Key technologies that maintain the accuracy of CNC crankshaft grinders
A crankshaft grinder does not stay accurate because it was built accurate. It stays accurate because five systems keep correcting it. This page explains the mechanism behind each one, the numbers that matter, and when a shop should not bother paying for them. Written for engineers and buyers who specify crank, cam, and journal grinding work.
In this article
- 1
- 2
- 3
- 4
- 5
- 6
What actually drifts when you chase the accuracy of CNC crankshaft grinders
A crankshaft is a long, unbalanced shaft with offset journals. When the wheel contacts a rod journal, the part is not spinning about its own centerline. It swings on the main journals, and the mass hangs off one side. That single geometry fact drives almost every accuracy problem on this machine class.
Four things move over a shift. The wheel face loads up and loses its form. The workhead and wheelhead grow with spindle heat. The steady rest and center height shift as the bed warms. And the dressing diamond wears, so the wheel no longer cuts the profile the CNC thinks it is cutting.
None of these are sudden failures. They are slow drifts, measured in microns per hour. The accuracy of CNC crankshaft grinders is therefore not a static specification. It is a rate: how many microns of size and roundness you lose per hour of cutting, and how fast the machine can correct itself.
Two error families matter. Size error is journal diameter drifting away from nominal. Form error is lobing, taper, and out-of-round on the same journal. Size error usually comes from wheel wear and thermal growth. Form error usually comes from dressing condition, steady rest setup, and wheel balance.
A grinder that holds ±0.005 mm on a 60 mm journal for a full shift is doing three things at once. It is dressing on a schedule tied to material removal, not to clock time. It is holding the workhead and coolant at a stable temperature. And it is measuring the part before the size walks out of tolerance, not after.
Dressing, wheel form, and the accuracy of CNC crankshaft grinders
Dressing is the single most powerful accuracy lever on a crank grinder, and the cheapest. A dull or loaded wheel does not cut; it rubs. Rubbing raises normal force, pushes the workpiece away from the wheel, and produces size scatter plus lobing on the journal.
The fix is depth-controlled dressing with a CNC-dressed profile. A typical CBN or aluminum oxide wheel for cast iron cranks is dressed every 8 to 20 parts, depending on stock removal per journal. Diamond infeed per pass runs 0.01 to 0.03 mm, with two spark-out passes at the end.
Do not dress by timer alone. A wheel grinding 0.4 mm of stock per journal loads far faster than one grinding 0.15 mm. Tie the dress counter to cumulative metal removed or to spindle load, and the wheel face stays consistent across the batch.
Dressing also sets the effective wheel diameter, and the CNC needs that number. Modern grinders measure the wheel after each dress with an in-machine probe or an acoustic gap sensor. If the control assumes a diameter that is 0.05 mm off, wheel feed positions shift and the first part after dressing runs oversize.
Wheel balance is the other half. An unbalanced wheel at 35 m/s spindle speed creates a once-per-revolution force that shows up as a two-lobe or three-lobe pattern on the journal. Balance the wheel on the machine arbor after mounting, and re-check after any wheel change.
Thermal stability and the accuracy of CNC crankshaft grinders
Grinding puts almost all its energy into heat, and a fraction of that heat goes into the machine structure. A wheelhead spindle that grows 15 μm in Y direction moves the wheel center by 15 μm. On a 50 mm journal, that is 0.03 percent of the diameter, and it appears directly as size drift.
Good machines handle this three ways. They pre-load the structure with a warm-up cycle before the first part. They run coolant through the bed and spindle housing to hold a set temperature. And they compensate in the control using spindle growth models or a reference probe cycle.
A practical floor rule: let the machine idle with coolant and spindles running for 30 to 45 minutes before grinding critical journals. Then check the first article and the tenth article. If size drifts more than 5 μm between them, the machine is still warming, not cutting.
Coolant temperature matters as much as structure temperature. A 2 °C rise in coolant raises the workpiece temperature and changes the ground diameter through thermal expansion. Hold coolant within ±1 °C of setpoint on crank work; that is a tighter band than most milling cells need.
Ambient swings hurt too. A shop door opening in winter can move the bed by several microns over an hour. For work at ±0.005 mm, keep the grinder away from doors and direct sunlight, and avoid running heavy grinding next to a machine that just started cold.
Workholding, steady rests, and the accuracy of CNC crankshaft grinders
A crankshaft is flexible in bending. Push a rod journal with 200 N of grinding force and the shaft bows between the main journals. The wheel then cuts a barrel-shaped journal instead of a cylinder. This is why crank grinders use steady rests and often a follow rest on long throws.
Center height error is the most common setup mistake. If the work center is 0.1 mm above or below the wheel centerline, the ground journal becomes slightly oval, and the error grows with journal diameter. Set center height with a gauge, not by eye, and re-check after any workhead adjustment.
Steady rest contact pressure should be just enough to control deflection. Too light and the journal chatters. Too heavy and the rest pads burnish the journal and imprint marks. On most crank work, hydraulically clamped rests with adjustable pad force work better than fixed shoes.
Drive dogs and face plates also matter. A loose drive lets the part slip and index slightly between passes, which shows up as taper. Index the part positively and check runout on the main journals before grinding, not after.
For a one-off or small batch, the setup time for full crank fixturing may exceed the grinding time. That is a real trade-off. Below roughly 10 parts, it is often cheaper to machine the crank on a 5-axis mill-turn center to ±0.02 mm than to build grinding fixturing.
In-process gauging and closed-loop size control
The last line of defense is measurement during the cycle. An in-process gauge rides on the journal as it grinds and feeds the actual diameter back to the control. When the target size is reached, the control retracts the wheel. This removes wheel wear and thermal drift from the size equation.
Gauging heads for crank journals typically resolve 1 μm and read at 10 to 50 Hz. The control uses the reading to trigger a spark-out phase, usually 2 to 5 seconds with no infeed, so the wheel springs back and the final diameter settles.
Not every job needs this. In-process gauging adds cost, setup, and coolant management. For journals at ±0.02 mm, post-process sampling every fifth part is enough. For ±0.005 mm or a CpK requirement, closed-loop gauging is usually the difference between holding the tolerance and sorting scrap.
Post-process measurement still matters. Even with gauging, check roundness and taper on a sample with a bench micrometer or a roundness tester. Gauging controls diameter; it does not tell you the journal is lobed.
Record the data. Size trend per part, dress events, and coolant temperature plotted together explain most accuracy complaints before anyone touches the machine.
Which accuracy technology fits which job
Match the control method to the tolerance band and the batch size
| Job condition | Method that pays off | Method that is wasted |
|---|---|---|
| Journals at ±0.02 mm, batch over 50 | Scheduled dressing plus sampling | In-process gauging, coolant chilling |
| Journals at ±0.005 mm, CpK required | Closed-loop gauging plus chilled coolant | Timer-based dressing only |
| One-off or under 10 parts | Mill-turn to ±0.02 mm | Full crank grinding fixturing |
| Long throws, flexible shaft | Steady rest plus follow rest | Rigid workholding alone |
| High stock removal per journal | Dress counter tied to metal removed | Dress counter tied to clock time |
| Cast iron cranks, roughing passes | Balance wheel on arbor after mount | Profile dressing on every pass |
The trade-off in one line
If the print says ±0.02 mm, dress on a schedule and sample the parts; if it says ±0.005 mm with a CpK number, pay for closed-loop gauging and chilled coolant, and keep the machine warm before the first cut.
Questions engineers ask next
How often should the wheel be dressed on a crank grinder?
Tie the dress counter to cumulative metal removed rather than to clock time. On cast iron cranks with 0.4 mm of stock per journal, that typically lands between 8 and 20 parts. Diamond infeed of 0.01 to 0.03 mm per pass with two spark-out passes keeps the wheel face open.
If size starts drifting upward between dress events, the wheel is loading. If roundness gets worse, suspect balance or the dressing diamond, not the dress interval.
Can a CNC crankshaft grinder hold ±0.005 mm without in-process gauging?
It can, but only under stable conditions: warm machine, chilled coolant within ±1 °C, frequent dressing, and a skilled operator sampling parts. The risk is that drift is invisible until the sample catches it.
For a tolerance this tight with a CpK requirement, closed-loop gauging is the practical answer. It takes wheel wear and thermal growth out of the size loop.
Does coolant temperature really change the ground diameter?
Yes. A 2 °C rise in coolant raises the workpiece temperature, and steel expands about 11 μm per meter per degree Celsius. On a 50 mm journal that is small, but once you stack it with spindle growth and wheel wear, it becomes visible at ±0.005 mm.
Hold coolant within ±1 °C of setpoint on crank work. That band is tighter than most milling cells need.
Why does the first part after dressing run oversize?
The control is still working from the previous wheel diameter. After dressing, the wheel is smaller, so the same commanded infeed removes less material and the journal finishes large.
Fix it with an in-machine wheel probe or an acoustic gap sensor that updates the diameter after every dress. If the control assumes a diameter that is off by 0.05 mm, the error shows up on the very first part.
When is grinding the wrong process for a crankshaft?
Below roughly 10 parts, the fixturing and dress setup usually cost more than the parts are worth. A 5-axis mill-turn center can hold ±0.02 mm on journals without building crank-specific tooling.
Grinding wins when the tolerance is tighter than ±0.02 mm, the batch is repeatable, or the surface finish has to reach Ra 0.2–0.8 μm on a hardened journal.
What causes lobing on a crank journal?
Lobing usually comes from wheel imbalance, a loaded or glazed wheel, or steady rest pressure that is too light. An unbalanced wheel at 35 m/s creates a once-per-revolution force that prints a two-lobe or three-lobe pattern.
Balance the wheel on the arbor after mounting, dress with controlled infeed, and set rest pressure high enough to control deflection without burnishing the journal.
Send us your crankshaft print
Upload the drawing and we will return a quotation and a free DFM analysis within 12 hours, with the grinding method matched to your tolerance band.
12-hour quote100% inspection
Follow GreatLight
We publish setup notes, tooling trials and inspection data from the factory floor.
