How to Judge the Accuracy of a Machining Center

• 1
  Key takeaways
• 2
  Why the accuracy of a machining center needs physical checks
• 3
  Geometry and level: the base of accuracy of a machining center
• 4
  Spindle checks: runout, taper contact and thermal growth
• 5
  Positioning, repeatability and backlash on each axis
• 6
  Test cuts that show the real accuracy of a machining center
• 7
  What changes the accuracy of a machining center over time
• 8
  Step by step: how to judge the accuracy of a machining center
• 9
  Check, instrument and typical pass limit

A datasheet quote tells you what the builder measured in a temperature-controlled room, on a fresh machine, with no chips on the table. That number is a starting point, not a promise about your parts. The accuracy of a machining center depends on the machine plus the fixture, the tool, the material and the operator. Any of those can eat the tolerance before the spindle turns.

We run the same sequence on every machine at GreatLight: geometry, spindle, positioning, thermal behavior, then a real test cut. The order matters. If the bed is twisted, every later measurement carries that error and you end up chasing a spindle problem that does not exist. Fix the foundation first, then the moving parts.

Engineers often ask for a single number. There is no single number. You need a set of measurements, each with its own limit, and you need to know which limit applies to your part. A bracket with a ±0.1 mm clearance hole does not need the same machine as a hydraulic manifold with a ±0.005 mm bore.

This article gives the checks, the instruments, the typical limits and the failure modes. Use it when you accept a new machine, when you investigate a drifting process, or when you compare two suppliers on paper and want to know what to ask them.

• 1
  Check the machine, not the brochureAsk for raw measurement records, not a summary sheet.
• 2
  Match limits to the partTighter limits cost more time and money. Do not over-specify.

Start with the foundation. Put a precision level (0.02 mm/m or finer) on the table in both X and Y directions. On a machine with a 1,000 mm table, a 0.02 mm/m reading means 0.02 mm of tilt per meter. If the table is out by 0.04 mm/m, the column leans and every Z move drags a little sideways.

Next check squareness between axes. A granite square and a dial indicator on the spindle will show X to Y deviation over the table length. For a general-purpose vertical machine, 0.01 mm per 300 mm is a reasonable target. Above 0.02 mm per 300 mm, circular interpolated bores will come out oval and you will see it on a CMM report.

Check the table surface itself with an indicator on a known straight edge. Dings, rust and a worn center area are common on used machines. If the table dips 0.03 mm in the middle, every part is thinner there and no amount of tool compensation will fix the whole surface evenly.

Do not skip this because it takes two hours. A laser calibration on a crooked machine buys you nothing. We level new machines after they sit on the floor for a few days, because concrete and castings settle.

• 1
  Level limitAim for 0.02 mm/m or better on both axes.
• 2
  Squareness limitUnder 0.01 mm per 300 mm for general milling work.

Spindle error shows up directly in the part. Measure runout with a dial test indicator on a clean test bar held in the taper. Inner taper runout of 0.005 mm or less is good for general work; 0.002 mm is what you want for finishing cuts with small tools. Measure at the nose and 300 mm out, because the far reading also shows alignment error.

Taper contact matters more than most people think. Blue the test bar and check the contact pattern. You want even contact over 80% of the taper area, heavy near the large end. A spindle with poor contact will hold a tool differently after a tool change and your Z depth will wander a few hundredths of a millimeter.

Thermal growth is the quiet killer. Run the spindle at production speed for 30 to 60 minutes and re-measure a fixed point on the table. A typical 40-taper spindle grows 0.01 to 0.03 mm in Z over that time. If the machine has no thermal compensation, warm it up before the first critical cut and keep the warm-up routine identical every morning.

Listen while you do this. A rumble at low speed can mean bearing damage. A high-pitched whine at 12,000 rpm is often normal, but a new noise after a crash is not. Record the sound or note the spindle load at a fixed speed, then compare it next month.

• 1
  Runout target0.005 mm near the nose, 0.002 mm for fine finishing.
• 2
  Taper contact80% or more, heavier at the large end.
• 3
  Warm-up30–60 minutes at production speed before critical cuts.

Positioning accuracy is how close the machine gets to a commanded point. Repeatability is how close it returns to the same point many times. Repeatability usually matters more for production, because tool offsets and fixture offsets assume the machine lands in the same place every cycle.

Use a laser interferometer for linear axes and a ballbar for circular motion. Laser linear positioning of 0.005 mm over 500 mm is a solid result for a modern vertical machine. Backlash should be under 0.005 mm; on a worn machine, 0.01 mm or more shows up as chatter marks and mismatched walls on climb-cut passes.

Run a ballbar circle at 300 mm diameter and 2,000 mm/min feed. Circularity of 0.010 mm or better is good, and you should see the same value in both directions and at two feed rates. A machine that looks fine at 500 mm/min and fails at 5,000 mm/min has a servo tuning or mechanical stiffness problem, not a leveling problem.

Repeat the positioning test at the start and end of a shift. If the numbers move by more than 0.01 mm, find out why before you blame the operator. Common causes are a loose coupling, a dirty linear scale, or a ballscrew that needs preload adjustment.

• 1
  Linear positioning0.005 mm over 500 mm is a good target.
• 2
  BacklashKeep under 0.005 mm on each axis.
• 3
  Ballbar circularity0.010 mm or better at 2,000 mm/min.

Instruments measure the machine. A test cut measures the process. Cut a part that has features in all three axes: a stepped block with a bored hole, two milled faces, a slot and a set of drilled holes. Use the same material, tool and fixture you plan for production. Aluminum 6061 is fine for a basic check, but if you run stainless 316 or 17-4PH, test in that material too.

Measure the finished part on a CMM or with a micrometer and bore gauge. Look at the numbers, not the surface. Bore roundness, step heights, slot width and hole position all tell you something different. A round bore in the wrong place points to positioning. A bore in the right place but out of round points to the spindle or the interpolation.

Run the test cut at three times of day: cold start, mid-shift and after four hours of running. Compare the measurements. A drift of 0.02 mm over a shift is common on machines without temperature control. If your tolerance is ±0.01 mm, that drift will produce scrap unless you warm up, compensate or control the room.

Keep the test part. Mark the date, the machine and the measured values on it. When a customer questions a dimension six months later, that block is your baseline and it costs nothing to store.

• 1
  Test in the real materialAluminum only is not enough for stainless or titanium work.
• 2
  Use a CMM reportSurface finish hides position errors. Measure the features.
• 3
  Repeat over a shiftCold, mid-shift and four-hour readings reveal drift.

A new machine and a five-year-old machine are different animals. Guideway wear, ballscrew wear and bearing wear all move the numbers. The rate depends on load, coolant, chip control and maintenance. A machine running light aluminum parts for 4,000 hours a year ages slower than one hogging 4140 steel every day.

Coolant and chips are the usual culprits. Fine chips under the way covers act like grinding paste. If the covers are torn or the wipers are hard, replace them. A 20-dollar wiper protects a 20,000-dollar ballscrew. We check way covers and wipers on a monthly schedule, not yearly.

Temperature control in the shop matters as much as machine temperature. A 10 °C swing between morning and afternoon moves steel parts and the machine frame. For work at ±0.005 mm, we keep the finishing area within a few degrees and let parts stabilize before final measurement.

Finally, track your own data. A simple log of the test-cut bore diameter every week will show a trend long before parts go out of tolerance. Trends are cheaper to fix than scrap. When the trend moves, check level and backlash first, then call for laser calibration.

• 1
  Wear is load-dependentHeavy steel cutting wears guideways faster than light aluminum.
• 2
  Protect the waysInspect covers and wipers monthly, replace when worn.
• 3
  Log the numbersA weekly test-cut reading catches drift before scrap.