I've been chassis tuning with alignment, anti-roll bars, and shocks for a about 30 years, now. The steady state understeer/oversteer balance of a car is determined by weight distribution and front/rear roll stiffness distribution, but suspension architecture can play a major part in transient behavior. Most production cars are set up to have mild to significant understeer. Less understeer is usually rewarding to a skilled driver, but a neutral car can sometimes transition to oversteer at the limit, which is the reason why most OE suspension calibrations have plenty of "safe and sane" understeer designed in. Basic suspension architecture has a significant influence on transient behavior and some are better than others with many IRS designs being more twitchy than a well located solid axle.

Because of its higher front weight distribution, the 190E 2.6 has more inherent understeer than a 2.3, and having owned both, the difference is noticeable. My former '84 190E 2.3 was closer to neutral (less understeer) than the 2.6 without having any tendency to snap into oversteer. For cars equipped with both front and rear anti-roll bars, a slightly larger rear bar will increase rear roll stiffness and move dynamic response closer to neutral. Since the anti-roll bar's contribution to roll stiffness varies as the fourth power of its diameter, even a 1 mm change is usually noticeable.

Roll steer is usually a secondary effect and is a function of the suspension architecture and detailed design. A small amount of roll understeer can be desireable by having the rear wheels steer slightly into the direction of the turn with increasing roll, which adds stability if it is not excessive. At the front, roll understeer is achieved with geometry that causes the front wheels to steer opposite the direction of the turn will increasing roll.

Achieving slight roll understeer with a solid axle suspension is fairly easy by the arrangement of the control linkage and bushing compliances. Roll understeer is achieved on IRS designs with slight toe-in with jounce and toe-out in rebound, but it should not be overdone as lifting the throttle will alter pitch moment, and this can lead to trailing throttle oversteer, especially on designs that have poor pitch control. This behavior is somewhat inherent in semi-trailing arm suspensions because they have fairly large toe and camber change with vertical suspension movement along with relatively poor anti-squat and anti-lift characteristics, which together consititute pitch control.

The five link rear suspension pioneered on the 201 models maintains close to constant toe, camber, and track through most of the suspension travel range, and has very good pitch control. It is essentially "neutral", which yields both stability and good response to small changes in roll stiffness, without introducing any untidy secondary effects, and whatever roll steer is desired for stability can be designed into the front suspension.

My two favorite rear suspensions are the Mercedes five link and the "torque arm" solid axle in my Cosworth Vega. Both have excellent wheel geometry and pitch control and can be tuned to have minimal understeer without a tendency to snap into oversteer in nine-tenths driving situations or if you have to drop the throttle in mid corner.

My experiences with semi-trailing arm suspensions and three and four link IRS designs has been less than satisfactory. These all exhibit transient oversteer due to track, camber, and toe changes, and/or poor pitch control, and if they are tuned to be fairly neutral at seven-tenths, they exhibit a tendency to snap into oversteer at nine-tenths, especially if you lift off the throttle.

Considering that free body in space has six degrees of freedom, and it is generally desirable for the rear suspension to limit wheel movement to one degree of freedom, a five link design makes perfect sense.

Duke