CNC Machining How Much Jump Is Too Much on the Tool Holder

• 1
  Key takeaways
• 2
  What runout actually is and where it comes from
• 3
  CNC machining how much jump is acceptable per holder type
• 4
  How to tell runout is the problem and not something else
• 5
  What excessive jump costs you on a real job
• 6
  Step by step: measuring and reducing tool holder runout
• 7
  Holder type vs. typical TIR and best use

Runout is the total indicated reading (TIR) of a rotating cutting edge against a fixed reference. If a 12 mm end mill spins with its edge sweeping a 12.016 mm circle, TIR is 0.016 mm. The spindle centerline has not moved; the tool has. That distinction matters because runout is a stack of small errors, not one big one, and you can only fix it if you know which layer is contributing.

The stack starts at the spindle taper. A worn or bruised taper face seats the holder slightly off-axis. The holder itself adds its own error, and a collet nut or a worn collet segment adds more. Finally the tool shank and the grind of the flutes contribute whatever they contribute. Most shops see 0.002–0.004 mm from a clean spindle, 0.003–0.008 mm from the holder, and up to 0.010 mm from a collet that has been used past its life.

The reason engineers care is not cosmetic. In CNC machining, how much jump you carry sets your effective chip load per tooth. A two-flute cutter running at 0.05 mm per tooth with 0.020 mm TIR cuts 0.060 mm on one flute and 0.040 mm on the other. The heavy flute wears first, the light flute rubs, and the surface finish turns inconsistent along the same pass.

Runout also moves the cutting edge off the programmed path. On a boring operation that shows up directly as bore size error. On a profiling pass it shows up as wall taper. It is rare for runout alone to blow a ±0.005 mm tolerance, but it eats the margin you need for thermal growth, tool wear, and fixture movement later in the run.

• 1
  Spindle taperTypical contribution 0.002–0.004 mm when clean and undamaged.
• 2
  Holder bodyTypical 0.003–0.008 mm; shrink-fit and hydraulic holders sit at the low end.
• 3
  Collet and nutUp to 0.010 mm once the collet has lost its spring and grip.
• 4
  Tool shank and grindUsually under 0.005 mm on a good carbide shank, worse on cheap imports.

There is no single universal number, but there is a practical band per holder family. For general milling in aluminum with a 10–20 mm cutter, keep TIR under 0.010 mm. For finishing passes in steel or stainless, target 0.005 mm or better. For reaming, boring, and any operation where the tool diameter sets the feature size, target 0.003 mm. Those are working limits, not absolutes, and they assume a clean spindle and a tool with at least 3× diameter of engagement.

Holder type sets the floor you can reach. A standard ER collet chuck in good condition lands around 0.008–0.010 mm. A precision ER collet with a matched nut gets to 0.005 mm. A hydraulic holder reaches 0.003 mm and damps vibration at the same time. A shrink-fit holder reaches 0.003 mm and holds it over a wider speed range. Side-lock holders are the weakest link at 0.020–0.050 mm, which is why they belong on roughing, not on a Ø8 mm finishing cut.

Overhang changes the consequence even when TIR does not change. A tool held 30 mm out deflects roughly twice as much as the same tool held 60 mm out, for the same side load. If your holder reads 0.008 mm at the gauge line, that error is magnified at the cutting edge by the length of the tool. This is the reason a job that ran clean in a short holder starts chattering after a deep pocket forces a long reach.

Material matters too. Aluminum tolerates runout because it cuts easily and the heavy flute just removes more metal. Titanium, Inconel, and hardened tool steel do not. In those materials the heavy flute takes a shock load, heat concentrates on one edge, and tool life falls fast. If you are cutting titanium, treat 0.005 mm as your ceiling, not your target.

Runout has a signature. One flute of a multi-flute cutter wears or chips while the others look fine. Surface finish shows a repeating pattern that matches the flute count, not the feed. Bore size comes out oversize by a consistent amount that does not drift with tool wear. Slot width varies between the two sides of the same pass. If you see two or more of those at once, measure TIR before you touch the speeds and feeds.

Compare that with a feed or rigidity problem. Chatter from low rigidity gets worse as you increase depth of cut and often improves when you raise speed. Runout does not care about depth of cut; it is there at 0.2 mm and still there at 2 mm. Tool wear gives you a slow, monotonic drift in size across hundreds of parts. Runout gives you a fixed offset that appears the moment you change the tool.

The cheapest test is a dial indicator on the flute. Set the tool in the spindle, bring the indicator tip onto the cutting edge, and rotate the spindle by hand. Read the difference between the high and low points. Do it on the flutes, not on the shank, because the shank can be perfect while the grind of the flutes is not. A 0.002 mm indicator is enough for this; do not try to judge runout by eye or by sound.

One more check: swap the tool only, not the holder. If TIR follows the tool, the shank or the grind is at fault. If TIR stays with the holder, clean the taper and re-seat. If it stays with the spindle, the machine needs a taper regrind or a spindle inspection. That single swap separates three very different repair paths.

The visible cost is scrap. A reamed hole that comes out 0.015 mm oversize is scrap, and it is often discovered at final inspection after the part has already been through several operations. On a part with a ±0.005 mm bore, that is the whole tolerance gone before you have accounted for thermal drift or tool wear.

The invisible cost is tool life. A cutter running 0.020 mm TIR in stainless can lose 30–50% of its life compared with the same cutter at 0.005 mm. That shows up as more tool changes, more offsets to manage, and more chance of an operator missing a size shift. On a long run the tooling bill alone can exceed the scrap bill.

There is a finish cost as well. If a customer specifies Ra 0.8–1.6 μm and your surface reads Ra 2.5 μm because one flute is doing the work, you either slow the finishing pass down or add a second operation. Both eat cycle time. Fixing the holder is usually cheaper than adding a pass.

The last cost is schedule. Runout problems tend to surface at the worst time, in the middle of a run with a delivery date attached. A holder that is cleaned and checked before the job starts costs a few minutes. A holder that fails mid-run costs a re-setup, a re-inspection, and possibly a partial re-machining of parts already in the fixture.