building a chassis...suspension geometry

[williammcc]

ok here goes...i'm building a mark 2 escort for rallying but i don't like going with the tried and tested (i've saw too many millingtons blowing off rear tyres in a straight line due to the fact that the suspension systems they use are designed for about 240 bhp instead of 440) so i'm planning on doing things a little different firstly using a zakspeed bodykit (the old group 5 one) to make room for 17x10 rims (most mark 2s are on 13 inch rims) so that should be a step in the right direction but makes all the standard suspension mounts obsolete.

the rear is a solid axle so going for a 6 link with short top bars longer bottom bars converging below the anti-squat line to give some wieght transfer (exact heights will be trial and error to get a balance between front and rear grip) with a watts linkage mounted below the axle to bring the rear roll center back down nearer the ground

now the front is where it gets entertaining...using a double wishbone set up with everything custom so everythings up in the air i've looked at so many things about suspension my head is starting to spin but what i've already worked out is

i need about 20 degrees of bumpsteer from full compression to full droop (simple to sort with steering rack position)
i need positive ackermans so the steering arms should line slightly infront of the rear axle
i need to sort the camber and castor but that can be sorted on the uprights (ie. steering angle introduces camber)
the hard part is the front suspension

i intend using a double wishbone front end but this is where i'm struggling... can someone correct me if i'm wrong as everything i seem to find online says i am but i can't understand why

parallel wishbones should end up with a fixed roll center in the middle of the car at ground levels (ie. parallel lines meet at infinity so the instant center is halfway between the two wishbones an infinite distance from the car so the line returning from the instant center through the contact patch can be regarded as a flat line at ground level?right? so front roll centre will stay on the ground regardless of corner roll and stay in the middle of the car

alternatively using non parallel wishbones i can have the front roll centre moving side to side durring cornering?ie.in a left hand bend move the roll centre left outside the car (increases the distance between the cog and the roll center increasingg tyre load when cornering and producing additional load on both tyres thus increasing grip) and the opposite on a right hand bend (haven't actually considered wishbone angles for this scenario so i'm not sure its possible but i'm more interested in getting my head round the theoretical first then i can decide how best to employ it)

another alternative is using a roll centre above the cog cornering forces will act on the cog about the roll centre thus loading the inside wheel instead of the outside?

i don't really care if everything i've said is 100% wrong as long as someone can explain why it's wrong i've had enough people tel me things won't work and when asked why the only answer is "because it's different" thanks for your help

[Tommy Cookers]

....mark 2 escort for rallying
the rear is a solid axle so going for a 6 link with short top bars longer bottom bars converging below the anti-squat line to give some wieght transfer (exact heights will be trial and error to get a balance between front and rear grip) with a watts linkage mounted below the axle to bring the rear roll center back down nearer the ground

now the front is where it gets entertaining...using a double wishbone set up with everything custom
i need about 20 degrees of bumpsteer from full compression to full droop (simple to sort with steering rack position)
i need to sort the camber and castor but that can be sorted on the uprights (ie. steering angle introduces camber)

parallel wishbones should end up with a fixed roll center in the middle of the car at ground levels (ie. parallel lines meet at infinity so the instant center is halfway between the two wishbones an infinite distance from the car so the line returning from the instant center through the contact patch can be regarded as a flat line at ground level?right? so front roll centre will stay on the ground regardless of corner roll and stay in the middle of the car

alternatively using non parallel wishbones i can have the front roll centre moving side to side durring cornering?ie.in a left hand bend move the roll centre left outside the car (increases the distance between the cog and the roll center increasingg tyre load when cornering and producing additional load on both tyres thus increasing grip) and the opposite on a right hand bend
another alternative is using a roll centre above the cog cornering forces will act on the cog about the roll centre thus loading the inside wheel instead of the outside?

afaik and broadly speaking .....
changing roll centre position does not change weight transfer
it changes the amount of weight transfer going through the springs relative to the amount going through the wishbones
eg having a freakishly high roll centre will not load the inner wheels overall
it will appear to load them by compressing the inner springs but invisibly unload them equally through the wishbones
(similarly full anti-dive/squat geometries work by routeing 0% of the dive/squat moment through the springs)

ie there's no free lunches

and since front spring compression in roll is resisted by spring compression at the rear
handling balance ie front:rear load distribution will be upset if eg the front roll centre moves much with roll or bump
the reason why tradition says that roll centres should not move much
and that non-parallel wishbones will be needed to combine this with the desirable amount of 'camber gain' in roll

tradition also says roll steer should be small

[Tim.Wright]

20deg of bumpsteer is about 200 times more than you should start with. The car would be undriveable with that much bumpsteer.

[Greg Locock]

To amplify Tim's comment, imagine the car rolls by just 1/4 of the suspension travel. That will steer the front wheels by 5 degrees, which is almost their peak cornering slip angle.

[williammcc]

i've already figured that putting the roll centre outside the wheelbase won't work...inspiration hit me today as i realized doing so will result in the cornering forces acting closer to inline with the line between roll centre and cog meaning less of the force being applied downwards so it wont give any great gains

20 degrees of bumpsteer is common usually a 5 degree toe in at full droop going to a 15 degree toe out at full compression....it sounds extreme but those figures are over 7 inches of travel and in reality only perhaps 3 or 4 inches of that travel is used under normal circumstances...on gravel where a larger ammount of travel is used the larger bumpsteer angle toes the wheels out on heavy braking giving a wider area to stop the car

i get what your saying about the wieght being transfered allong the wishbones as opposed to the shocks and springs but surely more wieght tranfer through the spring meants more wieght on the tire?

[Greg Locock]

Lateral geometric roll centre migration is, to my mind, one of those things that people fret about without ever being able to demonstrate an advantage, or a disadvantage, separate from the the other implied effects. I am still amazed that you think so much roll steer is appropriate.

[Tim.Wright]

I agree with Greg. I too have never seen any reason why the lateral roll centre migration needs to be so critically controlled.

On top of that its always motorsport engineers and street tuners who are the worst offenders when it comes to agonising about roll centre movements but being completely unable to explain why. When you take into account that:
A. The car doesn't roll about the geometric roll centre
B. The geometric roll centre isn't even valid for determining the elastic/geometric LT split at high LatAcc
C. The roll centre lateral movement depends on what roll movement you impose (symmetrical or asymmetric roll) and what coordinate system you use (body or ground).
D. The geometric roll centre becomes undefined (divide by zero error) as it passes through the ground.
E. Close to the ground the geometric roll centre is numerically unstable and can produce massive roll centre movements vertically and laterally for very small roll angles which are completely meaningless.

Its becomes pretty clear why there is little reason to be worrying so much about how it moves laterally or vertically in an incorrectly imposed roll movement using an unknown coordinate system in order to calculate the wrong elastic load transfer using an incorrect set of equations which for most racecars are mathematically undefined in many normal operating conditions.

I'd recommend that you forget what you've "learned" until now about the geometric roll centre location and instead consider it as a representation of the relative jacking between the left and right wheels. In other words, the slope of the last lines that you draw from the contact patch to the IC's are the only important things to look at. With a RC above ground these line will be sloping up towards the centre of the car and this means that the jacking forces will oppose the load transfer forces that the springs see and therefore it will reduce the roll angle. BUT the load transfer will still exist through the links. This load transfer you can't visually see but this is why it's not correct to state that more spring compressions = more load transfer.

Regarding the bumpsteer - theres no way that much bump steer is acceptable. How have you come to the conclusion that it's "common"?

[Tommy Cookers]

well said, Mr W !!

another case of 'no free lunch' ? .....
there was Trebron suspension, a passive mechanical linkage system that banked the car into the corner
but the contact load benefits in-corner would have a cost on corner entry
because the banking work (done against inertia) must come from tyre action when the tyre is already subject to great demands

[Jersey Tom]

20 degrees of bump steer would be acceptable if you were describing a kilometer of suspension travel. Unless you're really clever with it, best practice is typically to make your bump curve as close to zero as possible.

Put it this way - the steering wheel is for steering the car, yeah? So let it do so on its own, not random bumps and pitch and roll and heave.

As for roll center this or that - IMO it's all intellectual masturbation unless you can put some numbers on the forces involved and whether they'll be trivial or not.

[Tim.Wright]

Bear in mind too that 4-5deg of toe angle at 100km/h will make you corner at around 1G.

[williammcc]

thanks for the input guys....to be quite honest i've been messing about with this stuff for a few years and any information you find on it is clear as mud and contradicting something else to the point i had to just start making drawings and working out what was happening for myself with basic force diagrams, which is all well and good until you get into a real life scenario and realize that really its pointless as there are simply too many factors involved to factor them all in without some real world suspension design experience and a fancy computer program or 2.

as for the bumpsteer it was measured off a car which was set up by a suspension expert who used to work for brabham i think it was but i'll recheck it....or perhaps start with close to zero as i can space the rack up to grain bumpsteer

[Tim.Wright]

You sure he wasn't quoting the bumpsteer in minutes? 15' toe out = 0.25deg and 5' toe in = 0.083deg. Even these numbers would be relatively high for a circuit car.

I'd start close to zero and only rely on it to "patch" certain problems that don't have another solution. Bump steer works with the delay of the body movement so it effectively gives another steering input on top of what the driver gives but with some delay.

In any case you will need to make the rack height adjustable to account for manufacturing tolerances otherwise you will never get the bumpsteer you want.

What are you using to analyse the kinematics? There are a few cheap options on the market these days.

[Jersey Tom]

as for the bumpsteer it was measured off a car which was set up by a suspension expert who used to work for brabham i think

Doesn't matter who it was or where they worked, if it doesn't make sense to you - don't do it. The racing world is rife with misinformation, misinterpretation, and "experts" who are clueless.

to be quite honest i've been messing about with this stuff for a few years and any information you find on it is clear as mud and contradicting

Exactly to my point above. "So-and-so says this." "It says to do this in a book I read." Blah blah blah. Question everything you read.

i had to just start making drawings and working out what was happening for myself with basic force diagrams, which is all well and good until you get into a real life scenario and realize that really its pointless as there are simply too many factors involved to factor them all in without some real world suspension design experience and a fancy computer program or 2.

Even with software and experience, you will never be able to perfect everything. That's true in engineering in general but especially in racing. Some ideal "optimized" car - no time for that or obsessing over minutiae. Compartmentalize and make sense of what you can, break it down and figure out which items are most important to spend time on and which are probably small. Prioritize, get done what you can get done, and get to the track.

[williammcc]

that may make sense minutes instead of degrees actually

i do question everything and try to figure how it works for myself but with limited experience sometimes what makes sense in theory with the limited understanding i have doesn't work in practice (or even in theory when someone explains the other factors) hence the reason i won't be building a mark 2 escort with this set up (they're just too expensive to cut up and get it wrong) a 206 is near enough the same wheel base and with scrap prices at the moment scrap cars would near give a shell away to get it out of the way so build it into it as a test bed once everything's dialed in convert it over to the mark 2 there's going to be a bit of error due to different rates of chassis flex, cog height and over all weight but it should be close enough to sort with the built in adjustment