Originally Posted by Projekt on S2000turbo.com

1) Assume a static weight distribution of 60% Front and 40% Rear on a car like a G3 Integra prepared for SCCA Club Racing. Because most of us with these chassis' run the similar bars, and because the rear spring motion ratio is better than the front, a given spring rate on the rear produces a higher wheel rate than it does on the front. We can therefore continue for the purpose of discussion to refer to the NA setup as high rear bias and the JDM as high front bias.

2) Under steady state cornering the Front/Rear distribution of lateral force will reflect the static weight distribution and will not change in response to longitudinal acceleration except where the friction circle is overloaded by driver input. This means that the front axle has to do 60% and the rear 40% of the work of cornering (lateral force generation). This along with the stacked workload of the fronts (primary braking and acceleration) is what kills our front tires and has us looking for ways to improve front grip in all modes.

3) What we refer to as the non-linear relationship between vertical force on a tire and the traction it develops is one of our most important tools. Explained simply it's like this: A tire has X grip loaded with A pounds load. That tire has less than 2X grip when loaded with 2A pounds. What this means is that any pair of tires will develop maximum grip as a pair when their load is split 50/50. As the split biases the total grip declines.

4) Acceleration causes weight transfer, visible effects of which are roll under lateral acceleration and pitch (dive or squat) under longitudinal acceleration. Assuming identical acceleration, reduction in roll or pitch resulting from spring or bar changes alone DOES NOT indicate a reduction in weight transfer.

5) Weight transfer is a function of the height of the center of gravity, the track width or wheelbase, and the acceleration. Total weight transfer is reduced by lowering the center of gravity, widening the track, lengthening the wheelbase, or reducing the acceleration. We are not frequently interested in that last option.

6) Under lateral acceleration the end of the car with the highest roll stiffness will transfer the highest percentage of its inside cornerweight to the outside wheel. At any point in time the chassis has ONE roll angle - although our chassis' are not perfectly rigid torsionally, this assumption is necessary. Combined lateral and longitudinal acceleration gives diagonal weight transfer.

7) Our biggest problem in going faster is the problem of UNDERSTEER. We combat this by first maximizing front end grip, and only then sacrificing rear grip to find balance.

Cool What we call the friction circle, an application of Mohr's circle, is an analytical tool to explain what we can expect from a contact patch (or the sum of their generated force). Say we can achieve 1g braking, and 1g cornering. We cannot simultaneously achieve both - we'd call such a thing well outside the circle. We can achieve 1g total combined - perhaps 0.7g braking and 0.7g cornering (the square root of 0.7 squared plus 0.7 squared being equal to 1).

9) A front wheel drive car really does literally tow the rear wheels behind it. The front tractive vector of a FWD is generally in the direction of the cars desired heading. The rear tractive vector of a RWD is only so under considerable oversteer (which is the way to go on dirt with virtually any configuration, but not pavement where it will usually mean that the front tires are underworked to the detriment of total grip and maximum speed - sorry drifters.)

10) It might then be relatively straightforward to come to the conclusion that we might want the rear to have the higher roll stiffness. North American cars are typically prepped this way. The correctness of the basic case is unassailable in the most general terms, and corraborated by enormous practical evidence from competition. Smart people do it.

11) It's not necessarily intuitive. Among the encountered superstitions are ideas like "the heavy end gets the heavy springs", and "a car on 4 wheels is working better than one on 3 wheels". Furthermore, there are Japanese DC2's running higher roll stiffness in the front, and, when asked about the difference, Spoon techs say they can't understand why we do the opposite. By reasonable accounts they are not dummies.

12) One of the difficulties we run into with the typical NA setup is available front bump travel. The G3 Integra at NA racing ride heights of 5-5.5 inches has very little shaft travel left (less than 1.3 inches, and even if you had more shaft left, you'd still have upper a-arm interference issues with Skunk arms). We generally wind up as low as we can (at our legal ride height - if any) without bottoming the outside front in the turn in zone and generating massive understeer. In such an event we can either increase ride height (and go slower than we were hoping), or we can increase spring rates to get a total roll resistance that will keep us off the bump stops.

13) Among the few orthodoxies from old books that can be discarded are the particular bounds on Natural Frequency. We've been taught that somewhere not far beyond 2 cycles per second is the limit. Practical examples worldwide suggest that 3cps front and 4cps rear on these cars represent the more useful bounds.

14) Returning to what happens on the track - a rear stiff car will unload the inside rear faster than it unloads the inside front. At some point all of the inside rear weight has been transferred to the outside rear. A rear camber setting is selected so that at the terminal roll angle of the chassis sufficient grip is generated by the outside rear alone. This is typically achieved on our cars at something less than the camber required to generate the maximum potential rear grip. As the outside rear decambers in roll, and converges on its terminal camber, it is generating greater and greater grip - both as the vertical force increases from weight transfer, and as the dynamic camber improves. At the kind of rear spring rates common to both NA and JDM setups, dynamic rear camber change during rear weight transfer in pure roll is approx +1 degree (decambering). In any case the front decambers approx 50% faster than the rear in pure roll while all four wheels are on the ground. Since we rarely have pure roll without pitch effects we should note that the combined effects camber curves are a more complex story that must be examined in cases. Most interesting among these is drop throttle behavior. The rear decambers fast in combined roll and droop. This points to the usefulness from this perspective of high rear spring rates in general as common to both setups. Generally speaking, as we converge on max lateral acceleration we converge on optimal dynamic camber - assuming we're doing our job as a race engineer. In general camber change is not something to be minimized for it's own sake (the chassis performance envelope is a multi-dimensional topography - very little about it can be changed without affecting other perhaps more important factors) - the JDM guys don't do what they do simply to control front dynamic camber.

15)If we suppose a limit imposed by drivability on the rear spring rate (and it's amusing to suppose that a rate change giving less than a tenth of an inch difference in static deflection might be over said limit), there is not much supposition involved in determining that diminishing returns in rear roll stiffness are quickly reached.

16) At lower ride heights with the production tub configuration, and higher grip levels than can be achieved on typical US DOT race tires, there is no escaping the conclusion that front roll stiffness MUST be increased beyond levels used by NA racers. Some mitigation of this need could be realized by increasing available front wheel travel with a more compact upper a-arm (or modification of the inner fender / tower). If we're going to go faster than a NA car by using stickier rubber and a lower center of gravity, we will be FORCED to increase the front spring rate. The Japanese arrived at their setup the old fashioned way - they found that they went faster. But, if the track is rough, the ride height must go up, the rates down, and the bias rearward again - or else mechanical grip will be sporadic and the car will be undrivable.

17) This illustrates a common problem. One of the big obstacles to understanding this stuff is the haphazard way in which we assimilate new information and practical experience into our knowledge base. We rarely have all the information we need, and many are sufficiently ignorant of the mechanisms as to preclude any real understanding - they match incompatible cause and effect with regularity and are incapable of recognizing contradictions when they inevitably arise. A nice whitewash with "personal preference" masks their confusion.

1Cool So where does that leave this "argument". I stand by everything I've written to date as it was in the context of our ride heights and conditions, and nobody advocating use of the JDM rate bias made any qualification for conditions. Furthermore nobody made a case for how the JDM rate bias could make any sense - and I certainly provided every provocation for someone to do so. So, only an idiot would now use the JDM setup under typical US conditions, and only an idiot would be stumped by an understeering NA setup at 3 inch ride height, and only an idiot would think that that I'm just a closed minded *******.

19) The special case does not disprove the rule - Physics survives another puny attack - 20 year old books need not be discarded - worship of either NA or JDM tuners is unwarranted - and if you really want to go fast you gotta use your head.

Scott, who hopes you've enjoyed and benefited from this exercise....may I never call YOU an idiot......"I'm ready for my close-up Mr DeMille"......


This follow up framed some key conclusions too:


Take an ITR with an NA setup (800F/1100R) on DOT tires. It develops enough lateral acceleration such that when trail braking / turning in it unloads the inside rear completely and uses up almost all available outside front shaft travel.

Then increase the lateral acceleration. This might be for a case with grippier tires, or a lower ride height. The lower ride height brings down the cg, but the roll centers move in approx 1:1 relationship on these cars so the overturning moment arm is about the same. The lower cg means lower total lateral weight transfer at the same lateral acceleration, but we do this to go faster - the lateral acceleration will be higher and the weight transfer will be not much lower than before and likely significantly higher. WE NOW HAVE LESS FRONT SHAFT TO DO THE ROUGHLY THE SAME OR GREATER AMOUNT OF WORK WITH. So in the case where we have lower ride height AND grippier tires we have significantly increased TOTAL weight transfer.

If we try to drive that NA car under these new conditions it will be spending alot of time on the outside front bumpstops and understeering since everything is already coming off the inside rear, the increase in total weight transfer must come off the inside front - and go to the outside front in cornering phases prior to runout. This is the key thing to look at: as we increase total weight transfer on a car that already is capable of unloading the inside rear the difference can only come off the inside front. Increasing rear roll stiffness will do nothing to change this fact, and will only unload the inside rear earlier in the cornering sequence - and where rear wheel rates are so high already as to give deflections well under an inch these timing differences will be very slight.

So we are left with having to increase total roll resistance with the only tool left to us - the front roll resistance. The front roll resistance MUST GO UP. For a variety of reasons we would do this with the front spring rates.