72 replies to this topic

[Wolf]

Let's assume one is one is about to design a suspension for a light sportscar and is free to do so with very little constraints. Further let us assume that double wishbone susbension (front and rear) will be considered a first choice, and that it will have some suspension travel (just to be safe that it's not confused with suspension of a modern race car )...

I know I'm asking much but I would be very obliged if you kind people were to help me with 'few' things.

1 Geometry
What are primary considerations when choosing a geometry (one is given a free choice for suspension pick-up pionts, length of wishbones, & all)? What are secondary? And what are 'no-no's? I know it is a big question and a lot of ground needs to be covered...

2 Adjustability
Due to possible multi-purpose nature of the car (road, track, gravel) and to allow to compensate for design flaws the suspension would need to be maximally adjustable (within reasonable amount of complexity). What are best ways to make it as versatile and tuneable, and what factors should be given a priority (that will influence handling the most)?

3 Dimensioning
Now, this is a tricky question- how does one dimension various suspension components? First thing that pops into my mind is that at max. wheel load in steady state stress should be lower than endurance limit (with safety factored in). Seems simple enough, but how does one account for dynamic and impact loads? I would assume they would be related to strength rather than durability. Is there a factor of some kind, like max load that suspension would have to cope with, without breaking, is 5x greater than static load, ore something like that?

I'm sure if we get things started here, I may have a lot more questions as we progress (oops, now this isn't the best way to encourage you...)., but I think this is a handful to start with.

Thanks in advance.

P.S. I reckon that suspension that would fit my generic description would most likely be something like '60ies GP cars (even 'inboard' springs are acceptable)- does anyone have nice detailed drawings of something like that.

[phantom II]

I wish I wrote a post like this post before I started my roaster project. Every part and system was a challenge.


According to Gordon Murray's book. Each car must have 8 design objectives that must be stated and defined. Road, track and gravel is a stretch. Buy a rally car.
What budget, Production or one off,What race series, passengers,what power to weight, What weight and rigidity,Parts sourcing, What engine transmission, Packaging, etc, etc, etc.???
If it breaks, redesign. My car on 'What do Members Drive' thread is designed for street and SCCA solo one and autocross events. LS2 405 HP V8, 1800lbs, push rod coil/overs,BMW M3 diff and hubs. USDOT crash requirements. 50 state smog. Weight regidity, low CG, Power to weight objectives. No exotic materials. Mild steel tubular space frame and fiberglas. Speed limiter 40mph. 52% rear weight bias. 255 rear tires.

Do this before you proceed.

Originally posted by Wolf
Let's assume one is one is about to design a suspension for a light sportscar and is free to do so with very little constraints. Further let us assume that double wishbone susbension (front and rear) will be considered a first choice, and that it will have some suspension travel (just to be safe that it's not confused with suspension of a modern race car )...

I know I'm asking much but I would be very obliged if you kind people were to help me with 'few' things.

1 Geometry
What are primary considerations when choosing a geometry (one is given a free choice for suspension pick-up pionts, length of wishbones, & all)? What are secondary? And what are 'no-no's? I know it is a big question and a lot of ground needs to be covered...

2 Adjustability
Due to possible multi-purpose nature of the car (road, track, gravel) and to allow to compensate for design flaws the suspension would need to be maximally adjustable (within reasonable amount of complexity). What are best ways to make it as versatile and tuneable, and what factors should be given a priority (that will influence handling the most)?

3 Dimensioning
Now, this is a tricky question- how does one dimension various suspension components? First thing that pops into my mind is that at max. wheel load in steady state stress should be lower than endurance limit (with safety factored in). Seems simple enough, but how does one account for dynamic and impact loads? I would assume they would be related to strength rather than durability. Is there a factor of some kind, like max load that suspension would have to cope with, without breaking, is 5x greater than static load, ore something like that?

I'm sure if we get things started here, I may have a lot more questions as we progress (oops, now this isn't the best way to encourage you...)., but I think this is a handful to start with.

Thanks in advance.

P.S. I reckon that suspension that would fit my generic description would most likely be something like '60ies GP cars (even 'inboard' springs are acceptable)- does anyone have nice detailed drawings of something like that.

[Greg Locock]

Good point phantom.

Ok here's my list:

lower arm should be horizontal at the design position

arms should be longer rather than shorter

excessive use of fancy angles at the inboard ends (antisquat, semi trailing etc) is for crazy people. It is a mechanism. Designing a mechanism that tries to lock up is silly.

First find your steering rack

next find your spindles/knuckles.

next find your halfshafts

Once you have that lot then you are left with remarkably few choices. basically, decide what camber gain you want, and your roll centre height, aim at zero bump steer for all intents and purposes, and don't forget to package your sta bar.


Strength of parts

In the old days we used 3 -2-1 for circuit cars and prototypes. That is 3g vertically, 2g longitudinally, 1g laterally.

I'd be inclined to increase them a fair bit for a road or rally car. something like 5-4-2 will cope with potholes and the occasional kerb strike.

Those are still rather light compared with a production car.


Software

See http://www.eng-tips....aqs.cfm?fid=768

[Wolf]

Thanks guys, very helpful replies.

Phantom- yes, gravel is a bit of stretch... Actually, I'm toying with the idea of designing a cyclecar (light 2-seater with motorcycle engine)*. In light of that (even if I lay my hands on 1000cc engine I'll be stuck with no more than 110 Nm torque) I'll want very light car and as little drag as possible. Basically it will be an open-wheel two seater looking very much like '61-'65 GP cars, but with wet lay-up GRP monocoque.

* TBH, I started out with the idea of using this car as a test-mule for *my* suspension (very simple and novel idea) just to prove wrong most people who looked at it and refused to believe it would work as I say it will, but I'm too young to die () so decided to be conventional about suspension. [mode=mad scientist]Even though it *will* work as I say it will... [/mode] And in light of survival instincts, I may even dispense with originally planned fuel bags around driver and passenger a-la Lotus 25.

[Wolf]

Speaking of oddities*, I don't think I've seen it ever done (probably with a good reason )- but what would happen if one was to use a torsion spring to mount one of the wishbones and use that instead of conventional coil spring? And I'm quite sure one could also provide damping in similar fashion (maybe adjustable friction shocks that AU used on their GP cars)...

* sorry to go O.T. in my own thread, but this is one of the bestest sites I've ever seen- Museum of retro technology.

[hydra]

Greg,
I've seen you mention this several times before several times already, and I've always wanted to ask - why is it so important to keep the lower arm(s?) horizontal at the design position? You mentioned something about steering effects once but I couldn't figure out why... Also close to perfectly horizontal can you get away with? Is this important for the rear as well?


I would imagine that torsion bars would be better than coils due to unsprung weight, packaging, and stiction issues, but that's my $0.02 worth

[Greg Locock]

Wolf - look at a Ranger (Courier) or Explorer, I'm pretty sure they all use torsion bar +double wishbone.

For a race car it might be a nice idea as it is dead easy to change ride heights, but changing springs is a PITA. Use a lever type shock absorber from an MGB (or a linkage) and you'll have a very low bonnet line and loads of room for AWD.

Hydra - I don't remember the steering argument. Lateral compliance is one reason, if the arm is inclined then you get some jacking which is softer than the bushes in compression. Horizontal arms are good because the kinematics curve tend to be smoothest when the arms are horizontal - for that reason I am no great fan of short arms, which tend to have very small sweet spots. This is all predicated on linearity being the most important thing, which over the years not everyone would have agreed with. If that is all sensible then it is even more important at the rear. It is much easier to upset the driver with the rear suspension than the front, stability wise.

[Wolf]

Greg, I intended to use springs like this on my aforementioned weird suspension- would they be easier to replace*? The idea was that it would (perchance) be easier to change flexing plates than to carry set of different stiffness bars.



* I just noticed it throws the easy ride height adjusment out the window, but one could sock in wishbone via Z-profil into inner bar and that would sort it (methinks)

EDIT- and of course this spring would be much shorter to provide reasonable travel than conventional bar.

[Engineguy]

Like bump steer, bump induced track width change is bad.


When the lower A-arms' nominal position is horizontal, the lateral displacement of the outer ball joint as a function of vertical travel is minimal.


The further the arm is from horizontal, the more lateral movement per unit of vertical travel... so if you start from non-horizontal, the lateral element can get significant. Although we can wipe this out at the contact patch by manipulating the upper A-arm arc, it may contradict what we'd otherwise prefer to do to the camber.


Of course the longer the lower A-arm the less lateral change mess we have to deal with.


Long arms, where possible, save us from fighting a lot of unwanted geometrical drama.

[Deepak]

Fantastic Engineguy...

Makes things a lot easier to understand.

[hydra]

OK I get it now... You didn't mention how much lower arm inclination you can get away with. Does +/- 5deg sound reasonable?


Some other good guidelines when designing a sports car suspension from scratch include the following:

have the lower ball joints as low to the ground as possible (I think ~120mm is the legal limit for a road car) to make the instant center as low as possible to minimize tyre scrub in jounce. The outer LBJ should also be as outboard as possible.

lower arm length should be around 6-7 times the peak transient jounce deflection in roll at max lat accel. This is a function of the roll gradient (1.8-2.5 deg/g) and overshoot (around 25%). So typically for sports cars this ends up being 60-80mm

Upper arm inclination should be set according to acheive an initial camber change rate of around ~1deg/inch. Does that sound reasonable?
camber change rate in deg/in = arctan (1/fvsa)

Upper arm length is typically around 0.6-0.7 x that of the lower arm, it determines the rate of change of the camber curve.

At least one of the wishbone "legs" should be splayed horizontally in top view to maximize lateral stiffness

Geometric roll center height, isn't that important BY ITSELF in my opinion. Minimize your instant center height and set your front view swing arm to give you the camber change rate you want and live with whatever RCH you end up with. Fortunately, this usually ends up being pretty low, but I don't think it matters much whether its 50mm off the ground or 75mm off the ground...

As Greg said, one is better off dispensing with the "anti's" as they cause the suspension to lock up causing nasty unpredictable handling characteristics.

Personally, I think KPI should be minimized (have the UBJ as far out as possible and mount the caliper horizontally), and scrub radius should be around 25-30mm, although I suspect its importance is secondary.

I still think that torsion bars are superior to coil springs in almost every way. They're lighter overall, easier to package, allow easier ride height adjustment, have less unsprung weight, and (should) give less stiction in the suspension. They also allow you to have a slightly higher motion ratio if you're using a conventional damper. Ideally though, I think they should be coupled with some sort of rotary damper a la Sachs SRS (Ferrari F2004) or Suzuki's TL1000R bike. Why don't we see more of these around?



I don't mean to hi-jack this thread, but I've got an academic question or two for all you suspension gurus out there...

What would result from running negative scrub radii and/or negative KPI using virtual ball joints? The way I see it KPI doesn't matter if its positive or negative, it should still have the same effect ... Assume we're still talking about a lightweight RWD sports car.

Also, for a given total mechanical trail , what's the difference, if anything, between trading caster angle for side view steering axis offset or vice versa?

[soubriquet]

Originally posted by Wolf
Speaking of oddities*, I don't think I've seen it ever done (probably with a good reason )- but what would happen if one was to use a torsion spring to mount one of the wishbones and use that instead of conventional coil spring?

In this case you would have something that looked very much like a 115 series Alfa. That is, Alfetta, Giulietta (nuova), Milano/75 or GTV.

[soubriquet]

Here:

[Greg Locock]

Wolf. that's a nice drawing, but what is it of? a sort of torsional leaf spring? Have you done the sums?

Hydra effect of neg scrub using virtual ball joints is remarkably dull. Neg scrub's main virtue is that under split mu conditions the steering wheel torque from traction/braking forces is reduced. Our lot have a theory that it helps with straight line stability on rough surfaces (disturbance rejection). I shrug and give it to them. I believe the Golf Mk I (re)introduced it in order to improve brake balance.

I've never tried negative KPI, so far as I'm concerned one virtual ball joint is one too many, going to two would be four times as much work. Having said that on a race car you might enjoy being able to put the virtual king pin exactly where you want it, but you'd be breaking new ground (well, I'm sure people have tried negative KPI, just not racers and not since 1935 on a production car .... which will bring a hail of examples (good)).

I think if you use the same material and design for minimum weight then a torsion bar and a coil spring should weigh exactly the same.

I've kind of gone round the loop on RCH. Its effect is subtle, there is no two ways about that - a 25 mm change doesn't seem to me to be subjectively important. There again I am a terrible evaluator of subjective steering and handling, as I've never had to do it seriously.

Castor angle is a feedback obfuscator and is bad. Mechanical trail is a feedback obfuscator and is bad. Some of each is a necessary evil. Hmm, that wasn't much help! It wouldn't surprise me to learn that SVLO is important in its own right, but I can't rememebr any theories about it off hand that are applicable to race or sports cars. A lot of my stuff is to do with very low speed and parking efforts.

[bobqzzi]

Originally posted by Wolf
Speaking of oddities*, I don't think I've seen it ever done (probably with a good reason )- but what would happen if one was to use a torsion spring to mount one of the wishbones and use that instead of conventional coil spring? And I'm quite sure one could also provide damping in similar fashion (maybe adjustable friction shocks that AU used on their GP cars)...

* sorry to go O.T. in my own thread, but this is one of the bestest sites I've ever seen- Museum of retro technology.

Mnay cars have had torsion bars mounted to the lower wishbone- 1960s dodges, many Chevy pick-up trucks and the Porsche 911 come to mind. It has some attration from a packaging perspective, but offers some diffculty with mounting and stiction and bushing wear. Changing springs can be a bit time consuming, and the availabilty of good quailty torsion bars in various rates is spotty at best.

[hydra]

Great stuff Greg...

So you don't particularly care for having a virtual (upper) ball joint? I can see how its advantages are minimal on a sports car, but it gives you that little extra design freedom..

The way I visualize it, 5 degrees of negative KPI should have the exact same effect as 5 degrees of regular KPI, when it comes to self-centering and steering camber loss, right?

A torsion bar should be slightly lighter than an equivalent coil spring, due to the d/D effect (basically the higher this ratio for a given spring rate the lower its natural frequency/structural efficiency), but the difference isn't THAT big. The main weight savings from going to torsion bars is in unsprung weight

So caster angle in and of itself is a bad thing? I was under the impression that the opposite was true (especially if you want to minimize KPI)

Also, how much camber gain do you think should be designed in to a suspension like this? I know that tire data is everything here, but a ballpark figure would be nice...

[imaginesix]

Wolf:
Just want you to realise that if you support the car using a torsion spring attached to a wishbone, the wishbone will have the be very much stronger in order to transfer the weight of the car to the wheel hub. I don't see that there are any weight / packaging advantages to this layout.

Greg:
In your list of steps, the first item you recommend locating is the steering rack. Is that due to packaging concerns of a front-engine car? Would you still make this your first step if there was plenty of room up front to loacate the steering wherever it was needed?
I always thought most desireable steering geometries could be acheived with many combinations of steering rack/steering arm/steering knuckle geometry, which would make the steering the last design consideration.

[Wolf]

Imaginesix- I thought the torsion spring is best to be loaded only by torque, and all other loads in this case would be transfered through bushings. Even older suspensions (Porsche's, that I know of, and other German cars) had torsion springs running through a tube that was carrying structural loads. (My uneducated™ guess about the steering rack was because of bump steer- this way one could make sure it's minimal/nonexistant)

Greg- yes, a leaf spring of a sort, but I haven't done any calculations.

[Greg Locock]

I'm far too hungover.
Hydra
Lazily I'd assumed that the effect on unsprung mass is equivalent... I agree, wow, that is worth having, a typical coil spring is say 4 kg, half is unsprung, and the m/r is say .7 so that's about 1 kg of unsprung mass. OK, not worth killing for, but probably the easiest 1kg in the suspension. Not that I've seen anything that proves that a low unsprung mass is any great improvement. The trick would be designing the spring abutment so that the total system weight is no heavier than the conventional coil spring system.


Late edit: I can't acually figure out the 3d geometry at the moment, I think the following is correct!

I think castor is a necessary evil, for returnability,straight tracking and damping. For a road car it gives you a camberangle /toeangle gain, which frankly we don't need. It stands the inner wheel up, which is crazy, while increasing the negative camber on the outer wheel, which is good. Having said that the castor on 'my' suspensions is 7 degrees or so which is far too much for a race car I think. Castor angle introduces an error signal into the steering whereby the tie rod can 'see' changes in contact patch vertical force. I'd love to build a few different castors and see what it really does, the Ford Suspension Design Bible (written by John Miles) says don't screw with it, which merely whets my interest.

Is the main effect of KPI that it gives you symmetric camberangle /toeangle gain? that is, it stands both tires up? that's bad for the outer wheel. It again provides an error signal like castor. To be honest in my suspensions KPI is an outcome, not a key design target, but if you have any insights that'd be great. The other effect is returnability, where you are jacking the car up as the wheels turn. Interesting, I'd assumed that castor was more important, but that tends to fight the roll stiffness of the car, not the weight.neg kpi. hmmmm. Sorry my brain can't handle that today.

In my opinion it is very hard to talk about KPI without considering scrub, trail and castor at the same time.

For a race car wouldn't you want 100% roll compensation, ie if the car rolls by 2 degrees wouldn't you want at least 2 degrees more neg camber on the outside wheel? And a bit more for luck? I'd agree that is an awful lot, for a road car. We couldn't use that due to the tire wear implications for different loads, I think, and rough road stabilty would suffer a bit.

Imaginesix, I didn't mean the rack location, although even on a front engined rear wheel drive car the packaging space is limited, I meant the actual hardware, ie the length over the ball joints, rack travel, and the input shaft orientation.

[phantom II]

Wolf:

----------------------------------------------------------------------------------------------------------------------------------
Phantom- yes, gravel is a bit of stretch... Actually, I'm toying with the idea of designing a cyclecar (light 2-seater with motorcycle engine)*. In light of that (even if I lay my hands on 1000cc engine I'll be stuck with no more than 110 Nm torque) I'll want very light car and as little drag as possible. Basically it will be an open-wheel two seater looking very much like '61-'65 GP cars, but with wet lay-up GRP monocoque.
------------------------------------------------------------------------------------------------------------------------------

What speeds are you looking at? Is it just s fun car? What materials will you use? What power to weight? Since my events rarely exceed 120 mph, drag is not a primary consideration. You can't have low drag if the wheels are sticking out in the wind. My drag Coef. is .75. Nasty nasty. But I can bleed brakes quickly at the event.

Greg:
--------------------------------------------------------------------------------------------------------------------------------
Ok here's my list:

lower arm should be horizontal at the design position

arms should be longer rather than shorter

excessive use of fancy angles at the inboard ends (antisquat, semi trailing etc) is for crazy people. It is a mechanism. Designing a mechanism that tries to lock up is silly.

First find your steering rack

next find your spindles/knuckles.

next find your halfshafts

-------------------------------------------------------------------------------------------------------------------------------


Yes, I agree with the above. I use a VW Golf II rack. 2 turns lock to lock and is always available. I used Corvette wheel bearings hubs before but now I use BMW 3 series. Vette rotors bolt straight onto them. They ar lighter than the Vette units.
I make my own al uprights and I use Heims. M3 or 3 series diffs are light and strong. Can handle 600hp and there are plenty at the junk yards. Half shafts also.
Another important thing is cooling. How long are the events you plan on running? Air must come in fast, slow down over the radiator and exit fast paying careful attention to things that may be in the way of flow such as alternators etc. I use 405 Hp 50% of the time in very hot weather. Wolf will have to look for suitable stuff for his project.

Greg:

-------------------------------------------------------------------------------------------------------------------------
Once you have that lot then you are left with remarkably few choices. basically, decide what camber gain you want, and your roll center height, aim at zero bump steer for all intents and purposes, and don't forget to package your sta bar.

--------------------------------------------------------------------------------------------------------------------------------
Next and most important are tires. Everything you do is determined by tire properties. The lower the aspect ratio the better for on track pressure adjustment to suit conditions in low speed events. The wheel is ligher also. I load the tire on my press against an angled plate to simulate side loads and observe deflection and contact patch at various tire pressures. I then cut a section out of the tire and mount it to the rim so that I can change the suspension to optimize the contact patch. The chassis has no springs at this time. Castor is determined by turn in contact patch. The high profile tire will cause more jacking because the LBJ is higher and the trail is longer but I will turn in quicker than a Lotus Elise. Low roll centers and low Cof G eliminates ARBs. Weight, remember? Binding also.
Zero Ackerman and dive and squat geometry because of unwanted coupling. I will use anti squat on a live rear axle because it is incidental in roll understeer geometry with a 3 link set up.. I will beat a Lotus Elise in tight spots because I can kick the back out easy with the gas pedal because of the engine being in the front. The Lotus will push into the bushes. Push, push in the bush..
I have one exhaust down the rhs and one cat. because of corner weights and total mass considerations. I have an upper and lower gas tank. The lower one is plastic and is molded under the pax seat. The upper one is used just to get to the event. No PS, no P brakes, abs etc because of weight. The battery goes to the back and close to the ground on the right also. Packaging all this is a challenge. A Viper, Vette, Turbo Porsche will beat me above 140MPH but who cares. That is why I say you must define your design objectives. Maybe this info will help. Corner weights must be the same left and right. What screws that up in my case is a weight gain of 25lbs this year. I got to change my beer.


hydra:
-------------------------------------------------------------------------------------------------------------------------
OK I get it now... You didn't mention how much lower arm inclination you can get away with. Does +/- 5deg sound reasonable?

--------------------------------------------------------------------------------------------------------------------------

Why would you want it other than horizontal?

Hydra:
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Some other good guidelines when designing a sports car suspension from scratch include the following:

Have the lower ball joints as low to the ground as possible (I think ~120mm is the legal limit for a road car) to make the instant center as low as possible to minimize tyre scrub in jounce. The outer LBJ should also be as outboard as possible.
-----------------------------------------------------------------------------------------------------------------------
It also reduces or eliminates jacking with high castor angles.

Hydra:
------------------------------------------------------------------------------------------------------------------------
lower arm length should be around 6-7 times the peak transient jounce deflection in roll at max lat accel. This is a function of the roll gradient (1.8-2.5 deg/g) and overshoot (around 25%). So typically for sports cars this ends up being 60-80mm

Upper arm inclination should be set according to achieve an initial camber change rate of around ~1deg/inch. Does that sound reasonable?
camber change rate in deg/in = arctan (1/fvsa) Upper arm length is typically around 0.6-0.7 x that of the lower arm, it determines the rate of change of the camber curve.

-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

In my springless chassis jig above I roll the car and lock it, then with the Heims, I determine the length of the upper A arms according to the contact patch. Hands on, old school yet effective...

Hydra:
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
At least one of the wishbone "legs" should be splayed horizontally in top view to maximize lateral stiffness
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
The lower the roll center, the loads are taken up in the lower A arm before the spring.


Hydra:
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Geometric roll center height, isn't that important BY ITSELF in my opinion. Minimize your instant center height and set your front view swing arm to give you the camber change rate you want and live with whatever RCH you end up with. Fortunately, this usually ends up being pretty low, but I don't think it matters much whether its 50mm off the ground or 75mm off the ground...

As Greg said, one is better off dispensing with the "anti's" as they cause the suspension to lock up causing nasty unpredictable handling characteristics.

-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Right. Plus it is easier to diagnose and isolate handling problems and therefore corrections..

Hydra:
------------------------------------------------------------------------------------------------------------------------------------
I think KPI should be minimized (have the UBJ as far out as possible and mount the caliper horizontally), and scrub radius should be around 25-30mm, although I suspect its importance is secondary.

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

A another thing you have to purchase are wheels. Depends on the offset to achieve zero scrub. Zero scrub if you have o ABS.

Hydra:

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------


Hydra:
-----------------------------------------------------------------------------------------------------------------
I don't mean to hi-jack this thread, but I've got an academic question or two for all you suspension gurus out there...

What would result from running negative scrub radii and/or negative KPI using virtual ball joints? The way I see it KPI doesn't matter if its positive or negative, it should still have the same effect ... Assume we're still talking about a lightweight RWD sports car.

Also, for a given total mechanical trail , what's the difference, if anything, between trading caster angle for side view steering axis offset or vice versa?
----------------------------------------------------------------------------------------------------------------------
Ehrr, Greg.....?

[hydra]

hydra:
-------------------------------------------------------------------------------------------------------------------------
OK I get it now... You didn't mention how much lower arm inclination you can get away with. Does +/- 5deg sound reasonable?

--------------------------------------------------------------------------------------------------------------------------

Why would you want it other than horizontal?

I figured it would be very difficult, if not impossible, to ensure that the lower arm is EXACTLY horizontal, especially under dynamic conditions. So there has to be some margin for error... Furthermore, if we are allowed to tilt the lower control arm a little, we can lower the instant center even more, giving us a little more design freedom...


Also, it would be nice to hear your thoughts on camber gain and roll gain values. Remember, we're not talking about an all-out race car here, just a nice high-po sports car

[phantom II]

If not horizontal it should be higher at the wheel than the chassis like the upper A arm but only to limit roll center gain.. The camber gain should come from the upper A arm. This also lessons the effect of jacking due to castor on turn in. I don't think jacking is a bad thing because it loads the opposite rear which helps late braking without ABS and shoves out the bottom of the wheel out on turn in.
If you did not have to purchase 'off the shelf' parts, then you can get onto your CAD and have carte blanche with the design and these values can be anything you want including spring rates. It is a black art, especially if you use OEM parts. The spring rate is not what you calculate it to be anyway in my experience. This is backyard stuff for me as well as entertainment. For proper information, consult Greg or RDV. Ben seems to be learning a lot these days.

You get yourself a wheel that is readily available. BMW after market wheels have rather large negative offsets because of their Mc Pherson struts and their inherently poor camber gain. Since I use BMW or Vette hubs, I don't have much choice of wheel offsets so the design is compromised from that standpoint.

Anyway, once the wheel and tire are selected, you begin with the contact patch and work backwards. BJI is determined by the following: How wide your trailer is and aesthetics. Place the wheels were you think they look cool. The length of the lower control arm is as long as possible from the frame to the ball joint which should be as far inside the rim as you can get it with rotor and caliper mounted to the upright. The upright is fabricated at this point with BMW rear hubs for both front and rear suspensions which share geometry and parts because it is easier.



No spindles please. The bearing is flush with the upright so what ever the angle of the upright will also be the camber. You will need some pretty deep wheels for zero scrub. No ABS and a 6 liter V8 engine without power steering is the reason why. Wolf's car will weigh at least 1000lbs less than mine so his considerations are different but contact patch still rules. Scrub may be different.. The power to weight should be the same as mine with this car which sounds like the car Wolf has in mind.



This kit car has the Hiabusi engine and looks pretty light. I can get more details of it if you want. Those English are a strange lot.

Then you place the upper ball joint along a line drawn from the center of the static contact patch and as close to the wheel as possible. This keeps inclination to a minimum. Just how high? well, play with your cad 2000 to get the camber gain you want. The gain keeps the wheel at 1 1/2' neg. from static to max roll. There is no rule for this gain value other than maintaining a specific camber in roll. Just keep moving the two points till you get that gain. Then you make a fully adjustable 'test' A arm. Decide where you will mount the A arm to the frame . See that you have an inch up and down adjustment (total 2") both ends and start with the geometry you got from the computer. It might be right on your first go. Load a real wheel and tire as I have described and make a template of the carcass profile at different pressures and cambers. Cut a 6"section of the tire and mount it to the rim with glue and shape it according to the template. Have fun moving the chassis around on a table and marvel at your creation.
Castor is determined as I described before and adjusted with Heims. With 2 turns from lock to lock, I'm not too concerned about centering but you end up with about 6' castor. Tire pressure for fine tuning. Tires make all the difference. Change tires and everything changes.
I think with 110 NM of torque and a well designed suspension, Wolf can run with an open diff. Now there is a story that deserves to be told. My adventures with a M3 differential.

I don't know how to get the shocks right. Something for another thread.

Originally posted by hydra



I figured it would be very difficult, if not impossible, to ensure that the lower arm is EXACTLY horizontal, especially under dynamic conditions. So there has to be some margin for error... Furthermore, if we are allowed to tilt the lower control arm a little, we can lower the instant center even more, giving us a little more design freedom...


Also, it would be nice to hear your thoughts on camber gain and roll gain values. Remember, we're not talking about an all-out race car here, just a nice high-po sports car

[Supercar]

I think after PII expressed all that was on his mind, others had to pause. I too took a break. Now I have a question:

Does the lower control arm really have to be horizontal? I think the only criteria for zero scuff is that the virtual swing arm center (and the roll center too) must be on the ground. What do you think?

A related question: Will the world really come to an end if the roll center crosses the ground level in roll?

A related observation: Many recent sports cars have their front roll centers slightly below ground, and their rear roll centers slightly above ground. Ben said that this is to run stiffer springs, but even some of those that can run 3rd springs still do that. Do you think this could be to minimize tire scuffing in the rear, where consistent traction is more important?

[Greg Locock]

No, the arm does not /have/ to be horizontal... but it is a good place to start for a production car ,where linearity is probably the fundamental requirement.

I'm not sure zero scuff (that is track change in jounce?) is an overriding concern - the wheels are rolling so to some extent small amounts of track change can be tolerated. I vaguely remember a recommendation to keep it below 10%, which is not a hard target to achieve.

RCH's effects are fairly subtle in my book, so no, the world won't end if it bobs around above and below ground level, but it is causing a sign-change in its effect which will confuse the driver and/or chassis somewhat. I don't have any experience of near-zero height roll centres though. I suppose one reason for a certain amount of cynicism is that we used to have a great handling live rear axle car, with a rear RCH of around 307 mm, and then plonked an IRS in, with an RCH of around 120 mm, and didn't change the front suspension one bit, and the handling with the IRS was fine as well.

[Supercar]

Hehehe... I completely forgot about the sign change when the roll center crosses the ground level. Thanks for reminding.

Here is another dilemma then. The roll center that is already low would go down even further in jounce, at a rate of about -4 inch/inch of jounce, and may go below the ground level while braking and turning. We do not want that to happen. So, we should probably raise it 2-4" above ground, or lower it 0.5-1" below ground. I cannot be just above the ground when braking and turning, else the jacking forces will change direction and then the world may accidentally end. It appears that it is good to avoid front roll center heights between 0.5 and 1.5". Has anyone heard of this?

[soubriquet]

FWIW, 105 Series Alfa's have double wishbone front suspension, with the roll-centre well below ground level (I've done the geometry on this). Rear suspension is live axle with lower trailing arms and a "T" upper arm. I don't know how to do the the geometry for the rear, but I believe the roll centre is fixed at the top joint (please correct me if wrong). Springing is very stiff at the front and very soft at the rear. When cornering hard in otherwise steady state, bumps induce a very noticeable roll-rock motion. I've always attributed this to the centre of mass moving up and down relative to the roll axis. Is this a reasonable explanation?

[soubriquet]

Sorry for being so trivial, but can anyone explain how to find the roll centre for a three-link live axle?

[Greg Locock]

What sort of 3 link? two lower longitudinal links and a V above the diff? At the V above the diff.

If you can sketch it and it is in Milliken then it is easy. Otherwise, think about where it can apply lateral forces to the body.

Incidentally we use 4 different methods to find RCH. They don't agree particularly well. I think I'll start a zoo.

If anyone is interested:

Lateral motion of cp when cp is raised

Change in vertical force at cp when a lateral force is applied (I like that one, practically)

Some horrible calculation based on the SAE definition for Force Based Roll Centre

The usual geometric calculation, ie find the IC and then take the intersection of the line joining that to the CP and the mid plane of the car. What happens when the IC is outboard of the wheel?

Not only do they not agree, in some circumstances they move in different directions in response to a given change in hardpoints.

And it is much harder for leaf springs. Guess what we're working on this month!

[soubriquet]

Originally posted by Greg Locock
What sort of 3 link? two lower longitudinal links and a V above the diff? At the V above the diff.

If you can sketch it and it is in Milliken then it is easy. Otherwise, think about where it can apply lateral forces to the body.

Thanks. That's where I had presumed it to be, but I've learned that what is intuitively obvious is not always correct.

And it is much harder for leaf springs. Guess what we're working on this month!

Hmm. Isn't it time for Ford to move on from leaf springs?

[imaginesix]

For a fiberglass single-leaf spring I'd say it's time more OEs moved back to leaf springs, à la Chevy Corvette.
Greg can fill us in on the details, but they should no longer impose a weight or friction penalty over coils but in fact bring improvements in those areas along with packaging advantages.

[Powersteer]

Originally posted by Engineguy
Like bump steer, bump induced track width change is bad.

I would not mind some, but very little depending on tyre profile. It could settle in the tyre before the whole car's weight leans into the tyre. It would also widen the track over bumps and as track distance expand going over a bump it might absorb some of this with strethcing.



EDIT: This on the rear wheels of course. On the front it would sharpen the steering but like you said, bump steer.

[Greg Locock]

Hang on, bump steer is useful. Admittedly, as a starting point you design for zero bump steer, but that's because the optimum is only a couple of mm away. Nonetheless it is a crucial part of the setup. If you think you've got zero bump steer then you are kidding yourself, you will have some, and since toe-in in bump is oversteer, you do't want to go there (usually) so you end up with some toe out in bump just for safety's sake. Bear in mind we are talking fractions of a degree over the suspension travel.

[imaginesix]

Did you meant the reverse; toe-out in bump is oversteer, so set bump on the toe-in side of zero for safety?
The same way the deflection of trailing-arm suspensions tends towards toe-out, thus oversteer?

[imaginesix]

Originally posted by Powersteer
I would not mind some, but very little depending on tyre profile. It could settle in the tyre before the whole car's weight leans into the tyre.

...or unsettle an already loaded tire through a mid-corner bump?
In addition to which, helping to settle a tire during corner transition risks sending incorrect feedback signals to the driver as tire grip may not feel like it is coming on in a progressive manner.

[Greg Locock]

I was talking about front axle, rear axle is the opposite - although to be honest I prefer to leave the rear axle as neutral as I can make it, which is a bit wimpy, it ought to be possible to be more aggressive than that.

[Powersteer]

Originally posted by imaginesix

...or unsettle an already loaded tire through a mid-corner bump?
In addition to which, helping to settle a tire during corner transition risks sending incorrect feedback signals to the driver as tire grip may not feel like it is coming on in a progressive manner.

By that time the geometry won't be pushing out the tyre anymore hence, very little, just enough to settle the tyre wall, specifying tyre profile.

[Greg Locock]

Is there any value in thinking of track-change in roll as a fake toe? it momentarily increases the slip angle of the tire as it is happening. Is that good? It seems like it might be useful, as it only affects transients not steady state.

I'll still get shot for using it, because of disturbance rejection and tirewear.

[imaginesix]

Originally posted by Greg Locock
I was talking about front axle...

Oh yeah, I should have realised that.

Originally posted by Powersteer
By that time the geometry won't be pushing out the tyre anymore...

Oh yeah, I should have realised that.

[Wolf]

I thought to inquire about bushings- my thinking (not often correct) is to go for the lightest and simplest solution. The way I figure it, they should be there to reduce friction, and not to mess up the geometry too much by deforming under loads.

Would PTFE inserts in steel wishbones, on steel bolts, (probebly even PTFE shims next to wishbone) do the trick?

[hydra]

McLaren F1 uses a bushing arrangement that is "25 times stiffer radially than axially" to give good wheel control and ride. You might want to look at something similar...

[Greg Locock]

25 is a bit hard to do, 15 is easy. There is typically a tradeoff with durability and size as you go to these more extreme ratios (a basic suspension bush is around 5). You may run into problems with the coning rate of these bushes, which is the torsional rate about the axis of the arm. That will be very high.

Your biggest problem is just finding out the rates of a given bush. Those who know won't tell you and it is pretty difficult to work out, even if you know what grade of material is used.

To that extent you might have more luck finding an aftermarket parts supplier and writing to them.

[Supercar]

Originally posted by Wolf
Would PTFE inserts in steel wishbones, on steel bolts, (probebly even PTFE shims next to wishbone) do the trick?

Sure they will do. This is how many production-based racecars do it. Not sure what you meant about the shims though.

[Powersteer]

New Porsche RS Spyder use bending-leaf anti-roll bars, i have never seen this. Do they choose this packaging or it has a physical advantage.

[Greg Locock]

Have you got a picture? The FSAE cars often use a flat blade in their a/r bars, as they can then fine tun the balance very rapidly.

[Powersteer]

Originally posted by Greg Locock
Have you got a picture? The FSAE cars often use a flat blade in their a/r bars, as they can then fine tun the balance very rapidly.

That would hold the key, easy access and tuning. Sorry, none of the new RS Spyder yet. I'll try look out for the 2006 model.

[Ben]

Originally posted by Powersteer
New Porsche RS Spyder use bending-leaf anti-roll bars, i have never seen this. Do they choose this packaging or it has a physical advantage.

All Lola Champ Cars have bending leaf ARBs. The actually T-bar is very rigid to minimise lost motion at the third spring.

Ben

[Engineguy]

Originally posted by Greg Locock
25 is a bit hard to do, 15 is easy. There is typically a tradeoff with durability and size as you go to these more extreme ratios...

The McLaren F1 isn't doing it in the same bushing. The A-arm bushings offer basically (hard) radial compliance only... but the chassis member they pivot from is itself linkaged to the frame such that it can easily travel purely and only fore/aft against soft resistence. Pretty elaborate for a "light weight is king" project but Gordon Murray thought it prudent... racecar-like lateral and vertical (non) compliance combined with Cadillac longitudinal compliance desired for a road car.

[Greg Locock]

Oh, well in that case I can claim prior art. Lotus' SID, and the fat bottomed Elan, both started out using a raft concept (essentially a subframe with some attention paid to centres of percussion) to seperate out the local geometry effects from the globally desired compliance ones. For that matter Jag did much the same for many years. Philosophically for Lotus this was a development of the approach taken on the M300 (son of Etna, daddy of the EB110), which did use a mechanism on the wishbone pivots.

But in practice 15:1 in a bush is a lot smaller, simpler, cheaper, less rattly, and more reliable.

And I agree, by clever control of compliances you can end up with a lighter structure overall - it no longer has to react such large spike loads. One of our tests applies 80 kN to one suspension arm - in previous designs this was halved due to geometrical and compliant effects. (ie I screwed up this time round, whoops).

[shaun979]

From Brammo's Ariel Atom site...

The system features outboard rod ends all round for total adjustability and inboard rubber/metal bushes for compliance and comfort. Coil over Bilstein dampers front and rear extent the adjustability of the car allowing the Atom to be quickly set up for road or track use or to individual preferences.

There are also photos showing the outboard rod ends. Didn't Carroll Smith say outboard rod ends was a no no since they are put into bending? Did Ariel just go with really beefy rod ends or what? Are they any examples of any real race car that ran/runs outboard rod ends?

[Greg Locock]

Well, I've got 20000 cars a year running around with outboard rod ends, and Mercedes seem quite happy with them as well.

I think the point is, designing the interface is a little tricky. If you do some tests then you'll validate the design, but for a one-off car, who is going to do that?

Incidentally, if you are trying to design a suitable joint in an aluminum arm or or spindle then prepare for a massive redesign. The wrenching loads on the taper joint will make a mockery of your calculations. Those massive washers are there for a very good reason.