Bearing Preload and Clearance: A Decision-Based Selection Guide
Bearing selection becomes a guessing game the moment clearance and preload enter the drawing. Both terms describe the tiny internal gaps that decide whether a spindle runs silent for years or hums for three months and dies. The decision is not one number on a datasheet; it is a chain of choices among clearance class, fit, preload method, and assembly procedure, and every wrong link produces a different, usually expensive, failure.
This guide is built as a decision sequence. It defines the two parameters cleanly, answers when preload is mandatory, compares the three preload methods, ties clearance class to shaft and housing fits, works one calculation to a number you can defend, and finishes with an assembly checklist. By the end, the clearance-preload decision becomes a documented process instead of an engineer’s instinct.
1. The Two Parameters, Without the Jargon
Clearance is the total internal play between the rolling elements, the raceways, and the cage when the bearing is unmounted. It is quoted as a radial clearance class, with C2 tighter than normal, normal being the default, and C3 and C4 progressively looser. Preload, by contrast, is the deliberate application of a small negative clearance: the bearing is assembled so the rolling elements are squeezed between the raceways before any external load arrives.
The two are related but not interchangeable. A bearing selected with the right clearance can still run poorly if fitted into a housing that removes the clearance through interference. Conversely, a bearing with nominal clearance can behave like a preloaded unit if its outer ring is pressed into an oversize housing bore. The effective clearance after mounting is what matters, never the class marked on the box alone.
2. When Is Preload Mandatory? The Decision Table
Preload is not always right. It stiffens the bearing, removes clearance-based play, and prevents rolling elements from skidding at light load, but it also adds friction heat and reduces life if overdone. Use the decision table to find whether your application belongs to the preload side.
| Application Signature | Preload Needed? | Reason |
|---|---|---|
| High-speed motor with light radial load | Yes, light | Prevents skidding and ball slip |
| Spindle requiring axial and radial stiffness | Yes, medium | Removes play for positioning accuracy |
| Bearing carrying heavy constant load | Usually no | Load dominates; preload adds pointless heat |
| High-speed, high-temperature application | Yes, controlled | Counteracts thermal expansion of shaft |
| Low-speed, load-uniform gearbox | No | Clearance protects under shock and thermal growth |
| Precision rotary table and indexers | Yes, rigid | Eliminates back-and-forth positioning backlash |
The pattern: preload is the tool of choice when the load is light relative to the bearing capacity, the speed is high, or the application demands positional repeatability. Where steady heavy load already presses the rollers into the races, preload mostly manufactures heat.
3. Choosing the Preload Method
Once preload is required, pick the method. The three practical approaches differ in how precisely the squeeze is controlled and how easy they are to maintain on the floor.
| Method | How It Works | Accuracy | Best For |
|---|---|---|---|
| Position (locknut) preload | Nut tightens to a measured position | Good, needs feel/measure | Spindles, moderate volume |
| Constant-pressure (spring) preload | Spring stack holds near-constant force | Self-compensating | Variable thermal or speed range |
| Matched-pair tapered bearings | Factory-ground faces set preload | Excellent, repeatable | High-precision spindles |
Spring preload tolerates thermal change because the force stays roughly flat as the stack compresses. Locknut preload is cheaper and stiffer, but over-tightening is the leading assembly error, so it demands torque control and a measurement step. Matched pairs are the least forgiving to set up and the most forgiving in service.
4. Clearance Class Versus Mounting Fit
The clearance marked on the bearing is not the clearance in the machine. Interference fits shrink the internal clearance: pressing the outer ring into the housing squeezes the outer race inward, and pressing the shaft into a tight inner-ring bore expands the inner race outward. Both effects reduce the running clearance, and on small bearings the reduction can consume the entire normal class.
| Operating Condition | Suggested Clearance | Suggested Fits |
|---|---|---|
| Normal general-purpose | Normal (CN) | j6 shaft / H7 housing |
| High-speed or light load on shaft | C3 | k5 shaft / J7 housing |
| Heavy shock load, both rings tight | C3-C4 | m6 shaft / N7 housing |
| Thin housing, aluminium housing | C3 | k6 shaft / M7 housing |
| After-mount clearance critical | Compute remaining | Validate with feeler or gauge |
The guiding rule: select clearance so that the after-mounting clearance stays positive for clearance-class operation, and know the reduction quantity. A common first-principles estimate is that roughly 70-80% of the maximum interference reduces the radial internal clearance. That single 0.8 factor converts most clearance mistakes into a preventable one.
5. Worked Calculation: Set a Preload You Can Justify
For a locknut-preloaded angular contact pair used in a small spindle, the recommended preload force is proportional to about 5-8% of the dynamic load rating C for light preload, rising to 10-15% for medium. Take a pair with C of 24 kN. Light preload is thus about 1.2-1.9 kN; medium is about 2.4-3.6 kN. Choose medium at 3.0 kN for a stiff positioning spindle.
Convert force to locknut torque with T = F x (d/2) x tan(alpha + rho), where alpha is the thread lead angle and rho the friction angle. For a 35 mm thread with effective radius near 16 mm, lead angle about 4 degrees and friction angle about 9 degrees (mu of 0.16), tan(13) is about 0.23. T is then about 3000 N x 0.016 m x 0.23, roughly 11 N-m. Apply that torque, then verify actual preload by measuring starting friction torque or by gap measurement on a feeler gauge, because friction in real threads scatters 15-25% on either side of the calculation.
6. The Seven Selection Errors That Show Up in Fault Reports
Most bearing failures blamed on material or lubrication trace back to one of seven selection or mounting decisions. The table lists them with the signature symptom and the correction.
| Error | Signature Symptom | Correction |
|---|---|---|
| Ignoring fit-induced clearance loss | Overheating soon after start | Compute post-mount clearance |
| Over-preload by locknut | Hot running, brinelled path | Torque control plus verification |
| Wrong clearance for speed | High-frequency vibration | Shift to C3 for high speed |
| Mixed bearing pairs in one assembly | Asymmetric wear | Use matched sets |
| Preload set cold only | Lockup after warm-up | Account for thermal growth |
| Too much spring preload | Noise, high idle temperature | Resize spring stack for min force |
| No preload verification step | Intermittent positioning drift | Add feeler-gap or torque test |
Two of these errors deserve extra attention. Over-preload by locknut is the classic warranty driver: the assembler tightens until it feels right, the bearing skids and cooks the lubricant, and the failure is misattributed to grease. And preload set cold only is the subtle one, appearing hours after startup as thermal growth increases the squeeze beyond the elastic range of the assembly.
7. Assembly and Verification Checklist
Print this list and keep it at the bearing assembly station.
- Confirm clearance class suits speed and load after mounting, not before.
- Clean and verify shaft and housing fit tolerances with a ring gauge.
- Mount with the correct press ring; never drive through the rolling elements.
- Set locknut preload with an angle reading or torque wrench, not feel.
- Verify preload by starting torque, axial play, or matched-pair gap.
- Measure running temperature at first hour and record the trend.
- Re-check preload after thermal soak; adjust for measured expansion.
8. The Sequence in Practice: A Spindle Retrofit Story
A machine-tool rebuilder in a mid-size shop kept replacing a grinding-spindle bearing every four months. Each replacement used a normal-clearance pair, a k5 shaft, and a locknut tightened by hand. The fault report always read the same: overheating then seizure. The engineer finally measured the post-mount clearance and found it negative; the k5 shaft and the hand-tightened nut together had removed far more clearance than the normal class ever contained.
The corrected configuration used a C3 pair, a j5 shaft to add back margin, and a locknut torqued to a calculated value verified by a starting-friction measurement. The spindle survived its first year and its first thermal cycle without a single temperature alarm. The fix cost only a different clearance class and a documented torque; the years of recurring overheat had been caused by an undocumented assembly habit.
The lesson is that clearance and preload decisions are not datasheet numbers to copy. They are the output of a sequence: choose the operating regime, compute the fit effect, select the method, set the value with a calculation, and verify on the bench. Skipping any link moves the failure from the drawing to the customer.
Rule of thumb that prevents most errors: after-mounting clearance is the real design parameter; select the bearing class, the fits, and the preload together against it, never in isolation.
9. Conclusion
Preload and clearance are two ends of one decision, and neither can be chosen while ignoring the other. Use the decision table to decide whether preload belongs in your application, choose the method that matches your stiffness and thermal needs, compute the fit-induced clearance loss, and always verify the final setting on the assembled unit. Document the torque, the feasibility measurement, and the running temperature so the next engineer does not have to rediscover your mistakes.
Master this one sequence and the most common spindle and bearing failures stop being mysteries. They become preventable, measurable, and explainable, which is exactly what a maintenance-friendly machine should be.
10. Extra Math: Estimating the Fit-Induced Clearance Loss
To predict post-mount clearance you need the interference numbers. When the inner ring is fitted on a solid steel shaft, the radial expansion of the inner ring bore, in micrometres, is about 0.6 to 0.8 times the effective interference. When the outer ring is fitted in a solid housing, the ring compression is about 0.7 to 0.8 times the effective housing interference. The effective interference is the nominal fit value minus the smoothened surface roughness, typically a few micrometres.
Example: a 6205-class bearing, bore 25 mm, with a k5 shaft whose maximum interference is about 13 micrometres and a housing H7 whose maximum interference is zero. The shaft side alone removes about 0.7 x 13, roughly 9 micrometres, from the radial internal clearance. A normal-class 6205 has a radial clearance range of roughly 5-20 micrometres, so the low end of the population is already negative after mounting. That is exactly the case that over-preload looks like, and exactly why the retrofit story in section 8 switched to a C3 class.
| Fit | Typical Max Interference (m) | Clearance Removed (m) |
|---|---|---|
| k5 shaft, 25 mm bore | 13 | About 9 |
| m6 shaft, 25 mm bore | 25 | About 18 |
| N7 housing, 52 mm OD | 24 | About 17 |
| M7 housing, 52 mm OD | 15 | About 11 |
Run this check on paper before ordering the bearing class. It converts a datasheet clearance into a machine-reality clearance, which is the only number the rest of your design team can trust.
11. Glossary
| Term | Meaning |
|---|---|
| Radial internal clearance | Total play between elements and raceways, unmounted |
| Preload | Orchestrated negative clearance applied before service load |
| C2 / CN / C3 / C4 | Tight-to-loose clearance classes |
| Post-mount clearance | Remaining internal play after fits and assembly |
| Matched pair | Factory-selected bearings with matched face grinding |
| Starting friction torque | Torque required to begin rotation, used to infer preload |
With the decision table, the method comparison, the fit math, and the verification checklist in hand, clearance and preload become decisions you can defend in a design review and repeat on the assembly line. That repeatability, not the bearing price, is what keeps the spindle in service.
12. The Final Acceptance Criterion
Before a spindle or gearbox goes out the door, the preload and clearance decision should pass one acceptance gate: run the assembly on a test rig for one hour including a thermal cycle, record the bearing housing temperature trend, and measure vibration at the operating speed. The housing temperature should stabilise rather than climb past about 20 degrees above ambient for a sealed, lightly loaded pair. The vibration signature should contain no skidding spikes at light-load start-up, and the positional indexing repeatability should land inside the tolerance that the preload was sized to deliver.
Document the measured starting torque or axial play alongside the calculated preload so future maintenance can compare the baseline against a rebuilt unit. This single acceptance gate turns a wheel-of-misfortune bearing decision into a controlled, auditable engineering output, and it is the same gate that catches the over-preload, the cold-set-only mistake, and the fit-induced clearance error before they ever meet a paying customer.
