CNC Machining Portland: How the Process Works and When to Use It

• 1
  What CNC machining Portland shops actually do inside the machine
• 2
  Three-axis, four-axis, or five-axis: what the extra axes buy you
• 3
  Reading a tolerance callout like a machinist
• 4
  How material choice changes the cut
• 5
  Where CNC machining stops being the right answer
• 6
  Choosing a machine and process route

CNC machining is subtractive. A rotating cutting tool moves along a programmed path and shears material away from a solid block. The controller reads G-code, drives ball screws and linear guides, and keeps the tool tip on a path calculated from your CAD model. Nothing about the shape is formed by a mold or a die. That is why the process handles geometry that casting and stamping cannot.

The physics matter more than the machine brand. Cutting speed, feed per tooth, and radial engagement decide whether the tool shears metal or rubs it. Rub too long and heat builds at the edge, the tool wears fast, and the surface tears. A typical aluminum roughing pass runs at 200–400 m/min surface speed with 0.05–0.15 mm feed per tooth. Steel drops to 80–150 m/min. These numbers shift with hardness, coating, and coolant.

Every cut leaves a tool mark. The height of that scallop is the surface roughness, measured in Ra. A sharp finishing insert running a light stepover can hold Ra 0.8–1.6 μm. Push the feed and you land at Ra 1.6–3.2 μm, which is fine for a bracket but not for a seal face. Roughness is a process choice, not a machine default.

This is where CNC machining Portland buyers should focus their questions. Ask how the shop sets feeds and speeds, not how many machines it owns. The answer separates a real process engineer from a reseller.

A three-axis machine moves the tool in X, Y, and Z only. The part stays put. This covers most prismatic work: plates, housings, brackets, and manifolds with features reachable from one direction. Setup is simple and rework is rare. If your part has holes on four sides, each side needs its own setup, and every setup adds a position error.

A four-axis machine adds a rotary table, usually around the X axis. Now the part can index to a new face without being unclamped. That removes one class of error and one class of labor. Shafts with cross-drilled holes, hydraulic blocks, and parts with features at 90° to each other are natural fits. Tolerance stacks tighten because the datum does not move between operations.

Five-axis machining adds a second rotary axis. The tool can tilt, so it approaches the part from an angle instead of straight down. This matters for two reasons. First, undercut geometry and contoured surfaces can be cut in one setup. Second, a short, stiff tool can reach deep features that would need a long, flexible tool on a three-axis machine. Short tools chatter less, so the finish and the dimensional result improve.

The tradeoff is cost and programming time. Five-axis toolpaths take longer to simulate and verify. For a simple plate, the extra axes add nothing. For an impeller, a turbine blade, or a medical implant with organic curves, they are the only way to hit the print.

• 1
  3-axisPrismatic parts, one dominant face, tight budget
• 2
  4-axisShafts and blocks with features around a single axis
• 3
  5-axisContoured surfaces, undercuts, deep cavities, one-setup accuracy

A tolerance is a promise about how close the finished size sits to the nominal. On a drawing, ±0.005 mm (±0.0002 in) is a tight callout. It is achievable on a rigid machine with a stable setup and temperature control, but it is not free. Every tight dimension needs its own inspection step and its own fixturing thought.

General tolerances cover the rest of the print. Most shops apply a block tolerance such as ±0.1 mm for unmarked dimensions. If you mark every dimension ±0.01 mm, the shop must inspect all of them, and the quote reflects that. A better practice is to mark only the dimensions that actually control fit or function. The rest can ride on the block tolerance.

Geometric callouts change the game again. Flatness, perpendicularity, and position tolerances are measured on a CMM, not with calipers. A true position of Ø0.05 mm forces the shop to control the datum scheme, the fixture, and the machine alignment together. That is fine, as long as the callout is real. A position tolerance tighter than the assembly needs just adds cost.

Materials move. Aluminum expands about 23 μm per meter per °C. A 300 mm part that warms by 5 °C during roughing grows by roughly 0.035 mm before finishing starts. Shops that hold ±0.005 mm on large parts either rough, cool, and finish, or they control the room temperature. Ask which approach a supplier uses before you accept a tight quote.

Aluminum 6061-T6 is the default for prototypes and many production parts. It cuts fast, holds a good finish, and takes anodizing well. 7075 is stronger but gummier and more prone to leaving a torn surface, so it usually needs sharper tooling and lighter passes. 2024 sits between them and is common in aerospace brackets.

Stainless 303 is the free-machining grade and behaves well on a lathe. 304 and 316 work-harden if the tool dwells, so the shop must keep the feed up and never let the cutter rub. 17-4PH adds strength after heat treatment, which means the part is often machined in the annealed state and then aged. That aging step shrinks the part slightly, and the shop must plan for it.

Titanium TC4 (Ti-6Al-4V) conducts heat poorly, so the cutting edge absorbs almost all of it. Tool life drops and feeds must come down. Inconel is worse. Both are machinable, but cycle times are several times longer than aluminum and the tooling cost shows in the quote. Plastics behave differently again: POM and PEEK machine cleanly, while ABS and PP tend to burr and need sharp tools plus air blast to clear chips.

The point is simple. Material is not a dropdown. It decides feeds, speeds, tooling, and sometimes the machine itself. Tell the shop the alloy, not just the family.

CNC machining wins when the part is complex, the volume is low to medium, and the geometry needs to be exact. It loses when the part is simple and the volume is high. A stamped bracket or a die-cast housing will beat a machined one on unit cost once the tooling is amortized. The crossover depends on the part, but a few thousand units is a common tipping point.

Deep pockets and thin walls are another boundary. A pocket deeper than about four times its cutter diameter needs a long tool, and a long tool deflects. Thin walls below 0.5 mm flex under cutting force and chatter. Both are doable, but the shop has to slow down, and the cost climbs. Sometimes a redesign that relaxes the wall or the depth saves more money than any supplier negotiation.

Surface finish has a ceiling too. Mirror finishes below Ra 0.2 μm usually need polishing after machining, not a better cutter path. And features smaller than the smallest available tool simply cannot be cut. A 0.5 mm internal corner needs a 0.5 mm cutter, which is fragile and slow.

None of this is a reason to avoid the process. It is a reason to bring the shop in early, while the geometry is still soft and cheap to change.