You want to bolt something to a printed part and have it hold up to being taken apart and put back together. So you print a hole, drive a screw straight into the plastic, and it works. Once. The second or third time you back that screw out, the thread strips, the plastic crumbles, and now you have a part with a useless hole in it. Printed plastic threads are fine for a one-time assembly and terrible for anything you will service.
Brass heat-set inserts fix this, and they are one of the highest-payoff upgrades a functional print can get. But they only work if the hole is the right size and the wall around it has enough meat to survive installation. Get the hole 0.3 mm wrong and the insert either spins in place or splits the boss the moment it goes in. This post covers the pilot hole diameters that actually work, how to size the boss, when an insert beats the alternatives, and how to correct a bad fit without remodeling the part.
What is a heat-set insert, and why use one?
A heat-set insert is a small knurled brass sleeve with a machine thread on the inside. You push it into a printed pilot hole with a heated soldering iron tip, the brass melts the surrounding plastic just enough to sink in, and as the plastic cools it flows into the knurls and locks the insert in place. What you are left with is a real metal thread inside your plastic part.
The reason to bother is durability. A screw threaded directly into PLA or PETG chews up a little plastic every time it goes in and out, so a joint you will open more than a couple of times wears out fast. A brass insert gives you a metal-on-metal thread that handles repeated assembly, takes far more torque before stripping, and spreads clamping load into the plastic through the knurls instead of concentrating it on a few printed thread crests. If a part has a lid, a battery door, a bracket you will unbolt, or anything a user will open, that is where an insert earns its keep.
What size hole do you need for a heat-set insert?
The pilot hole has to sit in a narrow window: bigger than the insert's smooth leading tip so it starts straight, but smaller than the knurled outer diameter so there is plastic for the knurls to bite into and displace. Too big and the insert spins with no grip. Too small and the brass shoves too much molten plastic ahead of it, cracking the boss or leaving the insert standing proud.
There is no single universal number, because insert brands differ in knurl geometry and outer diameter. Always check your insert's datasheet first. That said, the common tapered brass inserts most makers buy (CNC Kitchen and equivalent M-series inserts) cluster around these starting-point hole diameters in PLA and PETG:
| Thread size | Pilot hole diameter | Typical insert OD | |---|---|---| | M2 | 3.0 to 3.2 mm | ~3.2 to 3.5 mm | | M2.5 | 3.4 to 3.6 mm | ~3.5 to 4.0 mm | | M3 | 4.0 mm (start here) | ~4.0 to 4.6 mm | | M4 | 5.6 mm | ~5.6 to 6.3 mm | | M5 | 6.4 mm | ~6.4 to 7.1 mm |
M3 is the size you will use most, and 4.0 mm is the number worth memorizing. Model the hole at nominal and let installation heat do the rest. One important wrinkle: FDM prints holes slightly undersized because the nozzle over-extrudes on tight curves, so a hole you model at 4.0 mm often comes off the bed closer to 3.8 mm. That works in your favor here (a touch of extra grip), but it is exactly why the test-print step later matters, and it is the same hole-shrink effect covered in our guide to 3D printing tolerances and making parts that fit.
How thick should the boss around the insert be?
Sizing the hole is only half the job. The insert expands the plastic outward as it goes in, so the boss (the raised cylinder of plastic the insert lives in) has to resist that outward push without splitting.
Two rules keep a boss out of trouble:
- Wall thickness: at least 1.6 mm of plastic all the way around the hole, and ideally closer to twice the insert's outer diameter for a boss that takes real load. For a 4.6 mm M3 insert, that means a boss around 8 to 9 mm across. Thin-walled bosses are the number one cause of cracked inserts.
- Hole depth: 1 to 2 mm deeper than the insert is long. When you press the insert in, it pushes a small slug of molten plastic ahead of it. That plastic needs somewhere to go. A blind hole cut exactly to insert length has no relief, so the displaced plastic either stops the insert from seating flush or forces the boss to bulge.
Also give the boss enough perimeters and infill to actually be solid where the insert sits. A boss printed at 15 percent infill is mostly air, and an insert melted into air does not grip. Three or four perimeters, or a locally solid region, is what you want around an insert.
Heat-set insert vs the alternatives
An insert is not always the right answer. Here is how it stacks up against the other ways to get a fastener into a printed part.
| Method | Reusable? | Strength | Effort | Best for | |---|---|---|---|---| | Heat-set brass insert | Yes, many cycles | High | Iron + a few seconds each | Anything you will unbolt repeatedly | | Screw straight into plastic | A few cycles | Low to medium | None | One-time or rarely-opened assembly | | Tapped plastic thread | A few cycles | Low | Tap the printed hole | Light-duty, no hardware on hand | | Captive nut in a pocket | Yes | High | Model a nut trap, drop nut in | When you already have nuts and want zero tools |
The short version: for a joint that gets opened more than two or three times, a heat-set insert or a captive nut is worth the extra step. For a bracket that gets bolted once and never touched, a screw straight into the plastic is genuinely fine. Match the method to how the part will be used, not to what feels most "proper."
How do you install a heat-set insert cleanly?
Set a temperature-controlled soldering iron to roughly 200 to 250°C for PLA and PETG (a little hotter for tougher materials), fit a tip that matches the insert bore or use a dedicated insert tip, and let it come up to temperature. Rest the insert on the hole, bring the iron down into it, and let the heat do the work: press slowly and straight down, keeping the insert square to the hole, until it sits flush or a hair below the surface. Pull the iron out, and do not touch the joint until the plastic has cooled and set, usually 20 to 30 seconds. A crooked insert almost always means you pushed before the plastic was soft, or the boss was too cold. Rushing is the enemy.
How Meshra helps you dial in the boss
Here is where designing the part parametrically, instead of downloading a finished mesh, saves you a reprint. In Meshra you describe the part in plain language, including the mounting boss, and it writes real parametric CAD (CadQuery on the OpenCascade kernel) and builds an exact solid, not a triangle mesh. You could ask for "a mounting bracket with two bosses, each with a 4 mm hole for an M3 heat-set insert," and get a solid part with those holes actually modeled.
The payoff shows up when your first test print is not quite right. Every meaningful dimension Meshra generates, including that hole diameter and the boss size, becomes a slider in the parameter drawer. If the insert spins because the hole came out a touch loose, you drag the hole diameter down 0.1 or 0.2 mm and the same code re-executes deterministically with the new number: the hole shrinks, everything else in the part stays exactly where it was, there is no AI call involved, and it is instant and free on every plan. If the boss cracked, widen the boss the same way. That is a five-second slider drag instead of re-modeling the part or fighting a mesh editor.
Meshra also gives you a couple of sanity checks before you ever slice. The printability summary reads the loaded part and flags a wall thin enough to be a printing problem (for example "2 walls may be too thin to print reliably, estimate only"), which is a useful safety net if you accidentally made a boss wall too skinny. Be clear about what that is: it is a print-reliability check on the geometry, not a calculation of whether your specific insert's expansion force will crack the plastic, so it does not replace the 1.6 mm and 2x rules above. When the part is right, export STL for your slicer, or STEP if you want to open the boss in another CAD tool later. For more on how that slider-driven, deterministic re-execution works under the hood, see from a sentence to a printable part.
FAQ
Can I just screw directly into a 3D-printed hole?
Yes, for a joint you will assemble once or open only rarely. A self-tapping or machine screw cuts its own thread into the plastic and holds fine the first few times. It is repeated disassembly that kills a printed thread: each cycle shaves a little plastic until it strips. If a part gets opened often, use an insert.
What temperature should I set my soldering iron to?
Around 200 to 250°C for PLA and PETG works well. You want the brass hot enough to melt plastic smoothly as you press, but not so hot that it scorches or oozes. Materials like ABS and nylon take a bit more heat. If the insert goes in with almost no resistance and plastic pools everywhere, drop the temperature; if it stalls and refuses to seat, raise it.
My insert spins in the hole. What went wrong?
The pilot hole was too big, so the knurls never got a grip. Reduce the modeled hole diameter by 0.1 to 0.2 mm and reprint the boss. This is exactly the kind of fix a parametric part makes trivial: change one number and re-run, rather than re-modeling.
Do I need to design the hole undersized to allow for shrinkage?
No, model the hole at the nominal pilot diameter from the chart. FDM already prints circular holes slightly undersized on its own, which usually lands you in the right window. Print a test boss with your actual material and settings, install one insert, and adjust from there if needed.
Where to start
If you have a functional part in mind that needs to be bolted together and serviced, describe it in the Meshra builder with the boss and hole called out, print a single test boss first, and tune the hole diameter with a slider once you have felt how your insert seats. The template gallery has a range of brackets, mounts, and enclosures with parameters already exposed if you want a starting point instead of a blank prompt. Free and Maker plans include AI generations for this kind of iteration; see pricing for current allowances, or create an account to save your part and dial in the boss after your first insert goes in.
A stripped plastic thread is not a reason to give up on printed fasteners. Size the hole from the chart, give the boss enough wall, press the insert in with patience, and when the first one is not perfect, fix the one number that is wrong instead of starting over.