CNC Milling Brass: 7 Essential Tips to Avoid Costly Mistakes and Maximize Precision
Brass cuts fast, which is exactly why small parameter errors turn into scrapped parts before anyone notices. This guide is written for design engineers, process engineers and buyers who need to judge whether a shop's brass process is under control. Read it and you can trace most defects back to grade choice, feed and speed, tool geometry, chip evacuation or inspection.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
Brass milling defects: symptom, cause, action
Three columns, one row per defect. Start at the symptom you can see on the part or hear at the spindle.
| Symptom | Likely cause | What to do |
|---|---|---|
| Short tool life, shiny wear land | C36000 run with steel or aluminium feeds | Raise surface speed to 150–300 m/min |
| Built-up edge on the cutting edge | Too low a feed per tooth for the alloy | Feed 0.05–0.15 mm/tooth, never rub |
| Stringy chips wrapping the cutter | Wrong grade or missing chipbreaker | Switch to C36000 or add a chipbreaker |
| Rough finish above Ra 1.6 μm | Small radial stepover and dull edge | Stepover 40–60% of Ø, change inserts early |
| Chatter marks on thin walls | Weak workholding, long tool overhang | Shorten overhang, add support, reduce axial depth |
| Bore size drifts over a run | Thermal growth and tool wear | Warm up spindle, check at fixed intervals |
Grade selection and cutting parameters decide the whole job
Brass is a family, not one material. C36000 free-cutting brass carries 2.5–3.5% lead, which is why it breaks chips cleanly and machines at high surface speed. C27400 and C28000 are stronger and lead-free or low-lead, but they form longer chips and push more heat into the tool. C110 copper and beryllium copper sit at the other end: gummy, prone to built-up edge, and slow to machine. Picking a grade for corrosion resistance and then running it at free-cutting speeds is one of the most common and most expensive mistakes in cnc milling brass.
Match the grade to the feature, not to the catalog. A valve body with a deep 6 mm bore in C36000 will hold ±0.005 mm all day. The same bore in C27400 needs more passes, more coolant and a slower feed, because the chip does not break and the tool rubs instead of cutting. If the drawing specifies a lead-free grade for a drinking-water or medical application, accept the cycle time cost and plan the toolpath for chip control from the start.
Feeds and speeds for brass are not a single number. A 6 mm three-flute carbide end mill in C36000 runs comfortably at 200–300 m/min surface speed with 0.08–0.12 mm per tooth. Drop to 0.03 mm per tooth and the edge rubs, work-hardens the surface and builds up material. Push past 0.20 mm per tooth on a small cutter and the tool deflects, the wall springs back and the finish shows scallops.
Depth of cut controls stability more than speed does. A radial stepover of 40–60% of the cutter diameter with an axial depth up to 1.5× diameter removes material quickly and keeps the load even. On a thin 2 mm wall, cut that axial depth in half and accept the extra passes. Chatter marks cost more to polish out than the passes cost to run.
- 1Free-cutting gradesC36000 for high volume, tight bores, short chips
- 2Lead-free gradesC27400 and C28000 need slower feed and stronger chip control
- 3Copper and beryllium copperTreat as gummy alloys: lower speed, sharp edge, plenty of coolant
Tool geometry, coatings and chip evacuation
Brass does not need a coating for wear resistance the way steel does, and the wrong coating causes more harm than good. AlTiN and TiAlN films are hard but chemically active against copper alloys at high temperature, and they tend to attract built-up edge. Uncoated micro-grain carbide with a polished rake face is the baseline for cnc milling brass. Where edge life matters on long runs, a thin diamond-like carbon or a polished PVD layer helps, but only if the edge is already sharp and the flute is open.
Geometry matters more than the label. Two or three flutes with a 10–15° helix and a positive rake cut brass with low cutting force. A high-helix cutter designed for aluminium pulls long chips out of the cut and throws them across the table, which is fine for C36000 and bad for C27400. For small bores under 6 mm, use a two-flute cutter with a short flute length; the extra room in the flute is what lets the chip leave.
Chip evacuation is the part most shops get wrong. Brass chips are heavy, they carry heat away with them, and they pile up in pockets and around fixtures. Recutting a chip on a finished surface leaves a gouge that no polishing will remove. Air blast at 4–6 bar plus through-tool or flood coolant keeps the pocket clear. On deep pockets, program a peck or a helical entry rather than a straight plunge, and never let the cutter sit in one spot while the spindle is running.
Fixtures trap chips. A soft-jaw vise with no relief at the bottom will pack brass dust under the part and lift it by 0.02–0.05 mm between operations. Cut chip relief grooves into soft jaws, blow the vise clean between parts, and check the seating face every few cycles. This one habit prevents more size drift than any parameter change.
- 1CoatingUncoated polished carbide first; avoid AlTiN on copper alloys
- 2Flutes and helix2–3 flutes, 10–15° helix, positive rake
- 3CoolantAir blast 4–6 bar plus flood; never cut dry in a deep pocket
Surface finish, deburring and post-processing
As-machined brass finishes land around Ra 1.6–3.2 μm with a fresh edge and a stable setup. Getting to Ra 0.8–1.6 μm is mostly about the last pass: a light radial stepover, a sharp insert and a constant feed that does not pause in a corner. Below Ra 0.8 μm you are polishing, not milling, and it is usually cheaper to bead blast or tumble the part than to chase the number on the machine.
Brass burrs are soft and sticky. They roll over instead of breaking off, so a hand deburr tool smears them along the edge. A controlled vibratory tumble with ceramic media removes them evenly and reaches internal edges that a scraper cannot. On sealing faces and O-ring grooves, specify the edge break on the drawing: 0.1–0.2 mm × 45° is enough for most brass parts and costs nothing extra if it is in the model.
Plating changes size. Electroless nickel adds 0.01–0.025 mm per surface, which closes a tolerance band fast. If a brass part is going to plating, machine the pre-plate size down and state it on the drawing. The same applies to anodizing on aluminium and to any conversion coating: tell the machine shop what happens after milling, or the finished part will not fit.
Handle brass parts gently after the last operation. It scratches easily, and a single fingerprint left before clear lacquer or plating becomes a permanent mark under the coating. Bag parts individually on cosmetic surfaces and keep them off bare steel benches.
- 1Last passLight stepover, sharp edge, constant feed
- 2DeburrVibratory tumble with ceramic media for internal edges
- 3Plating allowanceElectroless nickel adds 0.01–0.025 mm per surface
5-axis workholding and complex brass geometry
Complex brass parts with compound angles, deep pockets on several faces and thin ribs are the reason 5-axis machining exists. Doing them in three setups on a 3-axis mill adds two chances to lose position and two chances to scratch a finished face. One setup on a simultaneous 5-axis center holds the datums together and keeps the wall thickness consistent around a curved rib.
The trade-off is stiffness. A trunnion table has less support than a solid vise on a 3-axis machine, so cutting forces must come down. Reduce the axial depth on long-reach cuts, keep tool overhang under 4× diameter, and use a shrink-fit or hydraulic holder rather than a collet chuck where the reach allows it. Brass cuts with low force, which is why 5-axis works well here, but a 100 mm reach tool will still chatter if the parameters ignore the setup.
For parts under 200 mm, a Ø400 mm rotary table with a self-centering vise handles most brass work in one or two operations. For larger housings up to 4,000 mm, the work moves to a gantry-style machine and the process plan is built around access rather than around the part envelope.
Program 5-axis brass with the same chip logic as 3-axis. Helical entries, peel milling on deep pockets and a constant chip load through corners. The machine can move in five axes, but the cutter still needs a place for the chip to go.
- 1Setup countOne 5-axis setup replaces three 3-axis setups
- 2Reach limitKeep tool overhang under 4× diameter
- 3HolderShrink-fit or hydraulic for long-reach cuts
Inspection, documentation and what to ask a supplier
Brass parts drift slowly, not suddenly. A bore that starts at 10.00 mm may sit at 10.01 mm after 200 parts because the tool wore and the coolant warmed up. Inspecting only the first and last part of a run hides that drift. Measure a sample at fixed intervals, log the values, and adjust the offset when the trend appears, not when the part is already out of tolerance.
The inspection method has to match the tolerance. A ±0.005 mm bore is not verified with a caliper. Use a bore gauge or a coordinate measuring machine, and state the method and the environment on the report. Brass expands roughly 19 × 10⁻⁶ per °C, so a part measured straight off the machine at 30 °C and again in a 20 °C inspection room will differ by about 0.01 mm on a 50 mm feature. That is the whole tolerance band.
Ask for the material certificate, the inspection report and the first-article record before the run starts, not after a problem appears. A shop with ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022 systems can produce those documents as routine output. A shop without them will produce them only when you ask, which is a signal about the process behind the part.
Finally, treat the DFM review as part of the quote. If the drawing has a 0.5 mm internal corner on a 12 mm deep pocket, a shop that tells you the cutter cannot reach it is worth more than a shop that quotes it without comment.
- 1SamplingMeasure at fixed intervals through the run, not just first and last
- 2MethodBore gauge or CMM for ±0.005 mm; calipers are not enough
- 3DocumentsMaterial cert, inspection report, first-article record
Step by step: from drawing to first good part
Work through this order on any new brass part. Skipping a step usually shows up two operations later.
- 1Confirm the alloy and temperCheck the drawing against the material cert. C36000, C27400, C28000 and C110 machine differently; if the grade changes, re-plan feeds before cutting.
- 2Set the DFM baselineReview wall thickness, internal corner radii and depth-to-diameter ratios. Flag any pocket deeper than 4× the cutter diameter and any wall under 1.5 mm.
- 3Choose the cutter2–3 flutes, 10–15° helix, positive rake, uncoated polished carbide. For bores under 6 mm use a two-flute short-flute cutter.
- 4Set feeds and speeds150–300 m/min surface speed in C36000, 0.05–0.15 mm per tooth, radial stepover 40–60% of Ø, axial depth up to 1.5× Ø on stable setups.
- 5Fix the chip pathAir blast at 4–6 bar plus flood coolant. Cut chip relief in soft jaws. Use helical entry and peel milling on deep pockets.
- 6Machine the first part with a warm spindleRun the spindle at cutting speed for 10–15 minutes first. Measure the first part only after the machine has reached thermal steady state.
- 7Inspect and lock the offsetsMeasure critical features with a bore gauge or CMM, log the values, and lock the offsets. Then sample at intervals through the run.
- 8Plan post-processing before the last cutIf the part will be plated, anodized or lacquered, machine the pre-finish size and confirm the edge break on the drawing.
Brass milling questions engineers ask
Why does my cutter wear out so fast in brass, even though brass is easy to machine?
The usual cause is running steel or aluminium parameters. Below about 150 m/min surface speed, the edge rubs on the brass instead of cutting it. The rubbing work-hardens the surface, raises cutting temperature and wears the edge quickly.
Raise the surface speed to 150–300 m/min, keep the feed per tooth at 0.05–0.15 mm, and check that the cutter is not sitting in the cut while the spindle runs. A polished, uncoated carbide edge helps as well.
Can I machine lead-free brass the same way as C36000?
No. Lead-free grades such as C27400 and C28000 form longer, tougher chips and generate more heat at the edge. Reduce surface speed by roughly 20–30%, increase coolant flow, and add pecking or helical entry on deep pockets.
Expect a longer cycle time. The trade-off is real and should be quoted, not hidden.
How do I stop chatter on a thin brass wall?
Reduce the axial depth of cut and shorten the tool overhang. If the tool is longer than 4× its diameter, the setup is the weak point, not the parameter.
Add support behind the wall where the geometry allows, and use a shrink-fit or hydraulic holder. Cutting the axial depth in half and doubling the number of passes is often faster than polishing chatter marks out.
What surface finish can I expect from a normal brass milling operation?
As-machined brass typically lands at Ra 1.6–3.2 μm. With a fresh edge, a light stepover and a stable setup, Ra 0.8–1.6 μm is achievable on the last pass.
Below Ra 0.8 μm, plan on a finishing process such as tumbling, bead blasting or polishing rather than trying to mill to the number.
Does plating affect the tolerance on a brass part?
Yes. Electroless nickel deposits about 0.01–0.025 mm per surface, so a 20 mm shaft can grow by up to 0.05 mm on diameter. That is larger than a ±0.005 mm tolerance band.
Machine the pre-plate size down and state the plating thickness and the final dimension on the drawing so the shop can compensate.
What should I send with a brass part RFQ?
Send the 3D model, a 2D drawing with datums and tolerances, the alloy grade with temper, the surface finish callout, and the post-processing steps that follow machining.
Mention the quantity band and whether the part is a prototype or a production run. A shop can then return a DFM review and a quote, and you can compare processes rather than just prices.
Send us the brass part and the drawing
We review the model, flag the features that will not machine cleanly, and return a quotation with free DFM analysis within 12 hours. Prototypes and 10,000+ part runs both start on the same process plan.
12-hour quote100% inspection before shipmentNo minimum order quantityNDA on request