Every 3D printed machine eventually reaches the same moment. A stepper, a servo, a linear rail or a lead screw arrives from a catalogue with tolerances you did not choose, and the job of the printed part is to hold that hardware still while something else is allowed to move. It is the least glamorous thing a printed part does and the one that most reliably fails, because the printed bracket carrying a motor is doing three things at once: it locates the motor, it resists the torque the motor produces, and it sits at the point where the whole machine's alignment either works or does not.
This article is about that class of part. Not gears, which have their own treatment of module and backlash, and not shafts or hubs, which have their own article on getting torque into a part. This is about the stationary side: the mounting face, the slot, the boss, the bracket that ties a rail to a frame, and the block that supports the free end of a screw. These are the parts where a designer stops being a designer and starts being a machinist, and where the wrong instinct produces something that looks perfectly printable and is quietly destroying the machine it was built for.
The motor mount is a torque reaction member, not a cradle
A NEMA 17 motor with a 40 mm body produces something in the region of 0.4 to 0.5 N·m of holding torque when energised, and a good deal more in the instant before it stalls. That torque does not disappear into the driven shaft. It is reacted at the mounting face, and the arm between the mounting face and the shaft axis is only about 15 mm. A 0.5 N·m reaction over a 15 mm lever is roughly 33 N of force trying to peel the bracket off its fixings, applied continuously while the motor is energised and cyclically every time the shaft changes direction.
A printed cradle that wraps around the motor body looks stronger than a flat plate with four holes, and it is not. The wrap does nothing about the reaction torque. It adds material where the stress is already low and adds thermal mass around a motor that is already the hottest small object on the machine. The four fixings carry the load no matter how thick the wrap is.
| Mounting approach | What it actually resists | Where it fails |
|---|---|---|
| Flat plate, four bolts, motor cantilevered | Torque, via bolt tension and friction at the face | Plate flexes; bolts lose preload as plastic creeps |
| Plate with an integral pocket around the motor flange | Torque plus side loads from belt pull | Pocket floor must be flat, or the motor sits off-square |
| Full wrap or cradle | Very little extra torque capacity | Traps heat; complicates assembly; hard to print flat |
| Printed standoff, steel through-bolted to frame | Bolt tension carried by steel, not plastic | The standoff must be in pure compression |
| Aluminium L bracket plus printed spacer | Everything structural is in metal | The plate must be square to the frame it bolts to |
If you take one rule from this article, take this: the printed part should locate the motor, and steel should carry the bolt tension wherever you can arrange it. A printed boss acting as a spacer in pure compression is doing a job plastic is good at. A printed boss being stretched by a bolt is doing a job plastic is bad at, because the bolt cannot hold its tension in a material that creeps at 40 °C.
Slots beat holes, and a slot only works in one direction
Nearly every catalogue motor and rail has slotted mounting holes, and nearly every printed bracket should too, for a reason that has nothing to do with convenience. Printed hole positions come out with an error that is systematic: the part shrinks radially, the machine's own axes are not perfectly square, and the hole itself is slightly small because a round hole printed vertically is a polygon with a smaller inscribed circle. A round hole in a printed bracket therefore acts as a hard constraint that drags the motor off centre, and you have no adjustment left to fix it.
A slot gives that error somewhere to go. But the slot is directional, and the direction matters more than the length.
- Slot along the belt direction so belt tension can be set by sliding the motor, and add a screw or a cam to hold the position once it is right. A slotted motor mount with no tensioning device is a mount that will drift.
- Never slot both holes of a pair in both axes. Two slots at right angles give the motor four degrees of freedom and no location at all. One slot plus one round hole is the correct pattern: one hole locates, one slot allows.
- Slot the printed part, not the metal. Milling a slot into an aluminium bracket is a machining job; printing a slot with a location shoulder is free.
- Size the slot for the bolt plus a printed washer face. A slot 0.3 mm wider than the bolt shank will bind as soon as the bolt is tightened at a slight angle. Give 0.5 mm to 0.8 mm on the width and put the adjustment range in the length.
The corollary is that a printed slot needs a flat bearing face under the bolt head, and that face must be printed, not bridged. A slot that spans an unsupported gap will sag in the middle and the bolt will cock over. Orient the part so the slot's bearing face is on the bed, or design the slot into a wall thick enough that the roof of the slot is only a millimetre or two of bridging.
NEMA 17 is not a standard, it is a family
The 17 in NEMA 17 refers to a 1.7 inch face, which is 43.2 mm, and the bolt circle is 31 mm on four M3 threads. What the name does not tell you is that body length varies from 20 mm to 60 mm, that shaft diameter varies from 4 mm to 5 mm, that shaft length varies by several millimetres between manufacturers, and that the boss diameter around the shaft is not consistent. A printed bracket designed around the motor you have on the bench will not fit the motor you buy next year.
| Dimension | Nominal | What varies in practice | Design allowance |
|---|---|---|---|
| Mounting hole spacing | 31 mm square | Plus or minus 0.2 mm, and square versus bolt circle is ambiguous | Slot 2 mm of travel per hole |
| Bolt thread | M3 × 0.5 | Depth 3.5 to 5 mm | Through-bolt with a nut, or a heat-set insert |
| Pilot boss diameter | 22 mm typical | 21 to 23 mm, or absent entirely | Do not rely on it for location; add 1 mm clearance |
| Shaft diameter | 5 mm | 4 mm on short bodies, 6.35 mm on some servos | Couple through a flexible element, never a rigid printed sleeve |
| Body length | varies | 20 to 60 mm | Never enclose the motor body |
| Flange corner radius | varies | Square or rounded corners | Cut the pocket corners generously |
Do not let the pilot boss locate anything
The pilot boss deserves a note. On a machined assembly the boss centres the motor and the bolts merely clamp. On a printed assembly, treating the boss as a locator means the motor position depends on a printed circular pocket being round and on size, which it will not be. Use the four fixings and a slot to locate, and give the boss pocket a clearance of a full millimetre so it never fights you. This is the same logic as tuning tolerance for a repeatable fit: the constrained features should be the ones you can adjust, not the ones that happen to be round.
Bolts do not stay tight in plastic, and that is arithmetic
The bolt is steel and the bracket is polymer. Under a preload, the polymer relaxes. At 60 °C, near the glass transition of PLA and well within the working range of a motor that has been running for an hour, the relaxation is fast enough to measure in hours. A joint that was tightened to 0.6 N·m on Monday will be loose by Friday, and then the motor starts rocking on its face, and then the belt starts walking off the pulley.
Geometry that survives preload relaxation
There is no thread-locker and no torque value that fixes this, because the problem is the material, not the fastener. What fixes it is geometry and material choice:
- Give the bolt a long elastic length. A bolt that passes through 10 mm of printed plastic plus a washer has a short, stiff grip. A bolt that passes through a printed boss and clamps a metal plate against the frame has a long grip and keeps its tension. Where the plastic must be in the stack, make it a spacer in compression with the bolt's stretch happening in the metal above it.
- Heat-set inserts, not printed threads. The insert moves the thread into brass, and brass creeps far less than PLA. The insert still relies on the surrounding plastic for its grip, so give it a wall at least the insert diameter in thickness, and follow the insert geometry rules that keep a printed thread serviceable.
- Use wide flange washers. A printed face with a bolt head on it will dish. A 10 mm washer under an M3 head spreads the load over eight times the area and stops the dishing, which is what actually loses the preload in the first place.
- Choose a material with a higher heat deflection temperature for anything attached to a motor. A mount that reaches 55 °C in service should not be PLA. ASA or ABS-CF both hold their shape at that temperature; a glass-filled engineering grade holds it better still.
This is the same mechanism described in creep, fatigue and preload, applied to a fastener instead of a loaded beam. The bolt does not snap. It stops being a bolt.
The belt pulls the motor sideways, and the bracket was designed for torque
Almost every printed motor mount on a hobby machine carries a belt, and almost every one of them was designed as though the only load were the reaction torque. Belt tension on a typical GT2 setup runs 20 to 40 N per side. That force acts at right angles to the mounting face, at the end of the motor shaft, on a lever of 20 to 30 mm. The resulting moment is comparable to the motor's own torque reaction, and it is constant rather than intermittent.
What that means in practice is that the motor pivots about its mounting face, the face's lower edge digs into the printed bracket, and the bracket cracks along a layer line starting at the corner of the pocket. The classic symptom is a hairline crack radiating from the corner of a motor pocket, appearing after weeks rather than hours.
Three cheap changes to the belt-pull load path
Three changes fix it, and they are all cheap:
- Put an abutment under the motor. A rib or a step that the motor body lands against, carrying the belt-pull moment in compression rather than through the bolts.
- Orient the bracket so the belt-pull moment bends it across layers, not between them. Layer adhesion is the weak plane, and a bracket whose pocket floor is printed as a bridging roof is a bracket whose weak plane sits exactly where the moment lands.
- Add a tensioner so the belt does not have to be tight. A printed idler on a slotted arm lets you run the belt at the low end of its tension range and take up the difference with a roller. The load path then goes through the idler, not through the motor's mounting face.
Where the geometry allows, the best answer is to stop asking the printed part to carry the belt moment at all: bolt a metal plate to the frame, and use the printed part only as the pattern that positions the bolt holes. This is not giving up on printing. It is using printing for what printing is good at.
Rails and screws: the printed part is the alignment, whether you admit it or not
A linear rail and a lead screw both want to be parallel. The rail's carriage runs on hardened balls in a ground groove and will tolerate a surprising amount of misalignment by simply wearing out faster. A lead screw cannot tolerate misalignment at all; a screw that is 0.5 mm off parallel to its rail will bind, whine, and consume its brass nut in a few hundred hours. On most hobby machines the printed parts are what determine that parallelism, because the frame extrusions are straight and the printed brackets are what set the offset.
Four rules follow, and they are all consequences of the printed bracket being the alignment feature.
| Interface | The printed part's job | Rule |
|---|---|---|
| Rail to frame | Sets the offset and the parallelism | Print a locating shoulder that registers on the frame's machined face, not on its extrusion profile |
| Screw bearing block | Holds the supported end concentric with the nut | Float one end. Never fix both ends in printed plastic |
| Rail end stop | Absorbs impact at the end of travel | Elastomer bumper, not a printed face |
| Carriage plate | Spreads load across the carriage's four mounting holes | Full contact face, no bridging over the carriage |
Fix one end of the screw and float the other
The screw support is where nearly everyone goes wrong. A lead screw is supported at the motor end by the coupler and at the free end by a bearing block. If both ends are held rigidly in printed parts, then any error in the distance between them, any frame flex, and any temperature change all go straight into bending the screw. The screw then whines in one direction of travel and is quiet in the other, which is the signature of a bent screw rather than a worn nut.
The fix is the one machinists have used for a century: fix one end and float the other. Hold the motor-end bearing rigidly, and let the free-end support float in the plane perpendicular to the screw, with the constraint applied in the axial direction only. A printed block with an oversized bore and a slip fit around the bearing's outer race, located axially by a shoulder, does exactly this. It costs nothing and it turns a bent screw into a straight one.
Rails are more forgiving but not infinitely so. The printed bracket should register on the frame with a shoulder rather than relying on the bolt holes, because bolt holes in printed plastic are always slightly oversized and the rail will settle into whatever position the bolt clearance allows. A shoulder that the rail's back face slides against, printed at a right angle to the mounting face, is the cheapest alignment feature you can add. The same principle appears in how tolerance stacks up across a multi-part assembly: errors accumulate in series, and the cheapest way to break a chain of errors is to make one feature do the locating properly.
Rigidity is a property of the joint, not of the bracket
A printed bracket bolted to a frame with four M4 bolts is not a rigid bracket. It is four small steel bolts, a fairly flexible polymer plate, and two mating surfaces that are in contact over perhaps 10 percent of their nominal area. The contact patch is what determines the stiffness of the joint, and printed surfaces are not flat.
Print a 60 mm square plate and measure the flatness of its largest face. You will find 0.1 to 0.3 mm of waviness on the top surface, more on a part printed in a warm chamber, and more again if the part has been annealed. Two such faces bolted together touch at their high points and flex into contact under load, which is why a bracket that feels rigid on the bench rings like a tuning fork in service.
Three levers on joint stiffness
Three things improve joint stiffness, in order of value per unit of effort:
- Increase the bolt spacing, not the bolt size. Joint stiffness scales roughly with the square of the spacing between fixings. Moving two bolts 20 mm further apart does more than going from M3 to M5.
- Give the joint a locating feature so the bolts only clamp. A shallow printed tongue and groove, or a shoulder, prevents the micro-slip that eats both stiffness and preload.
- Face the printed part if it matters. A minute on a flat plate with 240 grit turns a wavy surface into a flat one, and flat contact is the single largest contributor to joint stiffness. The rules for designing tolerances in rather than slicing them out apply here too: flatness is a feature you design and then confirm, not one you hope for.
If the joint carries a dynamic load, remember that stiffness and damping are different properties. A very stiff printed bracket with no damping transmits everything to the frame, which is how a machine ends up humming and shaking itself apart. Where the choice exists, prefer a slightly softer bracket plus an elastomer pad over a maximally rigid one.
Material choice for machine hardware, in plain terms
Machine parts live in a harder environment than display parts. They see continuous small loads, warm motors, occasional impact, and they are expected to hold dimensions for years. The table below is ordered by how well the grade holds a bolted joint at 60 °C, which is the property that actually decides whether the mount survives.
| Material | Heat deflection | Creep under bolt load | Where it belongs on a machine |
|---|---|---|---|
| PPS-CF | very high | very low | Motor mounts, hot-end brackets, anything near a heated bed |
| PA6-GF25 | high | low once dry | Structural brackets; must be dried and printed hot |
| PA612-CF15 | high | low, less moisture-sensitive | Machine parts in humid workshops |
| PET-GF15 | moderate | moderate | Rail brackets well away from heat |
| ASA-CF | moderate to high | moderate | Outdoor and warm-chamber machine parts |
| ASA | moderate | moderate | General machine brackets, good UV and heat |
| ABS-CF | moderate | moderate | Motor mounts on enclosed printers |
| ABS | moderate | high | Light brackets only; will relax under bolt load |
| PETG | low | high | Non-structural covers and guards |
| PLA+ High Speed and PLA Matte | low | high | Jigs and setup aids that never get warm |
Nozzle and moisture footnotes that outweigh the ranking
Two footnotes matter more than the ranking. First, glass and carbon filled grades need a hardened nozzle; a brass nozzle will be oval after a few hundred grams of glass-filled material, and the resulting extrusion inconsistency looks exactly like a flow calibration problem. A tungsten nozzle or a hardened steel one pays for itself, and the high flow nozzle geometry helps because these materials want to be printed hot and fast. Second, every polyamide on this list must be dry. A wet spool prints a part with voids that cut the bolt bearing area by a third, and no amount of design rigour survives that. Keep the spool in a drying box and the numbers in this table stay true; skip that step and the table is fiction, exactly as described in why moisture tolerance differs by material.
The printed parts around the motor, not the motor mount
Once the motor is located and the rail is aligned, there is a small ecosystem of printed parts that make the machine work, and they are all easier to get right than the motor mount.
Cable management. A motor cable moving through a chain of printed links is a cable bending millions of times. The printed part's job is to set the bend radius, not to grip the cable. Grip it hard and the strands break inside the insulation where you cannot see them, and the fault appears as an intermittent lost step that you will spend a week blaming on the driver. Leave the cable loose in the link with 20 percent slack, and set the radius above ten times the cable diameter.
End stops and sensor flags. A printed flag passing a photo-interrupter needs to be opaque, parallel to the sensor face, and thin enough not to scrape. Print it in a dark material, at 0.8 mm, with the thin dimension in the layer plane. A flag printed with its thin dimension stacked in layers will delaminate at the tip within a few hundred cycles, and the machine will start homing in the wrong place.
Tensioners and idlers. A printed idler pulley running on a printed axle is a bearing without a bearing. It will work for a while and then it will squeal, because the two surfaces are the same material at the same roughness. The sliding surface rules for printed bearings apply in full: dissimilar materials, a generous clearance, and a shoulder that keeps the pulley from walking. A drop of PTFE-loaded grease doubles the life for almost no cost.
Guards and covers. These are the parts where a printed bracket's flexibility is an asset rather than a liability. A guard that flexes a little absorbs impact. A guard that is rigid cracks at the screw holes, because it is the least loaded part on the machine and therefore the one designed last and with the least care. Where the cover must seal against dust, follow the groove and face geometry that holds pressure rather than tightening a warped printed flange against a flat plate.
Printing the part so that the design survives
None of the geometry above compensates for a part printed in the wrong orientation. Machine brackets have a predictable set of failure directions, and each one maps to a layer direction.
- Motor mount: print with the mounting face on the bed. The bolt face is then flat, the pocket walls run vertically, and the belt-pull moment bends the walls across layers rather than peeling them.
- Rail bracket: print with the rail's registration shoulder vertical, so it is a printed wall rather than a bridged overhang. A shoulder printed as a bridge has a radius you did not design.
- Screw support block: print with the bearing bore vertical. A horizontal bore becomes an oval with a flat on top, and the bearing rattles in it.
- Anything with a slot: print with the slot's bearing face on the bed and the slot running vertically. This is the orientation that makes the slot an actual slot rather than a sagged bridge.
Wall count matters more here than anywhere else, because these parts have bolts passing through them. Three walls is the practical minimum for a bracket with a bolt hole; five is better. Infill beyond 40 percent contributes almost nothing to bolt bearing strength, since the load is carried in the shell and in the material immediately around the hole. This is the counterintuitive result already covered in why adding plastic often makes a printed part weaker, and it applies most strongly to small, bolt-heavy parts.
For the parts that must be dimensionally exact, a filament buffer and a dry spool are not luxuries. Inconsistent feed shows up as slightly under-extruded walls, which in a bracket means a bolt boss that is 0.2 mm thin on one side, and that is where it cracks.
Commissioning: what to check before you trust the machine
A newly assembled motion system has four failure modes and each one has a cheap test. Do these before running a job you care about.
- The rock test. Power off, grip the motor body and try to rock it. Any perceptible movement at the mounting face means the bolts have already lost preload or the face is not flat. Fix both before running.
- The hand-turn test. With the belt off, turn the screw or the shaft by hand through the whole travel. It should feel identical in both directions. A stiff spot in one direction is a bent screw or a misaligned rail, not a tight nut.
- The marker test. Draw a line across the bracket and the frame at the joint, and another across each bolt head and the bracket. Run the machine for an hour, then look. Any line that has moved is telling you which fastener is failing.
- The temperature check. After an hour of continuous motion, feel the motor mount with the back of a finger. A mount that is hot to the touch is above 50 °C and is creeping. That is the moment to change material, not the moment to drill a cooling hole.
Then, a week later, check the marker lines again. Printed machine parts do not fail on the first day. They fail in the second week, quietly, and the first symptom is usually a print that is slightly out of tolerance rather than an obvious mechanical break. The habit of measuring parts rather than judging them by eye is what catches this before it costs a batch.
When the printed bracket should not exist
There is a real category of machine part for which the correct answer is metal, and pretending otherwise wastes time and breaks machines. The list below is not a list of failures. It is a list of decisions made early that save the print for the work it is genuinely good at.
- Any mount whose bolt would need more than about 2 N·m to hold. The plastic will not maintain that preload, ever.
- Any bracket that sets the alignment of two precision components at once, such as a rail and a screw in the same block. Split the part and let metal do the datum.
- Anything bolted to a motor that runs above 70 °C at the flange. No unfilled polymer holds a joint there.
- Any part whose failure mode is the machine falling. Safety-relevant structure belongs in metal, and a printed guard is a guard, not a structure.
- Any part you would have to print in a material you cannot dry, and therefore cannot print correctly.
The inverse is just as true, and it is where most of the value is. Set-up jigs and drilling guides, cable routes, sensor flags, covers, spacers, adapters between two catalogue parts whose holes do not line up, transition pieces for a frame you are modifying: every one of these is a part that would otherwise take a week at a machine shop and does not need to hold a bolt preload for a decade. The rules for brackets that do not crack at the screw hole are all you need for that whole category. The printed bracket's best moment is not when a motor is bolted through it. It is when it makes a metal part fit a frame it was not designed for.
Ten rules, ordered by how much they save you
- Let steel carry the bolt tension and plastic carry the compression. If the bolt stretches plastic, the joint is temporary.
- Print the mounting face flat on the bed. Every downstream error starts with a face that was bridged.
- One round hole plus one slot. Two slots is not adjustment, it is a free body.
- Give the bolt a washer, not more torque. Dishing loses the preload that torque was trying to create.
- Do not locate the motor on its pilot boss. Locate on the fixings and give the boss a millimetre of clearance.
- Put an abutment under the belt-pull moment. The motor body is a convenient place to put a rib.
- Fix one end of a screw, float the other. A screw constrained at both ends is a screw you are bending.
- Use a shoulder to locate a rail, not a bolt hole. Bolt clearance is not an alignment feature.
- Choose the material by heat deflection temperature, not by tensile strength. A motor mount fails by relaxing, not by tearing.
- Mark the joint and look at it a week later. The failure will be there before the print quality is.
A machine is not made of movement. It is made of the things that do not move, and everything that seems to move is measured against them. The printed parts on a motion system are, almost without exception, on the side of the things that do not move, which is why they deserve the same attention as the gears. If you want the material side of that equation to be the easy part, the glass and carbon filled engineering grades do for a bracket what a machined part does for an assembly: they let you stop thinking about it. Print the mount, bolt the motor, mark the joint, and look again next week. That is the whole discipline.