OpenCameraLab
Making

Printing threads that actually fit

A thread that binds and a thread that rattles usually have the same cause, and it is rarely the model file.

By Domenico Caldesi · 4 August 2026 · 11 min read

A printed adapter fails in one of three ways. It will not start on the lens at all. It goes on two turns and jams hard enough that getting it off takes pliers and some bad language. Or it goes on freely and rattles. Most people meet all three before they meet a good one, and the model file is usually identical every time.

Threads are where printed camera hardware stops being forgiving. A cap that is 0.2mm oversized still works. A hood that is 0.2mm undersized still works. A thread that is 0.2mm out in either direction is scrap, and since almost everything interesting in this hobby involves screwing plastic onto metal, it is worth understanding why before you burn an afternoon on it.

Your printer does not make the hole you asked for

The dominant error is not resolution and it is not the model. It is that extruded plastic does not land exactly where the toolpath says it will.

Molten filament leaving a nozzle swells slightly and gets squashed into the layer below, and the result is a bead a fraction wider than nominal. On an external wall that pushes the surface outwards. On an internal wall it also pushes the surface inwards, into the hole. So printed holes come out undersized and printed posts come out oversized, typically by 0.05 to 0.15mm per side depending on the printer, the material and how fast you are running.

For a male thread going into a female thread, both errors work in the same direction. The peaks are too fat, the valleys are too shallow, and the thread binds. This is why the most common symptom is a thread that starts, gets tighter, and then stops.

Every slicer has a setting for this. It is called horizontal expansion in Cura, XY size compensation in PrusaSlicer and elephant foot's less famous sibling in various others. Set it negative — start at −0.06mm — and the slicer shrinks the toolpath to compensate.

You need to measure it rather than guess. Print a 20mm cube, measure it with calipers across each axis, and print a plate with a row of holes of known diameter and measure those too. The cube tells you outside error, the holes tell you inside error, and they are usually not the same number. Twenty minutes of this fixes threads permanently, and also fixes about half the other fit problems you have been living with.

Shrinkage is a separate error and it scales

Thermoplastics contract as they cool. Rough figures: PLA around 0.3%, PETG 0.4 to 0.6%, ABS and ASA 0.7 to 0.8%, nylon can exceed 1%.

Percentages are the important part. On a 10mm feature, 0.5% is 0.05mm and nobody notices. On an M42 thread — 42mm across — 0.5% is 0.21mm, which is the entire fit tolerance. This is why the same adapter model that fits perfectly in PLA binds in PETG, and why people conclude the file is wrong when the file is fine.

Slicers expose this as a scaling compensation, often per material. If you switch material and your threads stop fitting, this is the first place to look, before you touch clearance.

The first layer flares, and threads start at the first layer

The first layer gets squashed into the bed on purpose, to make it stick. That squash spreads it sideways, so the bottom half millimetre of every print is slightly wider than the rest. On most parts this is invisible. On a thread that begins right at the build plate, it means the first turn — the one that has to start the engagement — is the fattest turn on the part.

Three cross-sections of a printed thread base: as modelled, as printed with the first layers flared outwards, and with a chamfer removing the flared materialAs modelledAs printedfirst turn bindsWith a chamferstarts cleanly
The first layer is squashed into the bed deliberately, so the bottom of the part is wider than the rest. On a thread that starts at the plate, that makes the first turn the fattest one.

Two fixes, and they stack. Turn on elephant foot compensation, which most slicers have and which shaves the first few layers back. And design or add a small chamfer at the base of the thread, 0.5mm at 45°, which removes the affected material entirely and gives the thread a lead-in.

If you are printing someone else's model and it binds on the very first turn but runs freely afterwards, this is almost always the cause, and a few seconds with a craft knife on the bottom edge solves it.

Print it standing up

A round thread should be printed with its axis vertical. This is not a preference.

Printed lying down, the thread is built from stacked layers running across the circle, so the cross-section becomes visibly polygonal, the underside needs support, and support scars land exactly on the surface that has to slide. You also lose the roundness that the fit depends on, because the part sags slightly on the overhanging half.

Standing up, each layer is a smooth ring, and the only compromise is that the thread's helix becomes a staircase in Z. That staircase is controlled by layer height, and this is the one place where dropping to 0.1mm or 0.12mm is worth the extra print time. An M42 thread has a 1mm pitch; at 0.2mm layers you get five steps per turn, which is coarse enough to feel. At 0.1mm you get ten, and the thread runs smoothly.

Nozzle size matters less than people assume. A 0.4mm nozzle handles a 1mm pitch fine. A 0.6mm nozzle struggles, because the thread crest becomes narrower than a single extrusion and the slicer starts dropping features.

Clearance, and which side to put it on

Once the printer is calibrated, you still want deliberate clearance, because plastic against metal needs somewhere for debris to go.

For a printed male thread into a metal female, 0.15 to 0.25mm on the diameter is a sensible starting range. Tighter than that and it binds as soon as the surface picks up any dust. Looser and the lens will wobble, which on an adapter translates directly into the lens sitting slightly off-axis.

Put the clearance on the printed part, always. It is the sacrificial component and the cheap one. Never modify a model to make it a tighter fit on a camera mount in the hope of removing play, because the failure mode when a plastic thread seizes in an aluminium mount is that you are now working on a camera with a stuck object in it.

Test the thread, not the adapter

The mistake that costs the most time is printing the whole part to check the fit. An adapter is a two hour print. The thread is the first 8mm of it.

Cut the model down to a short collar containing only the threaded section — most slicers can do this without going back to CAD — and print that. It takes eight minutes. Print three of them with different compensation values in one go, labelled, and you will find your number in a single session rather than over a week.

Keep the good coupons. When you switch filament brands, printing one is a two minute check on whether your calibration still holds.

Cleaning up and running it in

A fresh printed thread has burrs, stray strands and a slightly furry surface. Take a moment before it goes near a camera.

Run a fingernail or a nylon brush over the crests to knock off the obvious debris, then blow it out. Compressed air, or a blower bulb. What you are avoiding is a small hard fragment of plastic ending up between the threads, or worse, falling into the mount and eventually onto the sensor.

Then run it in against the real part, slowly, backing off whenever it tightens. A printed thread beds in over the first few engagements as the high points get flattened. It should get smoother; if it gets tighter, stop, because you are galling it.

A dry PTFE lubricant helps and does not attract dust. Oil and grease do the opposite and have no business anywhere near a lens mount.

The general rule: if a printed part will not go on with light finger pressure, it is not going to be fixed by more force. Take it off, find out why, adjust, print another. They cost about forty pence.

When it still will not fit

Working through in the order that finds problems fastest:

  1. Is the first turn binding but the rest free? Elephant foot. Chamfer the base.
  2. Does it bind progressively, tighter every turn? Cumulative pitch error, which means scaling. Check your shrinkage compensation for that material.
  3. Does it not start at all? Diameter. Increase negative XY compensation by 0.03mm and reprint the coupon.
  4. Does it rattle? Too much compensation. Back it off.
  5. Does it go on smoothly but sit crooked? Check that the flange face is flat. A slightly warped first layer tilts the whole part, and on an adapter that tilts the lens.
  6. Does it feel gritty? Debris, not fit. Clean it and try again.

Fit is one of two things that have to be right on an adapter. The other is how far the lens ends up from the sensor, which is a completely separate problem with completely separate causes, and it is the subject of why cheap lens adapters lose infinity focus. A perfectly fitting adapter of the wrong thickness is still a paperweight.

Build one of these

Keep reading