Suspension designing from scratch

[dtype38]

Originally Posted by ubrben

My philosophy was something like the following:

- Long swing arm length at the front to keep the wheel upright during braking
- Large caster angle (we had 8 on my second car to get camber compensation in corners
- Shorter rear swing arm length (as much as half if you can package it) for camber compensation.

Ben. Can you run that one by us again in a tiny bit more detail please? On my own front engine RWD car my front swing arms are quite short because of the wide engine between them, but the rear ones are long because they are mounted on the diff. Is this bad?

I optimised my own front suspension by setting up the swing arms for zero camber change in static roll. I set up with only abot 1/2 deg of neg camber static so that is doesn't go too negative under braking, and use around 6 deg of camber to add a little more negative camber for turn-in and cornering. How would having longer arms help?

Originally Posted by ubrben

Keeping the wheel upright on the brakes is less important at the rear because they only generate 30% of the force

Isn't it more important to keep the wheels upright when they're lightly loaded. If you have 30% braking available, but only use 15% because your tyres are only 1/2 in contact, its quite a big sacrifice in braking and stability.


Sorry, not trying to hi-jack this one. Just very interested to learn more.

[Goran Malmberg]

Dtype 38, I fully understand your question, and I am not the one you ask here, but this is a very complex matter to say the least. Discussions like this mostley tend to be like there is a general answer. I hav been writing a book about A-arm geometry, that ONLY discuss A-arm geometry. It is obvious then that an answer in a few line is quite impossible. And as many people is making comments in a forum, some 200 pages is inf act nessesary if the discussion is not going to be a two person conversation.

However, the book starts of with a car that is totally neutral. I has the same track width as wheelbase and roll centre at ground level, no anttidive and no antisquat and finally no (antiroll) jacking. Infinite A.arms, so ther is no cambergain and no Sai and caster angle or scrub distance.
This car will roll as much as it pitch when subjected to the same G-fors in those directions.
Ok?
I the add the normal principles of a smaller TW then WB. What happens then? As the TW is shorter the car transfer more weight in rol than pitch an we got more roll deflection. Therfore we have to ad a ROLLBAR to the car to retain the numbers the same. And the rollbar rollstiffness should represent the change in TW. Then we are back to "normal" again.

I then keep adding normal car "buildups" like this to describe what happens. And in the ent the thing get so complicated that we will be awhere of the compleability of the situation.

Oh, excause my English as usual...
Goran Malmberg

[dtype38]

Your English is fine Goran. And yes, its far more complex than can be covered in a few lines of text. But one lives in hope

[GordonG]

DType

I think the confusion is that Goran is talking about *swing arm length* and you are talking about the length of your wishbones (A arms). Swing arm length is the theoretical length that swing arms would be (if the car had suspension similar to the rear of an early Beetle) to give the same camber change as the combination of the upper and lower wishbones in your double wishbone setup.

So, a long swing arm length indicates that there will be little camber change in roll/bump, but can be achieved with wishbones of any length.

I hope I've clarified things rather than muddying the waters.

G

[Goran Malmberg]

Originally Posted by GordonG

DType
I think the confusion is that Goran is talking about *swing arm length* G

We could say swing arm length, or instant centre length, as when A-arms is parallell there is no instant centre intersection of projection lines.
Then we have Rc at ground location.Ok, we kan have Rc at ground even with unparallel A-arms, but then other things starts to happen.

If the engine is in the way of lenghtening the A-arms, then moving the spindle in to the wheel is good. then we need a spindle with small Sai angles and great possitive offset wheels. Probabley even larger diameter wheels in order to be able to turn.

One problem with More "ZERO" everything is that camber goes positive under roll, So either we have to stiffen the roll rate, but better still, lower the CGH. The advantage is an easier to tune car, that is more predictable and that stays more neutral under all circumstances.Even braking and acceleration traction stays free from bad camber changes.

Well, I bet sooner or later somone com in with an objection here, as this might sound controversial to some extent. But we will see if I can handle this.
Goran Malmberg

[dtype38]

Ahaaaa.... well that's cleared that up!

Seriously though, I have no idea what the back end of a beetle looks like, but I think I get the drift Gordon. I think my design is trying to give swinging arms that effectively pivot on the centre line of the car in cornering. Unfortunately they're less successful at doing that under braking.

Goran, as you know I'm not a great fan of "from first principle" design. Too much like hard work and using brain power for my liking It's not that I think it isn't a valid approach, but I have so many built in limitations on what I can do with my actual car (I can't change the layout/chassis/wheel positions/etc/etc) that I need to concentrate on practical adjustments to what I have. In this case I'm looking at the pros and cons of dumping my Jaguar IRS back end and fitting a double wishbone system. It's a lot of work and would need to have some realisable benefits in terms of traction and cornering to make it worth while. Hence my interest in the whole subject.

[Goran Malmberg]

Originally Posted by dtype38

Goran, as you know I'm not a great fan of "from first principle" design. Too much like hard work and using brain power for my liking It's not that I think it isn't a valid approach, but I have so many built in limitations on what I can do with my actual car (I can't change the layout/chassis/wheel positions/etc/etc) that I need to concentrate on practical adjustments to what I have. In this case I'm looking at the pros and cons of dumping my Jaguar IRS back end and fitting a double wishbone system. It's a lot of work and would need to have some realisable benefits in terms of traction and cornering to make it worth while. Hence my interest in the whole subject.

As you already got the rear end geometry pretty fixed, my "theory talk" is more of just diskussing things, mixed up with the tread head line, "a clean sheet of paper design". I am right now very much in to matching front and rear geometry for a similar weight transfer at both axles, so in your case, I cant realy tell what to do with the front without knowing exactley what the rear look like. Of course I know the Jag axle but I never look that close to it.
Goran Malmberg

[TEAM78]

Firstly thanks for your input Goran. Its very interesting what you have written and have given me more to think about. Im using MSC ADAMS for modelling the vehicle and im doing front /rear suspension and steering
design for a FS car. Im aware that some people may argue that parallel wishbones dont place the RC on the ground but is doesnt exist. To me its not important really as its not a practical thing and most unlikely to use parallel wishbones.

I am right now very much in to matching front and rear geometry for a similar weight transfer at both axles[Quote]

what do you mean by this?
Does this refer to longitudinal and lateral weight transfer?
How do you determine where the weight distribution should be. should it always be 50/50? as this is prefered for maximising cornering speed.
If you have a 55/45 weight distribution biasing the rear. what do you use to balance this out??

[nopanon]

@TEAM78

Your desired weight distribution is the one that will make your car as quick as possible around a given circuit with the tires you have available

As far I know it works something like this:

If you first consider the case with identical tires on all four corners, you should get a neutrally steered car if you have a 50/50 weight distribution and equal roll stiffness front and rear. If you shift the cg towards the rear, the rear tires will be more heavily loaded, and thus due to load sensitivity can't produce an equally high friction coefficient as on the front axle.(at least not at the same slip angle) thus giving an oversteered car.

To asses this you can shift the roll stiffness distribution towards the front(change springs and or antirollbar). Then the front end will counteract more of the overturning moment due to cornering, you will get more load transfer in the front and thus due to load sensitivity less grip. Now you can again have a neutrally steered car, but with slightly less cornering power.

What you have gained though(for a rwd car) is more acceleration and breaking potential as we now have more weight on the driven wheels and will get a more even weight distribution during breaking. It is thus a compromise that will depend on your track and your tires. Having wider tires in the rear can make it less of a compromise shifting the weight backwards. A role of thumb is that your weight distribution should be the same as your tire width relation front and rear for a neutral car, but that really depends a lot on the construction of the tires. On a rwd car it is defenitely benifitial to have the weight shifted towards the rear. The obvious example is almost any formula car, as we all know they usually have more weight and wider tires in the rear.

What is the optimal tire/weight distribution combination for an fs car I don't know, but I guess it is quite close to 45/55 f/r, maybe a bit more difference. Get some tire data and start up matlab!

[Lukin]

I agree with what nopanon says, but have a few additions. The ideal weight distribution will change with each circuit.

The way I see it, it's a trade off between ideal distribution for grip (ie equal horizontal/vertical loads front and rear like you said) and the gain from having more to the rear (in terms of braking and acceleration).

There is a paper for a F1 car by Pat Symonds (among others) and for Suzuka the ideal weight distribution is 44/56 and for Barcelona it is 43.5/56.5.

I guess the best example would be in comparing say, Surfers Paradise to Barcelona. At Surfers you can sacrifice mid corner 'steady state' (I say that with caution) for being good under brakes and getting excellent drive so I would say more rearward and sacrifice some balance. At Barcelona however, I wouldn't fancy driving a car with bad balance.

That's how I see it anyway.

[dtype38]

Originally Posted by nopanon

To asses this you can shift the roll stiffness distribution towards the front(change springs and or antirollbar). Then the front end will counteract more of the overturning moment due to cornering, you will get more load transfer in the front and thus due to load sensitivity less grip. Now you can again have a neutrally steered car, but with slightly less cornering power.

Now that surprises me! Doesn't doing that put a lot of faith in the torsional rigidity of the chassis? Not having any in depth knowledge of the subject, I reasoned (incorrectly?) that to get a neutral car in cornering I should set the roll stiffness up at each end roughly in proportion to my estimate of the CoG height times the weight at that end. So for my front engine rwd car I (approx 55/45 f/r dist) I've got stiffer front suspension than at the rear rear. By your argument should I have it stiffer at the rear? Wouldn't that make it rather tail happy in the corners, not to mention traction issues?

Or am I misunderstanding you?

[nopanon]

I think you are understanding me correctly
First, to be able to accurately play with roll stiffness and other setup requires a reasonably stiff chassis. But having a soft chassis will only lessen the effect of a roll stiffness change, it will still be there.

Yes, by adding roll stiffness in the rear relative to front you will get less grip in the rear(but gain some in the front) in this simplified vehicle model in steady state cornering. If you don't, your car will in my opinion be understeered with your current set up if your tires front/rear are the same (In reality tires might go to excessive camber without for example an anti roll bar and you can in some situations gain grip by adding one, at least in my experience)

If it becomes tail happy or not should depend on how much grip you remove. Adding stiff anti roll bars can make your car very abrupt in it's behavior. I've no experience with fwd cars, but from what I've heard they are prone to understeer as the front tires should do pretty much it all. Thus, it is maybe not possible making a fwd car handle neutrally in all situations by the means of car setup. I'm not sure what you intend to do with the car, or how you want it to perform, but I would look for setup inspiration among other fwd cars with similar weight distribution that operate in similar conditions.

Hope this answers your questions

[Goran Malmberg]

Both nopanon and Lukin is on a similar track as my "zero" car. We want the same weight per area of rubber contact to the ground, and this can be achieved in a number of different ways depending on the sideeffects wanted. Tire area accordinly to weight distribution is on of the more neutral condition and rollbar according to track width to wheelbase relation is another one.
Camber is another factor that does the same as different loaded tires but within tha same contact patch. Slightley cambered wheel is creatin "weight transfer" with in the sam tire creating less µ number.

The reasom I say this obvious camber stuff is to put it in the same department as weight transfer.

Allright, the rollcentre! Roll centre has been disscussed upp and down and appear very confusing. The problem from my side is to explain my point of view in a few lines, with that said I am not claiming to be the one that is right, but I have a fairliy good ground for part of my theory.

Some statements...
With parallell A-arms the car rolls around a centre at ground level, vertically under cgh.

No matter the intersection of forcelines, (they might even be parallell without any intersection), the car rolls pretty much around and right under the cgh at a height where the forceline intersect the cgh vertical line.

With a more cambercompensative (more angeled shorter A-arms) the forcelines is changing very uneven from side to side. When calculating plus and minus jacking the outside wheel sees more load and therefore more + jacking. But in some cases the forceline of the unloaded wheel is raising so much that it despite is lower load become influencial in creating – jacking.

Jacking has two components, the jacking itself as the name imply, that lifts the car straight up.
This alter the cgh.
And, the unsprung force that hit the tire contact patch very brute, creating peak forces that brake tire grip loose.
Very bad situation that can not be adjusted via the suspension sytem as the force is strictley geometrical. Of course, the car can be balanced neutral by altering rollcentre but the higher the rollcentre the more sensitive to road obstructions the car will become.

Another thing is to calculate wheelrate. The scrub distance, SAI angle and upper A-arm angle
is a big factor for wheelrate. This phenomenon is not really brought up to daylight for some strange reason, but what about numbers like 50% reduction in Wr? The change is not even linear, so we got different motionratio depending on wheel level location. Which means that the car is getting a different behaviour depending on pitch mode. If the A-arm is parallel and the same lengh, we are having a perallellogram and then the same wheelrate independent from the mentioned factors.
I am not saying that we should use this “parallel equal lenght”A-arms, fare from it. But, the more extreme we go in the unparallel design the more important front and rear geometric balance will become. In other words we must see to such phenomenon as scrub and Sai even for the rear geometry.

I better stop here not to float away on the subject.
If anyone of you guys want to see a few images of my testing, give me a mail in privat and I should send it as an attachment. This is only for those already in this tread.
hemipanter@brevet.nu
Goran Malmberg

[nopanon]

Hi Göran!

Interesting post and I have a few questions:

On the subject of the pressure distribution within the tire contact path determining the tires overall friction coefficient I wonder if you know of any other factors involved here? For example FSAE tires seem relatively insensitive to camber within normal operating range for lateral force generation, while still quite sensitive to normal load. See this tire data from goodyear: http://www.racegoodyear.com/pdf/sae/20x65/page29.pdf

Regarding the jacking force discussion: The roll centre is still the force coupling point and if we can make it behave in an equally nice manner throughout suspension movement we shouldn't have more jacking related problems with a strongly camber compensating suspension compared to one with equal length parallell arms?

Regarding the influence of suspension parameters on the wheelrate I think you are on to something really important, hadn't thought of that.To fully asses this and see the influence on the ride rate I guess we would also have to know about the tires radial stiffness as a function of camber. Then there is also the possability to tune the wheelrates with your damperlinkage. A slightly progressive wheelrate is probably desired from a road holding point of view. When building my FS suspension I was not able to acheive constant wheelrates throughout suspension travel, but rather a 15-20% increase from min to max wheel position. I'll admit that I would have preferred to have an even smaller change if possible and still having the same suspension characteristics.

Cheers
Nopanon

[dtype38]

I'm following most of this, but could one of you please tell me what you mean by "wheel rate"?

Anyway, the waters seems to have reached a steady state of muddiness at the mo (for me anyway) with clarifications just about keeping up with complications. Soooo sorry to stir up the mud again, but can we discuss the effect on the suspension parameters of tyre profile and wall stiffness. I bring this up because on my car I have the same tyres all round at 225/60/R15. These are quite deep and soft walled. For my suspension design these tyres allows me tolerate quite adverse front camber under braking because they simply "squish" down onto the tarmac. I do find though that in cornering my turn in and front end grip is helped by using quite a lot of caster to add negative camber to my outer wheel (and reduce it on the inner wheel). I reason that this is because my soft sidewalls need camber assistance to stay fully attached at high lateral load. I suspect, however, that if I was running rubber bands with stiff sidewalls my car would be a pig. I'd probably lever the outer edges of the tyres off the ground under breaking and similarly the outer tyre during cornering. Sooooo... for the specific case at the start of this thread, don't we need to start with the characteristics of the rim/tyre combination before planning the suspension?

[Goran Malmberg]

Originally Posted by nopanon

Hi Göran!

1
Regarding the jacking force discussion: The roll centre is still the force coupling point and if we can make it behave in an equally nice manner
2
throughout suspension movement we shouldn't have more jacking related problems with a strongly camber compensating suspension compared to one with equal length parallell arms?
3
A slightly progressive wheelrate is probably desired from a road holding point of view. When building my FS suspension I was not able to acheive constant wheelrates throughout suspension travel, but rather a 15-20% increase from min to max wheel position. I'll admit that I would have preferred to have an even smaller change if possible and still having the same suspension characteristics.

Cheers
Nopanon

I am leaving the tire (pressure) Q for the moment, I didnt get the pdf file to open up.
1
The thing with the rollcentre here is that it is the point at which the car roll. Sounds like the cake on the cake. But, the forceline for the loaded wheel is the one that governs this roll action the most. The inside wheel may get a steeper forceline, say 5 degree for the loaded wheel and 6 for the inside wheel if the static angle is 4 degree at 1 g of cornering force. This means that the forceline of the inside itersect the middle load line at a higher location. From that standpoint we should have two rollcentre which is impossible. The load governs the actual Rc location. When calculating we simpley use the angle of the forceline times the actual load for the wheel in question. So we got four calculations here. We acually dont use the Rc in the calulation. We can then just withdraw those forces from the total Cgh arm from the ground, torque.
Rolcetre serves another purpose. we can draw a roll axis line from front to rear. If then the weight distribution is more to the rear the 90 degree angle from the rollaxis up to cgh becomes shorter than if we have had a 50-50 distribution. We got different antiroll numbers depnding on weight distribution then.
Am I dizzy?
2
It is a bit difficult to designe a cambercompensative solution that fills the bill, but of course a good layout make things better. And as I said, even the wheelrate enters the picture.
3
The thing is that the upper A-arm angle is creating a decrease in wheelrate, so it get softer and softer. This can be compensated by coilover mountings or rocker ratio in a pushrod system, but the thing is that one gets involved in a number of tricky situations.

However, we shouldnt overdramatice the situation. We are discussin this on a theoretical basis to understand how things works. We can go pretty fast with a good cambecomp-rsetup.

Goran Malmberg

[Goran Malmberg]

Wheel rate is the springrate at the wheel. How manny N it takes to move the wheel 1mm in bump direction.
Goran Malmberg

[Lukin]

To add my thoughts on the roll stiffness distribution side of the discussion.

I think a lot of advice (not just here) stems from rules of thumbs. Not that they never work, but are often car specific. Us engineers often find a number that works and latch onto it, sometimes even if it has no physical explanation. For example, from data I like the ratio of front roll gradient (deg/g) to rear roll gradient. It's not a physical number, but the car can be good in one range and rubbish in another.

Anyway, one such rule of thumb is to have the front roll percentage 5% higher than the weight distribution. So for your car dtype38, that would be about 60% of your roll stiffness from the front. But hey, that might not work, you might have to find your own number for countless hours on the computer and at the track.

Do you have a mechanical balance sheet? I have one from Claude Rouelle I could e-mail; anyone who wants it say I and include your e-mail address. It's good, but pretty simple in that it doesnt take into account rising rate and ride height effects, but it's a start. A guy from work made a great one for our car and it works so damn well.

After you know your distribution (or before, depending on how you work) you also need to know how much actual roll resistance do you want want and where do you want it from (ie springs or bars)?

PS I don't know the answer to this one either, but I don't think many do!

As pointed out before roll resistance is good as it reduces the roll angle to reduce any unwanted camber and bump changes due to high suspension travel (did someone say Porsche!?) as well as the apparent sharpness increase. After all, it takes less time to get to 1 degree roll than 4. Der.

Not that I'm saying anything most people don't know. But stiffer springs kill your traction and roll bars become a nightmare in one wheel bumps. So your magic number (as Claude puts it) will cary with the track as will the contribution from each element and the overall roll stiffness.

I don't have too much to add on the RC/kinematics side of things, still learning about that. It's most definately an off season Susprog/RCVD project for me.

[Goran Malmberg]

Originally Posted by dtype38

These are quite deep and soft walled. For my suspension design these tyres allows me tolerate quite adverse front camber under braking because they simply "squish" down onto the tarmac. I do find though that in cornering my turn in and front end grip is helped by using quite a lot of caster to add negative camber to my outer wheel (and reduce it on the inner wheel). I reason that this is because my soft sidewalls need camber assistance to stay fully attached at high lateral load. I suspect, however, that if I was running rubber bands with stiff sidewalls my car would be a pig. I'd probably lever the outer edges of the tyres off the ground under breaking and similarly the outer tyre during cornering. Sooooo... for the specific case at the start of this thread, don't we need to start with the characteristics of the rim/tyre combination before planning the suspension?

Sorry, I am probabley the one mudding the water to much. I also see I am turning the words around which dosent help.

Anyway, We have a difference between front and sideway traction. in braking directions the tire stays round so the actual pitch angle dosent matter. And the contact patch is wider than long. We could read about how shape affect the situation in a book of Paul Haney.

Back your situation, what Sai angle do you have? Might it be 7 dgr (or more)?

Tires and rims is what governs a LOT concerning spring setting and even geometry. We can not draw geometry lines without knowledge about these things.


I would like to throw an eye on those "balance sheet papers". My email is hemipanter@brevet.nu

Goran Malmberg

[ss_collins]

The ideal weight distribution will change with each circuit.

And each driver. The optimum engineering setup or design may not be the most driver friendly, the driver must have confidence and drivability built in - sounds obvious but its amazing how some commercially available racecars I have driven do not give you this. A flawed setup/design with a happy driver is probably going to be quicker than a optimal setup/design with an unhappy driver

[ss_collins]

Thoughts/questions without answers

Have you worked out the rate of deformation and possibly spring rate of the sidewalls the tyres you plan to use?


From out bin marked loony ideas and Hydrostic drive comes:
And to really screw this up and pour a whole rally stages worth of mud into the water, have you looked at the two slightly baffling but effective Formula Vee suspension types, Z-Bar (roll inducer) and Zero Roll. I have some pics of the two types that I will put on the Racecar website in the next week or so.

Also look at the Deep Sanderson - Lawrence Link.

[Goran Malmberg]

Originally Posted by ss_collins

The ideal weight distribution will change with each circuit.

And each driver. The optimum engineering setup or design may not be the most driver friendly, the driver must have confidence and drivability built in - sounds obvious but its amazing how some commercially available racecars I have driven do not give you this. A flawed setup/design with a happy driver is probably going to be quicker than a optimal setup/design with an unhappy driver

ABSOLUTLEY right so. Ther must be a communication with the engineer and the driver. Both must have some understanding from one anothers side to comunicate.
We can look at ads for many sports and race cars where we can read that the car is objetively amazingly fast. While I want to see a car the can be suited to specific demands. No wonder many people get the wrong idea...
Goran Malmberg

[ubrben]

Originally Posted by ss_collins

Thoughts/questions without answers

Have you worked out the rate of deformation and possibly spring rate of the sidewalls the tyres you plan to use?


From out bin marked loony ideas and Hydrostic drive comes:
And to really screw this up and pour a whole rally stages worth of mud into the water, have you looked at the two slightly baffling but effective Formula Vee suspension types, Z-Bar (roll inducer) and Zero Roll. I have some pics of the two types that I will put on the Racecar website in the next week or so.

Also look at the Deep Sanderson - Lawrence Link.

To call a Z-bar a roll inducer is a little odd. It stiffens the vertical mode of the wheel pair and therefore it depends on whether you've connected a front and rear wheel or a left and right wheel. If you had longitudinal Z-bars between a left front/left rear and right front/right rear wheels you'd increase your roll stiffness (but not your warp stiffness).

A Z-bar between left and right wheels at the same end of the car is another name for a third spring as seen in just about every modern racecar.

Zero roll is a crutch for the fundamentally crap swing arm rear suspension of the Beetle/FVee car and shouldn't really be an aim for any other race car. Very stiff roll mode would be desirable though if it came with a soft warp mode, that's what the Crueat system described recently in RE does.

Care to explain what use the Lawrence Link system is?

Ben

[ss_collins]

It was used on the Deep Sandersons in the 1960's and is a fascinating concept, possibly obselete - possibly not, we featured it in our May issue V15N5. It proved to be far more effective than the more conventional suspension on the Lotus 18.

http://www.lawrence-tune.co.uk/lawrence_link_1.html

"Ive never liked this business of anti roll bars lifting the inside wheel to make the outside wheels work properly. I call that building a four wheel motorbike. As you get into a corner you are only running on two wheels - Somebody once said an anti roll bar is quick way to prop up an otherwise inadequate suspension"

"a low roll centre that was completely fixed and would not migrate no matter what the suspension did"

[Goran Malmberg]

Originally Posted by ss_collins

"Ive never liked this business of anti roll bars lifting the inside wheel to make the outside wheels work properly. I call that building a four wheel motorbike. As you get into a corner you are only running on two wheels - Somebody once said an anti roll bar is quick way to prop up an otherwise inadequate suspension"

"a low roll centre that was completely fixed and would not migrate no matter what the suspension did"

When the car roll, the middle of the rollbar is not in motion, and could therfore be concidered an independent spring for each wheel in this particular roll situation. But of course, it has sideffects.
Goran Malmberg