PCB Milling: CNC Prototyping Without Chemicals

Milling a circuit board is the fastest way to turn a schematic into something you can solder, and it is the only common prototyping method that produces a board without chemicals. A CNC machine cuts the copper away from the traces instead of etching it, which makes the process clean, fast and entirely unsuitable for production.

How PCB Milling Works

The machine holds a rotating cutter and moves it along toolpaths generated from the copper layers of the design. The cutter separates the traces by removing a thin channel of copper around each one, a technique called isolation routing, and then drills the holes and cuts the board outline.

Because material is removed mechanically, the achievable geometry is limited by the tool rather than by a photolithographic process. That single fact explains both the speed of the method and its limits.

Isolation Routing and Trace Separation

Isolation routing produces each trace as an island surrounded by a milled channel. The channel width must be at least the tool diameter, so the spacing between traces cannot be smaller than the cutter. A smaller cutter allows finer spacing but breaks more easily and cuts more slowly.

The copper remaining between traces is removed in passes that gradually clear the unwanted areas. Flat end mills of larger diameter remove bulk copper, while a sharp V-bit finishes the isolation outline.

Tool Selection and Geometry Limits

V-bits cut with their tip, and the depth of cut determines the width of the channel. That relationship is the source of the method’s main weakness: any variation in board flatness changes the depth of cut, which changes both the channel width and the trace width.

Flat end mills of 0.8 to 2 mm are used to clear large areas and to cut the outline. Cutting tools down to 0.2 mm are available, and the practical minimum feature size for trace and spacing generally lands between 0.15 and 0.25 mm depending on the machine, the tool and the operator’s patience.

Board Flatness and Fixturing

Flatness decides whether the job succeeds. A warped blank changes the depth of every cut across the board, producing traces that are too thin in one area and isolated not at all in another.

Two techniques address this. The first is to skim the surface of a sacrificial bed so that the blank sits on a known flat plane. The second is to probe the surface at several points and let the control software compensate for the height variation as it cuts. On double-sided boards, both sides must then be aligned to each other.

Double Sided Alignment

Double-sided milling requires the bottom layer to be registered with the top. The usual approach is to drill two or three tooling holes through the blank and the bed at the start, then flip the board onto pins that reference the same holes for the second side.

The remaining error comes from the thickness of the material and from the flip itself. Any misalignment shifts the pads on one side relative to the other, which matters most for through-hole parts with tight spacing and for anything that needs a via to connect the layers.

Vias and Layer Connections

A milled board cannot be plated, so vias are mechanical rather than metallurgical. The usual solution is to press a piece of wire or a rivet into the hole and solder it on both sides, or to place a component lead through the hole where the circuit allows.

For boards with many connections between layers, that becomes the dominant labour. It is one of the reasons milled boards are usually single-sided or simple double-sided designs, and why complex prototypes move to a fabricated board.

What Milling Does Well

Turnaround is measured in hours rather than days, which makes it ideal for a design that must be tested today. There is no chemistry, so a machine can run in an office or a laboratory. The process also handles unusual substrates, including materials that would be difficult to etch.

Iteration is the real benefit. When a change can be cut and assembled the same afternoon, a design team tests more variants and finds problems earlier than a team waiting for a fabrication house.

Where Milling Falls Short

Fine pitch components are the hard limit. A 0.5 mm pitch device needs traces and spacing that a milled board cannot reliably hold, and the same applies to microvias, controlled impedance and multilayer constructions.

Solder mask is usually omitted, which means the copper is exposed and the board is harder to assemble cleanly. Some shops apply a printed mask, but the registration is coarse compared with a fabricated board, and the result is a prototype that behaves differently from the production version.

When to Choose Milling or Fabrication

Choose milling for a first article with through-hole or coarse SMT components, for a mechanical fit check, or for a design that must be tested before a fabrication cycle can be scheduled. It is also the right answer for a board with an unusual substrate or a very small quantity.

Choose fabrication as soon as the design involves fine pitch, controlled impedance, more than two layers or a solder mask. The comparison is not really about cost per board but about whether the prototype represents the production process, and reviewing design and fabrication practice alongside manufacturable design guidelines makes that judgement easier.

Milling Versus Etching for Prototypes

A milled board and an etched board of the same circuit differ in more than appearance. Etching produces uniform trace widths determined by the artwork, a solder mask with tight registration and plated holes that connect layers reliably. Milling produces variable widths determined by the depth of cut, no mask unless it is added separately and no plated barrels.

The practical consequence is that a milled prototype validates the circuit and the mechanical fit but does not validate the fabrication. A design that works when milled may still fail when produced because the etched version has different parasitics, a different finish and a different via structure.

Where the purpose is to prove the concept, that limitation is acceptable and the speed is worth it. Where the purpose is to confirm that the design can be manufactured, the prototype has to come from a fabrication house, which is why multilayer prototype requirements are worth reading before ordering one.

Machine Maintenance and Consumables

Milling consumes tooling quickly. Cutters blunt, break on a bad plunge and produce progressively worse geometry as they wear. A worn cutter effectively has a larger tip, which widens the isolation channel and thins the traces, so a job that worked last week may not work this week with the same settings.

Collet cleanliness, spindle runout and the condition of the sacrificial bed all affect the result as much as the toolpath does. Replacing the bed surface regularly and keeping a log of tool life turns an unpredictable process into a repeatable one, which is what makes the method genuinely useful for iterative development.

FAQ

Can a milled board be used for production? Not practically. The process is slow, tool wear changes the geometry across a batch and the achievable feature size is far above a fabricated board. It is a prototyping tool.

Do I need a solder mask on a milled board? It helps, particularly on a dense board, but applying it accurately is difficult. Many designs use a milled board with no mask and rely on careful assembly and a short service life.

What is the smallest trace a milled board can hold? Around 0.15 to 0.25 mm in practice, depending on the machine, the cutter and how flat the blank is. Attempting to go below that usually produces traces that vary in width across the board, which makes the result unpredictable rather than merely difficult.