Heat-Set Insert, Screw Hole and 3D-Printed Thread Calculator

Hole calibration and manufacturer dimensions

For nominal-bore recipes, $d_{CAD}=d_{target}+\Delta_d$, with $\Delta_d=d_{CAD,test}-d_{printed,test}$ measured for the selected hole-axis orientation. The predicted pre-installation bore is $d_{print}=d_{CAD}-\Delta_d$. This is a local additive calibration near the tested diameter, not a shrinkage law for all sizes. Leave existing slicer compensation in place when measuring.

The insert catalogue is CNC Kitchen's Dimensions & Design Guidelines, attached s7, page 2: length, knurl diameter, nominal bore and minimum radial wall W measured from that bore. The catalogue's blind-hole allowance is 1 mm below the insert. The adjustable nominal-bore trim is a coupon-based design change, separate from printer correction.

Bambu's highlighted PLA/PETG BT2, BT2.5 and BT3 recommendations in its linked test chart (s5–s6) are CAD hole sizes. They pass through unchanged except for explicit manual tuning. The selected correction predicts what that CAD size prints; it is not added to the Bambu size. The chart is limited to a hole axis perpendicular to the bed. Torque installation/removal columns in that chart are not torque-out capacities and are not used.

For the EJOT DELTA PT nominal-bore recipe, the manufacturer's guide s3, page 9 gives a usual bore ratio of 0.80, with selected technical plastics up to 0.88, a boss OD starting at twice screw diameter, and an illustration with half a screw diameter below engagement. Page 10 independently lists a 3.2 mm bore for the 4 mm DELTA PT screw. This guidance is for that screw family in thermoplastics and needs FDM trials.

Printed internal metric threads

Following the 60° basic-profile geometry named in ISO 68-1:2023, $H=\sqrt{3}P/2$, $D_2=d-3H/4$ and $D_1=d-5H/4$. The internal major diameter is d. Add the total diametral clearance c to all three internal diameters, then add the hole correction once to obtain CAD dimensions. Pitch is unchanged. The drawing is our own construction from these formulas; no standard table or figure is reproduced. This does not implement ISO 965 tolerance classes, an external screw root profile or thread gauging. Check against the standard in force for a specified metric joint.

Boss and solid walls

The boss recommendation is the greater of the geometry starting point and $D_{env}+2n_{target}w$, where $D_{env}$ encloses both the physical hardware/root and the complete CAD cavity (including the printed-thread CAD major diameter). The insert geometry starting point retains the catalogue wall beside both the CAD bore and predicted physical bore by using the larger diameter; the twice-screw-diameter starting point for other self-tapping recipes is a design heuristic, not a Bambu boss specification. Here w is extrusion line width. The required pre-installation slicer count is $\lceil n_{target}+(D_{root}-d_{print})/(2w)\rceil$.

Only the contiguous radial envelope of the hole's perimeters is credited: $D_s=\min(D_{boss},d_{print}+2nw)$. Material displaced or removed inside the installed knurl or thread-root envelope is not credited. Thus $A_{net}=\pi(D_s^2-D_{root}^2)/4$. The wall chart holds the chosen boss size fixed while varying perimeter count.

Coupon scope and strength screening

Numerical forces and torques are withheld by default, including intermediate force/torque results. Enable them only after entering a matching coupon scope and conservative stresses. The coupon method, polymer and hole-axis direction must match; hardware/root outside diameter and full engagement must agree within 0.001 mm for rounding. The physical pre-installation bore must lie inside the measured interval, and the current boss and perimeter count must meet the calibrated support minima. Invalid printable geometry also disables strength. The default coupon interval is a single 4.00 mm bore; it is inactive until representative testing is confirmed. Changing the bore to 4.59 mm cannot retain a 4.00 mm calibration. Do not widen the interval without measurements; use the weakest justified stresses across it. The wall sweep always shows geometric solid-wall thickness, not an uncalibrated force curve.

The equivalent interface shear area is $A_i=\pi d_i L_e q$. Inserts and self-tapping screws use q = 1, so their entered effective stress must already represent their particular knurl/thread form, pilot, installation and orientation. Printed threads use a half-pitch smeared area, reduced by remaining radial overlap: $q=\tfrac12\max(0,1-c/(d-D_1))$. This is an explicit screening assumption, not a validated printed-thread stripping correlation.

The axial failure estimate is $\min(\tau_a A_i,\sigma_{net}A_{net})$. The interface torque estimate is $\tau_r A_i d_i/2$. The elastic annular boss estimate is $2\tau_b J/D_s$, with $J=\pi(D_s^4-D_{root}^4)/32$. The smaller torque governs. Axial and rotational interface stresses, and printed boss shear stress, are separate inputs. Convert N·mm to N·m explicitly. Each screening load is its failure estimate multiplied by the user-selected service retention and divided by the additional strength divisor.

CNC Kitchen's 2026 hole-size tests (s1) reported about 1400 N peak pull-out and roughly 90% retention at 4.2 mm CAD with an approximately 0.25 mm print undersize. These are contextual reference measurements, not fitted universal curves. The 2019 tests (s2) reached 3–4 N·m before M3 bolt heads failed; those are censored lower bounds on the tested inserts' resistance, not measured insert torque-out values. Neither publication validates the default stress assumptions or our boss model. No numerical result is enabled merely by selecting one of those insert sizes.

Independent strength comparisons and their limits

SPIROL's 2026 insert guide (s9), PDF pages 10 and 22–23, provides dimensions and independent pull-out/rotation results under stated test conditions. The calibration configuration is INS19 M3 short: OD 4.62 mm, length 3.56 mm, drilled bore 3.99 mm, 780 N axial extraction and 1.5 N·m rotation. These training values alone give $\tau_a=780/(\pi\,4.62\,3.56)=15.09568$ MPa and $\tau_r=2(1500)/(\pi\,4.62^2\,3.56)=12.56717$ MPa.

Two different configurations are reserved for comparison, without resetting those stresses. INS19 M3 long (OD 4.75 mm, length 5.74 mm, bore 3.99 mm) predicts 1293 N interface pull-out and 2.557 N·m rotation versus the independently tabulated 1445 N and 3 N·m. INS19 M4 short (OD 6.25 mm, length 4.70 mm, bore 5.61 mm) predicts 1393 N versus the independently tabulated 1670 N; the net-section assumption is non-governing in that comparison. The shared 20% comparison band is an explicitly coarse model-approximation criterion, not measurement uncertainty, a working-load margin or proof of universal accuracy. The checked values were not used to calculate the stresses.

The source specifies heat-installed inserts in unfilled moulded nylon 6/6, bosses at least twice insert OD, and drilled holes. The benchmarks use that minimum boss size (9.50 or 12.50 mm). The defined endpoints are extraction of the insert and rotation in plastic; the guide does not identify microscopic fracture paths or exact host conditioning. The M3 long case checks interface scaling, not the generic annular boss limit. These comparisons do not validate FDM layer bonding, infill anchorage, printed threads or every insert family.

The FDM experiments by Kastner et al. (2023), doi:10.3390/jmmp7010042, s8 pages 3–12, used 26 × 26 × 16 mm GreenTEC Pro specimens, an 8 mm vertical pre-hole, M6 Tappex MULTISERT BN37885 inserts and a 2 mm/min pull-out test. Their observed wall-to-infill separation and interlaminar damage demonstrate failure modes outside this simple annular model. Their material and stresses are not silently substituted for PLA, PETG, ABS/ASA or PA.

• The hidden starting stresses (10 MPa for axial interface, rotational interface and bulk shell shear; 20 MPa for shell tension) are placeholders for data entry. They do not produce numerical strength results before matching coupon calibration is confirmed. They are not material presets.
• A coupon interval consists of measured physical bore diameters, not CAD dimensions. A single tested diameter uses equal lower and upper limits. The 0.001 mm matching tolerance is for rounding, not a manufacturing tolerance or permission to extrapolate.
• Determine effective axial stress only from an actual interface pull-out mode using the equivalent shear area. Determine rotational interface stress from an actual rotation endpoint with consistent torque units. Use separate printed-material tensile and shear data for the shell. A boss/base or wall-to-infill failure does not identify interface strength.
• Use the weakest justified stresses over the entire declared interval and matching process conditions. The selected material, orientation, hardware/root diameter and engagement must match. The current boss and perimeter count cannot be below the tested support minima. The scope fields cannot verify whether the user has performed the claimed tests.
• Service retention (starting scenario 0.50) and the additional divisor (starting value 3) are user-supplied design assumptions after calibration, not temperature, moisture or lifetime correlations. PA must be tested at its service conditioning.
• Only the hole's contiguous solid perimeter envelope is credited; sparse infill and outer boss perimeters are not treated as solid. Inspect actual sliced paths. External boss OD is assumed to print as modelled; both CAD and predicted physical cavity envelopes are protected in sizing.
• Full engagement excludes entry recesses, chamfers, tips and runout. Add their depth separately. Blind-hole floor bending, boss-to-part attachment, infill debonding, radial splitting, brass/screw failure and layer delamination are not calculated.
• Equivalent uniform interface stress, the printed-thread half-pitch overlap multiplier and the homogeneous annular shell are screening approximations. The SPIROL holdouts test limited interface size scaling in moulded nylon; they do not establish a universal FDM material law.
• The section is drawn upright for clarity. The hardware's internal bore and self-tapping thread marks are schematic; printed internal thread geometry follows the stated basic-profile formulas.

Geometry sizing supports FDM/FFF parts made from unfilled PLA, PETG, ABS/ASA and PA. Manufacturer insert walls are retained around both CAD and predicted physical bores, even for negative corrections. The complete CAD internal-thread major diameter must fit inside a printable wall, not just its minor opening. Bambu CAD pilots cover its tested PLA/PETG BT screws with a Z-axis hole; EJOT nominal pilot ratios apply to DELTA PT and require printed-part trials.

Strength is unavailable unless the declared coupon method, polymer, orientation, hardware/root OD, engagement, measured bore interval and minimum support match. It is also suppressed for invalid printable geometry. No generic published bore interval is assigned to all M3 inserts. A bore immediately below the knurl OD is not sufficient evidence of usable interlock. Extending the tested interval without data is not calibration. The starting stress numbers and confirmation switch are not independent evidence.

The two SPIROL holdout cases compare measured force and rotation from different insert configurations against stresses fixed using another configuration. The 20% band is a coarse model comparison, not confidence bounds. The source describes moulded PA6/6 with drilled bores and a defined extraction/rotation endpoint; it does not establish FDM strength, host conditioning, microscopic failure path or the annular shell's capacity. The CNC Kitchen 2026 force curve motivates withholding extrapolation; its 2019 bolt-head failures remain lower bounds, not measured insert torque-out capacities.

Gauge local hole diameters in the actual print orientation. Severe sagging or ovality cannot be represented by one additive correction. Inspect layer bonding, installation damage, wall-to-infill attachment, base bending, head bearing, pull-through, preload, fatigue and repeated assembly separately. Torque-out is not a seating-torque recommendation. No combined-load qualification, automatic temperature limit, creep lifetime or moisture model is provided. Printed basic profile geometry is not an ISO 965 tolerance class or a gauged fit; check against the standard in force. Resin, fibre-filled and metal printing and safety-critical qualification are outside this model.