Setting bearing clearance on an industrial gearbox comes down to two numbers that must agree at operating temperature, not on the bench: the axial clearance you measure cold on a taper-roller pair, and the internal radial clearance of the spherical roller bearing carrying the shaft's radial load. Taper rollers run best with a small preload or near-zero axial clearance because their contact geometry demands it; spherical rollers run best with a positive running clearance because they self-align and must tolerate misalignment. The target condition is an assembled cold axial clearance for the taper pair of roughly 0.02–0.10 mm on small gearboxes and 0.05–0.15 mm on medium units, adjusted upward for the predicted thermal expansion of the shaft, housing, and bearing rings β€” with the spherical roller bearing's internal clearance selected from ISO 5753-1 so that its residual clearance at operating temperature stays positive. The two clearances cannot be solved independently; they share a shaft, and every shim you add at the taper end moves the spherical bearing's position too.

Below is the bench sequence, the ISO clearance classes, and the specific places where gearbox rebuilds go wrong.

Step 1: Establish the Shaft and Housing Expansion Before You Touch a Shim

Thermal expansion drives the answer, so compute it first. Steel expands about 11.5 Γ— 10⁻⁢ per Β°C, so a 300 mm gearbox shaft heated from 20 Β°C to 70 Β°C grows roughly 0.17 mm axially β€” larger than the entire taper clearance window you are trying to set. That means the "clearance" you measure cold is not the clearance the bearing sees at running temperature.

Compute the differential, not the absolute. The shaft expands, the housing expands, and the bearing rings expand at close to the same coefficient, but the shaft's length and the housing's center-distance usually differ. The governing number is the difference between the shaft's axial growth and the axial growth of the bearing seats in the housing. On a typical foot-mounted helical gearbox running 50 Β°C above ambient, that differential lands around 0.03–0.08 mm over a 200–400 mm span depending on housing material.

Ask the gearbox maker for the operating temperature rise and the shaft and housing materials before proceeding. If you cannot get them, use the nameplate ambient rating plus a 45 Β°C rise for the sump as a working estimate, then confirm after the first oil sample at running temperature. Do not use the bearing manufacturer's catalog running clearance as the assembled target; that figure assumes the bearing idling in a free housing, not clamped between a steel shaft and a cast-iron gearbox wall.

Write the expansion number down next to the bench. Every shim decision downstream references it. Rebuilders who skip this step set to catalog clearance, run the box, and pull the taper rollers out blue-black at 200 Β°C because the shaft took the preload away by growing into the bearing.

Step 2: Read the ISO 5753-1 Internal Clearance Class Off the Nameplate

Every spherical roller bearing carries an internal radial clearance class in its designation. ISO 5753-1 defines the radial clearance for unmounted bearings; the common suffixes are C2, CN (normal), C3, C4, and C5, with clearance increasing in that order. A 23040 C3 bearing has a specified unmounted radial clearance greater than a 23040 CN, and the amount depends on the bore and diameter series.

The clearance class exists to absorb thermal growth, so a gearbox with a hot shaft and a cast-iron housing normally calls for C3 or C4 on the spherical rollers. Gearbox makers often already specify it. Check the drawing before ordering. Fitting a C2 or CN where a C3 was specified is one of the most common rebuild mistakes, and it shows up as skid marks on the outboard race within a few thousand hours.

Bearing size (bore) CN unmounted radial C3 unmounted radial Typical gearbox application
60–80 mm per ISO 5753-1 table wider than CN by 20–35 Β΅m small foot-mounted helical unit
90–120 mm per ISO 5753-1 table wider than CN by 30–50 Β΅m mid-size parallel-shaft reducer
130–200 mm per ISO 5753-1 table wider than CN by 50–90 Β΅m large bevel or planetary unit

The exact micrometre bands live in the ISO 5753-1 table and in the bearing maker's catalog, and the amount differs by bore and series. Do not guess them. Read them off the catalog page for the specific bearing part number. The class letter alone tells you the band, not the number.

Remember that unmounted clearance shrinks by roughly 0.6 to 0.8 of the interference fit when the bearing is pressed onto the shaft and into the housing. A 23040 C3 pressed with 0.04 mm interference on the shaft and 0.05 mm in the housing loses around 0.06 mm of radial clearance before any thermal load is applied. Account for that in the residual clearance calculation, not after the fact.

Step 3: Choose the Taper-Roller Target: Near-Zero Axial or a Defined Preload

Taper roller bearings are set axially, and gearboxes usually run them at a controlled preload rather than any positive clearance. The reason is geometry: taper rollers need line contact across the full roller length, and a small preload keeps the roller seated against the cone's large rib. Loose axial clearance lets the roller skew, edge-load, and generate the characteristic smeared, matte finish mechanics call false brinelling.

The target depends on housing stiffness and gear mesh forces. For small to medium foot-mounted gearboxes with a gear-driven shaft, a cold setting of 0.00 to 0.05 mm preload (recorded as a light turning resistance) is common, achieved by shimming tighter as you trial-fit. For a gearbox rebuilt in a cold shop, setting to 0.02–0.05 mm cold clearance and letting the run-in preload develop is often safer than attempting cold preload the housing cannot tolerate. The change from clearance to preload across the operating range is exactly the shaft-growth differential you calculated in Step 1.

The taper bearing makers publish the same guidance through their catalogs. SKF's rolling-bearing handbook and Timken's engineering manual both recommend running taper roller bearings with preload at the operating temperature for gearbox and machine-tool spindle applications, and both give preload as a function of the required axial displacement number rather than a stated millimetre figure. That is the honest answer: the target cold setting is application-specific, not a universal constant. Match the maker's method for that bearing series.

If the gearbox has only one taper roller bearing and the second support is the spherical roller, the pair acts like a located/non-located arrangement. The taper roller fixes the shaft axially; the spherical roller lets it float axially as it grows. Set the taper before you measure the spherical clearance measurement because the taper position determines where the shaft sits axially.

Step 4: Measure the Cold Axial Clearance with a Dial Indicator and a Hand Load

Once the shaft is in the housing and the spherical bearing is fitted, mount a magnetic-base dial indicator with its stem on the shaft end, zero it, and push the shaft axially by hand against the taper roller bearing while watching the needle. Read the travel. That number is the cold axial clearance of the entire shaft stack including the taper pair.

For a more repeatable reading, use two indicators on opposite sides of the shaft end and average them, which cancels end-face runout. Apply an axial push of roughly 50–100 N by hand on small gearboxes, or use a dial pusher or hydraulic load for larger shafts. Do not use a hammer. A sharp axial tap seats the roller and gives a false clearance reading because the load relaxes the seat.

Record the reading at three shaft positions (0Β°, 120Β°, 240Β°) and use the largest value. Runout in the bearing seat produces a measured clearance that varies sinusoidally; the maximum is what governs because that is the position where the shaft will rock into the clearance at speed. A single reading can be low by 0.02–0.03 mm on a shaft with 0.05 mm seat runout.

Check the indicator direction and stem reach before trusting the number. A magnetic base on a warm casting creeps during the measurement. Set the zero after the indicator has sat on the bed for 5 minutes, not immediately after mounting.

Step 5: Shim the Taper End, Then Recheck the Spherical Float

Shim stock under the bearing end cover or behind the outer-ring shoulder moves the taper roller position, and the amount of shim required equals the target cold setting minus the measured cold clearance. If you measured 0.12 mm and the target cold clearance is 0.04 mm, remove 0.08 mm of shim. Simple arithmetic, done on the bench.

After shimming the taper end, recheck the spherical roller bearing. If the spherical roller is the non-located bearing (its outer ring slides axially in the housing), the shim change at the taper does not move its radial clearance. If the spherical roller shares an axial seat with the taper bearing or is itself located, the shim change moves its clearance too, and you must re-measure the radial clearance of the spherical roller with feeler gauges through the oil port or with a bore gauge, depending on whether the housing allows access.

Confirm the spherical roller's residual radial clearance. Subtract the shaft interference and housing interference from the catalog unmounted clearance, subtract the thermal growth in the radial direction, and verify the result is positive and above the manufacturer's minimum running clearance for that bore. At the CN class, a 100 mm bore spherical roller loses roughly 0.05–0.07 mm to interference fits and needs a positive running clearance of at least 0.01–0.02 mm at operating temperature. If the arithmetic goes negative, switch to C3 and re-measure.

Step 6: Run the Box Cold for 30 Minutes and Recheck the Taper Setting

A cold run at no load, driving the gearbox by hand or by a slow motor, equalizes the shims and seats the taper roller against its rib. After the run, bring the shaft to top dead center as before and recheck axial clearance with a dial indicator. Expect the cold reading to have dropped by 0.01–0.03 mm because the rollers have seated. That drop is the reason the trial-fit step matters: a taper set to nominal preload cold will often run in with no measurable clearance at all after the rollers bed in.

If the reading has collapsed to zero, do not re-shim immediately. Bring the box up to operating speed and temperature, run at least 30 minutes at load, then take an oil sample and check the bearing surface temperature. A taper roller bearing running in preload at 60–70 Β°C sump temperature with a 10–15 Β°C differential between outboard race and sump is normal. A bearing running 30 Β°C over sump or hotter is running too tight, and that is when you go back to the shims.

Recheck the spherical roller after the first run as well. Its radial clearance should still show measurable feeler-gauge entry at start-up. If it will not take a 0.02 mm gauge at room temperature, it is either the wrong class or already too tight for the housing fit.

Step 7: Choose Between the Two Bearing Types When the Gearbox Has Both

Most industrial gearboxes mount one shaft in a taper roller pair at the drive end and one shaft in spherical rollers at the load end, so the choice usually is not "which bearing" but "which shaft arrangement." Get that arrangement right and the clearance numbers follow from it.

Use case Taper roller pair Spherical roller bearing
Gear-driven shaft, high axial load, rigid housing First choice; sets shaft axial position and carries gear thrust Second choice; limited axial capacity unless in a floating arrangement
Load-end shaft, misalignment under load, long bearing span Acceptable but sensitive to housing alignment First choice; self-aligns up to the maker's listed misalignment limit
Shaft needing free axial float for thermal growth Not suited unless the second bearing is a floating unit First choice; outer ring slides in housing and takes up growth
Low-speed output shaft with heavy radial load Not preferred for pure radial load First choice; higher radial capacity per bore than a taper pair
Shaft with frequent cold starts and contaminated environment Preload helps but marginal on contamination Sealed or open with heavy preload in greenfield designs

Pick the arrangement first, then the clearance. If both bearings must be on the same shaft (a "box of gears on one shaft" design), the maker's drawing already tells you which is located and which floats. Never substitute a different bearing type on a rebuild without the gearbox maker's written approval; the gear mesh spacing depends on the radial and axial position that the original bearings gave the shaft.

The one real drawback of running both types together: the spherical roller cannot share the axial load without its outer ring being located, and locating it forfeits the self-alignment feature under thermal growth. That is a design compromise you accept, not solve on the bench.

Step 8: Convert the ISO Table Through the Interference Fit and Thermal Rise

The most useful calculation you can do at the bench is a three-line clearance audit. Start with the unmounted internal clearance from the ISO 5753-1 table for the specific part number. Subtract the interference-fit loss from the shaft fit and the housing fit, which the bearing maker's catalog gives as a roughly 0.6–0.8 multiple of the interference value, or read it from the manufacturer's fit conversion table. Add the radial thermal growth of the shaft and rings at operating temperature minus the housing's growth. The result is the running clearance, and it must be positive for the spherical roller and near zero or slightly negative for the taper roller.

For a 120 mm bore C3 bearing with a stated unmounted radial clearance of 0.14–0.20 mm from the catalog, subtract about 0.07 mm for interference and add about 0.03 mm for radial thermal expansion: the running clearance lands at roughly 0.10–0.16 mm. That is comfortably positive. The same bearing in CN with 0.08–0.12 mm unmounted, minus 0.07 mm interference and plus 0.03 mm thermal, lands at 0.04–0.08 mm, still positive but with much less margin. On a hotter or larger gearbox, the CN band can go negative, and that is the specific case where the rebuild fails.

Every number above comes from the catalog and arithmetic, not from a formula that pretends to universal truth. The bearing maker's catalog is the authority for the exact fit conversion; different manufacturers use slightly different housing material assumptions. Read the conversion from the manufacturer's own table for the specific part number and housing material.

Step 9: Check the Bench Against Known Failure Signatures

The bench numbers are only as good as the failure signatures they eliminate. Three signatures tell you the clearance was wrong and tell you which direction it went.

Spalling on the outboard half of the taper roller race, with a matte, smeared finish and no deep pits, means the bearing ran at excessive preload β€” the shaft grew into it or the shim was too thin. Add shim stock to the end cover, targeting the differential you calculated in Step 1. The typical recovery is 0.03–0.08 mm more shim, then recheck after a cold run.

A polished skid mark across the taper roller cone's large end with a bright band and no spalling means most of the bearing carried no load β€” excessive clearance. Remove shim, run the box in at light load, and recheck the axial reading. On a gearbox that has been rebuilt once, this signature often appears alongside a spherical roller that is also running loose, a sign the whole shaft stack was set from a stale drawing.

On the spherical roller, skid marks on both raceways and a heavy wear band at the outer-ring guide lip mean insufficient running clearance, usually from the wrong class or a shrunk housing fit. Switch to the next clearance class and re-measure with feeler gauges after the shaft is seated. If you cannot access the roller via the oil port, remove the housing end cover and use a bore gauge through the exposed race.

Any one of these signatures is enough to justify re-opening the box. Two together almost always mean the clearance was set cold without the thermal differential. That is the expensive lesson: the bench reading was right, the thermal arithmetic was wrong.

FAQ

What is a typical cold axial clearance target for a taper roller bearing in a gearbox?

Most industrial gearboxes in the 50–300 mm shaft size range are set at a cold axial clearance of 0.02–0.10 mm, with the larger end of the band for medium units and a light preload accepted after the first run-in. The governing number is not the clearance itself but the predicted shaft-growth differential; a box with a 0.05 mm cold target on a 250 mm shaft that grows 0.13 mm will run in preload unless the housing compensates. Ask the gearbox maker for the target, and if you cannot get it, set to the smaller end of the band and recheck after the first 30-minute run.

Can I use a C2 spherical roller if the gearbox drawing specifies C3?

No. ISO 5753-1 clearance classes exist to absorb thermal growth, and a C2 band is narrower than CN. A C2 bearing loses most of its residual clearance to the interference fit before the first cold-run, and it will show skid marks on the outer race within a few thousand hours. If a C3 is what the drawing delivers, order the C3. If you cannot source a C3, that is the trigger to consult the gearbox maker rather than substitute a tighter class.

How do I correct clearance if the dial indicator reading does not match my calculation?

First verify the indicator is not on a creep-prone magnetic base and that the shaft-end runout is not contaminating the reading. Take readings at three rotational positions and use the maximum. If the reading still disagrees with the calculated target, the most likely cause is that you are measuring the whole shaft stack, not just the taper pair's clearance β€” a loose spherical roller seat or a floating bearing position adds to the reading. Isolate the taper pair with a second indicator on the bearing outer ring.

Does the spherical roller bearing need to be pulled to check its running clearance?

Not always. With the end cover off, a feeler gauge through the oil port can confirm that the spherical roller's outer ring still takes a 0.02 mm gauge at room temperature, which is the practical test that it is not too tight. For large bores where the oil port does not reach the roller, remove the end cover and use a bore gauge or an internal measuring instrument from the exposed race. The running clearance audit is the reliable method, and it works without removing the bearing.

What if a rebuild starts with a 0.15 mm cold axial clearance and the taper rollers smoke on the first run?

That cold number is the wrong target for that shaft. Recalculate the differential expansion from Step 1 and add shim to bring the cold setting down to the calculated target. Do not chase the smoke by tightening the taper end further without re-measuring the spherical roller's residual clearance, which will go negative if you take up shim on a located bearing. The smoke is the visible end of a shaft that grew into the bearing, and the fix is a reshim plus a second cold run.

How much does interference fit shrink a spherical roller bearing's internal clearance?

The fit-loss is a fraction of the interference, usable only for the specific bearing size and catalog table. In practice, a 0.04 mm shaft interference typically consumes 0.05–0.06 mm of radial clearance, and a 0.05 mm housing interference adds roughly 0.02–0.03 mm depending on wall thickness. The exact figure must come from the bearing maker's fit conversion table, because it depends on the housing material and the ring section. Never apply a single multiplier across every size and material.

Should the taper roller pair run with preload or a small positive clearance?

At operating temperature, near-zero or slight preload is the normal gearbox setting; a positive clearance at speed lets the roller skew and edge-load. The cold setting that produces preload at temperature depends on the shaft-growth differential, so the cold bench reading is a means to an end, not a fixed standard. Match the maker's method for the specific bearing series, and confirm by the first 30-minute run.

What is the most common cost from getting clearance wrong on a rebuild?

The bearing replacement itself is not the main cost; the shaft and housing are. A preloaded taper that smokes typically leaves heat discoloration on the shaft seat and a polished mark on the outer-ring housing bore, and a machined shaft or a sleeved housing can run a significant fraction of a new gearbox price. The bearing set is recovered at generally less than a tenth of that. The cheap insurance is a thermal differential calculation and a recheck after the first run, which costs under an hour of bench time.