Buyers ask for radial clearance out of habit. On a deep groove bearing that habit is correct — CN, C3, C4 are exactly the right language. On a double row angular contact bearing it is the wrong question, and asking it can hide the failure that actually matters.
The dimension that decides whether one of these bearings runs smoothly, runs hot, or seizes is axial clearance. This article covers what sets it, what happens at the two ends of the range, and how to write it into a specification so a supplier cannot quietly interpret it their own way.
Why the controlling dimension changes
A double row angular contact bearing behaves like a pair of single-row angular contact bearings mounted face to face, at a contact angle typically in the 25–30° region. That geometry has a consequence: the balls do not sit at the bottom of a groove taking pure radial load. They sit on an angled contact line, and the two rows push against each other.
Because the contact is angled, internal freedom shows up most clearly as axial movement of the inner ring relative to the outer ring. You can measure that directly and repeatably. Radial play, on this geometry, is a derived and much less informative quantity — small radial numbers can coexist with wildly different preload states.
So the industry measures and controls what the geometry actually presents: axial clearance, usually written Ca.
What actually sets the axial clearance
Three ring dimensions and one ball dimension combine. It is worth listing them because only one is the ball, and the ball is the one most people assume is the whole answer:
- Inner ring groove diameter (the groove bottom on the shaft ring)
- Outer ring groove diameter
- The distance between the two groove centres — measured separately on the inner ring and on the outer ring, and they are not the same number
- Ball diameter, by selected group
Point 3 is the one that surprises people. On a real double row bearing the inner-ring groove centre distance and the outer-ring groove centre distance differ deliberately. That difference is what creates the contact angle.
From a 5203-2RS sample we measured and reverse-engineered in our own shop (sample data, 2026-06-18 measurement, not a specification):
| Item | Measured |
|---|---|
| Ball diameter | 6.35 mm (1/4″), 8 per row |
| Inner ring groove centre distance (Mi) | 6.5497 mm |
| Outer ring groove centre distance (Mo) | 6.77 mm |
| Mo − Mi | 0.2203 mm |
| Axial offset per row | 0.11015 mm |
| Groove radius, average | 3.2867 mm |
| Groove curvature ratio R/Dw | 0.5176 |
That 0.22 mm difference between the two groove centre distances is not an error and not a tolerance. It is the design. Change it and you change the contact angle, and with it the axial clearance and the load-sharing between the rows.
This is also why you cannot fix a clearance problem on a double row bearing by changing the ball size alone. The ball moves the clearance, but the groove centre distances set the angle the ball works at.
What goes wrong at each end
Too tight — negative clearance
If the assembled bearing ends up with negative axial clearance, the balls are preloaded against the raceways before the bearing has done any work. Turn it by hand and you feel it immediately: a distinct dragging, notchy resistance instead of a free spin.
What follows is predictable — high running friction, heat that has nowhere to go, contact stress permanently elevated, harsh noise, and drastically shortened life. On a double row bearing this is the single most damaging assembly error, because everything upstream can be perfect and the bearing still fails.
Too loose
Excess clearance shows up as axial play in the mechanism, and the load zone narrows — fewer balls share the load, so contact stress on the loaded few rises. You also get more noise and, in reversing applications, impact as the balls take up the slack each time direction changes.
The industry window
For a standard C0 double row angular contact bearing, a typical finished-product release window for axial clearance is on the order of 6–23 µm. That is a release window, not a target: a competent factory works to a tighter internal window than it ships to, leaving margin for measurement error and for the small changes that occur when closures are fitted.
Note the units. This is a micron-level control problem on a part whose outside diameter is tens of millimetres. It cannot be judged by hand.
When it must be measured
Here is the point most specifications miss entirely: fitting the closure can change the clearance.
Pressing a steel shield into the outer ring shoulder applies force to the ring. Press on the wrong face, or too deep, and the ring distorts. A bearing that measured correctly before closure fitting can be out of window afterwards — and by then the closure is in, so nobody re-measures.
The discipline that prevents this is straightforward. Axial clearance is measured on 100% of units after ball loading and before the closure goes on, under a defined load, turning the inner ring two to three turns first so that every ball seats properly in the groove bottoms before the reading is taken. Units outside the internal window go back for re-matching with a different ball group — they do not proceed to assembly and get sorted out later.
Two details in that sentence do real work:
- Under a defined load. An unloaded reading on an angled contact is not repeatable. Different operators will get different numbers on the same bearing.
- Turn it first. A ball that has not seated in the groove bottom reads as clearance that isn't there.
How to specify it
Put these five things in the enquiry and the quotation becomes comparable between suppliers:
- Axial clearance range in microns — not a clearance class letter borrowed from deep groove bearings
- Measured before or after closure fitting — state which; they are different numbers
- The measuring load used to take the reading
- Sampling — 100%, or a defined sample
- Whether you want the readings back as inspection evidence
Point 2 is the one that separates a supplier who controls this from one who does not. If they cannot immediately tell you at which step they measure, they are probably measuring after closure and reporting whatever it says.
Working with us
If you are specifying a non-standard double row bearing and are unsure what axial clearance your mechanism needs, send the application conditions along with the outline: load direction and magnitude, whether the load reverses, speed, temperature range, mounting fits, and how much axial play the mechanism can tolerate. Fits and temperature both consume clearance in service, so the target has to be set against the installed condition, not the free bearing.
FAQ
Can I just ask for C3 on a double row angular contact bearing?
C2/CN/C3/C4 are radial clearance classes intended for deep groove bearings. On a double row angular contact bearing, specify an axial clearance range in microns instead.
What does negative axial clearance feel like?
Turned by hand, the bearing feels dragging and notchy rather than free. It is an immediate reject — it means the balls are preloaded before the bearing has taken any load.
Does the seal or shield change the clearance?
The closure does not set clearance, but fitting it can distort the rings. That is why clearance should be verified before the closure goes on, and why the specification should say which step the number refers to.
Why measure under load?
On an angled contact, an unloaded reading is not repeatable. A defined load with the inner ring turned two to three turns seats the balls so the reading means the same thing every time.
Can ball size alone fix a clearance problem?
Ball group selection shifts the clearance, but the contact angle comes from the difference between inner and outer groove centre distances. If the angle is wrong, changing balls will not fix it.