Torque Vectoring AWD: Active Side-to-Side Power Distribution

Written by Anthony Calhoun, ASE Master Tech A1-A8

Torque Vectoring AWD Explained

AWD systems used to have one job: get torque to all four wheels. That was enough. The problem is that sending equal torque to all four wheels is only slightly smarter than sending it to two. What actually makes a vehicle handle well — whether cornering hard or recovering from a slide — is controlling how much torque goes to each individual wheel, independently, in real time. That is what torque vectoring does.

Torque vectoring has moved from exotic supercar hardware to mainstream performance vehicles and now into EVs at every price point. As a technician, you will see these systems on your lift. When they fail, they fail in specific ways. This article covers how they work mechanically, how the electronics drive them, what breaks, and how to approach diagnosis and repair decisions without guessing.

What Torque Vectoring Actually Is

A standard AWD system controls the split between front and rear axles. A limited-slip differential controls the split between left and right wheels on one axle — but passively, based on speed difference. Torque vectoring goes a step further: it actively and independently controls how much torque goes to each individual wheel, side-to-side, in real time, based on what the vehicle needs at that exact moment.

Think about what happens when you corner. The outside rear wheel needs to travel a longer arc than the inside rear wheel. In a basic open differential, the torque splits equally and the outer wheel speeds up to compensate. In a torque vectoring system, the ECU can deliberately send more torque to the outside rear wheel during cornering. That extra torque on the outside creates a yaw moment — it pushes the rear of the car outward, rotating the vehicle into the turn. The result is a car that steers with the throttle in a controlled, predictable way rather than understeering into a wall.

Going the other direction — reducing torque to an inside wheel or braking it — has the same yaw effect. The direction and magnitude of torque vectoring determines whether the system is helping the vehicle rotate into a corner or stabilizing it against oversteer. A well-calibrated system does both, continuously, in fractions of a second.

How It Works Mechanically

There are three primary mechanical approaches to torque vectoring. Each has different hardware, different performance characteristics, and different failure patterns.

Brake-Based Torque Vectoring

This is the cheapest implementation and the most common on mainstream vehicles. It uses the existing ABS/ESC hydraulic system to apply brake force to individual wheels. When the ECU wants to reduce torque to the inside rear wheel during cornering, it briefly applies the brake on that wheel. The other wheels receive the engine torque by default — slowing one wheel has the effect of transferring drive to the other.

Brake-based systems are real and they work, but they have significant limitations that you need to understand:

• They can only reduce torque, not add it. You can brake a wheel to redirect drive elsewhere, but you cannot independently add torque to a specific wheel beyond what the drivetrain is already producing.
• They generate heat. Every braking event converts kinetic energy into heat at the rotor. Repeated torque vectoring interventions during aggressive cornering or performance driving accelerate rotor and pad wear significantly faster than normal driving would.
• They are slower to respond. Hydraulic brake actuation has more latency than a mechanically engaged clutch pack or a purpose-built electric motor. The difference is measurable in milliseconds, but at high speed and high cornering loads, milliseconds matter.
• They fight the driver. Applying brakes while the driver is accelerating out of a corner creates competing inputs. The system has to balance traction management against the driver's throttle intent, which limits how aggressively the intervention can be.

Brake-based torque vectoring shows up in vehicles marketed with names like "active torque management" or "electronic yaw control" where no dedicated mechanical hardware is present in the drivetrain. It is a software feature layered on top of standard ABS hardware. For light to moderate interventions on everyday driving, it works. For sustained high-performance use, it wears brakes and gets thermally limited.

Mechanical Clutch-Pack Torque Vectoring

True mechanical torque vectoring uses a rear drive unit or differential housing that contains clutch packs — multi-plate wet clutches — on each side of the rear axle. These clutches are actuated independently by the control module, allowing precise, continuous control of torque to each rear wheel.

The mechanical advantage of a clutch-pack system over brake-based vectoring is that it can both add and redirect torque without generating brake heat. It works by overdriving or underdriving one side of the differential output. The clutch packs act as variable coupling devices between the ring gear and each axle shaft. When the ECU applies pressure to the left clutch pack, it couples the left axle shaft more tightly to the ring gear, increasing the torque delivered to that wheel. The right side gets less. The variation can be subtle — a 55/45 split — or dramatic depending on the driving scenario.

Some designs use planetary gear sets to create a mechanical speed advantage between the input and the individual output shafts. By engaging different clutch packs against different planets in the gear set, the system can actually overdrive one wheel — spinning it slightly faster than the differential would naturally drive it — which produces a torque increase at that wheel. This is a more sophisticated implementation than a simple clutch coupling and allows more precise control.

Electric Motor Torque Vectoring

In EVs and hybrids with individual wheel motors or dual-motor axle setups, torque vectoring is software-defined. Each motor can be commanded to produce a specific torque output in milliseconds. There are no clutch packs to wear out and no hydraulics to actuate. The control system reads yaw rate, steering angle, lateral G, and wheel speeds and adjusts motor output continuously.

Some performance hybrids use a dedicated electric motor on the rear axle specifically for torque vectoring. The combustion engine drives the front wheels while the electric motor handles the rear, and because the electric motor can vary torque to each rear wheel independently (through the motor control unit), the system achieves active yaw control without any mechanical coupling hardware. Rivian's dual-motor and quad-motor setups and the Polestar 3 with Performance Pack work this way. So does the Toyota GR Yaris with its GR-Four AWD, which uses a dedicated high-voltage rear motor unit for torque distribution.

Systems by Manufacturer

Ford Focus RS — Rear Drive Unit (RDU)

The Focus RS uses a GKN-supplied rear drive unit that is the textbook example of mechanical torque vectoring done right. The RDU contains two independent wet clutch packs — one for the left rear wheel and one for the right rear. Each pack is actuated by an electric motor driving a ball-ramp mechanism. The result is fully independent, continuously variable torque delivery to each rear wheel.

In normal driving, the RDU sends most torque to the front axle. Under acceleration out of a corner, the rear clutch packs engage to vector torque to the outside rear wheel, creating the controlled rotation that makes the Focus RS so well-regarded for its handling. The system also enables the infamous Drift Mode by deliberately sending torque to both rear wheels simultaneously and allowing controlled oversteer. The RDU requires its own fluid — Ford specifies a dedicated synthetic fluid — and failure to use correct fluid or service at proper intervals destroys the clutch packs. This is not a suggestion. Wrong fluid has been documented as a direct cause of RDU failure in real-world warranty and repair cases.

Acura SH-AWD — Super Handling AWD

Honda's SH-AWD system uses a rear differential with electromagnetic clutch packs on each side. The system can send up to 70 percent of total drivetrain torque to the rear axle and then vector that torque left or right within the rear axle independently. It does this through electromagnetic clutch packs that vary their engagement based on ECU commands derived from steering angle, yaw rate, lateral G, throttle position, and wheel speed inputs.

The current generation SH-AWD (used in the MDX, RDX, TLX) adds an electric motor on the rear axle in Type S applications, which improves response speed and allows more aggressive vectoring during transient maneuvers. The system fluid is Honda-specific. Using a generic gear oil or generic ATF instead of the specified SH-AWD fluid causes accelerated clutch pack wear. Expect to see SH-AWD failures from improper fluid serviceability in shops that do not stock the correct product.

BMW xDrive with Torque Vectoring

BMW's xDrive is a front-rear AWD system that pairs with rear axle torque vectoring through the rear differential on performance models. The M5, M6, M8, and X5 M/X6 M use an active M differential that can bias torque between the rear wheels using independently controlled clutch packs integrated into the rear differential housing. The system reads the same sensor array — yaw, lateral G, steering angle, throttle — and can apply a locking force on either side of the rear diff.

BMW also applies selective braking via DSC (Dynamic Stability Control) on non-M xDrive models to simulate torque vectoring when the mechanical active differential is not present. On full M models with the active M differential, the clutch pack hardware is present and requires attention. BMW specifies differential fluid for these units separately from the standard rear differential fluid used on open differentials.

Audi Sport Differential

Audi's Sport differential is fitted to RS models and some S models with quattro AWD. It sits at the rear of the drivetrain and uses a planetary gear set with clutch packs on each side to overdrive the outside rear wheel during cornering. The key difference from a simple clutch coupling is the planetary element — it allows the system to actually spin the outside wheel faster than the drive input, creating a genuine torque addition rather than just a redistribution. This is mechanically more complex and produces a stronger yaw moment for a given amount of clutch engagement. The Sport differential fluid is separate from standard Audi quattro rear differential fluid and must not be substituted.

Electronic Controls — How the ECU Decides What to Do

Every torque vectoring system is only as good as the sensor inputs feeding it. The control module reads a stack of data simultaneously and computes a torque distribution command many times per second. The primary inputs are:

• Steering angle sensor: Tells the module what direction the driver intends to go and how sharply.
• Yaw rate sensor: Measures the actual rotation rate of the vehicle about its vertical axis in degrees per second. This is compared against what the steering angle predicts — discrepancy between intended yaw and actual yaw triggers intervention.
• Lateral accelerometer: Measures side-to-side G force. Combined with yaw rate and speed, this confirms the vehicle's actual cornering state.
• Throttle position / accelerator pedal position: The module factors in driver demand — aggressive acceleration out of a corner requires more vectoring to prevent understeer or oversteer depending on drivetrain layout.
• Individual wheel speed sensors: All four wheel speeds are monitored continuously. Speed differences between wheels indicate slip or cornering geometry and inform the torque distribution algorithm.
• Vehicle speed: Vectoring behavior at 15 mph is different from behavior at 75 mph. Speed scaling is built into the control map.

The module compares actual vehicle behavior against a reference model — a mathematical prediction of how the vehicle should be behaving given the driver's inputs and vehicle speed. When actual behavior deviates from the reference model, the system intervenes to correct it. This is the same fundamental logic as electronic stability control, but torque vectoring acts through drivetrain output rather than individual brake application.

Faulty sensor inputs corrupt this entire process. A wheel speed sensor sending erratic data will cause the module to see a false slip event and command torque vectoring when none is needed — or fail to command it when it is needed. A yaw rate sensor that has drifted out of calibration will generate a persistent mismatch between intended and actual yaw, causing unnecessary interventions or storing fault codes without obvious symptoms. This is why sensor integrity is the starting point of torque vectoring diagnosis.

Common Failure Modes

Clutch Pack Wear

Mechanical torque vectoring units use wet multi-plate clutch packs that wear over time. Wear rate accelerates dramatically with improper fluid, overheating from performance driving without adequate cooling, and extended service intervals. Worn clutch packs produce symptoms ranging from reduced vectoring effectiveness — the vehicle understeers where it used to rotate — to shudder, vibration on engagement, and eventually complete loss of torque distribution capability. By the time a driver notices handling degradation, the clutch packs are often significantly worn.

Fluid Contamination and Wrong Fluid

This is the most preventable and most common cause of premature torque vectoring unit failure. The friction materials in these clutch packs are engineered to work with specific fluid chemistry. That chemistry controls the coefficient of friction, the heat dissipation rate, and the clutch engagement feel. Using a standard gear oil, generic ATF, or even a correct-specification fluid from a different manufacturer in place of the OEM-specified product causes the friction surfaces to either grab incorrectly, slip excessively, or glaze over time.

Water contamination from a failed seal is equally destructive. A leaking axle seal or differential cover gasket that allows water ingestion will compromise the fluid and attack the clutch packs within a short period of operation. Inspect fluid condition during any drivetrain service on a torque vectoring-equipped vehicle. Dark, metallic, or water-contaminated fluid is a diagnostic finding, not just a maintenance note.

Control Module Failures

The torque vectoring control module — which may be integrated into the AWD control module, the ABS/ESC module, or a standalone unit depending on the application — can fail due to water intrusion, connector corrosion, or internal component failure. Module failures typically present as stored fault codes with no mechanical symptoms (the module is not commanding anything, so nothing is happening), or as erratic behavior where the system commands incorrect torque distribution at random.

Wheel Speed Sensor Inputs Causing Incorrect Torque Distribution

A faulty wheel speed sensor does not just affect ABS. On a torque vectoring system, it feeds bad data directly into the torque distribution algorithm. A sensor with an air gap issue, a damaged tone ring, or internal failure that produces an erratic or dropout-prone signal will cause the module to see a wheel spinning at an incorrect speed. The module will interpret this as slip and command vectoring torque to compensate — vectoring that is unnecessary and may actually destabilize the vehicle. In some cases, this generates no DTC because the wheel speed discrepancy is within a threshold that looks like normal cornering geometry to the module. Start with wheel speed sensor data in every torque vectoring diagnostic.

Service Requirements

Fluid service on torque vectoring units is not optional and is not interchangeable with standard differential service. Follow these principles:

• Use only the specified fluid. Every OEM specifies a fluid for their torque vectoring differential or rear drive unit. Ford Focus RS RDU uses a specific synthetic fluid — not standard gear oil. Acura SH-AWD has its own fluid. Audi Sport differential fluid is not the same as standard Audi rear diff fluid. BMW active M differential fluid is specified separately. Refer to the service information every time. Do not substitute.
• Service intervals are shorter than most technicians expect. Most torque vectoring units specify fluid changes every 30,000 to 60,000 miles depending on manufacturer and driving conditions. Severe use — towing, track driving, repeated aggressive cornering — shortens this interval. Customers who drive these vehicles hard need shorter intervals regardless of the maintenance schedule.
• Inspect fluid condition at every opportunity. When a torque vectoring-equipped vehicle comes in for any drivetrain service, pull a fluid sample if possible. Look for metallic particles (clutch pack wear), darkening (thermal degradation), and water separation or milky appearance (seal failure). These findings justify additional service or further diagnosis before a unit fails completely.
• Check for seal leaks during every inspection. Axle seals, fill plug seals, and cover gaskets on torque vectoring units are the first line of defense against contamination. A slow seep that looks minor can allow water ingestion during rain or car washing. Flag it and fix it.

Diagnostic Approach

Start with Scan Tool Data

A capable scan tool with manufacturer-level access is mandatory for torque vectoring diagnosis. Generic OBD-II will not give you what you need. You are looking for:

• Stored and pending DTCs in the AWD control module, ABS/ESC module, and any standalone torque vectoring module
• Live data showing torque distribution commands — what the module is commanding versus what is actually happening
• Individual wheel speed sensor live data — all four wheels simultaneously, at low speed in a straight line and through a slow turn
• Yaw rate sensor output — should read near zero on a straight road, should change proportionally in a turn
• Lateral G sensor output — same validation logic
• Steering angle sensor output — verify it reads zero when the wheel is centered, verify it tracks smoothly through full lock-to-lock

Compare all four wheel speeds on a straight, level road at a steady speed. All four should read within 1-2 mph of each other. Any significant discrepancy indicates a sensor, tone ring, or calibration issue. A wheel that consistently reads high or low is the diagnostic starting point.

Verify Communication

Torque vectoring modules communicate over the vehicle network — typically CAN. A module that has lost communication with the ABS/ESC module, the powertrain control module, or the instrument cluster will default to a safe state (usually disabling active vectoring) and store a communication DTC. Network faults that look like a drivetrain problem are common. Pull every module scan, not just the powertrain codes, before touching mechanical components.

Functional Testing

On supported platforms, use bi-directional controls to command the torque vectoring unit directly. If you can command clutch engagement and measure a response at the wheels (via wheel speed or torque sensors in the live data), the mechanical system is functional and the problem lies in the control logic or sensor inputs. If you command engagement and nothing responds, the fault is either in the module output circuit, the actuator itself, or the mechanical unit.

Repair vs. Replace Decisions

Torque vectoring differentials and rear drive units are expensive. A Focus RS RDU replacement runs $2,000 to $3,500 in parts alone depending on source. An Acura SH-AWD rear differential is in a similar range. Audi Sport differentials are $3,000 or more. Before recommending replacement, consider the following:

• Is the failure mechanical or electronic? A failed control module or damaged wiring harness is a fraction of the cost of a mechanical unit. Confirm the mechanical hardware is at fault before quoting a differential replacement.
• Remanufactured units exist for high-volume applications. Focus RS RDUs, Haldex units, and some SH-AWD rear differentials have remanufactured options at significant savings over OEM new. Verify the remanufacturer warrants the unit and that it comes with fresh fluid and any required seals.
• Fluid service first on ambiguous cases. If the complaint is reduced vectoring performance without hard mechanical noise or complete failure, perform a proper fluid service before recommending unit replacement. Degraded fluid causes clutch pack glazing that sometimes recovers partially after a fluid change. This is not a guarantee, but on a $3,000 repair, it is worth the $150 fluid service first.
• Clutch pack replacement is possible on some units in the right hands. Some torque vectoring differentials can be rebuilt with new clutch packs by shops with the tooling and training. This is not a general repair shop procedure, but specialty drivetrain shops do this work successfully. It may be worth a referral rather than a full unit replacement if the customer wants to save money and the rest of the unit is serviceable.

The Shift to Electric Motor Torque Vectoring

The trend is clear. As EVs and hybrids proliferate, mechanical torque vectoring with clutch packs is being replaced by software-defined motor torque control. Individual wheel motors eliminate the need for differentials entirely in some configurations. In others, a dual-motor setup on one axle — two motors sharing an axle — allows independent left-right torque control by simply commanding different output levels from each motor.

Electric torque vectoring is faster, more precise, and requires no fluid service. It does introduce new diagnostic considerations: motor controller failures, high-voltage wiring faults, resolver or encoder issues affecting motor position sensing, and software calibration requirements. The diagnostic approach shifts from mechanical inspection and fluid analysis toward scan tool data review, high-voltage system safety procedures, and software updates.

Vehicles already on the road with electric torque vectoring include the Rivian R1T and R1S (quad-motor), Polestar 3 with Performance Pack, Tesla Model S Plaid (three motors), BMW iX M60, Porsche Taycan (dual rear motors with torque vectoring), and Audi e-tron GT. More are coming every model year. A technician who understands the underlying control logic — sensor inputs, reference model comparison, and output commands — can adapt to these systems even as the hardware changes, because the logic stays the same.

The fundamentals do not change: the system reads where the vehicle is, compares it to where it should be based on driver inputs, and adjusts torque delivery to close the gap. Whether it does that with clutch packs or motor controllers is a hardware detail. Understanding what the system is trying to accomplish is how you diagnose it when it stops working.