"Here is a bearing. Can you make it?" arrives with no drawing more often than with one. The bearing may be obsolete, from a machine whose maker is gone, or bought through three intermediaries who each know less than the last.
This is a walk through one real teardown — a 7 × 19 × 10 mm double row sealed bearing from a customer sample — showing the measurement sequence, the derived quantities, and, more usefully, the two cross-checks that caught genuine errors before they reached a drawing.
Step 1: Envelope and what the sample is
Hand measurement first, before anything is dismantled.
| Item | Measured |
|---|---|
| Envelope | d 7 × D 19 × B 10 mm |
| Rows | 2 |
| Closure | Rubber seals, both sides |
| Balls | 3.175 mm (1/8″), 7 per row, 14 total |
| Cage | Nylon |
Ball count and cage material are confirmed by dismantling, not assumed from a catalogue. Catalogue sources for this envelope describe a double row type at 7 × 19 × 10, which matched — but a catalogue tells you what a nominally similar part should be, not what is in your hand.
Step 2: Profilometer, both rings
This is where a teardown stops being guesswork. A profile trace across each raceway gives groove radius, the distance between the two groove centres, and the profile deviation of each groove.
| Parameter | Inner ring | Outer ring |
|---|---|---|
| Bore / shoulder | 7.00 mm | 15.05 mm |
| Outside diameter | 11.05 mm | 19.00 mm |
| Groove bottom diameter | 9.81 mm | 16.10 mm |
| Groove centre distance X1 | 3.3123 mm | 3.4431 mm |
| Groove radius R1 / R2 | 1.6710 / 1.6802 mm | 1.6888 / 1.6854 mm |
| Profile deviation | 0.0064 / 0.0038 mm | — |
| Step feature Z1 | — | 0.2401 mm |
Two grooves on each ring, not one. That single observation settles the most important production question: this is ground as a double-groove part, and the grinding operations are planned accordingly.
Step 3: Derived geometry
Every derived value is kept as a live formula in the measurement file, so correcting one measured input recalculates everything downstream:
| Derived | Value | From |
|---|---|---|
| Radial section | 6.000 mm | (D − d) / 2 |
| Pitch diameter | 12.955 mm | (inner groove dia + outer groove dia) / 2 |
| Ball diameter estimate | 3.145 mm | pitch diameter − inner groove diameter |
| Inner ring wall (bore to groove bottom) | 1.405 mm | (groove dia − bore) / 2 |
| Inner ring groove depth | 0.620 mm | (shoulder OD − groove dia) / 2 |
| Outer ring wall under groove | 1.450 mm | (OD − groove dia) / 2 |
| Outer ring shoulder height | 0.525 mm | (groove dia − bore) / 2 |
Cross-check 1: does the estimated ball match the real one?
The estimate from pitch diameter minus inner groove diameter gave 3.145 mm. A second, independent estimate from groove conformity gave a range of roughly 3.15–3.23 mm. The dismantled bearing contained 3.175 mm balls — a standard 1/8 inch size, inside both estimates.
Two independent methods landing either side of the true value is the result you want. It means the estimation method is sound and can be trusted on the next teardown where the balls are damaged, missing, or the bearing cannot be dismantled without destroying it. Measured conformity worked out at about 0.528, a normal figure.
Cross-check 2: when the numbers refuse to agree
This is the part worth reading twice.
An early measurement recorded the outer ring bore as 10.5 mm. The inner ring shoulder outside diameter measured 11.05 mm. Those two numbers cannot both be true — the outer ring bore would be smaller than the inner ring outside diameter, so the bearing could not be assembled at all.
That contradiction was the signal. Re-measurement gave the outer ring bore as 15.05 mm, and everything resolved: about 2.0 mm of radial gap per side, a normal span for the seal to bridge. The 0.2401 mm step feature on the outer ring then made sense too, consistent with a seal-seat bore stepping up to a shoulder that clears the balls.
A second correction followed the same route: ring height was first recorded as 9 mm, then corrected to 10 mm, at which point the sample envelope matched the standard envelope for this size exactly — which in turn told us the part could be quoted against the standard type rather than as a fully bespoke design.
Neither error was caught by measuring more carefully. Both were caught because derived geometry has to be self-consistent, and it wasn't. That is the entire argument for computing pitch diameter, wall thicknesses and shoulder heights during a teardown rather than after it. A teardown that only records what the instruments say will carry its errors straight through to a drawing.
Step 4: What the geometry says about the design
The groove centre distances differ between rings: 3.4431 mm on the outer, 3.3123 mm on the inner, a difference of 0.1308 mm — an axial offset of about 0.0654 mm per row.
Combined with the groove radii and ball diameter, that offset implies a contact angle in the region of 17–20°. That is the signature of an angular contact design rather than a double row deep groove, and it means the grinding must reproduce that offset, not just the groove diameters.
As with any angle derived from a free sample, it is an inference from unloaded geometry. It sets the design intent for a sample plan; it is confirmed against measured clearance and contact track before anything is frozen.
What we hand back
- A measurement file with all measured values, live formulas for derived geometry, and the profilometer traces embedded next to the numbers they came from
- An explicit separation of measured, calculated and still to be confirmed — inferences are labelled as inferences
- A list of open items requiring physical confirmation
- A manufacturing drawing and sample plan once the open items are closed
If you have a sample and no drawing, send the sample and the application conditions. The measurement pass is the first deliverable, and it stands on its own even if you never order the part.
FAQ
Can you work from a sample with no drawing?
Yes. Dismantling plus profilometer measurement produces the geometry file that a drawing is then built from.
What if the bearing cannot be dismantled?
Ball size can be estimated from pitch diameter and groove diameter, cross-checked against groove conformity. On this teardown both estimates bracketed the true 3.175 mm ball.
How do you know it is angular contact rather than double row deep groove?
The two rings have different groove centre distances. That offset — 0.1308 mm here — creates the contact angle. A double row deep groove design would not have it.
How accurate is a reverse-engineered contact angle?
It is an inference from unloaded geometry, good enough to set design intent, not good enough to freeze a specification. It must be confirmed against clearance and contact track.
Do you need the whole bearing or just the rings?
The whole bearing. Ball count, ball size, cage type and seal arrangement all come from the assembled part.