Should I Buy a CNC Machine? The Engineering Trade-Offs

• 1
  Should I Buy a CNC Machine? Start With the Part, Not the Price
• 2
  Tolerance Band Decides the Machine Class
• 3
  Volume and Utilization: The Numbers That Break the Case
• 4
  What the Invoice Does Not Show
• 5
  Buy Versus Outsource: Match the Row to Your Situation

A machining center is a tool that removes metal along programmed paths. Buying one only makes sense if you can keep it cutting. The purchase price is the smallest number in the whole calculation. The larger numbers are floor space, power, compressed air, tooling, coolant disposal, programming hours and the person who sets up the vise at 7 a.m.

So the useful first question is not what a machine costs. It is what your part family looks like. A bracket with two holes and a face is a different animal from a housing with five intersecting bores, a 0.4 mm wall and a sealing groove on a curved face. Write down the worst part you expect to make in the next two years, not the easiest one. That single part sets the machine class you need.

The part also decides whether you need a fourth or fifth axis. If every feature is reachable from one direction, a three-axis mill with two vise setups will do the job. If the part has features on four or five sides, you are looking at either multiple fixtures or a machine that tilts and rotates the work. Fixture count is where hidden cost lives: each setup adds labor, adds stack-up error and adds a chance to scrap the part.

One more thing before the money talk. Ask whether the part geometry is stable. Parts that are still changing every week do not belong on a machine you just financed. They belong on a supplier who can absorb the revisions while you settle the design. Buy the machine when the drawing stops moving. Until then, you are paying interest on someone else's learning curve.

• 1
  Simple 3-sided partsA three-axis mill and one or two fixtures usually cover it.
• 2
  Features on 4–5 sidesCount setups first. Past three setups, a 5-axis machine wins on labor.
• 3
  Thin walls under 1 mmVibration and tool deflection dominate. Rigidity matters more than spindle speed.
• 4
  Designs still in fluxOutsource until the drawing freezes, then reassess the volumes.

Tolerance is where most first-time buyers get caught. A general machining tolerance of ±0.1 mm is easy on almost any modern vertical mill with a good operator. Move to ±0.025 mm and you need thermal stability, a warm-up routine, a controlled room and a probe or a skilled setup person checking datums. Move to ±0.005 mm and the machine itself is only part of the story.

The rest of the story is the environment and the metrology. A machine that sits next to a loading dock sees 8 °C swings across a shift. Steel grows roughly 11 μm per meter per °C. On a 300 mm aluminum part, a 5 °C swing moves the feature about 18 μm before the tool even touches it. You cannot inspect your way out of that; you have to control the room.

Then there is inspection. If you promise ±0.005 mm, you need a CMM or a high-accuracy measurement setup that can resolve a fifth of that band. A caliper is not a metrology tool at that level. Budget for the gauge, the granite plate, the temperature-controlled corner and the person trained to use them. Buyers forget this line item constantly.

A practical rule: match the tolerance you need to the tolerance you can verify. If you cannot measure it in-house, you cannot hold it in-house. In that case, outsourcing to a shop with the gauge and the certification trail is cheaper than buying both the machine and the metrology.

GreatLight holds ±0.005 mm (±0.0002 in) on production parts and inspects 100% before shipment, with raw material checks, in-process monitoring and final reports on request. That infrastructure is exactly what a small in-house setup lacks on day one.

• 1
  ±0.1 mmAny decent 3-axis mill with a competent operator.
• 2
  ±0.025 mmTemperature control, warm-up cycle, probe or careful setup.
• 3
  ±0.005 mmControlled room, CMM-level inspection, documented process.

A machine only pays back when the spindle turns. Count the hours it will actually cut, not the hours the shop is open. Fixture setup, tool changes, probing, deburring, cleaning and programming typically eat 50–70% of the wall-clock time on small runs. If you buy a machine for 20 hours of cutting a month, you bought a very expensive workbench.

Run the payback arithmetic honestly. Take the fully loaded hourly rate you would pay an outside shop, multiply by your annual cutting hours, and compare that to the annualized cost of the machine: loan payment or depreciation, floor space, power, consumables, tooling, maintenance and the operator's loaded wage. If the machine does not run at least a few hundred hours a year, the outside shop usually wins.

Volume also changes the process, not just the cost. At one to fifty parts, fixtures can be soft jaws and a vise. At a few hundred parts, you start justifying a dedicated fixture and a probe routine. At thousands of parts, you justify a pallet changer, a bar feeder or a mill-turn center so the part comes off complete in one cycle.

This is why the buy decision is really a volume decision in disguise. The geometry tells you which machine. The volume tells you whether the machine should exist at all. Answer the volume question first; it eliminates most bad purchases before the quoting stage.

• 1
  Under 200 cutting hours/yearOutsource. The machine cannot amortize its own overhead.
• 2
  200–1,000 hours/yearA used or entry-level machine can work if the part family is stable.
• 3
  Over 1,000 hours/yearDedicated fixtures, pallet systems and lights-out running start to pay.

The machine invoice covers the iron. It does not cover the foundation, the rigging, the three-phase power drop, the compressed air dryer, the coolant system, the chip conveyor or the first set of toolholders. Add spindle tooling, a probe, a vise or two, soft jaw blanks, a workbench, a deburring station and a computer with CAM software. That list is often 20–40% of the machine price on a small installation.

Then come the recurring costs. Carbide end mills, inserts, drills and taps wear out whether or not the machine is busy. Coolant needs changing and disposing of. Way lube, filters and belts show up on a schedule. Preventive maintenance on a production machine is not optional; a spindle rebuild costs more than a year of outsourcing.

Labor is the largest recurring line. A CNC machine does not run itself. Someone programs, sets up, loads, measures, deburrs and troubleshoots. If that person is a skilled machinist, you are paying a skilled wage for every hour the machine is idle too. If it is an apprentice learning on your dime, expect scrap and slow cycles for the first several months.

There is also the opportunity cost. Money spent on a machine is money not spent on design, tooling for a contract manufacturer or inventory. For a small team, that trade is often worse than it looks on a spreadsheet, because the machine consumes management attention as well as cash.

One underrated line: spare capacity. A machine sized for your biggest part sits idle most of the time, and a machine sized for your average part cannot make the big one. Outsourcing lets you buy capacity in whatever shape the next job needs.

• 1
  Installation and utilitiesRigging, foundation, power, air, coolant, chip handling.
• 2
  Tooling and workholdingToolholders, vise, soft jaws, probe, first carbide order.
• 3
  Recurring consumablesCutters, inserts, coolant, filters, way lube, maintenance.
• 4
  Operator timeProgramming, setup, deburring and inspection hours.