CNC Micro Machining: 7 Critical Mistakes to Avoid for Perfect Precision

• 1
  Micro machining defect triage table
• 2
  Treating micro work as scaled-down macro work
• 3
  Ignoring tool runout and spindle health
• 4
  Treating burrs as a finishing problem
• 5
  Averaging dimensions instead of controlling every part
• 6
  Under-designing workholding and ignoring thermal growth
• 7
  Choosing a supplier on price instead of process control
• 8
  Step by step: correcting a failing micro process

The most common of all cnc micro machining mistakes is assuming that a process proven on a 200 mm aluminum plate will scale down to a 3 mm brass part. It does not. At small feature sizes, the ratio of cutting edge radius to chip thickness changes, and the tool stops cutting cleanly. It starts pushing material instead.

A 0.2 mm end mill has an edge radius that may be 1–2 μm. Your target chip load might be 2 μm per tooth. The edge is the same size as the chip, so ploughing and work hardening take over. The symptom is a bright, smeared surface and rapid tool wear.

The fix is not a slower feed. It is a higher feed per tooth with a shallower radial engagement, plus a spindle that can actually reach the required speed. We run small tools on machines capable of far higher rpm than a general-purpose mill, because surface speed has to stay in a workable band.

Macro rules also break down on rigidity. A micro tool has almost no stiffness, so any runout becomes a bending load. That leads straight into the second mistake.

• 1
  Rule of thumbIf the chip load is smaller than the tool edge radius, expect rubbing, not cutting.
• 2
  Check firstAsk what rpm the shop can hold, not just what tolerance it advertises.

On a macro job, 10 μm of runout is often absorbed by the cut. On a 0.1 mm tool, it is the whole tolerance budget. Runout turns a two-flute cutter into a one-flute cutter that is also being bent sideways on every rotation. The result is a slot that is wider at the top, a poor floor finish, and tool breakage that looks random.

Spindle health matters just as much. Bearings with normal wear still pass a macro test, but the thermal growth and vibration they add show up directly on micro features. We measure runout at the tool tip, not at the holder taper, because that is where the cut happens.

Hydraulic and shrink-fit holders get runout below roughly 1 μm at the tip. Standard collets in good condition are usually 3–5 μm, and worse once chips get into the slots. For features under 0.5 mm, holder choice is not a detail. It sets the ceiling on what the process can hold.

Tool life should be tracked by feature count, not by minutes in cut. A micro tool that has made 400 slots may still look fine and already be cutting oversize.

• 1
  Measure at the tipRunout at the taper tells you nothing about the cutting edge.
• 2
  Log tool changesChange micro tools on a count, not on a visual check.

On a micro part, the burr is not a cosmetic issue you clean up at the end. It is often larger than the feature that produced it. A 0.15 mm slot can carry a burr of 0.05 mm on each edge, which changes how the part fits and how fluid moves through it.

Hand deburring is not an option at this scale. A scraper or abrasive pencil removes material you cannot see and leaves a radius you did not specify. The better answer is to control the burr at the cut: sharper tools, a small edge break built into the toolpath, and a controlled finish pass that removes the hanging material before it folds over.

Where a burr-free edge is a functional requirement, say so on the drawing. Otherwise the shop will apply the standard deburr and the edge will be visually clean but dimensionally undefined.

The same thinking applies to tolerances. A ±0.02 mm tolerance on a 0.4 mm feature is not a tight tolerance. It is a different part every time.

• 1
  Define the edgeState whether a 0.05 mm edge break is allowed, required, or forbidden.
• 2
  Watch the ratioTolerance should not exceed about 5% of the feature size.

Micro parts are often measured in small batches, and it is tempting to report a mean value and call the process capable. That hides the parts at the ends of the distribution, and those are the ones that jam an assembly or fail a leak test.

Dimensional tolerances on micro features behave differently from macro ones. Tool wear over a 500-part run can move a 0.3 mm slot by 4–6 μm, which on a ±0.005 mm callout is the entire window. The process has to be re-centered during the run, not just at setup.

We inspect 100% before shipment and keep the raw material check, in-process monitoring and final inspection as separate steps. Reports are available on request. If a feature is critical, in-process probing is worth the cycle time because it catches drift while the part is still in the machine.

A supplier that quotes a single best-part number rather than a distribution has not measured the process.

• 1
  Ask for the spreadMinimum and maximum, not the average, on critical features.
• 2
  Plan for driftExpect 4–6 μm of tool wear movement over a long micro run.

A micro part is usually thin, and thin parts move when you clamp them. A vise tightened to normal torque can bow a 1 mm wall by more than the tolerance before the tool ever touches it. The part springs back after unclamping, and the measured dimension no longer matches the cut dimension.

The fix is a fixture designed for the part, not borrowed from the shelf. That often means a dedicated pocket, low-torque clamping, adhesive fixturing for very thin plates, or machining from a thicker blank and trimming the support afterwards. It costs setup time and saves the run.

Thermal effects run in parallel. The spindle, the coolant and the part all warm up over the first 30 minutes. On a 0.5 mm feature, a 2 °C change in a 20 mm aluminum part is already a few microns of growth. Warm-up cycles and coolant temperature control are process steps, not luxuries.

Residual stress in the stock is the third leg. Removing material lets the part move, so a face that was flat at the start can bow after the last pass. Stress-relieved stock or a rough-then-finish sequence with a pause in between limits the damage.

None of this shows up in a CAD review. It shows up in the first article.

• 1
  Clamp lightlyIf the fixture marks the part, it is also bending it.
• 2
  Warm up firstRun 20–30 minutes before the first critical cut.

The last of the cnc micro machining mistakes is a purchasing decision, not a machining one. Micro work is sensitive to factors that a standard quote does not capture: spindle condition, holder quality, coolant temperature, tool change discipline and how the shop measures what it makes.

A shop that subcontracts its heat treat, its finishing and its inspection cannot control the chain. When a micro part fails, no single party has the full picture. Vertical integration is not a marketing line here. It is what makes a corrective action possible.

Ask how runout is measured, how often tools are changed, what metrology is used on features below 0.5 mm, and whether in-process probing is available. The answers tell you more than a price sheet.

We have run micro and small-feature work since 2011 from three wholly-owned plants, with 127 high-precision CNC machines including 16 simultaneous 5-axis centers. Quotation and free DFM analysis come back within 12 hours, and production can start within 24 hours.

• 1
  Ask about metrologyA shop without the right CMM or vision system cannot verify micro features.
• 2
  Ask about the chainHeat treat, finishing and inspection should be under one roof.