 350 dollars instead of the 200 or thereabouts it currently runs) and you're going to be more realistic and honest on the question's answer. Saying it isn't practical...even from when the post originated isn't being exactly forthright in this day and age- and disregards that a person MIGHT just be willing to spend the more money for the precision, etc.


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.