Printed gears, and the tooth that always went first

Where a printed gear fails, and why it is always the same tooth.

A printed gear fails at the same tooth every time because a filament printer restarts and ends every single layer's outline at roughly the same angular position around the gear, leaving a small vertical seam of slightly weaker, less continuous material running straight up through one specific point on the finished part, and whichever tooth happens to sit at that seam is quietly weaker than every other tooth on the gear before it ever meshes with anything.

Why the same tooth always failed

A filament printer tracing a gear's circular outline layer by layer has to start and stop extruding somewhere on each layer, and most slicing software defaults to placing that start-and-stop point at the same angular location on every single layer, purely because that is the simplest, most predictable way to programme the toolpath. Each layer's start-stop point leaves a tiny discontinuity, a small blob or a slight gap where the extrusion path rejoins itself, and stacking hundreds of these individually tiny discontinuities directly on top of one another builds a real, continuous seam running the full height of the part at exactly one angular position, while every other point around the gear's circumference benefits from a clean, continuous path with no equivalent weakness at all.

The rolled-paper-tube comparison

A cylindrical tube rolled from a flat sheet of paper and glued along one seam is, apart from that one seam, a smooth, continuous, remarkably strong surface, since the paper's own fibres run unbroken everywhere except along that single glued join. Press on the tube at any point away from the seam and it resists firmly, but press directly on the seam itself and it gives noticeably more easily, since the glued join is never quite as strong as the continuous paper surrounding it, a difference invisible until someone actually goes looking specifically at that one line running the tube's length. A printed gear's layer seam behaves exactly the same way, invisible in a casual glance at the finished part and consistently the first place to give under repeated load, precisely because it is the one line on the whole gear where the material was never truly continuous to begin with.

Why this problem hides until the gear is actually loaded

A gear with a seam running through one tooth measures correctly on every dimension a caliper can check, since the seam is a difference in the material's internal continuity rather than in its outer shape, which means the flaw is invisible to any inspection that only checks size rather than the tooth's actual strength under repeated load. Only once the gear is put to work, meshing thousands of times under real force, does the seam's weakness actually show itself, and by then it shows itself in exactly the same place every single time a given gear design is printed with the same default seam placement, since the seam's angular position is set by the slicer's default behaviour rather than by anything random about any individual print.

This is a genuinely different category of failure from the tooth-root fatigue covered earlier in this set for resin gears, since fatigue at a tooth root is a property of the material and the load acting on an otherwise uniform part, while a seam failure is a property of a specific, repeatable manufacturing choice that happens to have placed a weak line exactly where a load-bearing feature needed to be strong. Two gears cut from the same nominal design can therefore fail for entirely different reasons depending purely on which process made them, one from the material's own inherent limits, the other from where a slicer's software happened to draw a line nobody was watching.

The number that matters here

A gear tooth sitting directly on a printed seam can fail at a noticeably smaller number of meshing cycles than an identical tooth positioned anywhere else on the same gear, a gap large enough that printing the identical design several times and testing each one to failure reliably points to the same tooth position failing first every time, rather than the failures scattering randomly around the gear's circumference the way a genuinely uniform material would produce.

What this changes in practice

The practical fix is almost embarrassingly simple once the cause is actually understood, rotating the gear's design so the seam lands between two teeth in the gap where no load is concentrated, rather than directly through the middle of one, or configuring the slicer to randomise the seam's position from layer to layer so no single point ever accumulates the full stacked weakness in the first place. Either fix costs nothing in material or print time, which makes this a rare case where understanding the actual mechanism converts an apparently mysterious, intermittent-looking failure into a design decision that can be corrected permanently rather than merely worked around after the fact. Checking where a slicer intends to place its seam, before a single load-bearing part is ever printed, is a habit worth adding to any workflow producing repeated printed parts that will actually see real cyclic load in service.

What this does not explain

Not every printed gear failure traces back to a seam, since a tooth can just as easily fail from ordinary fatigue at its root, from being printed in an unfavourable layer orientation relative to the load, or from a material simply unsuited to repeated impact, all covered elsewhere in this era. The seam is worth checking specifically when a gear fails at the same tooth reliably across multiple identical prints, since a failure that instead scatters randomly around the gear's circumference is very unlikely to be a seam problem at all.

My key error with this

I printed gears for considerably longer than I should have, because a printed gear is enormously convenient, it arrives overnight, it costs almost nothing, and the first one always meshes well enough to look like the problem is solved. The teeth are where it comes apart, since a printed tooth is built from layers running across the exact direction the tooth is loaded in, and it fails there repeatedly and predictably no matter how the print is tuned. I kept adjusting parameters, changing orientation and thickening the tooth profile, each iteration costing another print cycle and another partial failure, when the real answer was that the process and the part were mismatched from the start. What replaced the belief is that if a plastic gear genuinely does the job, it should be cut from sheet where the material is continuous through the tooth, and that if a printer must be involved then the printed part should be carrying something other than gear teeth. I have not printed a gear since, and I would not recommend it to anybody.