Differentials: Why They Matter in Rally.

By Andrew Comrie-Picard for Inside Track

 

If you’re a keen observer, you might have heard comments by World Rally Championship drivers about their differential settings – tuning them, reprogramming them, and so on. Why are they so concerned about the diffs? The short answer is that more than any other single aspect of the car, they can make it handle beautifully or make it the worst terror on the road.

 

For quick review: a differential is necessary in a car because the outside and inside wheels go different distances when you turn a corner – the outside wheel has to go further. In a rear-drive car, if the rear axle were a solid connection between the two wheels, the car would be very hard to turn because one of the rear wheels would have to break traction in the corner. In order to allow the car to turn more easily – and not stress the tires and driveline – the wheels are allowed to move at different speeds by means of a set of bevel gears between the rear half-shafts. You can see the way they work if you jack up a typical car and turn one drive wheel in the air; chances are the wheel on the other side will turn in the opposite direction.

 

Now, the problem with having a set of bevel gears between the half-shafts is that when you go around a corner, the outside wheel has a lot of weight on it but the inside wheel does not, and if you punch the throttle you tend to spin the inside wheel and lose acceleration force. If this is happening on the front wheels, it affects the car’s response to steering input as well. To prevent this, a number of designs have evolved, and so we have the Limited Slip Differential (LSD).

 

Any LSD prevents slippage in one of four ways. Either:

(A) There is a set of squeezed plates in the differential working like a clutch on a manual transmission that slip when necessary, but resist total freewheeling (a “clutch-pack LSD”).

(B) There is an intelligent arrangement of worm gears that allows free differential action when no torque is applied to the input shaft but locks up when there is torque applied (hence TORque SENsing or “TorSen”).

(C) There is a set of non-squeezed perforated plates in the differential running in a bath of special fluid that thickens when the particle structure of the fluid is sheared (or, in rare systems, when an electric current is passed through the fluid). It sort-of “glues” the plates temporarily together, so that when the diff begins to slip, it tightens up (a “viscous coupling”).

Or (D), A car can actually have a fully open differential, but use electronic control of the brakes to simulate a limited-slip character to the diffs, as on the BMW X5.

 

There are street cars and rally cars on the road right now with all of these systems, and often with two or even three of them in the same car.

 

Three? Yes – an all-wheel-drive car needs differentials not only on the rear axle, but also on the front axle for the same reason as the rear. Also, there’s one in the middle of the car between the two axles for a similar reason: the front of the car actually travels a different distance than the rear of the car through a corner (under normal circumstances the front travels further). So we have to worry about three differentials and their behaviour.

 

How do the individual diffs affect the behaviour of a rally car? Without adding acceleration or braking to the mix (yet), a car with a tight rear diff will oversteer because the rear wheels will try to move at different speeds and so will break free and come around, a car with a tight front diff will understeer since the front wheels will break free for the same reason, and a car with a tight centre diff will generally understeer because the front wheels will tend to break free relative to the rear ones.

 

But here’s the rub: for maximum acceleration and steering predictability under power while sliding on gravel, you want the diffs to be tight when you’re on the throttle, but for best turn-in and predictability off the power, you want the diffs mostly open under braking. So we’re at the tricky part: you need the differentials to behave differently under braking than under acceleration. You also want them to tighten and loosen smoothly and predictably.

 

Now consider what I do when I’m left-foot braking through a corner: I come up to the corner, lift the throttle and press on the brakes to put weight on the front of the car and induce oversteer, then once sliding get on the throttle to continue the slide, then maybe to tighten the line apply a little brake to put weight to the front, then to straighten out the car and open the line I will accelerate away with throttle. That was a relatively simple turn, but it was turn, brake, accelerate, brake, accelerate just for a single constant-radius corner. The differentials had to start slipping to allow me to turn in, then tighten to allow me to put the power down predictably, then loosen to allow the brakes and the corner adjustment to work, then tighten again to allow acceleration and weight transfer to the rear to straighten out and accelerate away.

 

So how can diffs tighten and loosen? Well, some can and some can’t, and those that can do it in different ways. A viscous coupling of the “shear” type can only sense when the wheels are turning at different speeds and so cannot sense braking or acceleration. As a result, it is not a very predictable differential to use in a rally car, although the Subaru WRX (non-STi), the Mitsubishi Evolution (up to the Evo VI), and the Eagle Talon all have viscous centre differentials from the factory. Essentially, the diff will allow the front and rear wheels to move at slightly different speeds without resistance, but the moment there is a big speed difference the front and rear wheels will lock to each other. This is fine for an average driver in slippery conditions on the way to the ski hill, but not so good for competition. (Incidentally, for the hooligans amongst us, this is why these cars won’t do very good handbrake turns!)

 

A TorSen gear differential is very effective at locking under torque and opening under braking, and the actual moment of transition from locking to unlocking is fairly smooth and predictable. However, it has a flaw in that if one wheel (or end of the car) has no traction at all, it works like an open differential and all advantage is lost. The classic Audi Quattro had a TorSen center diff. We don’t see them in many cars today, as they are more expensive to produce for a street car than a viscous coupling, and they are not as sophisticated in their reaction as modern clutch-pack LSD for competition.

 

The really sophisticated modern rally diffs are all plated LSDs with clutch packs that tighten or loosen depending on either torque applied to the input shaft or on an external control. The “torque-reactive” version works on an eccentric cam that presses the plates together under torque, and it’s what I have in the front and center KAAZ differentials on my Mitsubishi Evo rally car. It works pretty well – the system can essentially lock all four wheels together when I’m on the throttle, and release them all from one another when I’m on the brakes. It’s not a perfect system, but it’s pretty good.

 

The perfect (or nearly perfect) system is the clutch-plate LSD with an external control not based only on input torque. That control works exactly like a manual transmission clutch, but the “leg” operating that clutch can be electro-magnetic or electro-hydraulic, and the current or fluid control can be by driver input or by computer. If it’s controlled by computer, it can integrate all kinds of dynamic data such as throttle position, brake position, individual wheel speeds, and lateral and linear acceleration. The most sophisticated center diffs have independent computers collecting all this data and controlling torque distribution, as in the Subaru WRX STi and the Mitsubishi Lancer Evolution VII. This is the now-famous “active differential,” and a car equipped with three active differentials working together can do an amazing thing: it can distribute the optimum amount of torque to each wheel to maximize overall acceleration, it can constantly change this distribution as the traction conditions of each wheel change, and it can maintain sharp response to braking and steering whether on or off the throttle. This is much more sophisticated than just “tightening” and “loosening,” although that is what is constantly happening to each of the three diffs in concert. And this is where we return to Gronholm or Burns talking about their diff settings.

 

Given the different surfaces on the various World Championship events and the different styles of the drivers who compete on it, the programming of the differential maps is a major activity for the teams. Monte Carlo’s mix of tarmac, snow, and ice requires a completely different differential setup from New Zealand’s fast and flowing gravel, and Cyprus’ tight, twisty, and rough roads require a different map again. Gilles Panizzi is known to have preferred the (now) old-fashioned mechanical clutch-pack diffs tightened by input torque, but now even he has gone to computer control and suddenly he’s a force to be reckoned with on gravel where he was previously fast only on tarmac (which has much more predictable traction and so needs less sophistication in the diffs). Notice that he slides less than he used to? Huge 4-wheel drifts are a thing of old technology with “passive” locking diffs. Active diffs allow drivers to be faster with less sliding. And here is one of the main points to take away: Peugeot has been the leading proponent and developer of active differentials since 1999, and in 2000 was the first WRC team to fit three fully active differentials to their rally car (except for Panizzi’s!). Their mastery of the diffs is one of the significant factors keeping them at the front of the World Rally Championship.

 

This technology is only beginning to be used on rally cars – and indeed on street cars – in North America, and rest assured that it will take the sophistication of rally competition here to another level. There was talk of banning active differentials from the US rally championship, but I have seen a number of cars that definitely have it. And with Patrick Richard’s new WRX STi in the Canadian Championship, we have an active centre differential running in Canada. Now I just have to get a Mitsubishi Evolution VII to get my own active centre diff…