suspension concept

If a suspension system is analyzed in the context of electrical design, I use the following method. The input signal characteristics are 4 channels representing the vertical load present at each wheel. The signal formed is a varying dc signal superimposed with a sine wave representing bump/rebound and is filtered through the use of a spring/damper.Dc amplitude represents transitional load from rest ,and varies with dynamic input due to roll, pitch and yaw. If a 'mechanical potentiometer' is added to directly control spring/damping rates over a wide range instead of remaining fixed,dc amplitude can be contolled more effectively,ie ride height change at each corner. The pattern relationships between the 4 channels represent transitional load transfer during cornering, acceleration and braking. If a second 'mechanical potentiometer' is used to adjust chassis ride height independant from spring/damping rate control,(normally done with antiswaybars and anti dive/anti squat geometry), then transitions can be actively controlled over a wider range than a passive reactive design. The relationship between load and ride height is that load directly influences ride height, but ride height has little to no effect on load ie if I change the ride height on my 4x4, I don't have to alter spring and damping rates.(dynamic loads will be altered,and require adjustments in design to optimize performance). Therefore, by separating these 2 functions and giving each a range of adjustment, with a simple processor control system, a dynamic suspension with much better performance is achieved. Comments?
 

Dave Bilyk

Dave Bilyk
Supporter
Not something I have looked into in any detail, but I do understand electro-mechanical analogy models where resistance is equivalent to damping, inductance to mass and capacitance to spring stiffness. It works fine for linear systems and any mechanical system can be represented in this way, but it gets a bit complex for real life non-linear systems but can be very informative. Way back in college we used to have analogue computers for that purpose, now I guess digital software is available for the job. Don't know what the state of the art is, but Citroen suspension systems (mainly hydraulic), F1 and rallying come to mind. So how do you translate this into a real world suspension system, Ferrofluids are used in some high end dampers aren't they, so that bit is easy, but how do you deal with the springs, active magnetic? I guess hydraulic would be too bulky.

Dave
 

Seymour Snerd

Lifetime Supporter
It's an interesting subject but I think your model is a little over-simplified. For example, "spring/damping rates" is not a single parameter; it is at minimum two. Even so, how in a mechanical context would you "control" spring rate?

Also, although I'm not sure it matters at this point, anti-roll bars and anti-dive/anti-squat geometry do not control "ride-height" any more than springs and dampers do.

It is not even approximately true that ride-height does not affect load. Imagine your 4x4 with a ride-height of 30 feet; load distribution in the direction of acceleration is clearly affected; you've changed position of the center of mass and that affects the relative load on the tires.

Finally, even if you could separate the control of basic suspension parameters into two time-varying scalars, the "processor control system" would be far from simple, but of course "simple" is in the eye of the beholder.

If you want to make an "electrical analog" analysis of suspension systems I think you would be better of going the traditional way whereby springs ("compliance") are modeled as capacitance, mass (the chassis itself, the un-sprung masses, etc.) are modeled as inductance, dampers (and other forms of friction) are modeled as resistance. Even with that you will have non-linearities that are difficult to model due to suspension geometry. You will also have difficulty acquiring modeling data for things like tire behavior, non-linear friction in the suspension pivots and bushings, both of which are highly non-linear and hard to determine empirically.

Finally, your text basically describes the "actuators" (outputs) but there is no discussion of the "sensors" (inputs). What are the inputs to your processor control system?

There has been some interesting work in this area over the years, a long time ago by Lotus and more recently in the upcoming S-class Mercedes, but both of these involve road surface "look-ahead" sensors. In the Lotuses this was by replacement or augmentation of the suspension system with hydraulic rams. In the Mercedes I believe it is via damper control, and may only be used at low speeds (e.g. speed bumps). And of course many modern cars have dynamic control over suspension damping. Another much simpler example is the anti-roll system in the Land Rover Discovery II that provides for a disconnected anti-roll bar when there is low lateral acceleration that becomes progressively "connected" as lateral acceleration increases as determined by accelerometers in the chassis. "Dynamic" ride height is controlled in a few cars pneumatic springs or hydraulic actuators although not very dynamically (AFAIK), i.e. sometimes for aerodynamic reasons and sometimes for ground clearance reasons. I don't know of any that provide direct dynamic control of spring rate but perhaps they exist.

I'd suggest you do a literature search on "active suspensions."
 
Kind of ,however I'm trying to reduce the need for a computer system to a minimum. I'm presently building a type of gocart utilizing a dirtbike engine...sofar in development I've been able to come up with a way to left foot clutch and shift while right foot simultaneous throttle and brake control, and manually control hydraulic actuators with a hand pump and valve...parking lot skidpad, acceleration and braking tests show promising results...
 
I'm changing linkage ratio to the coilover shock by moving the fulcrum point of the lever compressing it with a hydraulic cylinder perpindicular to the load path.Most suspensions (eg air ride truck systems) increase spring rate without changing damping rates...I sized the coilover units for highest anticipated load..when ride height changes are small,-I'm working with a range of 6 cm-using a cylinder and wedge at the chassis mounting point, CG changes are small, particularly when I raise the outside and lower the inside ride heights. Camber control looks promising (although you end up driving in circles almost to the point of vomiting trying to establish tire wear patterns,lol). I'm trying to come up with a strictly hydraulic control using a pendulum controlling a valve system -not optimum, granted, but if it's an improvement over existing design while remaining simple, cheap, durable and light, and if I can build it myself, where's the harm?
 
As well, the next step in development is aero..I've got a crude undertray made of aluminum sheet that's attached to the spindles with short nylon straps to maintain a constant height above road surface,and am adding a small centrifugal blower presently driven by an electric motor in an attempt to create a vacuum car..this is all spare time stuff as an alternative to watching television...
 
I'm not trying to step on any toes here, just trying to learn for myself with a 'can do' attitude...in the past, I've had success with original thinking...eg in 1979 while crewing for an oval track dirt stock car, I managed to obtain a reverse rotation GM marine 350 in need of a rebuild..I used a V belt drive and adapted the water pump pulley to a reverse rotation waterpump so the fan turned normally, and flipped the diff upside down..once the suspension was revised to work with the change in forces, the thing was so much better on corner exit..got caught halfway through the season by the track inspector noticing the car body reacted the wrong way while the driver was revving the engine in neutral in the pits (he was leaning on the fender)..had to put it back to normal as they made a rule up against it on the spot.. as a mechanic, I see many examples where a 'rube goldberg contraption' design can be replaced with a simpler design, and in my mind that's an improvement..the ultimate dream is to make this work in prototyping to the point where I can build a gt40 to work the way I think possible..
 
A good example of what I'm trying to accomplish is the Koenigsegg supercar freevalve system, in essence their design took the fixed values of camshaft duration and overlap etc. and made them widely adjustable with a relatively simple mechanical system and computer control..because the transitional changes in suspension load occur over longer periods of time as compared to engine speed, if I can create a hydraulic control system to do the job effectively without the need of electronics I'll simply use the best tool for the job...
 
If your design analysis tends toward that of electrical perhaps the apex solution would be an electro-mechanical one.

The System

I was never impressed with the sound of his audio systems but I need to give Dr. Bose credit for genius by thinking outside the box designing unique solutions in a multiple of fields of design. His suspension system not only reacts to the environment, it anticipates it in one axis. The brick wall to be encountered by the shade tree mechanic would be the complexity of the algorithms necessary to control the system and apply the correct signals to the four corners of the auto in the millisecond realm. It must have taken years to develop a program that was acceptable for a run-of-the-mill passenger car. If you possess a degree from an institution the equivalent of MIT you may have what it takes to design a proper suspension system that eliminates solely reactive metal components. I don't; therefore, I won't.
 
Ok,here we go again..I'm semi bed-ridden due to a sore back, so the delay in responses should be understandable..after looking at the Bose system, it appears to me to be an electro magnetic Macpherson style system. My idea is laid out in a different configuration than this..however,if I used a standard strut, but I created an arced track on the lower control arm with a cylinder connected to the chassis so I can push/pull the bottom of the strut sideways, then I change the effective spring rate by moving the bottom of the strut (lever fulcrum)..then, if I put a weight jack as the nascar guys call it on top of the strut (hydraulic) and the tire load increases, I can move the bottom of the strut outwards to 'stiffen' that corner as well as raise the chassis with the weight jack so my camber doesn't go all to he-double hockeysticks...if I put this system on all 4 corners, I think I'm getting somewher handling wise..anyway, back to bed...
 
Oh yea, I should also mention that I'm overcompensating with weight jack operation in opposition to body roll in order to gain camber to try and offset tire distortion from lateral load..... and if you lift it, you can lower it,so the end result I'm finding in my prototype testing is I'm starting to be able to get this thing to corner like a boat/aircraft/motorcycle effect... my system is an SLA configuration with a linkage to the coilover...
 

Howard Jones

Supporter
One thing to keep in mind. As the total down force value goes up, the LESS is the need to control suspension movement. The huge down force created by the F1 cars in the FW14 era led Williams to give up on controlling movement in the traditional sense and instead simply concentrate on ride height. So much more of the cars grip came from controlling undercar airflow and thus the increased down force that optimum ride height created, that it became the only thing to consider when deciding between trade offs such as body roll, and front to back weight transfer.

This was possible because the tracks had become so smooth. In the end the Lotus approach of trying to model all of the dynamic forces acting on the suspension and write software to adapt to the huge number of variables became moot.

Just set the ride height, create many multiples of the car weight in down force and plant the drivers right foot on the floor. As corner speed began to approach 5 G's, Nicky Lauda swears they exceeded even that number, the drivers began to see the limits of their ability to drive the car under the enormous physical stresses.

All that to simply say that the more you gain in down force the less all the suspension/body roll control work will be worth. It's almost as if you need to pick a direction, mechanical grip or areo, and follow it to an optimal solution.

The surface the car will be used on seams to be the deciding factor here.
 
Last edited:
If a suspension system is analyzed in the context of electrical design, I use the following method. The input signal characteristics are 4 channels representing the vertical load present at each wheel. The signal formed is a varying dc signal superimposed with a sine wave representing bump/rebound and is filtered through the use of a spring/damper
I think we're of an age! A class project I was in in the '60s tried to model the behavior of an aircraft nose wheel on landing using an analog computer. Much simpler than your 3 axis problem, but while you can set the computer up to perform the linear differential equations representing the system if you neglect inter-dependent variables, simple input forcing functions don't give you much useful data. I'm sure your computer is much better than ours was, but I'll pass on what we learned in case it might help you out.

What we found was that if the forcing function represented the ground, we ended up with a wheel hub position that followed it with a phase delay depending on the tires, which are spring/damper systems themselves. At some point you get a trampoline effect when the rebound from spring, tire and the forcing function act together, and the tire loses contact. At that point our analog system became discontinuous and we had to reset all the initial conditions on the op amps, re-zero the strip chart and start over. You could work around that for sine wave forcing functions, but creating real world shaped bumps with a Taylor series of summed oscillators was just way beyond our ability. Perhaps if you can implement a Fourier transform with the computer modules you have available now, and give it a shot. You've got a very complex system on your hands, but it's certainly worth tackling.
I loved the visualization and real-time results you get with an analog computer, but for me it was more a learning tool than one for design. Good luck and please post your progress on this.
 

marc

Lifetime Supporter
in the 60's a gm engineer toyed with anti dive/traction by using the sway bar concept front to rear. Today companies are using computer controlled hydraulic shocks to duplicate l.e. Audi. magnetic controlled fluid in ferrari and GM are more active suspension for phase delay. given enough computing power and ahead of wheel sensors this could be capable of eliminating delay.
 

Jim Craik

Lifetime Supporter
I had a 1967 Austin Cooper S that had "Hydrolastic" fluid suspension that connected the front/rear in a similar fashion

But then I also had a 1965 Morris Cooper S with the isolated rubber suspension and quite frankly, I could not feel any difference between them.
 
Back
Top