Almost every heat-set insert hole chart on the web repeats the same handful of numbers, and none of them say where those numbers came from. That is a problem, because the two things you actually need are not the same kind of fact. One is the geometry of a brass part, which is fixed and knowable. The other is what your printer does to a hole, which is neither.
So this page keeps them apart. The insert dimensions below were measured off the manufacturer's own STEP solids — the CAD files ruthex publishes for each part — using a script that reads the cylindrical faces directly. The FDM correction comes from the same solver behind the calculator. No figure on this page was typed in from another chart.
The three diameters on a heat-set insert
Pull an insert out of the bag and it has a stepped profile, not a single diameter. Three of those steps matter:
- Knurl crest — the widest point, the ridges that bite into the plastic. This is a hard ceiling: a hole this size has nothing left for the brass to grip.
- Pilot end — the plain step at the bottom. This is the part that drops into the hole so you can pre-seat the insert before melting it, and it is the diameter your hole should be sized against.
- Thread minor — the internal bore. Irrelevant to the boss, but it tells you the clearance hole needed in whatever part you are bolting on.
| Insert | Knurl crest | Pilot OD | Seat OD | Thread minor | Length |
|---|---|---|---|---|---|
| M2 | 3.571 | 3.200 | 3.200 | 1.623 | 4.00 |
| M2.5 | 4.600 | 4.000 | 4.400 | 2.075 | 5.70 |
| M3 | 4.600 | 4.000 | 4.400 | 2.529 | 5.70 |
| M4 | 6.300 | 5.540 | 6.000 | 3.332 | 8.10 |
| M5 | 7.100 | 6.400 | 6.400 | 4.234 | 9.50 |
| M6 | 8.627 | 7.950 | 7.950 | 5.035 | 12.70 |
| M8 | 10.081 | 9.500 | 10.000 | 6.779 | 12.70 |
The measured pilot diameters land exactly on the hole sizes published for M2 (3.20), M3 (4.00), M5 (6.40) and M8 (9.50). That agreement is the reason we size against the pilot rather than subtracting a guessed offset from the crest: the vendor's own recommendation is the pilot diameter, to the micron.
What to model in CAD
The finished hole should equal the pilot diameter. But an FDM bore prints undersize — the extruded bead is pushed inward around a curve — so the number you draw has to be larger than the number you want. That difference is the process allowance, and it is bigger on small holes because the curvature is tighter.
| Insert | Target hole | Allowance | Model this | Ceiling |
|---|---|---|---|---|
| M2 | 3.20 | 0.39 | see warning | 3.57 |
| M2.5 | 4.00 | 0.37 | 4.37 | 4.60 |
| M3 | 4.00 | 0.37 | 4.37 | 4.60 |
| M4 | 5.54 | 0.34 | 5.88 | 6.30 |
| M5 | 6.40 | 0.33 | 6.73 | 7.10 |
| M6 | 7.95 | 0.31 | 8.26 | 8.63 |
| M8 | 9.50 | 0.30 | 9.80 | 10.08 |
Ceiling is the knurl crest. Your as-printed hole must stay below it, and every row above leaves between 0.23 and 0.42 mm of margin. That margin is your entire tolerance budget, and it is worth knowing that a well-tuned consumer machine repeats a dimension to about ±0.08 mm at 0.2 mm layers — so the budget is real, but it is not generous.
The M2 insert has only 0.37 mm of knurl depth, and the process allowance a 0.4 mm nozzle needs at that diameter is 0.39 mm. The correction is larger than the feature it has to fit inside, so there is no hole that both accepts the pilot and leaves the knurl anything to grip. This is not a rounding artifact — it is the size where the process runs out of room.
Fit a 0.25 mm nozzle and the allowance falls to 0.25 mm, giving a 3.45 mm model hole with real margin under the 3.571 mm crest. If you are stuck with a 0.4 mm nozzle, drill the hole to 3.2 mm after printing, or design around M2 entirely.
How much can you trust the allowance?
An independent check is available, and it is a good one. CNC Kitchen ran pull-out tests on M3 inserts across a range of hole diameters from 3.6 to 4.6 mm and measured their printed holes coming out about 0.25 mm undersize on a well-tuned machine in PLA. Two things fall out of that:
- Their tested optimum for an actual hole was 4.0 mm, which is exactly the pilot diameter measured here from a different brand's CAD. Two independent routes, same answer.
- Their measured allowance was 0.25 mm against the 0.37 mm this model uses. Both are defensible; the real spread across machines and materials is roughly 0.25 to 0.37 mm, and this model deliberately sits at the loose, easy-to-install end of it.
Their tests also quantified the cost of going too large: strength held at about 90% of peak at the easy-install diameter, then fell off sharply as the hole approached the knurl crest. That is the shape of the trade-off. Slightly tight costs you a burr and some fiddling. Slightly loose costs you the joint.
Run your own numbers Enter the target hole, pick press fit and your material, and the calculator returns the diameter to model for your nozzle and layer height.Boss wall thickness and depth
A heat-set insert loads the boss in hoop tension, exactly like a press fit, and a thin wall splits on the first bolt. Keep at least 0.25 × the hole diameter of material around the bore, and make the pocket about 1 mm deeper than the insert so displaced plastic has somewhere to go instead of forming a burr under the thread.
| Insert | Model hole | Min wall | Min boss OD | Min depth |
|---|---|---|---|---|
| M2.5 | 4.37 | 1.09 | 6.55 | 6.70 |
| M3 | 4.37 | 1.09 | 6.55 | 6.70 |
| M4 | 5.88 | 1.47 | 8.81 | 9.10 |
| M5 | 6.73 | 1.68 | 10.09 | 10.50 |
| M6 | 8.26 | 2.07 | 12.39 | 13.70 |
| M8 | 9.80 | 2.45 | 14.70 | 13.70 |
Print bosses solid, or close to it. The pull-out figures everyone quotes come from samples with heavy walls and dense infill; at 15% infill the plastic fails around the insert long before the insert lets go, and you are testing your infill rather than the joint.
Does material change the hole?
Barely, and this surprises people. Running the same 4.00 mm target through every material in the model gives holes from 4.374 mm in PLA-CF to 4.434 mm in PP — a total spread of 0.060 mm. That is smaller than the ±0.08 mm repeatability of the machine printing it.
At fastener sizes, shrinkage differences are lost in process noise, so one hole size covers PLA through nylon. Material choice still matters enormously for whether the joint survives — PLA is brittle and creeps very little, PETG and ABS tolerate the hoop stress far better — but it does not change the number you draw.
Frequently asked
Why is my insert sitting proud of the surface?
The hole is too small, so the insert has displaced plastic upward instead of melting into a hole that could accept it. Increase the diameter, and use a flat tool to press the insert the last fraction of a millimetre so it finishes flush.
Why can I pull the insert out with pliers?
The hole is at or above the knurl crest, so the ridges never engaged. Compare your printed hole against the ceiling column above. Once a hole is oversize there is no installation technique that recovers the strength.
What temperature should the iron be?
Around 240 °C for PLA and PETG, using a tip that contacts the whole top face of the insert so heat reaches the knurls rather than just the rim. Too cool and the plastic is pushed aside instead of flowing into the knurl pattern, which is where the entire holding force comes from.
Should I just drill the holes instead?
You can, and it removes printer variation completely: drill to the target hole column rather than the model column. It does not make the joint stronger than a correctly sized printed hole, so it is worth the time only when you cannot characterise your machine, or for M2 on a 0.4 mm nozzle where printing the right hole is not possible.
Do these numbers apply to other brands?
The geometry table is specific to ruthex parts. The method is not: measure the pilot diameter of whatever insert you have, add the allowance for your nozzle, and check the result still clears the knurl crest. Most brands in this class are dimensionally close, but "close" is not a word you want inside a 0.3 mm margin.