That's just about the worst on-center test results plot I've ever seen and I would hesitate to blame the car. It takes a certain amount of skill to run this test but its easier if you realize that at 100 km/h its +-1.25 meters lateral deviation and every 5 seconds for the period. I had my test drivers play a Matlab generated sound recording of a base tone with the 5 second period that this SAE test procedure calls for. Just play it through the car stereo system.

That being said, I would also argue that the biggest effect would be where the yaw velocity is changing the most, NOT where its being shown on the graph.

I would have run a routine swept steer frequency response test on these conditions. I also use a Matlab generated chirp function to assist the driver to get a uniform PSD for getting good data.

Want the most dramatic results comparison ? Run it on an E46 BMW 5 series.

Want to see the absolute best way to study the vehicle dynamics? Check out my videos on YouTube:

www.youtube.com/watch?v=S4O7b0-epBo
This was run using a robotic driver to input the steer angle and a restraint bar to produce the yawrate from the measured sideforce transducer and the known speed.

Here's another way I've run such a test: An MTS Flat-Trac Roadwy simulator.
www.youtube.com/watch?v=SgrY-LN4GUs
and
www.youtube.com/watch?v=I2o1rxMGFdI

I don't have the mathematical analysis background that cibachrome does, but I have an "actual field experience" background, including working on my own roadrace motorcycle and a few others.

I can't help but think that the tests shown in that paper were chosen to minimize the bad side effects of the extra unsprung weight, and that if you expose the situation to real world conditions - washboard surfaces, choppy pavement transitions that exceed the capability of the tire compliance to absorb them and blow through the damper's low-speed valving, cornering near vehicle traction limits on irregular pavement - then the extra unsprung weight will indeed reveal itself as being "bad".

When cornering hard on choppy pavement, it's a whole lot easier to keep the tire in contact with the ground if the unsprung weight is kept down. Adding more is not going to help.

Brian: You have already exceeded the amount of sense required to judge these conditions. In MY DIRECT experience with Lotus (Starting with the 'Active Corvette Project') They don't use anything but a 'change parts and drive around' development process. Yet, the Lilliputians worship them because of some unfounded claims of mechanical and artistic superiority.

As I will repeat: that was the wrong test, the wrong portion of the crossploted data and the wrong conclusions drawn from the Oujia Board. Sort of like a Hollywood movie script though, don't you think ?

I don't usually plot SWT vs yawV, and having just checked both track data and ADAMS runs, I think I can see why. It is inherently rather a noisy plot as it compares two responses against each other, whereas both yawV vs SWA and SWT vs SWA are output vs input.

Also there is too much filtering on that data, my data is far less swirly than that.

Agree about the metrics, that red line looks like it is measuring a very strange property as you slow the swv in preparation for steering the other way.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?

Actually, there is some good information that can be obtained from a plot of SWT vs U*R(g). Testing is usually done at 100 km/h. Presuming that you can get a decent statistical description of the Steering Wheel Torque gradient (N-m/g) at 0.0g [TG0] AND a decent value for the Steering Sensitivity (g/100 deg SWA) at 0.1g, [SS.1] then a metric referred to as the "Steering Work Sensitivity" can be computed from 57.3*SS.1/TG0.

Values for typical passenger cars lie between 2.5 and 3.5. Anything out of this range is a bit out of place. High values (lets say 3.5 to 6 are twitchy or darty and too "light" fot the amount of steering gain the vehicle has. It could be too much power assist or too much steering gain. Vehicles with a work sensitivity below 2.5 are too "heavy" (not enough boost or not enough steering gain. Across the board, manufacturers typically produce sports cars that are NOT in the 2.5 to 3.5 range for various reasons. Manual steering, heavy parking effort, understeer is very low (meaning the gain change with speed is high), a very quick steering ratio, difficulty in blending low speed steering boost with high speed assist, heavy tierod loads, very light tierod loads, marketing image and the cost and availability of special steering gears, valves and pumps and good tires, blah, blah, blah, hopefully you get the idea.

But the cars (and trucks) that are the most fun to drive (nimble, sporty, easy, etc, have work sensitivities in the 3 to 3.5 range. Given this basic relationship: (there is an appropriate steering effort buildup for a given level of maneuverability) the wheelbase of most cars (2700 mm), the understeer of most cars (3.0 deg/g) and the steering gear ratios of most cars (16:1), you can quickly figure out what the boost rate for a good steering system ought to be to drive comfortably on the highway and get good ratings for it from MOST drivers). The engineering required to get an appropriate tierod load gradient (N/g) and a pump flow rate and a valve profile to deliver the parking effort and a means to roll the assist off as you pick up speed is where all the artistic handwaving (and a good simulation) begins.

Been there, done that.