CNC Machining of Toothbrushes: Where the Accuracy Comes From

• 1
  Why CNC machining of toothbrushes lives or dies at the hole
• 2
  Tool runout and micro-drilling limits
• 3
  Fixturing a thin, curved handle without crushing it
• 4
  Surface finish, burrs and cleanability
• 5
  When machining wins and when molding wins
• 6
  Machining vs molding for a brush handle

A manual brush head carries 20 to 40 tufts. Each tuft sits in a blind or through hole, typically Ø0.8 mm to Ø2.0 mm depending on filament diameter and tuft density. The bristle anchoring method decides how tight the hole tolerance has to be. Stapled tufts need a slightly larger bore so the anchor wire can pass. Hot-melt or in-mold tufting grips the filament directly, so the hole can run tighter.

What matters to the tufting machine downstream is hole-to-hole pitch, not the absolute diameter. If pitch drifts, tufts collide or leave visible gaps in the row. Pitch tolerance on a machined head can hold around ±0.02 mm across a 25 mm row. That number is set by the machine's positioning accuracy and by how the part is clamped, not by the drill itself.

Hole depth matters for a different reason. Too shallow and the anchor has nothing to bite. Too deep and the drill breaks through into the hollow neck, which turns the handle into a leak path for water and toothpaste. A depth callout with a flat-bottom drill and a controlled peck cycle is the usual answer.

• 1
  Ø0.8–2.0 mmTypical bristle hole range for manual brushes
• 2
  ±0.02 mmRealistic hole-to-hole pitch across a 25 mm head row
• 3
  Depth ±0.05 mmNeeded when the head wall is thin

Below Ø1 mm, the drill behaves less like a cutter and more like a flexible shaft. Runout at the tool tip multiplies. A 0.010 mm runout on a Ø0.8 mm drill means one flute does most of the work, the hole bell-mouths, and the drill walks on entry. On a machined brush head that shows up as a tuft that sits 0.03 mm off the row.

Spindle speed and feed must be matched to the drill's capability. A common starting point for Ø1 mm carbide in POM or ABS is 12,000–18,000 rpm at 0.02–0.05 mm per revolution. Go too slow and the plastic smears. Go too fast without adequate chip evacuation and the flutes pack, which is the usual cause of a snapped drill and a scrapped head.

Coolant choice depends on the polymer. ABS and PC tolerate air blast or a light mist. PEEK and other high-temperature plastics generate enough heat at the drill tip to soften the wall, so a chilled air or minimum-quantity lubricant setup keeps the hole round. We run 16 simultaneous 5-axis centers and a set of three-axis machines for this class of work, which lets us keep the micro-drilling cells separate from larger roughing.

• 1
  Runout under 0.005 mmTarget at the drill tip for holes below Ø1 mm
• 2
  Peck depth 1×DCommon starting peck for deep bristle holes
• 3
  Air or MQLPreferred over flood coolant on most polymers

A brush handle is a long, thin, curved part with a hollow or ribbed interior. Clamping force that would be fine on a bracket will bow the neck. The fix is to spread the load: soft jaws machined to the handle's outer profile, plus a support pin or low-melt fixture inside the neck if the wall is under 1.5 mm.

For a machined prototype, the usual approach is to cut the blank from a solid block, clamp it in a custom soft-jaw pocket, then drill and profile in the same setup. That keeps the head and the handle in one coordinate system. Moving the part between operations is where pitch errors creep in.

When the geometry has undercuts at the head or a compound curve along the thumb rest, a 4-axis or 5-axis setup avoids a second fixture. The rotary table indexes the part so the drill approaches normal to the surface. On our floor that means a Ø400 mm rotary table on a mill-turn center, which handles a 150 mm handle without re-clamping.

• 1
  Soft jaws to profileKeeps clamping stress off the neck
• 2
  One setupHead and handle in the same coordinate system
• 3
  Ø400 mm rotary tableIndexes long handles without a second fixture

A molded handle comes out of the tool with a skin. A machined handle comes out with tool marks, and every mark is a place where residue can sit. On a brush head this is not cosmetic. The area around the tuft holes and the neck seam are the two spots that collect paste and water.

Machined plastic typically lands at Ra 1.6–3.2 μm as cut. A light bead blast or a vapor polish can bring that to Ra 0.8–1.6 μm without rounding the hole edges, which matters because a rounded hole edge reduces tuft pull-out force. If the drawing calls for a smoother surface, polishing is done with a controlled process so the hole diameter does not grow.

Deburring is the step that gets skipped on prototypes. A broken edge at 0.05 mm × 45° is enough to remove the feather from a drilled hole without changing the functional diameter. Inspect under magnification, not by feel.

• 1
  Ra 1.6–3.2 μmAs-machined finish on most polymers
• 2
  Ra 0.8–1.6 μmAfter bead blast or vapor polish
• 3
  0.05 mm × 45°Typical deburr callout at hole edges

Injection molding produces the same handle in seconds. CNC machining takes minutes. So the question is never which process is better in the abstract. It is which process fits the stage of the program.

Machining wins for functional prototypes where the tufting pattern is still changing, for low-volume clinical or trial runs, for testing a new filament anchor geometry, and for one-off fixtures used to test a tufting machine. Machining also wins when the material is hard to mold, such as carbon-fibre filled polymer or a filled PEEK head for a specialist brush.

Molding wins once the design is frozen and the annual volume is in the tens of thousands. The tooling cost is the barrier. Below roughly a few hundred parts a year, a machined batch is usually cheaper than a tool, and the lead time is days instead of weeks. Above that, the tool pays for itself.

There is a middle path. Machined parts validate the design and the tufting process, then the same CAD data goes to a mold. If the machined head passes, the mold cavity is cut to match. If it fails, you have lost a few days, not a tool.

• 1
  Prototype or trial runMachining. Design still moving
• 2
  Under a few hundred a yearMachining. Tool cost is not justified
• 3
  Tens of thousands a yearMolding. Cycle time decides it