There's a good reason we're rolling on dubs in a bone stock G35. It's part of an overly-complicated attempt at testing different types of limited-slip differentials on the same car, on the same day, to gather a bunch of data we're not quite sure will truly reflect the differences between various types of limited-slip differentials (LSDs). But here goes.
Although we've described the inner workings of various types of LSDs ad nauseum (and occasionally installed and trialed one specific version on a project vehicle), we've never really tried a back-to-back test to figure out how each one performs and feels. Granted, it's a matter of taste and driving style-like drifting, aggressive track lapping or just being stupid on the streets-but, in theory, different diffs are better suited to different situations. The only way to prove this is to install each one and run it on a set course while data-logging wheel speeds, throttle position, steering angle and vehicle yaw-all under the same conditions.
The premise goes like this: get an open diff, an OEM viscous, a popular 1.5-way clutch type and a Torsen gear-type LSD for the same car, using the same set of tires. We've been planning to do this for years, but the monumental task of installing reliable independent rear-wheel speed and steering angle sensors made it a back-burner project.
It wasn't until we stumbled onto UpRev's Cipher data and diagnostic software for the Nissan 350Z, Infiniti G35, and most VQ35DE platforms that it became possible. The software runs on a standard PC and allows the computer to communicate with the vehicle ECU via an USB-to-OBDII dataport cable. Through this, the software can read in real time, most engine and stability control sensor channels at up to 20Hz, or 20 data samples, per second. In addition to recording and saving data to an Excel file, Cipher also allows temporary access to most CAN-BUS channels for anything from turning on wipers and lights to changing idle and injector maps.
We originally intended to use a 2005 Nissan 350Z with its independent rear suspension and single mechanical differential, but ended up using the more advanced G35. This car comes equipped with the Vehicle Dynamic Control (VDC) system which incorporates a steering angle and yaw sensor on top of the standard individual wheel-speed and ABS sensors found on the Z's Traction Control System (TCS.) Using OE sensors saves us from noise, calibration, accuracy or durability issues.
To get better data, we also slapped on a staggered set of 20x9 (front) and 20x10 (rear) Axis MOD7 wheels wrapped with Bridgestone's new top-of-the-line RE050A Pole Position tires. The larger rolling radius of the 20s adds some resolution to our data as a noticeably bigger wheel speed difference (if the outside and inside tire are turning at different rates when the diff isn't locking). The super-sticky Bridgestones also give enough resistance to load up the diff and make them work on a 100-foot radius turn. And they provide the most grip without stepping up to R-compounds and overloading the stock suspension and roll rates.
Off to the track then, with four 75-pound pumpkins (each housing a different LSD) prepped by Steve Mitchell at Nissan, our Racepak G2X telemetry recorder, and a support vehicle full of tools and an understandably disgruntled mechanic. Each differential was subjected to eight laps on our 200-foot figure-eight which simulates a decent mid-speed turn and gives ample opportunity to test turn-in, constant radius balance, and track-out. Along with the Cipher data, we also kept tabs on lap times and consistency. The fasted and average lap times for each diff are also listed. For you statistics buffs, we also calculated one standard deviation for the laps we ran. The consistency of the laps are more important than the fastest or average lap times as track conditions improved as the day progressed. We tested the diffs in the following order: Nismo Clutch type, Quaife, open and viscous.
Nismo 1.5-Way Clutch-Type LSDMost G35 or 350Z owners prefer a clutch-type LSD, on account of the large selection available and the huge hype in drifting. A 1.5-way LSD locks under acceleration and only partially locks off throttle. They are known to clunk and pop as the clutch plates grab and release, which is a drawback for some, as is the relatively abrupt engagement. Depending on how tightly it's packed or preloaded, this type of diff will cause a car to understeer at turn-in, since both wheels are already locked and pushing the car straight instead of letting it rotate. The advantage is that the touchy locking nature makes it very sensitive to throttle modulations-drifters like that.
Our Nismo unit came already broken-in with two of its ten clutch plates disengaged on each axle. This gives it 80-percent of its full locking strength (still tight when compared to our Project Z's diff, which is only at 30 percent). As predicted, it does push the car at turn-in, but adjusts the car's attitude easily with mild throttle inputs mid-turn. Had all ten plates been engaged, the on-power understeer would have been significantly more sever and the data much less like the stock viscous LSD (unfortunately, we forgot to record the vehicle yaw data for this).
Quaife Automatic Torque-Biasing (ATB) LSDQuaife's Torsen- or gear-type differential is usually preferred by most road racers. Without writing a dissertation on how the gears transfer torque, we'll just talk about how the unit performs. Unlike the clutch-type or viscous LSD, the Quaife isn't susceptible to wear on the clutch plates or degradation of the dilatant fluid used in viscous diffs. It should also provide the smoothest transfer of torque to the outside wheel and thus the least amount of understeer. Around the figure-eight, the Quaife-equipped car was by far the lightest on its feet, being easily rotatable, with quick smooth transitions from understeer to oversteer. The automatic torque biasing mechanically distributes power to the outside wheel, effectively making it feel similar to Mitsubishi's Active Yaw Control (AYC) system, which throws torque to the outside wheel to help rotate the car under throttle. Understandably, many racers prefer this type.
Stock Nissan Viscous LSDTo see how these aftermarket LSDs compare to factory offerings, we also collected data on the OE-style viscous diff. A viscous LSD uses the fluid viscosity of a dilatant fluid inside a sealed housing to control output shaft speeds. When the fluid is exposed to shearing forces caused by the two axles spinning at different speeds, it thickens up, forcing both output shafts or axles to partially lock and spin at the same speed. Viscous diffs are the most civil and the softest of the three types tested, since they act more like a torque converter instead of using solid mechanical connections to create lock. In theory, it requires the least amount of maintenance, but when exposed to hard-driving conditions, the dilatant fluid has a tendency to break down and degrade from excessive heat, losing its slip-limiting capabilities. Because of its non-mechanical nature, it's also less effective in transferring power.
Manufacturers choose the viscous diff for good reason. It's the most forgiving of the bunch, allowing time for reactions between over- and understeer. It also takes out the understeer/oversteer balancing act required to get around the figure-eight, making it a good choice for the street or weekend track driver with non-aggressive suspension and tires.
Open DifferentialAs a control, we also installed a pumpkin equipped with an open differential. Although not much slower in a competent driver's hands, the open diff requires huge on/off throttle inputs to keep the car in check. After turn-in, the car would continue to understeer massively through the turn. The only way to tuck in the front end is to lift off the throttle and wait for the weight to transfer to the front tires. This results in lurching between oversteer and understeer without ever achieving a decent neutral set.
One scenario we weren't able to test was how each diff reacted when one drive wheel was completely unloaded, such as when the inside wheel catches air if you round a rumble strip too aggressively. We've heard some types will function like an open diff under these conditions.
The ResultsAs eye-opening as the driving impressions were, the heart of this extravagant experiment is the data we've collected and the depth of detail it shows. For each diff, we show the difference between inside and outside wheel speeds (in mph) for both front and rear wheels. The difference in front wheel speeds around the 200-degree, 100-foot radius turn shows how much faster the outer wheel is spinning. Disregarding slip angle, this is a reference for what the rear-wheel speed difference should look like if there was an open differential installed. The rear wheel speed difference shows how much the wheels are locking. With any LSD, the difference in rear wheel speed should always be less than the front while the car is under power. If it's zero, then the two axles are fully locked. Since there are some erratic changes due to imperfections in the asphalt, we also overlaid a filtered and slightly smoothed representation of the difference in rear wheel speed.
One thing to look for is the smoothness of engagement and disengagement of the diff as it locks when the car enters and exits a turn. We also overlaid the throttle position data on top of the wheel speed data to show the magnitude of throttle modulations required to keep the car in balance. Looking carefully, you can gauge not only how much throttle input is required, but also how long it takes to lock after power is applied.
Two additional graphs show the steering angle and vehicle yaw data for each differential. The steering angle (amount the wheel was turned) illustrates how much the car understeers or oversteers at each stage of the turn. The consistency of the yaw rate throughout the turn also shows how well the car stays balanced.
It's a long, drawn-out geeky rant, but the story is the data.