Digger Barnes
|
| posted on 10/5/04 at 03:18 PM |
|
|
No offence taken.
It certanly has been an interesting deviation from the origonal post. As you say listening to what others have to say then doing some basic proofs (so
as not to take hear say as fact) takes your knowledge along in leaps at times when otherwise it may have been small steps.
Cheers
Gareth
|
|
|
|
|
pbura
|
| posted on 10/5/04 at 03:39 PM |
|
|
Syd, you can liken us to sh*thouse lawyers, but these vaunted experts with their GPS sensors and captive programmers aren't helping US too much
I've been thinking about the nature of that line from the contact patch to the instant center of the wishbones. A force applied above the line
compresses the tire, and if below the line would provide lift. The action is very similar to dive and squat behavior in the longitudinal plane, and
each tire is really on its own.
So when a roll center is fixed laterally, it means what? That downward force is being relieved from the inboard wheel* at the same rate that
it's being applied to the outside wheel*? If my understanding is correct, then I can imagine that the 'feel' would be predictable
and reassuring to the driver. But maybe it's not what will get the car around the corner the fastest, which may be your point.
Shee-it, this stuff is complicated. Admittedly, the geometric roll center model may be outmoded with the automotive elite, but it's all that us
hoi polloi have for now 
Gareth, I didn't know that you're loaded with some serious ammo, and I look forward to future developments  Hope you don't
charge too much!
Pete "Willing to Learn" B.
*Edit: Whoops, meant "springs".
[Edited on 10/5/04 by pbura]
Pete
|
|
|
Digger Barnes
|
| posted on 10/5/04 at 04:35 PM |
|
|
Pete watch this space it may take some time till I am happy with a final program (many months), but you should be getting a copy of the drap n drop
geometry stuff in a week or so.
As to geometrical lateral fixing of the roll centre in roll (I will only answer part of your question for now); if my understanding is correct it is
only possible to limit its movement with the geometry and not to absolutley fix it unless the geometry is set up as a paralellogram (which would be
cr@p) hence the fixing it at ground level at the center of the car (infinite IC's).
|
|
|
cymtriks
|
| posted on 16/5/04 at 08:51 PM |
|
|
Ok, the car does roll about the tyre contact patches.... BUT The feel and dynamic behaviour of the suspension is greatly influenced by the roll centre
height and the camber gain in roll.
Just because all loads ultimately go through the contact patch doesn't mean that how the main mass of the car is controlled and moves relative
to those patches isn't important.
Too high a roll centre results in the suspension trying to fold under the car making high speed cornering feel, and sometimes be, very disconcerting
especiall when accompanied by large camber changes as on swing axles. This effect is called "jacking". This is why sports cars usually
have low roll centres. This increases the tendency to roll but this is countered by the stiffer springs and anti roll bars on sports cars. The result
is a firmer ride but secure and predictable handling with good feel.
Too low a roll centre increases the suspensions susceptability to road imperfections in feel and action. This is why even F1 cars rarely go below
below ground level with most curent designs being around 20mm (so I'm told). The fact that the front roll centres of both the original elise and
the revised all independent Caterham are at 30mm suggests that this might be the practical lower limit for road sports cars.
FYI the original elise used off the shelf tyres. The latest version uses special tyres. All my data relates to the original version.
|
|
|
jcduroc
|
| posted on 16/5/04 at 10:48 PM |
|
|
quote:
Originally posted by cymtriks
Ok, the car does roll about the tyre contact patches....
A doubt suddenly assaults my thoughts: how does a rigid body roll about TWO different points (if each axle considered separately) or FOUR contact
patches?
Shouldn't it be around an axis?!
Joćo
JCM
|
|
|
crbrlfrost
|
| posted on 17/5/04 at 12:01 AM |
|
|
It does roll about an axis, one that passes through the two roll centers. Thats why it is more involved than simply deciding on a static roll center
and calling it a day. Cheers!
|
|
|
crbrlfrost
|
| posted on 17/5/04 at 02:42 PM |
|
|
Just a quick question, which books are you referring too? I'll grant you the most accrurate is looking at it from the newtonian physics point,
but to some extent, that is reinventing the wheel. Reiterating chassis-suspension interactions via forces can get downright confusing after a while
(done that) versus a geometric understanding (roll centers, axis, inclinations, etc). But point taken.
|
|
|
Bob C
|
| posted on 17/5/04 at 08:48 PM |
|
|
the way i'd rationalised this......
the car doesn't rotate round it's tyre contact patches, but each wheel does.
"Roll centre" has bugger all to do with mr newton, everything to do with a network of links.
If you let the wheel rotate sideways round it's contact patch, the chassis will rotate around an imaginary point that I call its roll centre.
What is so offensive about this model?
bewildered of knutsford.
PS very few models used to predict real world behaviour are absolutely correct at unrealistic extremes - otherwise they become too complex to be
useful. Replace suspension units with solid links? - then it's not actually suspension any more is it!
As I've said before - Syd could well be right & it's all hot air & doesn't matter a damn, but at least it's given me a
way to calculate where my wishbones should pivot & how long they should be - so I feel a little bit more comfortable than if I'd just
guessed (which was my original intention!)
|
|
|
andkilde
|
| posted on 17/5/04 at 09:30 PM |
|
|
ROTFL...
Poor Syd
OK, just to throw another uninformed opinion into the mix.
Syd, I gather that you're trying to expose "Roll Center Theory" for the convenient fiction that it is.
i.e. Someone, many moons ago, tried to come up with a simple explanation for the complex interaction between a vehicle and the road surface. From
this, the concept of roll centers at either end of a common roll axis came to be. A nice concept, one which has hard and fast rules and ways in which
to easily calculate (statically). But, it is flawed, as the geometric centers "float" about as a car actually rolls and to run through the
whole gamut of permutations and combinations actually requires some heavy duty math and computer power (starting to lose the simplicity factor) to
figure out the "whole picture". Also, tires (tyres?) have their own ideas of how to behave which isn't easily calculable.
Now, everyone's goal is to have loads of traction and "predictable" handling.
There is no arguement about what you need to achieve this -- keep the tires as close to square with the road as possible, point them in the correct
direction(s) and share the work (weight for simplicities sake) as equally amongst the four contact patches as possible. We'll call all this the
"Happy Tyre Place".
Now, let's look at what happens to a car in real life. Your car will flop about (a bit) from side to side and fore and aft as you accelerate,
turn and brake. While it's flopping about you want the tires to remain as close to the "Happy Tyre Place" as possible. To this end
we try to determine and control camber and pitching through "Suspension Geometry". Now, by this point, things are coming at us at a fast
and furious pace, we've got roll centers, anti-dive, uncle-squat, moments of inertia, et flipping cetera, ALL OF WHICH are mathematical fictions
(sometimes known as models) designed by smart people to try and explain complex things to the rest of us.
Truth is, keeping the weight (mass) low and as close to the center of the car as possible is good. Keeping your tyres upright while your car flops
about is good. Controlling the degree of flopping about is also good but, funnily enough, doesn't affect the handling much as long as you remain
near the "Happy Tyre Place". Too much (or non-standard) flopping about can affect your perception of the car's handling though.
Now to the part where trouble begins...
How do I design a good suspension?!?
Turns out our "convenient fiction" with roll centers low in front, slightly higher in rear, often (when combined with well distributed
weight) will provide good camber control in roll and predictable handling. That said, the roll centers are still a "fiction" and a
theoretical car with "perfect" roll centers can have poor camber control and handle very very badly if it happens that our models and
reality end up conflicting with one another. So, the most important thing is still the "Happy Tyre Place" and we have to be careful that
we end up building functional cars rather than disfunctional mathematical models.
Or, to paraphrase what I've seen written previously, bottom arms roughly parallel to ground, top arms, shorter, sloping down a bit towards
center of car. Drive it to see how it feels/works and muck about with it until it's right.
Just for laughs throw in kingpin angle intersects ground through centerline of tire (prevents tire being kicked offline by bumps) steering arms
configured for full Ackerman (helps with "turn-in" prevents scrubbing of inside wheel and some say let's the inner wheel do more
"work" ) and, make sure your inner tie rod arms proscribe the same arc as your suspension arms to avoid bump steer.
Sounds great, now let's all go do it 
Cheers, Ted (who has had a bit of Uni Physics in his past but still sucks at billiards)
[Edited on 17/5/04 by andkilde]
|
|
|
ceebmoj
|
| posted on 17/5/04 at 09:37 PM |
|
|
Not got much to add I started playing around with the idea of designing my one car over a year ago (how hard can it be I thought) and then I came on
sights like this and found that there where a lot of things that I did not know about and realised I would need to take in to account thanks to every
one who has taken there time to add to discussions like this even if you all disagree I think I have learnt a lot.
So 6 months or so ago I started working on a couple of simulations programs. A simple FEA i.e. for use by idiots like me and a suspension geometry
program also for idiots like me. Digger maybe we can swap some notes and mabby collaborate on this.
[Edited on 18/5/04 by ceebmoj]
|
|
|
MikeP
|
| posted on 17/5/04 at 10:36 PM |
|
|
LOL, great fun.
Is it possible there's some confusion between weight transfer and chassis roll? Certainly Syd and Newton are correct, the roll center does not
have any affect on weight transfer to the outside tires - that value is constant. It doesn't change regardless of using no springs or solid
links, a high roll center/axis or a low one.
But Syd's solid link model doesn't break the roll center / roll axis theory. The roll axis and roll center affect only the amount of
chassis roll and the tire's "happy place" as Ted says - for independent suspensions anyway. But they also affect the
distribution of the weight transfer between the front and rear tires.
I use this mental model - if I stiffen the rear (or shorten the roll couple), the softer front now compresses more than the rear in the turn. Just as
if the chassis was raised at the rear when stationary, the fronts get more weight and can generate more traction. The front gets tighter, the rear
loosens.
In the solid link case, the links prevent the weight transfer from changing front to rear, so the traction distribution remains constant too (not
counting braking or acceleration or bumps).
Syd's example triggers an interesting thought - his solid links are like converting the entire car to unsprung weight. Could stiffer springs be
like converting a higher proporation of the car to unsprung weight at that end of the car? And the front/rear weight distribution be modelled that
way?
|
|
|
cymtriks
|
| posted on 18/5/04 at 08:53 PM |
|
|
As echoed by others the important point of my earlier post is this-
"Just because all loads ultimately go through the contact patch doesn't mean that how the main mass of the car is controlled and moves
relative to those patches isn't important."
The fact that cars work well with such a narrow range of possible geometry and suspension designs suggests that suspension and its geometry is
important.
Syd, we know that the loads are ultimately reacted through the contact patches and that consequently a car ultimately rolls about them. However the
main mass of the car is still suspended and the geometry of the suspension influences how the contacts move relative to everything else and so
suspension design is important.
Have your contacts got anything to add to suspension design other than rubbishing our approach? If a different way of looking at the problem indicates
a better design then we'd like to know more. I'm fairly open to alternative views as i've tried to think this through from a contact
patch point of view myself. So far you haven't given much of a clue as to what you are getting at.
|
|
|
jcduroc
|
| posted on 18/5/04 at 11:08 PM |
|
|
quote:
Originally posted by cymtriks
As echoed by others the important point of my earlier post is this-
"Just because all loads ultimately go through the contact patch doesn't mean that how the main mass of the car is controlled and moves
relative to those patches isn't important."
The fact that cars work well with such a narrow range of possible geometry and suspension designs suggests that suspension and its geometry is
important.
Syd, we know that the loads are ultimately reacted through the contact patches and that consequently a car ultimately rolls about them. However the
main mass of the car is still suspended and the geometry of the suspension influences how the contacts move relative to everything else and so
suspension design is important.
Have your contacts got anything to add to suspension design other than rubbishing our approach? If a different way of looking at the problem indicates
a better design then we'd like to know more. I'm fairly open to alternative views as i've tried to think this through from a contact
patch point of view myself. So far you haven't given much of a clue as to what you are getting at.
Clap! Clap! Clap!
Suddenly, in the up till now silent theater, applauses were heard prizing the so much awaited "line"!...
Apart from apples that fall down on our heads, Syd, what do you have to tell us that can help everybody make better Locost's?
Longing for something really substancial...
Joćo
[Edited on 18/5/04 by jcduroc]
JCM
|
|
|
crbrlfrost
|
| posted on 19/5/04 at 01:23 AM |
|
|
I'm in complete agreement that "roll center theory" isn't exactly an exact science, but rather a model. As such, it provides a
descent way to layout an sla suspension without running throughan impossible number of combinations. And, just a thought, that model was developed by
a group of scientists that were employed by the likes of GM to NASA. The fact that the model is often oversimplified for the sake of beginners
doesn't make the model obsolete. A vehicle is a dynamic body, and thus anytime we analyse it statically, it becomes a gross oversimplification.
But it is a solid basis. But since this thread seems to have become a pissing contest, I'm sure none of this will matter. Oh, by the way, I have
worked the physics for small formula cars, and I still vouch that the "roll center theory" comes close, albeit some aspects slightly over
emphasized. Cheers!
|
|
|
andkilde
|
| posted on 19/5/04 at 03:02 AM |
|
|
quote:
Originally posted by Syd Bridge
Andekilde, if you have an education in physics, then your sarcastic remarks show that you wasted a lot of money to achieve little. I have a son in the
third year of his physics masters degree, and he can put the maths to this. So you must be able to?And you believe in antidive and anti squat!
He who laughs last, and all that...
[Edited on 18/5/04 by Syd Bridge]
Now Syd, play nice...
If you'll re-read my post carefully I'm (at least) half agreeing with you. All of the items I mentioned are mathematical models used by
folks to attempt to describe and control vehicle behaviour. Not once did I state or imply that the models were correct.
They are what they are though -- models are simplified predictors of complex interactions, and as such are doomed to fail miserably outside of a
tightly controlled range of situations.
Now, since you're Mr. Smartypants, why dontcha fill us peons in on the simple, derived from 1st principles method of designing the perfect
suspension with perfect camber control 
I'll tell you why! Because it really isn't all that simple
But! We're not going to go up against Ferrari for the F1 title, we just want stable, predictable specials to play and tinker with. So, for our
purposes, cobbling together some simple ideas stolen primarily from what's worked in the past for other folk is as close to perfect as
we're likely to ever need.
And for those of us silly enough to ask, WHY? It is useful to start with available models and understand that they are fallible and need to be
thoroughly tested to ensure that our tires are correctly positioned throughout their range of motion.
Cheers, Ted
p.s. FWIW, on paper, I'm brilliant at billiards, it's only when I pick up a cue that it all goes pear shaped
|
|
|
