GUIDES / 09CNC SOURCING GUIDES

ER collet clamping range: why the marked size is not what it grips

This page is about choosing the collet, not about measuring runout. It explains what the number on the collet means, what it leaves out, and which figures on a record have to be read in the right place before an order goes in. For the order of runout checks, see the A3 guide.

The number is a size, not a range

A collet marked 20 is described everywhere as a 20 mm collet, and the description invites the one reading that fails: that it holds a 20 mm shank. It does not. The marked size is the largest shank the collet will accept, and the band it actually clamps runs down from that number. A collet sold as 20 is a working 19 to 20.

An ER collet face with the series and size etched on the ring, beside a millimetre scale where the yellow band from 19 to 20 is labelled as what a 20 mm collet grips. — The etched number is the top of the band, not its bottom. The band below it is what the bore closes onto.

The same convention runs through every step of the series. The face carries the series and the size; the size is the top of the band, and the band is what the nut closes the bore onto. Ordering by the number alone gets a collet that is dimensionally correct and functionally wrong.

Why the band is one millimetre wide

The collet is a spring with a bore through it. Slots are cut along the body from alternate ends, so the material between them can deflect; when the nut drives the collet into the holder taper, the slots close and the bore shrinks with them. Nothing else in the assembly changes shape. The nut does not squeeze the bore directly, and the holder does not either.

Two face views of the same collet: on the left at rest with wide slits and the bore at nominal, on the right closed by the nut with narrow slits and a smaller bore. — The travel is the slit width. Closing the slots is the only way the bore gets smaller.

That is also the limit. A collet whose slots are already shut cannot close further, whatever the nut does, so the travel left in the design has to be small. One millimetre of diameter is what it leaves: enough to take up the difference between a nominal shank and the one actually in the machine, small enough that the slots never run out of room in normal use.

When the shank measures exactly nominal

Section through a collet and a shank at exactly nominal size, the shank filling the bore so that contact is a single line, with the slits still wide open although the nut is tight. — At nominal the nut is tight and the slits are still open. Nothing has been compressed.

Put a 20.000 shank into a 20 collet and the fit is a light sliding fit at rest. Tighten the nut and the collet has nowhere to travel: the shank already fills the bore, the slots stay open, and the force goes into the collet body instead of into the shank. The nut reaches its stop without ever closing the bore onto the tool.

What holds the tool is then whatever friction the fit happens to give. It feels right in the hand and it indicates acceptably at low speed, then the shank creeps in a heavy cut, usually when the tool is deepest in the work. Nothing about the feel of the nut tells you this has happened.

One millimetre down is where it works

Section of a collet drawn in onto a shank one millimetre under nominal, the slits part closed and a thick yellow contact band running along the bore. — Inside the band there is travel to use. That travel becomes contact along the bore.

Take the same collet and a shank that measures inside the band, and the geometry changes. There is now half a millimetre of travel on each side, the slots are part closed, and the walls meet the shank along their length instead of on a line. The collet is being used where it was machined to be true.

That is the condition the grade figure describes: contact along the bore, an elastic preload holding the shank, and the drive taken by friction spread over the grip length rather than by a few high spots. It is also the condition in which a collet lasts, because nothing is rocking on an edge.

Two ER collets on a white surface, one standing and one lying down, showing the slots, the front taper and the etched size on the ring. — The slots run in from alternate ends, and the size is etched on the ring. The band belongs to that size, not to the series name.

Below the band: closed, and still not touching

Section of a fully closed collet with its slits shut, standing clear of a shank that is smaller than the closed bore, so a visible gap remains on both sides. — Closed on itself, not on the tool. The gap stays whatever the nut is doing.

Go the other way and the failure is quieter. A shank below the bottom of the band lets the collet close all the way: the slots shut, the bore stops shrinking, and the walls are still standing clear of the tool. The nut is tight. Nothing is gripping. Tightening further only loads the collet against itself.

The tool is marked only near the front of the collet, because that is the only place it was ever touched.
Runout changes when the tool is taken out and put back, and changes again when the nut is indexed.
Small tools break early, and the break is at the shank rather than at the cutting end.
A heavy cut pulls the tool out instead of breaking it.

None of those symptoms belongs to a collet that is too large alone. That is why the band is worth checking before the spindle, the holder or the tool is blamed for a problem the collet started.

Two accuracy figures, measured in two places

Two panels. Left, a dial indicator on a collet mounted on a test arbor, with 0.015 mm above it. Right, an indicator on the cylindrical body of a holder, with under 0.005 mm above it. — One figure belongs to the collet on its arbor. The other belongs to the holder body.

Records in this catalogue quote collet accuracy three ways: 0.015 mm on the AA grade ER set, 0.008 mm on another collet set, and 0.005 mm on one more. Separately, the BT holder records quote a maximum runout below 0.005 mm and a taper tolerance below AT3. All of those are real figures, and none of them replaces another.

They are not interchangeable because they are not measured in the same place. The collet figure is a grade: the part is mounted on a test arbor and indicated at the collet, in a fixture, once. The holder figure is a maximum runout for the body, read on the body itself. One describes a part; the other describes a surface.

So a collet quoted at 0.015 mm does not make the holder worse than its own figure, and a holder quoted below 0.005 mm does not prove the collet. When two offers are compared, compare the same kind of figure, taken the same way, on the same part. A grade read at the collet and a runout read on the body are two measurements, not a ranking.

Hardness without a scale is not a number

Two catalogue cards side by side: the left reads 44-48 with the note that no scale is stated, the right reads HRC 56-58 with Rockwell C named, and a VS mark between them. — Same field name, two notations. Until the scale is named, the pair cannot be lined up against anything.

This category writes hardness in more than one form. The ER collet set record gives 44-48 with no scale named. The BT holder record gives HRC 56-58. The 65Mn sharpener collet records give HRC 55-60. Only some of them say what the number was measured on, and only some name the scale at all.

A bare pair is not comparable with a Rockwell C figure, and it is not comparable with the next supplier either, because the scale is unknown. The larger number is not automatically the harder part: without the scale, even the direction is not established. Two records can look like a difference in quality and be nothing more than a difference in notation.

The collet does not grip by itself

A collet between a fixed holder stop on the left and a nut with an internal cone on the right, with the axial push from the nut and two inward squeeze arrows on the collet. — The nut supplies the push; the holder taper turns it inward. The collet is wedged, not self-tightening.

The force path is short and worth following once. The nut is turned, its internal cone bears on the front of the collet and pushes it axially into the holder, the holder taper resists, and the only direction left for the material is inward. That inward movement closes the slots and the bore.

Every part of that path can spend the force without closing the bore: a taper with a chip on it or a film of dried lubricant, a nut face worn or mismatched to the series, a collet already bottomed in the holder, a nut with the right thread and the wrong cone. In each case the nut feels tight and the collet is not clamped, which is the same symptom as a shank at nominal and a different cause.

A carton of BT40 collet chucks in protective sleeves, with one chuck unwrapped in front showing the taper, the flange, the body and the nut. — A BT collet chuck: taper at the back, flange, body, nut at the front. The collet sits inside that nut, between two tapers.

What the records say, and what they leave out

One record in this category lists the series as ER11, ER16, ER20, ER25, ER32, ER40 and ER50, in spring steel, for milling machines, lathes and grinding machines. Others name 65Mn and 65rMn for the spring collets. The toolholder records add a balance range of 12000 rpm to 20000 rpm, a taper tolerance below AT3 and a case depth of 0.8 mm plus or minus 0.2 mm on the records that state it.

What they do not carry is the band for each individual size. The series is listed, the grade is listed, the material is listed; the band is not. It has to come from the marking on the collet in your hand, or from the supplier, before the order goes in. A page like this can explain the rule; only the part can tell you its own band.

A group of ER collets of different sizes arranged on a white surface seen from above, showing how the bore and the slot pattern change across the range. — One series, different bores. The size is on the ring of each one; the series name says nothing about the band.

The order to choose a collet in

A numbered six step list: measure the shank, read the band, confirm the accuracy figure, ask the hardness scale, match holder and nut, then seat and indicate. — Six checks, in this order. Only the first two decide whether the collet can hold the tool at all.

Measure the shank where the collet will grip it, not at the end of the tool.
Read the band from the collet marking, and check the shank sits inside it rather than on its top edge.
Confirm which accuracy figure is quoted and where it is measured before comparing it with another part.
Ask which hardness scale the record uses, and what it was measured on.
Match the holder interface and the nut to the series, and check the nut cone for wear.
Seat the collet in the nut, tighten, then indicate the tool before the first cut.

The list is short because the decision is short. Almost every collet that slips was the wrong band, and almost every wrong band was chosen from the size in the title rather than from the marking on the part.