Collet runout: how to read the spec without being misled

Almost every ER collet listing quotes a runout number. Almost none of them quote the distance it was measured at — and without that distance, the number tells you nothing.

A runout figure without a distance is meaningless

Runout is an angular error: the tool axis tilts slightly relative to the spindle axis. That means the deflection you measure grows with distance from the collet face. Measure one millimetre out and you get one number; measure fifty millimetres out and you get a much worse one — same collet, nothing changed.

This is why the standard specifies a gauge length for every test. DIN 6499 sets the mandrel length by tool diameter:

Mandrel Ø d (mm)	Gauge length L (mm)
1.0 – 1.6	6
> 1.6 – 3.0	10
> 3.0 – 6.0	16
> 6.0 – 10.0	25
> 10.0 – 18.0	40
> 18.0 – 26.0	50
> 26.0 – 30.0	60
> 30.0 – 34.0	80

Practical consequence. When two suppliers both claim "0.01 mm", check whether either states the gauge length. A figure measured at L = 6 mm and one measured at L = 40 mm are not comparable, even though both are real measurements taken honestly. If a listing gives no distance, treat the number as marketing copy and compare on material and build instead.

Accuracy classes you'll see on the market

Common label	Published runout	DIN reference	Where it earns its keep
Standard / Class 2	≤ 0.015 mm	DIN 6499 class 2	Roughing, general milling, most router work
Precision / Class 1	≤ 0.008 – 0.010 mm	DIN 6499 class 1	Finishing passes, slotting, decent surface finish
Ultra precision / AA	≤ 0.005 mm	above standard	Micro tools, high-rpm finishing, small-diameter end mills

These are the bandings most commonly published across manufacturers. Individual makers use their own grade names, so always read the actual millimetre figure rather than the marketing tier.

Where to stop spending

An AA collet in a spindle with its own 0.02 mm of bearing runout buys you nothing. Before buying a premium collet it's worth knowing what the rest of your machine contributes, because the errors add up:

Spindle bearings. Usually the largest single contributor on a router. Measure it once with the collet removed.
The holder taper. A speck of swarf between taper and spindle seats the holder off-axis. This is the most common cause of sudden, unexplained runout.
Clamping outside the collapse range. See the size chart — this produces instant, severe runout.
Nut condition. A dry or worn nut bearing drags the collet sideways as you tighten. Hence the value of bearing-sealed UM nuts.
Then the collet itself. Usually the last thing to spend money on, and often the easiest to fix.

Worth knowing before you upgrade anything. Clean the taper, use the correct size collet for the actual shank diameter, and torque the nut to something reasonable. On most hobby and light-production machines those three steps recover more accuracy than moving from a standard to an ultra-precision collet — and they cost nothing.

How to measure your own

You don't need a metrology lab. A dial indicator with a magnetic base and a known test mandrel gets you most of the way:

Seat a clean test mandrel in a clean holder, tightened to your normal torque.
Indicate at two points along the mandrel — say 10 mm and 50 mm from the collet face.
Note both numbers and always quote both. The pair tells you far more than either alone.
Repeat with a second collet size. If one is much worse, that collet is the problem; if all are equally bad, look at the spindle or taper.

What to buy, given all that

For most single-machine shops: one standard set for the roughing sizes you use hard, and one precision collet in the shank diameter you finish with. Standard across everything is false economy; AA across everything is wasted money.