Preload, ramp angles & friction faces: drive vs coast locking
Locking torque = preload + (friction faces × μ × radius / ramp geometry) × input torque. Lower ramp angles and more faces lock harder; drive and coast are set separately. A worked example puts real numbers on torque bias.
Preload, ramp angles & friction faces: drive vs coast locking: L1 Foundation explanation
Three hardware parameters define a plate/ramp LSD:
- Preload — a spring stack (Belleville washers) pre-clamps the plates. It gives a constant locking torque, present even at zero input torque (mid-corner, light throttle). Measured as breakaway torque: car on stands, one driven wheel held, torque to rotate the other. Typical measured race values ≈ 30–200 Nm depending on car and preference. Convention — read this twice. With the gearbox in neutral (case free) and one wheel held, the case turns at half the wheel speed and both clutch packs slip at half speed, so the reading equals one pack's friction torque. On track both packs work, and the left–right torque difference the diff can sustain at zero input torque is ≈ 2 × that breakaway reading. In this course always means that difference capacity (bias preload = 2 × neutral breakaway). Suppliers and teams use different conventions (in gear with the case held, the reading is already the full difference): always write the test method on the setup sheet.
- Ramp angles — the input cross-pins sit in V-shaped ramps. Under torque the pins push the pressure rings apart; the shallower the ramp (smaller angle, by the common convention), the larger the axial force for a given torque → more locking. Each side of the V acts in one direction: the drive face under power, the coast face under engine braking. Typical options: 30°, 45°, 60°, 80° (85° ≈ almost no torque-sensitive locking).
- Number of friction faces — locking capacity scales with how many plate surfaces share the clamp. More faces = more locking for the same ramps and more preload torque for the same spring clamp. Typical 4–12 faces in total (2–6 per side).
Simplified locking law:
with friction faces per side (per clutch pack; if a supplier quotes the total count, use half of it), plate friction coefficient (≈ 0.08–0.15 in oil), mean plate radius , ramp radius and ramp angle .
How locking is often described:
- Locking % = (ignoring preload): e.g. 35 % in drive, 20 % in coast.
- Torque bias ratio : torque the slower wheel can get relative to the faster one when the diff is slipping. k = 0.35 → TBR ≈ 2.1.
Behaviour summary:
| Change | Entry (coast) | Mid-corner | Exit (drive) |
|---|---|---|---|
| + preload | more stable, less turn-in | less rotation, more understeer | earlier lock, better traction at low torque |
| Lower coast angle | more stable on lift/braking | – | – |
| Lower drive angle | – | – | more traction, more power understeer → possible snap |
| More friction faces | more coast lock | more (preload torque scales with faces unless re-shimmed) — re-measure breakaway after any plate change | more drive lock |
Preload, ramp angles & friction faces: drive vs coast locking: L2 Practitioner explanation
Worked example (illustrative numbers). Plate diff with = 4 faces per side (8 in total), μ = 0.12, = 0.8:
Bias preload = 120 Nm (i.e. ≈ 60 Nm breakaway measured in neutral with one wheel held).
Exit, 2nd gear, diff input torque 3000 Nm: Nm. If the diff slips, the slower wheel can receive up to 1270 Nm more than the faster one.
Suppose the inside rear can only take 800 Nm before spinning (light, laterally loaded). With an open diff the axle gets 2 × 800 = 1600 Nm. With this LSD the inside wheel spins slightly, becomes the faster wheel, and the outside receives 800 + T_lock:
→ +73 % usable drive from the same tyres, with the extra on the outside wheel (pro-rotation yaw moment).
Entry, engine braking −600 Nm on coast ramp 60°: Nm. Preload is half the total. Change preload from 120 to 60 Nm and the coast locking drops by a quarter; change the coast ramp from 60° to 45° and it rises from 250 to ≈ 350 Nm.
Mid-corner, neutral throttle (≈ 200 Nm): Nm — preload dominates. This is why preload is the mid-corner rotation lever.
Trackside workflow:
- Measure breakaway torque cold, with the documented method (neutral or in gear, wheel, direction, speed of rotation, oil temperature) — record it on the setup sheet like spring rates.
- Adjust preload via spacer/spring stack; ramps via cross-pin housing / ramp rings; faces via plate stack (friction plates ↔ plain spacers). Most require diff removal: plan changes for the overnight work list.
- Re-measure after the first run: preload typically drops as plates bed in; then track it over mileage.
Preload, ramp angles & friction faces: drive vs coast locking: L3 Advanced explanation
Derivation of the locking coefficient. The diff case drives the cross-pins with total tangential force at the ramp radius, shared between the two pressure rings (each ring passes half the input torque to its side). A ramp of angle (measured so that smaller θ = larger axial component) converts each ring's share into an axial clamp on its own clutch pack. With friction faces per pack, each pack's friction torque is . When the wheels slip relative to each other, both packs slip at ; the power balance gives a left–right torque difference of :
The same power balance explains the breakaway convention of L1: in neutral with one wheel held, the measured torque , so — half of the difference capacity.
Manufacturers' exact geometry factors (whether faces on both sides share the full , pin friction, side-gear thrust) differ; use their locking tables for real numbers and this law for trends and sensitivity. Note the strong non-linearity in θ: going 45° → 30° raises by 73 %, 45° → 60° lowers it by 42 %, while an 80° ramp gives only ≈ 18 % of the 45° value.
Coast is not just "less of drive". On coast, input torque includes engine braking (gear-multiplied, so far higher in low gears), plus any downshift blip errors — an insufficient blip creates a large negative torque spike that the coast ramp turns into a lock spike exactly at turn-in. Stability complaints at specific corners after downshifts are often a blip/coast-ramp interaction.
Preload vs ramp choices — the non-linear trade-off. Preload is a torque-independent offset: relatively important at low torque (high gears, low-grip conditions, wet), relatively unimportant at high torque. Ramps scale with torque: the same ramp set behaves "more locked" in 2nd than in 5th gear in absolute Nm because wheel torque is larger. Consequence: a setup that is right in slow corners can be too free in fast ones and vice versa. Engineers therefore pick the ramp for the slow, traction-critical corners and use preload (plus aero/mechanical balance) to correct fast-corner behaviour, or vice versa depending on where the lap time is.
Temperature and wear. Plate μ varies with oil temperature and condition (≈ ±20 %), so drifts within a stint. Plate wear reduces spring stack compression → preload drops (commonly 10–30 % over a race weekend before re-shimming). A car that was "perfect on Friday" and "lazy on the exits on Sunday" may simply have a worn diff: check breakaway torque before chasing setup.
Preload, ramp angles & friction faces: drive vs coast locking: L4 Expert explanation
Limits of the linear law. is a quasi-static idealisation. A simulation or diff-design engineer must know where it fails:
- Static vs dynamic friction. Plates in oil have , and varies with sliding speed. A stuck pack breaks away at a higher torque than it then sustains. The torque drops at breakaway, which can trigger stick-slip (pit-lane chatter, judder in hairpins) and, on track, a small step in yaw moment at the instant the axle unlocks. LSD oils carry friction modifiers to flatten the curve; the wrong oil or degraded oil gives judder and a sharper unlock.
- Ramp (pin) friction and hysteresis. The cross-pin slides on the ramp with friction angle . While torque rises (pin climbing) the clamp is ; while it falls, friction holds the rings apart, up to . The locking curve is therefore a loop: locking lingers briefly after a lift. Small ramp angles amplify this, and a ramp with would self-lock.
- Bevel-gear separating forces. The side gears also receive an axial thrust from the bevel mesh that is proportional to the torque they carry. That adds torque-sensitive locking even with an 85–90° ramp, and it is why a nominally open diff shows a TBR of 1.1–1.3.
- Spring position and non-linearity. Belleville stacks have a regressive load–deflection curve. Whether plate wear (tenths of a mm of stack height) costs 5 % or 30 % of preload depends on where on that curve the stack is shimmed. Depending on where the spring sits in the load path, the preload either adds to the ramp clamp or is progressively overridden by it. Supplier locking curves often show a knee rather than a straight line.
- Drive/coast crossover. Each time passes through zero (lift-off, throttle pick-up, a gearshift), the pins cross the backlash between the drive and coast faces. For a few ms the locking falls to preload only, then the other ramp engages. That is a transient yaw-moment dip that a quasi-static model never shows.
Modelling consequence. For transient or driver-in-the-loop simulation, model the diff as a look-up with hysteresis, a static/dynamic friction pair and a three-state logic (slipping one way, stuck, slipping the other way). Identify it on a rig where input torque and slip speed are both controlled. Then validate it against measured rear wheel-speed differences, not only lap time. A diff that matches lap time but not the exit traces is matching for the wrong reason.
Key formulas
- bias preload = left–right torque difference at zero input (≈ 2 × breakaway measured in neutral, one wheel held) (Nm)
- k
- torque-sensitive locking coefficient (–)
- diff input torque (Nm)
- n
- number of friction faces per side (per clutch pack)
- plate friction coefficient (≈0.08–0.15)
- mean friction radius (m)
- ramp radius (m)
- ramp angle (smaller = more locking)
Trackside takeaways
Locking = preload + k × input torque; k ∝ faces × μ / tan(ramp angle).
Lower ramp angle and more friction faces = more locking; drive and coast faces are separate.
Preload dominates at low torque → it is the mid-corner rotation tool.
Worked example: k = 0.38 + 120 Nm preload turned 1600 Nm (open) into ≈ 2770 Nm usable drive.
Engine braking and blip errors pass through the coast ramp — entry stability depends on both.
Plates bed in and wear: measure breakaway torque routinely and record it on the setup sheet.
On your car
Preload is the usual adjustable (via spring/shim stack); ramp sets and plate count may be fixed by homologation or offered as a few options. Changes typically require diff removal — plan them for overnight work.
Check yourself
Changing the drive ramp from 45° to 30° (same plates) changes the torque-sensitive locking coefficient by roughly:
- Milliken & Milliken – Race Car Vehicle Dynamics — Ch. 20 Driving and Braking · limited-slip mechanisms and torque bias
- Other — Manufacturer LSD service documentation (e.g. Drexler, Hewland, Xtrac) · ramp-angle and preload-measurement conventions vary by supplier – check the manual for your unit
- Segers – Analysis Techniques for Racecar Data Acquisition — §4.2 Traction and Longitudinal Slip · evaluating differential work with left/right slip channels