# Printing screws - is the output usable? (M3 or M4)

I just want to ask if anyone has successfully printed a screw (M3 or M4). Is the printed output usable as a screw? What printer is capable of printing screws? I am using an M3D printer - is there a configuration to successfully print a screw that is usable?

Can anyone share a picture of the best 3D printed screw?

• It's better to print a cylinder and then tap it. – FarO Oct 28 '20 at 9:46
• @FarO: Disagree strongly, especially if you're printing the axis in the XY plane. The zigzag walls and how the shifted versions of them in each layer interlock are far stronger than a smooth cylinder printed in any orientation, even before you start cutting into it with a tap. – R.. GitHub STOP HELPING ICE Nov 5 '20 at 15:04
• @R..GitHubSTOPHELPINGICE good point, but interlocking requires parts of the thread to be smaller than the average one, areas which will have a weaker resistance. So it will lock better, but the surface which will then hold will be less than the optimal one. Also, screws don't rely on interlocking but on tension when the nut is tightened. Or am I making a mistake somewhere? – FarO Nov 5 '20 at 15:54
• On the other hand, if you print a cylinder with axis in the XY plane then tap it, all of the outermost extrusions will be cut repeatedly at spacing matching the pitch, only attached by their bonding to the next wall inward and some small surface with the (also cut) walls above/below, and the tension from the nut tightened on them will be encouraging them to compress along and break off from the next wall inward. Simple inline ascii art: /\/\/\/ vs -=-=-=-= – R.. GitHub STOP HELPING ICE Nov 5 '20 at 17:09
• @FarO: I see. I think that's best handled (see My Tech Fun's video linked in my answer) by printing the threads with slightly negative clearance then using a tap aligned with the printed thread, so as to remove well under a whole extrusion wall at all points, achieving both the maximum contact surface area and ideal extrusion/layer structure. – R.. GitHub STOP HELPING ICE Nov 6 '20 at 14:54

well... it's hard to imagine printing M3 or even M4 I haven't try but I haven't because I'm pretty sure it's not possible (on my printer of course)

but some time ago I've tried M8 which is of course way from your needs it was printed on 0.1mm layer height

it went ok into the nut without any problems but the strength is not very high I suppose

I know the quality is poor but even such bad photo shows issues

• I'd say it's not just your printer; no FDM printer is capable of doing stuff as fine as M3 or M4. – Tom van der Zanden Dec 10 '16 at 20:58

Yes. This question and most of the answers are old, and 3D printing has come a long way in the past 4 years or so. Alexandre Strube's bumping with one up-to-date answer inspired me to take another try at it, and my first attempt was an overwhelming success.

• Bolt printed in horizontal orientation (with support), nut vertical.
• Wall line width reduced from 0.4 mm to 0.3 mm (still 0.4 mm nozzle) - most important for the bolt or the thread profile is not achievable.
• Layer height 0.12 for bolt, 0.08 for nut (to give nearly 10 layer per revolution)
• Both offset 0.045 mm from ideal thread profile (at M6-M8 this is about the maximum allowable by ISO standard; at M4 it's surely excess but extrusion roundness of FDM probably compensates entirely anyway).
• 100% infill, achieved with high wall line count rather than zigzag fill

The nut and bolt fit tightly but can be turned by hand (if your fingers are small enough) with some force.

These were printed in cheap PLA on an Ender 3 with entirely stock hardware, but upgraded controller board (SKR mini E3 V2.0) and firmware (presently using Klipper).

As for whether they're usable, My Tech Fun has a video testing the strength of 3D-printed bolts in different orientations. His tests were of M10, but the horizontally-printed one was able to hold over 140 kg in a pulling test, making it (especially with multiple bolts) suitable for serious applications.

Since the question mentioned M3, I went back later and figured I'd go even smaller and give M2.5 a try... and it worked! This is pretty surprising considering the pitch is just over the nozzle size, but with 0.2 mm extrusion widths and 0.08 mm layers it came out usable:

• That looks nowhere near as good as with a properly sized nozzle for the extrusion widths, I believe the "extrusion width < nozzle diameter --> inaccurate flow" still stands. – towe Dec 4 '20 at 7:10
• @towe: which part are you talking about? I had an error in priming setup when I did these that messed up the heads a bit. – R.. GitHub STOP HELPING ICE Dec 11 '20 at 4:16

I just did.

I printed a 4 screws at .2mm and 20% infill. surprisingly, all screws worked with a plastic m4 nut I had from china (not 3d printed). The project itself 1 tells to print at 100% and 0.1.

When I started screwing the 4th one, I pushed the tool badly, and the screw head popped off.

After reading this thread, and experiencing what I just said above about the screw's head, I'm printing one more, this time at the maximum resolution and maximum infill. I'm also printing a nut to compare to this.

Even the ones with plastic flying around them worked well (see picture).

Ps: apparently the person can't make hexagons. I will redo the heads in proper hexagons, as well as the little wrench when I have time.

• It looks like you were bit by Cura's bad default Maximum Retraction Count and Minimum Extrusion Distance Window causing skipped retractions. Setting the window to 0 fixes it. – R.. GitHub STOP HELPING ICE Oct 26 '20 at 23:45
• At 0.2 mm layer height it's really surprising these work at all. Assuming M4x0.7, that's only 3.5 layers per revolution of the thread, and the thread is almost certainly wide enough (just due to layer height) to be out-of-spec. Normally you want at least 5-6 and preferably 8-10 layers per revolution (e.g. at most 0.2 and preferably 0.12 for M8x1.25). If you're getting by with less it's probably a matter of non-rigidity of the bolt and nut; with a metal nut, threading it on might actually cut a fix in your print. However.. – R.. GitHub STOP HELPING ICE Oct 26 '20 at 23:51
• ..it's nice that you bumped this question! Things have changed a lot since 2016, and it's nice to see how something everyone thought was impossible then is actually doable now, even on a \$200 printer. When I saw this bumped, before reading your answer, I took it as a challenge to print some, and I might still do that. – R.. GitHub STOP HELPING ICE Oct 26 '20 at 23:53

I've had success printing pins at that size, but not screws. At that size the pins were also exciedingly suceptible to stress, ie easy to break. At that scale the tolerances are a bit tight when printing that small and there will be small variances in precision with most printers. The best example of this is to try to print a lego piece, they dont snap together, some are slightly to big, and some are slightly to small and sometimes they will break.

Its just not possible on a FDM machine. for example using a layer thickness of 0.1 which is standard there would only be 5 layers of material between each peak (the pitch) for a coarse M3 and a fine M3 would not even have four layers (wikipedia on ISO thread) which is just not enough layers to resolve the required shape. The tensile strength of such a bolt even if you could print it would be abysmal.

Not very likely unfortunately.

M3D nozzle diameter is too wide to be able to print something that would print reliably, not to mention any FDM machine won't have the precision to build good threads. even 0.05mm layer heights won't run very well.

I've tried on a Form1 (SLA style printer) and even that set at it's max resolution wasn't able to print clean threads.

Vertically printed threaded bolts generally work better than you might expect, because (1) the layers are nearly parallel to the direction in which they have to slide, so the bolt can turn smoothly, and (2) when a typical bolt is sliced, the resulting layers are just circles, which are about the least problematic shape you could print.

Also note that, especially if you're printing a bolt to use with a metal nut (or vice versa), then as long as the thread is close enough, it will end up being crushed / sheared into the right profile; it's a bit like how you can drive a screw into timber without having to tap a thread.

In principle, the pitch of a printed bolt could be very fine; as long as the layer height is less than half the pitch (i.e. the Nyquist limit), and the spacing is correct, then the threads will engage. So if you can print 0.1mm layers, you have enough detail for M1 or smaller. Of course you're not going to print a bolt that small, but the point is, the limit is not resolution, it's the strength of the print.

The horizontally-printed bolt in @R.. GitHub STOP HELPING ICE's answer is impressive – if you can print that, then that's the answer. But even if you can't, you could definitely print a horizontal 3mm diameter cylinder, and then use a die to cut the thread on the finished part. It's not a "pure" printed solution, but you can get a cheap tap-and-die set from any auto parts store for less than the cost of a spool of filament.