Say the indicator on the shank reads 3 µm, yet the same flute on a 0.5 mm end mill keeps chipping first. The reading looks small and the cutter says otherwise. Toolholder runout can sit behind that argument, because a reading near the holder nose can look fine while the tip runs further off center.
How much is too much depends on the chip each flute is meant to take. On a small cutter that chip may be only a few microns thick, so a runout figure that means nothing on a 20 mm rougher can leave one flute doing nearly all the work. The way through is to measure at the cut, compare the reading with the chip load, then find which part of the stack it comes from.
First, sort what you’re seeing
| What you see | What it can point to | Check first |
|---|---|---|
| The same flute wears or chips first | Uneven load between flutes | TIR at the cut against chip load |
| Low reading near the nose, higher at the tip | Possible angular misalignment | Readings at two or three points |
| The reading changes each time you re-clamp | Seating, dirt or a tired collet | Clean, reseat and read again |
| Slots cut wider than the cutter | The tip orbiting off center | TIR close to the cutting edge |
| Low static reading, same flute still wears | Effects that only show at speed | Dynamic check, stickout, engagement |
Any of these can have another cause, so let the flute evidence and the reading agree before you change hardware.
1. Measure toolholder runout where the tool cuts
Harvey Performance’s guidance is to measure at the point where the tool will be cutting. BIG DAISHOWA reads the holder nose, the mid-shank and the tip; on a concentric setup all three come out the same and close to zero. More than one reading is what makes the check useful.
If toolholder runout grows toward the tip, that can indicate angular misalignment. In the tilted setup shown below, each extra millimeter of stickout adds to the offset at the cut. We cover what stickout does to stability in tool stickout and milling chatter.
The check itself is slow and manual: with the machine stopped under its safety procedure, you turn the spindle by hand and watch the needle. Flutes make a contact reading awkward, so a ground shank close to them gives you a contact reference. On miniature tools an indicator’s contact force can push the tool and spoil the number, which is why Harvey recommends a laser for them.
2. How much toolholder runout is too much?
No single number covers every cutter, so start with a ratio. Tormach’s rule of thumb is that tool life drops sharply once total indicated runout (TIR) passes about 20% of the chip load. On a two-flute 0.5 mm end mill at 0.005 mm per tooth (illustrative values, not a recommendation), that limit is 1 µm of TIR.
A simple geometric model shows why. Line the two flutes up on the offset and, while both are still cutting, the long flute’s peak chip is about the chip load plus the TIR. The short flute’s is about the chip load minus the TIR. At 5 µm of TIR in this example, the short flute stops cutting and the long flute’s peak chip is about 10 µm, twice the programmed feed per tooth.
Attanasio’s micro-milling study describes the same end state, with run-out high enough that just one flute cuts.
Two published figures show the scale. Union Tool recommends a runout control value of 5 µm or below for its small long-neck CBN end mills of 1 mm and under. BIG Kaiser’s Jack Burley, writing in Cutting Tool Engineering, gives an example: a three-flute 1/16 in end mill profiling mold steel at 0.0002 in per tooth (about 5 µm). It runs unbalanced at 0.0005 in of TIR (about 13 µm).
Lower runout showed up in tool life too. The same article reports a four-flute 10 mm carbide end mill cutting about 1.3 times as far when runout fell from 0.0006 in to 0.00008 in (about 15 µm to 2 µm).
Lowering the feed eases the load, not the runout
Lowering the feed can ease the overload on the long flute, but it doesn’t remove the runout. The chip gets thinner while the runout stays put, and below a certain thickness the edge rubs and ploughs. If that’s what you see, fix the runout instead of reducing the feed again. We go through that effect in engraving quality problems.
3. Split toolholder runout into spindle, holder, collet and cutter
Toolholder runout measured at the tip is the combined result of several parts, and one reading can’t tell you which. Tormach’s method separates them with a dial test indicator, working from the machine outward.
- The spindle taper. This reading is the spindle alone. One machine builder’s monthly service check allows 0.005 mm at the taper. On a precision test bar it allows 0.013 mm at the gauge line and 0.025 mm at 150 mm down. If the taper reads high after cleaning, look for damage before suspecting the holder; we cover taper problems in tool stuck in a VMC spindle.
- A precision dowel in the holder. This adds the holder and its clamping, with the cutter taken out of the picture.
- The taper inside an empty collet chuck. This separates the holder from the collet.
- The holder turned 180° in the spindle. A high point that follows the holder is a reason to start with the holder; one that stays put points toward the spindle. Spindle and holder errors add and cancel, though, so compare the size of the reading as well as its position, and confirm with a known test bar.
4. Recheck the parts that change at every setup
Tool assembly can introduce substantial toolholder runout. BIG DAISHOWA notes that small particles or oil in a holder can cause large runout. REGO-FIX’s advice in Modern Machine Shop adds a number: seating the collet in the holder before the nut can cause up to 0.001 in (about 25 µm). That’s five times the chip in our example.
The same advice asks for at least two-thirds of the collet’s gripping surface on the shank and a torque wrench in place of feel, since an over-tightened nut can deform the collet.
Collets wear out too: Tormach calls ER collets consumables and once traced a problem to a new one that was bad out of the box.
Repeating the check tells you whether the number is stable. Re-clamp the tool, reseat the holder and read again. When the value or the high point moves around, the cause is more likely seating, dirt or the collet than one permanently eccentric part.
5. When toolholder runout reads low and the same flute still wears
A hand-turned reading is a geometric check, and the cut happens at speed. BIG DAISHOWA tests reported in the same Cutting Tool Engineering article found machines accurate at 500 rpm with more than 0.001 in of runout at 30,000 rpm.
If the static number is low and one flute still takes the damage, try a suitable non-contact dynamic measurement. Then check stickout, radial engagement and tool deflection. If the flutes wear evenly and the parts are in tolerance, there’s no immediate reason to keep chasing runout for this fault.
When it’s the machine, and when it isn’t
- Start with the tooling as the source of the toolholder runout if the high point follows the holder, a new collet clears the problem, or the reading changes with every clamp.
- Start with the spindle if the taper reads high after cleaning, the high point stays put when you turn the holder, or a known test bar reads high in this spindle and fine in another.
- Look elsewhere if the reading is low and the wear is even. The cause is more likely in the chip load, toolpath or workholding. If the complaint is hole shape, milled holes that aren’t round walks through that test.
A checking rhythm for small tools
- At every assembly: a clean taper, collet and shank, and the torque the holder maker specifies, because each assembly can add a new toolholder runout error.
- On the first tool of a small-cutter job: a reading near the cutting end, written down with the high point, so you have a baseline to compare with when a flute chips.
- Monthly: the spindle taper and a test bar, the interval that machine builder’s service check uses, so you can spot a developing spindle problem.
What we check on our machines
Our vertical machining centers and our horizontal, gantry and engraving machines have their spindle runout checked before they ship. A 300 mm test bar goes in the spindle, and its far end has to run within 0.008 mm.
That check covers the spindle. Toolholder runout from the holder, collet and cutter you add still counts, so the reading near the cut stays on your setup sheet. The spindle taper also gets an air blast at each tool change, which helps keep chips off the surface the next holder seats on.
Small cutters also need speed. Even at 24,000 rpm, a 0.5 mm end mill only reaches about 38 m/min of surface speed. For work under 1 mm we’d start with the motor spindles on our CS engraving machines. For small tools mixed with general milling, we’d set up a V-series VMC with the 12,000 rpm direct-drive spindle.
If you’re specifying a machine for small-tool work, send us the material, the smallest cutter and the chip load you plan to run through our inquiry form. We’ll suggest a spindle and machine configuration and quote it.
Sources and Method
- Harvey Performance Company, Reducing Tool Runout, In the Loupe: where to measure, lasers for miniature tools.
- Tormach, Measure and Fix Spindle Runout: The Tool Life Killer: the 20% TIR rule of thumb and the measuring sequence.
- Union Tool, CBN-LRF 4 Flute High-Grade Long Neck Radius End Mills: 5 µm runout control value for φ1 and below.
- Jack Burley, Riding herd on runout, Cutting Tool Engineering (2010): chip-load example, end-mill test, runout at speed.
- BIG DAISHOWA, The Science of Tool Holder Concentricity and Cleaning is Key to Runout Control.
- REGO-FIX tips in Five Common Mistakes Shops Make with ER Collets, Modern Machine Shop (2023).
- A. Attanasio, Tool Run-Out Measurement in Micro Milling, Micromachines 8(7), 221 (2017).
- Haas Automation, Spindle – Taper – Maintenance – VMC: monthly service limits.
- Method: the two-flute chip split is a simplified geometric model with the flutes in line with the offset. Worked numbers are illustrative, and unit conversions are rounded.
