Start with what the part must do
Separate functional features from decorative ones. A bearing seat, locating face or gear tooth needs a different level of control from a cosmetic cover. Mark which dimensions affect movement, which surfaces touch, and which features must remain accessible during assembly. This prevents effort being spent on a flawless outer finish while the important interface remains poorly defined.
Choose a material for its behavior in service and during manufacture. Brass is often convenient to machine, but alloy composition matters. Aluminum can be light and easy to shape while presenting wear or galling concerns in some sliding contacts. Steel offers many strength and wear options but may demand different tooling and corrosion protection. Engineering polymers introduce stiffness, creep, temperature and moisture considerations. No material is automatically best because a part is small.
Leave yourself something to hold
Workholding locates a component and resists process forces. A miniature part may deflect under clamping pressure before cutting even begins. Thin walls can spring back after release, leaving a feature that measured correctly in the fixture but no longer does. A useful plan keeps material where it supports the work and removes temporary stock only after the critical operations are complete.
Consider whether related features can be produced in one setup. Turning a shaft and its bearing journal together can help preserve their relationship. Repeatedly moving a part between fixtures adds locating uncertainty. Soft jaws, collets or dedicated fixtures may help, but each must be suitable for the machine and the operation. Never improvise a holding method that puts hands near rotating tooling.
Small tools are not scaled copies of large tools
In micromachining, tool edge radius, runout and material microstructure can become important relative to the intended chip thickness. Runout means deviation of a rotating feature from its intended path. If one cutting edge does most of the work, the tool can wear unevenly or break. Very small drills and cutters offer little warning before failure.
The correct speeds, feeds, engagement and coolant strategy depend on the exact tool, material, machine and operation. Use manufacturer guidance and trained supervision rather than a generic miniature recipe. A machine with a high-resolution position display is not automatically capable of producing equally small features: stiffness, thermal behavior, spindle condition and process forces still matter.
Electrical discharge machining, including wire EDM, removes electrically conductive material through controlled electrical discharges. It can provide useful alternatives for difficult profiles, but wire access, corner geometry, surface condition and equipment capability impose constraints. It is not a universal replacement for milling or a way to machine arbitrary nonconductive materials.
Use additive processes where they fit
Material-extrusion printing can be excellent for layout models, handles and larger demonstration mechanisms. Layer adhesion, anisotropy, hole shape and surface texture can make it less suitable for a tiny bearing seat without further work. A printer’s nominal layer height does not state the dimensional accuracy of every feature.
Resin processes can resolve different kinds of detail, but orientation, supports, cleaning and post-curing influence the final part. Uncured photopolymer requires the protective measures, handling and disposal specified by the material supplier. A detailed printed gear is not automatically a durable gear. Evaluate tooth strength, contact conditions and dimensional stability before relying on it.
For genuinely microscopic systems, photolithography and related semiconductor processes can define structures in layers. Richard Feynman’s lecture “There’s Plenty of Room at the Bottom” is a useful historical entry point into small-scale thinking, but modern MEMS fabrication is a specialized discipline. Cleanroom processing is not simply workshop machining performed with a stronger magnifier.
Burrs can be functional obstructions
A burr that looks negligible on a larger component may block a small passage or prevent a miniature part from seating. Deburring must remove the unwanted edge without rounding a locating surface or enlarging a critical hole. Decide how an internal edge will be inspected before choosing a process that leaves it inaccessible.
Cleanliness also has a relative scale. A particle large enough to span a running clearance can change friction or jam a mechanism. Keep finished components protected and use cleaning methods compatible with their materials, coatings and lubricants. Do not direct compressed air at skin or use it casually to scatter fine debris around a workshop.
Technical note: a dimension is not a process capability
One successful part does not demonstrate a stable production process. Capability assessment requires suitable measurement, enough representative data and a process that is under statistical control. For a one-off project, record actual dimensions and the matching strategy without claiming repeatability that has not been established.
Assemble in an order you can reverse
Before making parts, rehearse assembly using a drawing or a nonfunctional mockup. Check tool access, fastener direction and whether a later part traps an earlier one. A tiny screw that can be inserted but cannot be tightened is a design problem, not a patience test. Provide references that prevent reversed or ambiguous placement.
Inspect motion after each meaningful stage. Add one constraint at a time, then observe whether the mechanism still behaves as expected. This makes faults easier to locate than completing the whole assembly and discovering that it binds. Record matched parts and any deliberate adjustments, and keep energy sources disconnected during mechanical checks.
A good first project is a hand-driven demonstrator with visible joints, generous access and replaceable links. Its value comes from a repeatable manufacturing and inspection route. Once that route works, reduce dimensions or add complexity one controlled change at a time.
Bench references
- Sandvik Coromant: machining knowledge — tooling and process principles; follow application-specific guidance.
- NIST: additive manufacturing — measurement and process research.
- Caltech: Richard Feynman resources — historical scientific context.
- HSE: work equipment and machinery — machinery hazards and controls.