Micro CNC Mill Basic Guide for Engineers

• 1
  Key takeaways
• 2
  What a Micro CNC Mill Actually Does Differently
• 3
  Tool Runout and Why It Sets Your Real Tolerance
• 4
  Speeds, Feeds and Heat in the Micro CNC Mill Basic Guide
• 5
  Workholding and Setup: Where Micro Accuracy Is Lost
• 6
  Which Materials Suit a Micro CNC Mill
• 7
  When a Micro CNC Mill Stops Being Economic
• 8
  Inspection and the Real Meaning of ±0.005 mm
• 9
  Feature Size vs Process Choice

A micro CNC mill is not a small version of a big VMC with the same rules scaled down. The spindle still turns at 20,000 to 40,000 rpm in most shops, but the cutter is now Ø0.5–3 mm. At that scale the edge radius of the tool is a real fraction of the chip thickness, so the cut no longer behaves like a clean shearing action. Part of the material rubs instead of shearing, and that rub turns into heat in a tool with almost no mass to absorb it.

The second change is stiffness. Tool deflection scales with the cube of the diameter. A Ø1 mm carbide end mill has roughly 1/64 the bending stiffness of a Ø4 mm tool of the same length. You cannot fix that with a heavier machine base. You fix it by taking shallow axial cuts, keeping tool overhang short, and accepting that the finishing pass may take longer than the roughing pass.

The third change is chip size. A micro mill produces chips measured in microns. They do not fall out of the flute. They sit in the pocket, get re-cut, and dull the edge. This is why air blast, mist, or a high-pressure through-spindle coolant matters more here than on a large part. The chip evacuation problem, not the cutting problem, is what usually drives the cycle time up.

None of this makes the process fragile. It makes it narrow. There is a band of part sizes and feature sizes where a micro CNC mill basic guide like this one is the right tool, and outside that band a larger mill or a different process wins on both cost and tolerance.

Runout is the total indicated reading of the cutting edge as the spindle turns. On a Ø1 mm two-flute end mill, 5 μm of runout means one flute cuts a chip twice as thick as the other. The overloaded flute wears first, the cutting force pulses once per revolution, and the wall finish shows a pattern that no feed or speed change will remove. Check runout with a dial indicator on the flute, not on the shank.

For micro work, hold runout under 3 μm if you can, and under 5 μm as a practical shop floor limit. That means shrink-fit or hydraulic holders rather than a standard collet chuck. A worn collet nut can add 10 μm on its own. This is a tooling decision, not a machine decision, and it is often the cheapest accuracy upgrade available.

Runout also sets the minimum chip load you can run. If the target chip load is 2 μm per tooth and runout is 6 μm, one flute is cutting 5 μm and the other is rubbing. Rubbing generates heat without removing material, and heat is what kills a Ø0.8 mm tool in a deep pocket. Keep the chip load above the edge radius of the tool, usually 1–2 μm for a sharp micro end mill.

The practical test is simple. Cut a test pocket in the same material and measure the wall with a micrometer or a CMM. If the wall tapers more than 10 μm over 10 mm of depth, the problem is tool deflection or holder runout, not the controller.

Surface speed is the starting point. In aluminum 6061, carbide micro tools run best between 150 and 250 m/min surface speed, which at Ø1 mm means roughly 48,000 to 80,000 rpm. Most spindles cannot reach that, so shops run lower rpm and accept a smaller chip load. The cut still works, but the heat has more time to soak into the tool.

Feed per tooth should stay above the edge radius. For a sharp micro end mill that means 1–3 μm per tooth in aluminum and 0.5–1.5 μm in stainless. Below that the tool rubs. Above about 5 μm the cutting force rises enough to deflect a long, thin tool. The window is narrow, and it is narrower in titanium and Inconel than in aluminum.

Axial depth of cut is where you buy back stiffness. A common starting point is 5–10% of the tool diameter for roughing, so 0.05–0.1 mm on a Ø1 mm tool. Radial engagement can be higher, 30–50% of diameter, because the tool is strongest in the radial direction. This is the opposite of what you would do with a Ø12 mm tool, and it is the single most common mistake in micro milling.

Cooling matters because the chip is small and the heat has nowhere to go. Air blast is usually enough for aluminum and plastics. For stainless, titanium and Inconel, use mist or through-tool coolant. Flood coolant alone often fails on micro tools because the stream pushes the tiny chips back into the cut instead of lifting them out.

A micro CNC mill can hold ±0.005 mm on a well-supported part. It cannot hold it across three re-clamps. Each time you unload and reload the part, you add a setup error that is usually larger than the machining tolerance itself. Design the part so the critical features are reachable from one or two sides, and keep the datum surfaces flat and accessible.

Thin walls are the other common failure. A 0.3 mm wall in aluminum will move when the tool passes it, because the wall has almost no stiffness. Support it with wax, a low-melt fixture, or leave a sacrificial rib that gets removed in the last operation. For walls under 0.5 mm, expect to take lighter finishing passes and to inspect between them.

Vacuum fixturing works well for flat plates down to about 0.5 mm thick, provided the plate is flat to begin with. For small prismatic parts, soft jaws machined in place to the part profile hold better than a vise with parallels. The jaws should be cut with the same tool and the same offsets as the part, so the clamping geometry matches the machining geometry.

Measure the part on the machine when you can. A touch probe or an optical tool setter lets you verify a critical feature without breaking the setup. If the feature is out, you can correct it in the same clamp. Once the part comes off the table, you are inspecting, not adjusting.

Aluminum is the easy case. 6061, 6061-T6, 7075 and 2024 all cut cleanly with sharp micro tools, produce continuous chips that clear well with air blast, and hold ±0.005 mm on stable geometry. 7075 is stiffer and gives a better wall finish on thin features, but it is more prone to chipping at the tool edge if the feed drops too low.

Brass and copper are also good. C36000 free-cutting brass machines fast and leaves a fine finish with almost no built-up edge. Copper C101 and C110 are gummy, so keep the chip load up and use a polished flute. Beryllium copper cuts well but needs dust control, since the dust is a health hazard.

Stainless 303 and 304 are workable but slower. They work-harden if the tool rubs, so the chip load must stay above the edge radius at all times. 316L and 17-4PH are harder still, and micro features in these grades usually need more finishing passes and more frequent tool changes. Titanium TC4 (Ti-6Al-4V) and Inconel are the hardest case in this list. They are machinable at micro scale, but the tool life is short and the process is best reserved for features that cannot be made another way.

Plastics behave differently again. POM and PEEK cut cleanly but tend to burr on the exit edge. PMMA chips easily, so use a sharp tool and a positive rake. Carbon fiber reinforced plastic is abrasive and will wear a micro tool quickly; diamond-coated tools help, but the cost per part rises. For any of these, confirm the material grade before quoting, because a small grade change can move the feed window.

The cost of a micro-milled part is driven by cycle time and tool changes, not by the material removed. A Ø1 mm tool cutting a 20 mm deep pocket may run at 0.05 mm axial depth, which means 400 passes. Each pass takes seconds, but the total adds up, and a single tool break in the middle of that pocket can scrap the part. Quoting a micro part from a 3D model without checking the depth-to-diameter ratio is how estimates go wrong.

A useful rule is to keep pocket depth under 5× the tool diameter, and under 3× if the tolerance is tight. Beyond that, chip evacuation gets unreliable, the tool deflects, and the wall tapers. If the design needs a deeper pocket, consider splitting the part, machining from both sides, or using EDM for the deep section and milling only the shallow features.

Part count changes the answer too. For one prototype, micro milling is often the fastest route because no tooling is needed. For a thin, flat part in the thousands, stamping or photo etching will beat milling on unit cost. For a small prismatic part with tight features, micro milling may stay competitive even at higher volumes, because the alternative processes need their own tooling and setup.

The honest boundary is this. Micro milling wins when the part is small, the features are defined by a few tight dimensions, and the quantity is low to moderate. It loses when the part is large, the features are deep and narrow, or the quantity is high enough that a dedicated process pays for itself.

A tolerance of ±0.005 mm is a process capability statement, not a promise on every dimension of every part. It applies to features that are accessible to the tool, supported by the fixture, and measured with an instrument that can resolve 1 μm. On a deep pocket with a thin wall, the achievable tolerance may be ±0.02 mm, and no amount of inspection will change that.

Inspection should match the feature. A micrometer works for outside dimensions. A bore gauge or pin gauge works for small holes. A CMM or optical comparator works for position and profile. For micro features, an optical system is often the only practical way to measure without touching the part and pushing it out of tolerance.

In-process checks reduce scrap. On a long cycle, stop after the roughing pass and check one critical dimension. If the wall is thick or the pocket is shallow, adjust the offset before the finishing pass. Catching a 20 μm error early is cheaper than finishing the part and finding it at final inspection.

Keep the reports with the parts. A dimensional report tied to the setup, the tool list and the machine gives the buyer a traceable record. That record is what makes a tight tolerance credible on the next order, not just on the first article.