A bearing's chamfer is the rounded corner where the ring face meets the bore and the outside diameter. It is not part of the designation. It sits in the dimension table, in a column headed r s min, and that column reads like a property of the bore: a 30 mm shaft takes a 30 mm bearing, and a 30 mm bearing has whatever corner 30 mm bearings have.

It does not. Two bearings with a 30 mm bore and a 55 mm outside diameter carry chamfers of 0.3 mm and 1.0 mm. Nothing about the shaft or the housing separates them. Their widths do: 9 mm against 13 mm.

The column is a floor, and it constrains the shaft

r s min is the smallest chamfer the bearing's own rings may carry. It is a limit on the shaft rather than on the ring: the fillet a turning tool leaves in a shoulder has to sit inside the bearing's corner, so the bearing's corner sets the largest fillet that shoulder may have. That is the whole job of the number, and it is why it matters when a bearing is replaced rather than when it is bought.

One bore, 5 chamfers

Outside diameter (mm)Width (mm)r s min (mm)Series
3740.2617
4270.3618
4790.3619
47170.3NA49
47230.3NA59
47300.3NA69
5590.3160
55131.060
62161.012 62
62201.022 222
72191.113 213 63
72271.123
90231.564

Every row fits a 30 mm shaft. 13 combinations of outside diameter and width, 17 bearings, and 5 different chamfer minimums running from 0.2 to 1.5 mm — a factor of 7.5 on one shaft diameter.

Take the pair that shares the 55 mm outside diameter: 16006 at 9 mm wide with a 0.3 mm chamfer, and 6006 at 13 mm wide with 1.0 mm. Same bore, same housing bore. The narrower one allows a shoulder fillet only 30.0% the size of the wider one's.

The width is the number that moves it

Across the library 91 bore-and-outside-diameter combinations appear on more than one bearing, and in 17 of them the chamfer is not the same:

Bore (mm)Outside diameter (mm)160: width / r s min60: width / r s minRatio
20428 / 0.312 / 0.62.00×
25478 / 0.312 / 0.62.00×
30559 / 0.313 / 1.03.33×
35629 / 0.314 / 1.03.33×
40689 / 0.315 / 1.03.33×
457510 / 0.616 / 1.01.67×
508010 / 0.616 / 1.01.67×
559011 / 0.618 / 1.11.83×
609511 / 0.618 / 1.11.83×
6510011 / 0.618 / 1.11.83×
7011013 / 0.620 / 1.11.83×
7511513 / 0.620 / 1.11.83×
8012514 / 0.622 / 1.11.83×
8513014 / 0.622 / 1.11.83×
9014016 / 1.024 / 1.51.50×
9514516 / 1.024 / 1.51.50×
10015016 / 1.024 / 1.51.50×

All 17 are the same two series, 160 and 60, and in every row the narrower ring carries the smaller chamfer — by a factor of 1.50× at the least and 3.33× at the most, on widths that differ by 4 to 8 mm. The width shrinks by 30.8% to 40.0% and the corner goes with it.

The two series agree at the small end. Up to 17 mm their chamfer minimums are identical; from 20 mm they part, and above that bore not one of the 17 shared bores has them meeting again.

Why the tables split there and not earlier is not in this data. What is in this data is that the split is total: above it, the thinner series never catches up.

The bearing type does not move it

The opposite test settles what the number depends on. 71 combinations of bore, outside diameter and width appear on more than one bearing type here, and not one of them carries two different chamfers.

1200 and 6200 are 10 × 30 × 9 mm and come from two different types; both are 0.6 mm. The same holds at the other end of the range, with 2320 and 22320 at 100 × 215 × 73 mm.

Put the two results side by side. Change the type and hold the envelope: the chamfer never moves. Hold the bore and the outside diameter and change the width: the chamfer moves by up to 3.33×. The number belongs to the envelope, and the envelope is three numbers, not two.

Which types carry the column at all

Bearing typeRecordsr s min columnIn the comparison
Deep groove ball213yes213
Drawn cup needle roller100——
Needle roller94yes94
Needle roller, no inner ring94yesno bore to group by
Self-aligning ball84yes84
Cylindrical roller55——
Spherical roller45yes45
Thrust ball41——
Tapered roller38——

436 of the 764 bearings in this database are in the comparison, across 4 types. The rest are excluded for two different reasons: 4 types have no such column at all, and one type has the column but no bore to group it by.

Inside a series it only rises

14 values cover 40 bores and 77 outside diameters. Sort any one of the 19 series by bore and the chamfer minimum never falls: 436 bearings, no exception in any series.

The whole ladder runs from 0.1 mm on 61700 — 10 × 15 × 3 mm — to 7.5 mm on 6360, 300 × 620 × 109 mm. A factor of 75.0 between the smallest and the largest corner in the library, which is why a single "typical chamfer" is not a number anyone can use. The full ladder, value by value with the bores each one covers, is on the ISO 582 page; this page does not print it a second time.

The widest spread in the library is at 70 mm: 7 values there, 0.3 mm to 3.0 mm, a factor of 10.0 on one shaft diameter. That is wider than the example above, and it is the same mechanism: on that one shaft, several series and several widths overlap.

What it costs when you substitute

Swapping a 6006 for a 16006 looks like a free substitution. The 55 mm housing bore and the 30 mm shaft are unchanged, the designation grows by one digit, and every boundary dimension on the drawing still matches.

The corner does not. The shoulder fillet that cleared 1.0 mm has to come down to 0.3 mm, and a fillet left at 1.0 mm interferes by 0.7 mm: the ring stands off its shoulder by that much and is then located by nothing. Going the other way — 16006 to 6006 — widens what the shoulder may carry, so it is the direction that never causes this fault.

A fillet is not a dimension anyone re-checks when a part number changes by one digit. That is the failure this column prevents, and it is why the number has to be read on the row being bought rather than on the bore.

Where this stops

  • 4 bearing types have no such column, and the reasons do not agree with one another — a document that splits the corner in two, a value given per ring, a table that contradicts itself, a catalogue that never printed the cell. Each type states its own reason on its series page, and this page does not restate them: the four sentences disagree with each other, and a summary written here would be a fifth version to keep in step.
  • The comparison covers 4 types. The 4 without the column are absent from it, which is not the same as being equal to anything.
  • 94 needle bearings with no inner ring carry a chamfer value but no bore — their mounting dimension is the shaft diameter Fw — so there is no bore to group them by and they are outside this comparison entirely.
  • No maximum is given here. ISO 582:1995 made the ceiling depend on bore, the earlier edition did not, and two editions cannot share one column, so that page records the minimums and the rule and names the edition to check.
  • Nothing about the shape of the corner. Both limits bound where the arc may lie, not what it looks like, so a chamfer measured between them is acceptable whatever its profile.
  • Not why the standard splits where it does. The split is reported above; it is not explained, because nothing in this data explains it.

Every bearing page prints its own r s min beside its boundary dimensions, so any row in the tables above can be checked against the page it came from.