How to Reduce the Radial Jump of the Tool in CNC Grinding
Radial jump of the tool is the runout you can still measure after the wheel is dressed and the spindle is warm. This guide is for engineers and setup machinists who need to bring tool radial runout down on a real job, not on a spec sheet. We cover how to measure it, where the error comes from, and which adjustments actually move the number.
Key takeaways
What Radial Jump of the Tool Tells You
Radial jump of the tool is the total indicated runout (TIR) of the cutting edge or wheel face relative to the spindle axis, measured in the radial direction. A single-point tool gives you one high spot per revolution. A multi-edge tool gives you as many high spots as there are teeth, and each tooth cuts a slightly different depth. That is why the same machine can hold ±0.005 mm on a boring bar and miss it on a 12-tooth cutter.
The practical consequence is not just size error. Radial jump changes chip load tooth by tooth. One tooth takes a heavy bite, the next rubs. You get uneven flank wear, chatter marks at the tooth-pass frequency, and surface finish that drifts from Ra 0.8–1.6 μm to Ra 3.2 μm or worse within a few dozen parts. On grinding work the same logic applies to the wheel: an eccentric wheel face burns the part on the high side and leaves it unground on the low side.
Measure TIR with the spindle at operating temperature, not cold. Mount a 0.001 mm or 0.0005 mm dial indicator on the table and touch the shank just below the holder, then repeat within 5 mm of the cutting edge. Record both numbers. The difference between them is bending or a bent shank. If the two numbers match, the error is in the holder or the spindle taper, not the tool.
One short sentence matters here. Runout is a stack, not a single fault. Spindle taper, holder, collet, nut, tool shank and the wheel arbor each add their own contribution. You cannot fix a 20 μm stack by polishing one part of it. Find the largest contributor first, correct it, then re-measure.
- 1Target for finishingKeep TIR at or below 10 μm on finishing passes of tight-tolerance work.
- 2Target for roughing20–30 μm is usually acceptable if the following pass removes the error.
- 3Measure hotCold spindle readings understate runout by several micrometers.
Where the Error Comes From
Spindle and taper errors are built in. Coaxiality error between spindle journals, bearing error, and taper eccentricity are set at machine assembly. An operator cannot remove them by adjusting a setscrew. What you can do is measure the taper directly with a test bar and log the value. If the bare taper shows more than 5 μm TIR, no holder change will save the job.
Mounting errors are where most shops lose accuracy. Chips, coolant residue, or a burr on a CAT40 or HSK taper face will tilt the holder by a few micrometers, which becomes tens of micrometers at the tool tip. Collet nuts torqued by feel, worn collet segments, and holders that have been dropped all show up as radial jump of the tool. Clean the taper with a lint-free wipe, stone any burr flat, and use the holder manufacturer's torque figure.
Grinding adds two more sources. First, the wheel arbor and flange. A flange face that is not flat, or a blotter that is torn, will tilt the wheel. Second, wheel balance. An unbalanced wheel generates a centrifugal force that grows with the square of spindle speed, and that force pushes the wheel off axis at every revolution. At 3,000 rpm a small imbalance is harmless; at 12,000 rpm it is not.
Cutting force closes the loop. Radial cutting force is the component of the total force pushing the tool away from the work. It bends the tool, deflects the holder, and excites vibration. Higher radial force makes existing runout worse, and worse runout raises the peak force on the high tooth. That feedback is why a job that ran fine at 0.3 mm radial depth can chatter at 0.6 mm with no other change.
- 1Spindle taperMachine-built error. Measure with a test bar, log it, do not chase it with setup changes.
- 2Holder and colletCleanliness, burrs, wear and torque. This is the highest-yield place to work.
- 3Wheel arbor and balanceFlat flange faces, intact blotters, balanced wheel and arbor as one assembly.
- 4Cutting forceRadial depth of cut, feed per tooth and wheel speed set the force level.
Process Adjustments That Lower Radial Force
The fastest way to reduce radial jump of the tool during the cut is to lower the force that pushes it off axis. Start with radial depth of cut. On a finishing pass, dropping radial engagement by 30–40 percent often removes chatter without any change to the spindle or holder. The trade-off is cycle time, so apply it to the finishing pass only and keep roughing aggressive.
Wheel and spindle speed matter too. Raising wheel speed while keeping feed per tooth constant reduces the chip thickness each edge takes, which lowers the peak radial force. On aluminum and mild steel, a 15–20 percent increase in wheel speed is usually safe if the wheel is rated for it. On hardened steel above 55 HRC, watch for burn instead and hold speed steady.
Feed per tooth is the third lever. Too low a feed lets the edge rub instead of cut, which raises force and heat. Too high a feed raises force directly. Aim for a chip load that produces visible, uniform chips on every tooth. If you cannot see chips on all teeth, the runout is already spreading the load unevenly.
Coolant direction deserves a mention. Aim the stream at the contact zone, not at the wheel periphery. Poor cooling on the high tooth causes localized thermal growth, which changes the effective runout during the pass. On a long grinding cycle this shows up as size drift from the first part to the tenth.
- 1Radial depthReduce 30–40 percent on the finishing pass to cut peak force.
- 2Wheel speedRaise 15–20 percent on aluminum and mild steel if the wheel rating allows.
- 3Feed per toothSet it so every tooth produces a visible chip. Rubbing raises force.
- 4Coolant aimDirect the stream at the contact zone, not the wheel face.
When You Cannot Get Below the Target
Some jobs will not reach single-digit micrometer runout, and it is worth knowing that before you spend a shift chasing it. If the bare spindle taper measures more than 5 μm TIR, the machine itself sets a floor. Grinding can still hold size if you compensate with in-process gauging or a finishing pass with low radial force, but you will not fix it at the holder.
Very small tools set another floor. A 1 mm or 2 mm shank has little bending stiffness, so radial force deflects it more than a larger tool at the same parameters. The usual answer is to reduce radial depth further, raise spindle speed to keep the chip load, and accept a slower cycle. Trying to hold the same parameters as a 10 mm tool will not work.
In-process thermal effects are the third limit. On tight-tolerance grinding, the wheel, workpiece and coolant all change temperature during the cycle. Runout measured cold is not the runout in the cut. For parts that need ±0.005 mm, warm up the spindle for 20–30 minutes, then take your measurements.
If the geometry is beyond what one setup can hold, that is a signal to split the operation. Rough on the machine with the higher runout, then finish on the tighter one. It costs an extra setup but it removes the argument about which correction to make first.
- 1Taper over 5 μmMachine floor. Compensate with gauging rather than chasing the holder.
- 2Shank under 3 mmDeflection-limited. Lower radial depth and raise speed.
- 3Tight toleranceWarm up 20–30 minutes before measuring anything.
Step-by-Step: Reduce Radial Jump of the Tool
- 1Clean and inspect the spindle taperWipe the taper with a lint-free cloth and check for fretting, scoring or embedded chips under a light. Stone any raised burr flat with a fine oilstone, then wipe again. A 2 μm burr on the taper face can become 15–25 μm at 100 mm from the gauge line.
- 2Measure taper TIR with a test barInsert a clean test bar and indicate it at 50 mm and 150 mm from the gauge line with a 0.001 mm indicator. Rotate the spindle by hand through one full turn. If the bare taper exceeds 5 μm TIR, log the value and plan around it instead of changing holders.
- 3Check holder and collet conditionLook for scoring inside the collet bore, a collapsed nut thread, or a worn taper shank. Replace any collet that has run more than its rated life or has visible polishing marks. Never mix collet brands in the same nut.
- 4Mount with correct torque and clean interfacesWipe the tool shank with solvent and let it dry. Insert to full depth, then torque the nut to the holder maker's figure, typically 80–120 N·m for a 20 mm collet nut. Under-torque lets the tool creep; over-torque bell-mouths the collet.
- 5Indicate the tool or wheel, not the holderTouch the indicator 5 mm above the cutting edge and rotate slowly by hand. Record TIR. Then move to the shank just below the holder and record again. A large gap between the two readings points to a bent shank, not a mounting problem.
- 6Balance the wheel and arbor as one assemblyMount the wheel on its arbor and flange, then balance the complete assembly, not the wheel alone. Re-check balance after the first dressing pass, because dressing removes mass unevenly. On vitrified wheels, re-balance whenever you change the wheel or move it to a different spindle.
- 7Dress, then re-measureTake a light dressing pass, 0.02–0.05 mm depth, at the normal traverse rate. Re-measure wheel face runout with the indicator. Dressing removes eccentricity from the wheel face, but it cannot correct arbor or flange tilt. If runout is unchanged after dressing, the fault is upstream.
- 8Cut a test diameter and verifyGrind a short test section at reduced radial depth, then measure size and roundness at four points. If size spread stays within ±0.005 mm and finish sits in Ra 0.8–1.6 μm, release the setup. If not, go back to the largest contributor from your measurements rather than adjusting everything at once.
Which Correction to Apply First
Match the symptom to the likely cause before touching the machine.
| Symptom | Most likely cause | First action |
|---|---|---|
| Runout constant after dressing | Arbor or flange tilt | Check flange flatness and blotter |
| Runout rises with spindle speed | Wheel or arbor imbalance | Balance the full assembly |
| TIR differs at shank and tip | Bent tool shank | Replace the tool, re-indicate |
| Chatter marks at tooth frequency | Uneven tooth load | Re-check TIR and chip load |
| Size drifts over a long run | Thermal growth at contact | Redirect coolant, check wheel speed |
| Bare taper above 5 μm TIR | Spindle taper error | Log it, plan around it, service spindle |
Fix the biggest contributor, then re-measure
Radial jump of the tool is a stack of errors. Cleaning the taper, torquing the holder correctly, and balancing the wheel arbor as one assembly removes most of it before you touch a single cutting parameter.
Frequently Asked Questions
What TIR should I aim for on a finishing grinding pass?
For finishing work held to ±0.005 mm, keep tool or wheel TIR at or below 10 μm. Above that, the runout starts to show up as size spread and finish variation across the part.
On roughing passes 20–30 μm is normally acceptable, provided the finishing pass removes the error. Measure after dressing, not before.
Does a balanced wheel guarantee low radial jump?
No. Balance controls the centrifugal force, but it does not correct arbor or flange tilt. A perfectly balanced wheel on a flange with a 10 μm face error will still show runout at the wheel face.
Treat the arbor, flange, blotter and wheel as one assembly. Balance the assembly, and re-check it after the first dressing pass.
How often should I re-indicate the tool?
Re-indicate after every tool change, after any crash or alarm, and at the start of a shift on tight-tolerance work. On long unattended runs, add a check at the midpoint.
The measurement takes about 2 minutes with a 0.001 mm indicator. That is cheaper than scrapping a batch.
Can coolant pressure reduce runout?
Not directly. Coolant does not change the mechanical runout of a tool. What it does change is thermal growth at the contact zone, which affects the effective runout during the cut.
Aim the stream at the contact point rather than the wheel periphery. On long cycles this reduces size drift from the first part to the last.
Why does my runout get worse as spindle speed increases?
That pattern points to imbalance rather than a mounting error. Centrifugal force from an unbalanced wheel or arbor grows with the square of spindle speed, so the deflection gets worse quickly as rpm rises.
Balance the complete assembly, then repeat the measurement at the speed you actually run.
Is runout the same as eccentricity?
They are related but not identical. Eccentricity is a geometric offset between two axes. Runout is what you measure with an indicator as the part or tool rotates, and it includes eccentricity plus form error and any spindle motion.
When you read TIR, you are seeing the sum of all of them. That is why the fix usually involves more than one component.
Send Us Your Grinding Job
Upload your drawing and we will return a quotation with a free DFM analysis within 12 hours.
12-hour quote100% inspection±0.005 mm tolerance
More grinding and machining notes
We publish setup notes, tooling trials and inspection data from the factory floor.
