What Holds CNC Machine Grinder Size Dimensions

• 1
  Key takeaways
• 2
  Stiffness: the loop that decides how far the wheel pushes away
• 3
  Thermal drift: why the first 20 parts differ from the last 200
• 4
  Wheel condition and dressing: the input nobody measures
• 5
  Feed resolution, slides, and what the control can actually command
• 6
  Metrology: the measurement decides whether the size holds
• 7
  Where the size specification stops holding
• 8
  Which element controls which size error

In grinding, the wheel does not cut to a commanded depth. It deflects the loop by an amount equal to the normal force divided by the loop stiffness, then cuts what is left. That is the whole mechanism. A machine with a 30 kN/mm loop and 300 N of normal force loses 0.010 mm to deflection before the first spark. If the force is steady, the loss is steady and the controller can compensate. If the force drifts, size drifts with it.

So the question is not only how stiff the machine is, but how stable the stiffness is over the cycle. Cast iron bases, ribbed columns and preloaded linear guides all raise the number. Hydrostatic or hydrodynamic ways add oil-film stiffness and damp vibration, which shows up in surface finish more than in size. On a small surface grinder holding ±0.005 mm, a 10 percent change in loop stiffness is often enough to move size by 1–2 μm across a batch.

Workholding is part of the same loop. A thin plate clamped on four corners bows between them. A shaft held in a three-jaw chuck changes roundness as the jaws seat. Magnetic chucks hold flat parts well, but they distort thin ones, and the part springs back after release. If size matters, the fixture has to be as stiff and as repeatable as the machine.

Practical check: grind a test bar, measure the taper and the size scatter, then repeat with a lighter spark-out. If the scatter shrinks when you reduce force, the loop is soft and you are fighting deflection rather than wheel wear.

• 1
  Loop stiffnessMachine plus fixture plus part, measured as a system.
• 2
  Force stabilitySteady force means steady size; dressing and coolant decide this.
• 3
  ClampingThin parts distort under clamping and spring back after release.

Steel grows about 11 μm per meter per °C. Aluminium grows about twice that. A 300 mm part that warms 3 °C during grinding is 10 μm longer than it was at the start. That is larger than the tolerance on most ground features. The heat comes from three places: the grinding zone, the spindle and slide motors, and the coolant as it circulates.

Machines handle this in two ways. The first is to remove heat at the source: flood coolant at the wheel contact, high-pressure through-spindle coolant, and time for the wheel to cool between passes. The second is to make the machine expand evenly. Symmetrical castings, coolant routed through the base, and temperature-controlled oil or water on the slides all help. Some grinders run a warm-up cycle before the first part so the structure reaches steady state.

In-process gaging closes the loop. The gage measures the part while it is still in the machine and the control adjusts for the next pass. That works well for cylindrical and centerless grinding, where the part is round and accessible. It works less well for interrupted surfaces or when the gage itself sits in a warm coolant stream.

A good shop rule: log the size of the first part, the tenth part, and the last part of the run. If the trend is monotonic, it is thermal. If it oscillates, look at dressing, coolant flow, or wheel balance before you blame the machine.

• 1
  Warm-up firstRun the spindle and coolant until size stops moving.
• 2
  Coolant temperatureHold it within 1–2 °C of ambient for tight work.
• 3
  Trend vs. scatterMonotonic drift is thermal; random scatter is force or balance.

A grinding wheel is a cutting tool that changes shape as it works. Grains dull, bond wears, and the wheel face loads with metal. Each of those changes the normal force, and force changes size. Dressing restores the face, but dressing too often wastes the wheel and dressing too rarely lets force climb until the part burns or goes oversize.

Dressing parameters matter as much as the wheel. A single-point diamond with a 0.02 mm depth of cut and a 0.15 mm/rev feed leaves one face; a rotary dresser at the same nominal setting leaves another. The dresser itself wears, so its position has to be compensated. On a CNC grinder, the dress cycle is a program with numbers, and those numbers need to be treated as process settings, not housekeeping.

Wheel speed and work speed interact with dressing. Higher wheel speed usually lowers force per grain, which reduces deflection and improves finish, but it also raises grinding-zone temperature. Higher work speed spreads the heat over more of the part. For most steel work, a wheel speed of 30–45 m/s and a work speed that keeps the specific removal rate moderate is a reasonable starting range.

If size drifts slowly upward across a run, check the dress interval and the dresser compensation before touching the machine geometry. In many shops this single change recovers more size control than any mechanical work.

• 1
  Dress intervalSet by force and finish, not by a fixed part count.
• 2
  Dresser wearCompensate the dresser position or the wheel face moves.
• 3
  Wheel speed30–45 m/s is a common range for steel grinding.

A grinder that can resolve 0.1 μm on the infeed axis still cannot hold 0.1 μm if the slide has 2 μm of stick-slip. Friction behavior, not encoder count, is usually the limit. Hydrostatic and aerostatic slides avoid stick-slip because they have no metal-to-metal contact. Roller and ball linear guides have low friction but a small non-linear region at reversal. Box ways with proper lubrication sit between the two.

The control side is simpler than people expect. Modern CNCs interpolate and compensate well. What they cannot do is know the wheel has worn. Wheel wear compensation, either by a wear model or by in-process gaging, is what keeps size on target across thousands of parts. Spark-out passes and constant-force grinding are two other ways to make the last few microns predictable.

Axis calibration matters for size, not just for position. A slightly misaligned wheelhead axis shows up as taper, and taper is a size error at one end of the part. Squareness between the work axis and the wheel axis has the same effect. These are setup errors, not control errors, and they are found with a test cut, not with a laser interferometer alone.

The useful question is: what is the smallest reliable infeed step on this machine, on this part, with this fixture? Measure it. Do not take the resolution figure from the brochure.

• 1
  Stick-slipOften the real limit on the smallest reliable step.
• 2
  Wear compensationKeeps size on target as the wheel shrinks.
• 3
  Geometry setupTaper and squareness show up as size error, not position error.

You cannot hold a tolerance you cannot measure with confidence. A micrometer read at 35 °C part temperature tells you the part is bigger than it will be at 20 °C. For a 300 mm steel part, every 1 °C of measurement error is about 3 μm of reported size error. For aluminium, about 7 μm. So a shop holding ±0.005 mm has to control measurement temperature, not just machine temperature.

Gage repeatability is the other half. If the gage contributes 2 μm of scatter, the process has to be centered well inside the tolerance to avoid scrap. A common target is a gage repeatability and reproducibility under 10 percent of the tolerance band. That usually means a temperature-controlled room, a calibrated setting master, and a measurement routine that the operator follows the same way every time.

In-process gaging is a different trade. It measures the part in the machine, which removes handling and thermal lag, but the gage sees coolant, swarf, and vibration. It works best where the surface is round and the contact is stable, such as cylindrical grinding. For flat or contoured surfaces, post-process measurement in a controlled room is often more reliable.

Measurement data is also how you find the cause of a size problem. Plot size against time, against wheel dress count, and against coolant temperature. One of those three will usually explain the drift. Without the data, the diagnosis becomes guesswork.

• 1
  Part temperature1 °C on 300 mm steel is about 3 μm of reported size.
• 2
  Gage R&RKeep it under 10 percent of the tolerance band.
• 3
  Trend dataSize vs. time, dress count and coolant temperature.

There is a point where grinding can no longer hold size on a given part, and it is usually a geometry or material problem, not a machine problem. A long thin shaft deflects under grinding force no matter how stiff the machine is. A thin disc warps as the two faces are ground. A part with interrupted cuts impacts the wheel and moves the size on each impact.

Material matters too. Soft, gummy aluminium loads the wheel and the size wanders. Hardened tool steel above 60 HRC grinds cleanly if the wheel is right, but burns easily if the coolant is weak. Titanium and nickel alloys generate heat fast and hold it, so they need lower removal rates and more coolant pressure, which slows the cycle.

The other limit is tolerance against feature size. Holding ±0.005 mm on a 20 mm diameter is normal cylindrical work. Holding the same tolerance on a 400 mm long unsupported shaft is not. The rule of thumb: as the length-to-diameter ratio climbs past about 10:1, deflection and thermal growth start to dominate, and you need steady rests, in-process gaging, or a different process.

The honest answer to what holds size is that a well-set-up grinder holds the tolerances its loop, its thermal control and its metrology support. Push past that and you are not grinding to size, you are sorting parts.

• 1
  L/D ratioPast about 10:1, deflection and thermal growth take over.
• 2
  Thin walls and discsClamping and grinding both distort them; expect to sort.
• 3
  High-heat alloysTitanium and nickel need lower removal rates and more coolant.