Last year in my more than you ever wanted to know about AWD systems article, I described the function of several types of passive all-wheel-drive arrangements. One of the most popular passive center differentials is the planetary differential. A planetary differential offers some advantages over the traditional bevel-gear differential, which is often used as the front or rear differential of a FWD, RWD, or AWD car — although some cars such as the DSM or manual transmission WRX use a bevel gear center diff.
Unlike a bevel gear differential which necessitates an equal torque split on both outputs, a planetary differential can be designed for completely symmetric through radically asymmetric output according to what the application requires. The 3000GT and Stealth AWD system has such a planetary differential, and I have an animation to help visualize it:

The major components of the differential are the ring gear (the outside circle), the sun gear (the inside circle), the planet gears, and the planet carrier (the middle ring which holds the planet gears). On a 3/S, the ring gear is the input, the sun gear is the front output (via a shaft to the front differential), and the planet carrier is the rear output (to the transfer case and rear wheels). This differential is made visually complex by the arrangement of the planet gears, which are necessary to make both outputs rotate in the same direction as the input.
Confused yet? Don’t feel bad — I find it hard to visualize without help. A while ago I set out to describe how the differential works using only a pen and paper. In researching the problem, my main resorce was Jeff Lucius’ excellent, comprehensive Illustrated Guide to the Mitsubishi 3000GT AWD system, without which I probably never would have asked many of the questions that prompted me to dig deeper. On this page, he dismantles the factory technical literature before coming to the following conclusion:
“…the claimed 45/55 front rear torque split for the AWD 3000GT/Stealth only applies when the VCU is not limiting differential rotation rates and 1) the transfer case is removed so that center output shaft can rotate at a different rate than the front output shaft, 2) the front wheels are slipping and so rotate slightly faster than the rear axles, or 3) the car is just beginning to move (analogous to either the carrier or sun gear being locked). Once the car is moving and there is the same traction at all wheels, torque is split evenly (50/50) between transfer case and front differential. “
In essence, he says, the differential only produces the manufacturer’s claimed 45/55 front-rear torque split when the wheels are not turning at all, or when one front or rear pair of wheels is spinning faster than the other. I could not reconcile this with how I felt the differential should operate. After a lot of head-scratching and finding answers I knew were wrong, I fell back on some 2D physics simulation software, using it to model the differential and simulate its behavior in various states.
In the first experiment the sun gear was set to resist less than the planet carrier while an input torque of 100ft-lb was applied to the ring gear. This is equivalent to a loss of front traction:
Simulation 1
The sun gear accelerates much faster than the planet carrier, but the output torque to each stays exactly the same throughout — 45ft-lb to the sun gear and 55ft-lb to the planet carrier.
In the second experiment the planet carrier was set to resist less than the sun gear while an input torque of 100ft-lb was applied to the ring gear. This is equivalent to a loss of rear traction:
Simulation 2
The planet carrier accelerates much faster than the sun gear, but again the output torque to each stays exactly the same at a 45:55 ratio.
In the final experiment I ran the simulation with both outputs resisting proportionally to the torque applied to them, so that they would be guaranteed to accelerate at the same rate. According to Jeff’s calculations, the differential should be operating at a 50:50 torque split whenever the outputs are rotating at the same speed. His logic, or so it follows, is that unequal torques must generate unequal wheel speeds. Specifically, he cites the conservation of angular momentum. The reason for this conclusion is simple: the model he uses does not consider both outputs as connected to something which forces them to turn at the same speed (the tires and road). This is not so unreasonable, considering how hard I found it to model the differential at all, but without this constraint one is forced to conclude that each output turning at the same speed must be the result of equal torque to both sides.
Simulation 3
In this simulation, the sun gear and planet carrier accelerate at the same rate from a stop, and the torque “felt” at each output is 45lb-ft out the sun gear and 55lb-ft out the planet carrier, corresponding to a 45:55 ratio.
In conclusion, the 45:55 front:rear torque split does occur when all wheels are turning at the same rate — in fact this is especially the case in the actual car because the viscous coupling operates on speed differential, resisting more as the relative speed between the two outputs becomes higher. Note that all of the models here are without any viscous coupling; only the differential itself is modeled.
Finally, I would like to give much thanks to Jeff Lucius and his beyond-comprehensive site that has made so much knowledge accessable to so many people. Its relevence extends well beyond the 3/S community, and I hope this information will be preserved long after people have stopped working on these impressive cars.
Technorati : 3000GT, AWD, Stealth, all wheel drive, center differential, planetary differential
Del.icio.us : 3000GT, AWD, Stealth, all wheel drive, center differential, planetary differential
Ice Rocket : 3000GT, AWD, Stealth, all wheel drive, center differential, planetary differential
Flickr : 3000GT, AWD, Stealth, all wheel drive, center differential, planetary differential
Zooomr : 3000GT, AWD, Stealth, all wheel drive, center differential, planetary differential
Buzznet : 3000GT, AWD, Stealth, all wheel drive, center differential, planetary differential
Riya : 3000GT, AWD, Stealth, all wheel drive, center differential, planetary differential
43 Things : 3000GT, AWD, Stealth, all wheel drive, center differential, planetary differential
on May 24, 2010 at 10:39 am Richard Romano wrote:
It is easier to think about this as a open differential with a fixed gear ratio down stream of one of the differentials.