32 replies to this topic

[hydra]

Hey all,
I've been wondering lately about the relevance (if any) of rear suspension caster, KPI, scrub radius, and caster trail/offset, not to mention fore/aft spring location on vehicle dynamics/handling on a SLA double-wishbone setup. I would suspect that these parameters are responsible for lateral/longitudinal force wheel compliance, in that scrub radius and caster trail cause toe-changes when longitudinal/lateral forces are applied. I would imagine that caster trail needs to be negative (i.e. behind the wheel) to give lateral force understeer (resulting moment wants to steer the wheel into the turn) whereas scrub radius should be minimized to minimize toe-out under braking. It might also be useful to have the coilover mount onto the rear upright (instead of on the wishbones), and have it do so aft of the rear wheels so that it preloads the upright/wheel in the toe-in direction - or is this a bad idea?

Here are some pics of what I had in mind
:











Comments? Critiques?

[imaginesix]

Can you separate the pictures into their own post so that we don't need to keep scrolling left and right to read the initial text? Or shrink the images.

Thanks.

[hydra]

done!

[phantom II]

Don't pay no mind to imaginesix, he just likes bossing people around. Tell him to delete his post.

The coil/over mount pivots on one axis so it can only be mounted on the A arm. There are enough axial loads on a double wishbone rear suspension without creating another.

KPI determines scrub radius and bearing loads. Caster at the rear in conjunction with bump steer in roll can increase toe-in and increase camber depending on the design geometry. Porsche 911s do this. The quality of the contact patch is the most important design criteria. Know the properties of the tire (easier said than done) and reverse engineer the design from them. All geometry is determined and limited by the tire. You can use your hydraulic press with some imagination to get some idea of the tire's properties with static loads. Make your lower A arms as long as possible and let your CAD do the rest.
The 94 ZR1 Corvette and the 96 GS Corvette had the same tire and wheel sizes all round, but the rear 11" wheels on the GS had a different offset by +1 1/2". This was good for 3 seconds on Hoosier slicks at Homestead. Both wheels were tried, everything else being equal. The whole story was the wider track on the ZR1 caused jacking and there was more turning leverage.
There may have been less discrepancy with Michelins due to their softer sidewalls but this is an example of scrub radius importance even with a poorly designed rear suspension in this case. The scrub radius will change depending on camber and tire pressure and multi axial loads.
A F1 car has the smallest contact patch on the straight and the biggest in a high speed turn even larger than under braking at the front. Under braking the front wheels have the most negative camber and the most negative scrub yet take most of the braking forces.
The rear tires contact patch remain relatively constant.
Your design must not only take into consideration what the tires can do, but also what you want them to do.
At the end, you can always mess with tire pressures and wheel offsets to cover some design discrepancies.
I think double A arms at the rear are a thing of the past anyway. Michelin tried a multi-link at the rear of a Renault F1 car and it was promptly banned. The Germans seem to have perfected it and the new Ferrari Superamerica front engine car has a multilink rear suspension.

Originally posted by hydra
done!

[imaginesix]

Thanks hydra.

Before the experts chime in, I'd say I generally agree with what you're saying which is that compliance steer is unavoidable so we may as well ensure that whatever compliance we end up with is not in the direction of instability. However you go too far by failing to locate the tire contact patch within the rigid 'triangle' (in side-view) formed by the wishbones and track rod outer mounting points. You end up with more compliance your way, even if it is in the right direction. You can get the best of both worlds by simply ensuring that the more rigid side of the 'triangle' is behind the contact patch so that the more compliant suspension member ahead of the contact patch ensures that you have toe-in from lateral force compliance. But you'll have less of it!

As for the shock mounting, I think you're getting yourself into trouble by thinking of the shock as a possible ally in the battle against compliance. If you get to the point where the shock mounting placement gives you any kind of benefit WRT compliance then you have way, way, too much compliance.

[hydra]

I'm not sure I understand why "the coilover mount pivots on one axis so it can only be mounted on the A arm"
Keep in mind that the setup I've envisioned does not allow the wheel to rotate about the vertical-axis (steer) and thus has no external toe-adjustment. Compliance effects (at least deliberate ones) are to be kept to a minimum, through the use of nylon bushings. Here is an illustration of the rear uprights off my donor car (which I will be modifying to suit my needs)




Due to packaging reasons, I'm pretty fixed as to what rear KPI/scrub radius I can run, at zero degrees and 65mm respectively. Control arms are also as long as they can be. The only real variables I can play with are
- caster angle
- caster trail
- lengthwise coilover location
- upper arm inclination

I ended up with pretty short front-view swing arms to try and get as much roll camber as I could to combat camber loss under deceleration (rear end lifts) Roll center height is a reasonable 117mm (front is 65mm) and it doesn't move around too much within the suspension's operating envelope. Car is mid-engined with a relatively tall quad-cam aluminum V8 drivetrain...

[hydra]

Originally posted by imaginesix
Thanks hydra.

Before the experts chime in, I'd say I generally agree with what you're saying which is that compliance steer is unavoidable so we may as well ensure that whatever compliance we end up with is not in the direction of instability. However you go too far by failing to locate the tire contact patch within the rigid 'triangle' (in side-view) formed by the wishbones and track rod outer mounting points. You end up with more compliance your way, even if it is in the right direction. You can get the best of both worlds by simply ensuring that the more rigid side of the 'triangle' is behind the contact patch so that the more compliant suspension member ahead of the contact patch ensures that you have toe-in from lateral force compliance. But you'll have less of it!

As for the shock mounting, I think you're getting yourself into trouble by thinking of the shock as a possible ally in the battle against compliance. If you get to the point where the shock mounting placement gives you any kind of benefit WRT compliance then you have way, way, too much compliance.

imaginesix,
My other motivation for mounting the (200lb/in if anybody's interested) coilover to the upright is packaging; my current iteration is probably the easiest to jig/fabricate, not to mention it allows for a chassis x-member that goes straight above the transaxle gearhousing, instead of arching around the bellhousing - lighter, stiffer, and easier to construct...

Right now, both upright pickup points are 4cm behind the wheel centerline axis, and the coilover is 8cm behind. I can always decrease these numbers a little if need be... Anybody have any suggestions as to what a good starting point for these would be? Tires are 225/45/15 slicks on 15x7" rims

[Greg Locock]

coil over to the spinde - good idea lets the arm bushes work properly. You might like to think about pointing it at the contact patch.

You want compliance understeer from the rear, so put the tie rod further behind the axle line and mount it stiffly using ball joints.

I've never found front suspension concepts like castor, kpi , scrub or whatever very helpful when designing IRS. Trail is useful to think about as it changes the lateral steer compliance, effectively.

I think about

lateral toe compliance

camber compliance

toe in braking

toe in acceleration

RCH

bump steer

roll steer

bump camber

I am no fan of antisquat and prodive

There's nothing much wrong with double wishbone rear ends, they are bit more complicated ($) than you really need, but work well. They can tie themselves in knots due to traction and braking, if you can extend your lower triangle forward by another foot you'll have an easier time.

[hydra]

Greg,
I'm glad you decided to chime in...

- Pointing the coilover at the contact patch (i.e. moving the coilover frame mount rearwards) would entail using heim joints or spherical bushings in the shock eyes instead of a hard plain bush... What's the advantage in doing this, and how much of a cant is needed?

- How foolish would it be to dispense with the toe-control link altogether, relying only on hard plain bushings instead of spherical ones for toe control?

- I have otherwise done everything I can to provide lateral force understeer in principle, with a "pure" suspension setup with no "anti"-s, one straight wishbone leg for maximum stiffness. What do you think of the #s specified; RCH, camber gain, etc... What about my caster trail starting point of 4cm and how it pertains to lateral force understeer? You should have a much better feel for these things than I do, especially regarding compliances and such..

- Extending the lower triangle forward is a good idea, I had splayed everything 45* to facilitate fabrication, but its not that big a deal. I should add that my control arms are to be fabricated 1.5mm stressed-skin boxed sections and so should be very stiff in both shear and torsion. Speaking of which, why don't we see this done more often?

[Greg Locock]

- Pointing the coilover at the contact patch (i.e. moving the coilover frame mount rearwards) would entail using heim joints or spherical bushings in the shock eyes instead of a hard plain bush... What's the advantage in doing this, and how much of a cant is needed?

It provides for minimal loads in the bushes and so on, I know it isn't important.

- How foolish would it be to dispense with the toe-control link altogether, relying only on hard plain bushings instead of spherical ones for toe control?

Lots of people have done it that way (Jag for example). I like the easy way a tie rod allows you to adjust bump steer.

- I have otherwise done everything I can to provide lateral force understeer in principle, with a "pure" suspension setup with no "anti"-s, one straight wishbone leg for maximum stiffness. What do you think of the #s specified; RCH, camber gain, etc... What about my caster trail starting point of 4cm and how it pertains to lateral force understeer? You should have a much better feel for these things than I do, especially regarding compliances and such.

I can't be bothered to read the fuzzy numbers in your screenshots, sorry. I'd guess you want lots of negative trail on a rear suspension.


- Extending the lower triangle forward is a good idea, I had splayed everything 45* to facilitate fabrication, but its not that big a deal. I should add that my control arms are to be fabricated 1.5mm stressed-skin boxed sections and so should be very stiff in both shear and torsion. Speaking of which, why don't we see this done more often?

Which? long based triangles or elaborate sections?

[hydra]

Greg,
To sum up, I'm have the following:
static R RCH 120mm
94cm SVSA length
14cm IC height
31.75cm LCA length
22.85cm UCA length
28cm spindle length
21cm vertical spread between control arms
0.83 motion ratio
0 KPI
62.5mm scrub
0 caster
4cm negative trail (this is probably the quantity we're most concerned with)
zero bumpsteer and rollsteer (theoretical)
64.6*/m camber gain in bump


I spent a lot of time minimizing camber loss under deceleration/corner entry, as this is a mid-engined design, and is prone enough to lift-off oversteer as it is.. I finally got this down to ~+0.6 degrees on the outer rear wheel during corner entry, assuming maximum transient roll and 0.3g deceleration. The added Camber gain during corner exit/acceleration should also help increase rear-end stability right?


Another thing I'm a little fuzzy about is how big (or should I say small) I can make the bushing. Due to packaging reasons I've gone with an outer diameter of 35mm and a bushing length of say 40-50mm, how do these numbers sound for a 800kg car?



Finally, I was talking about a stressed-skin boxed-section sheetmetal control arm, as seen on some production cars. I was wondering why we don't see this sort of thing more often on specialist low-production sports cars, as its light, cheap, very stiff, and easy to manufacture - using nothing more than a shear, a sheetmetal bender, and a spot welder.


PS - On a slightly unrelated note, does a caster trail of 23mm sound reasonable for the front (no PS)? Scrub radius is effectively locked in at 60-something mm, how bad is this?

[Greg Locock]

Last things first, you can make up some 5/8" tube arms and see how they compare with your fancy box ones.

For the front 23 mm trail sounds absolutely fine, possibly even a little excessive.

60mm scrub is neither here nor there, twice that would be drivable, as would half that. The car will be a bit twitchier under braking than if you had say 20 mm, but then you probably want to feel what the contact patch is doing.

For the rear two numbers caught my eye

64 deg /m camber gain is a lot. That's about 100% camber/roll. Neil Roshier convinced me that that is not the way to go, it'll hurt straight line traction too much. In fact he dug out a whole bunch of examples to make the point that less is more.

Secondly the -40 mm of trail is really 'only' -10 mm, since your pneumatic trail will be 30 mm. So, that isn't much.

Oh, you need to find out what bushes you can get easily, design around them.

[Fat Boy]

Originally posted by Greg Locock

I am no fan of antisquat and prodive

Agreed. Anti-lift can be useful to some extent depending on the problems that you're fighting. Particularly, it can increase entry stability if used sparingly. If straight line forward traction is _really_ a problem, I have used pro-squat. Without having an idea of what problems you might face, I'd tend to make thing readily adjustable.

Tires are such a huge chunk of the equation that if you don't have a good idea of what they're doing, everything else is only about a step and a half away from hand waving.

[NRoshier]

Hydra I am building/designing a car perhaps similar, but what do you intend to use yours for?

[hydra]

About the 64deg/m number, I placed more of an importance of camber under braking than under acceleration because the car is typically going to be subjected to 50-100% stronger forces under braking than when accelerating. Besides, MR cars tend to have a nasty habit of throttle-off oversteer as it is, yet another argument in favor of adding a lot of roll camber in the rear, which unlike the front doesn't have the benefits of camber gain in dive and camber gain in steer. Besides, I figured that the extra camber produced during corner exit, when the driver gets back on the car) might produce a little extra lateral grip, which is a stabilizing effect, or am I completely off here? I'd also be very keen on hearing Neil Roshier's views on this if possible...

If -40mm of trail isn't much then how about -60/70mm instead?

I should have mentioned that I'll probably be turning the bearings out of delrin/polyamide on a lathe so I'm free to use whatever dimensions I want (smaller being better in this case).


Oh, and Neil, the car is to be used for trackdays and the odd amateur hillclimb/autocross event. I'm shooting for a ~650kg dry weight and 350 naturally-aspirated, bent-eight mules!

[Fat Boy]

Originally posted by hydra
Besides, MR cars tend to have a nasty habit of throttle-off oversteer as it is, yet another argument in favor of adding a lot of roll camber in the rear, which unlike the front doesn't have the benefits of camber gain in dive and camber gain in steer. Besides, I figured that the extra camber produced during corner exit, when the driver gets back on the car) might produce a little extra lateral grip, which is a stabilizing effect, or am I completely off here?


Oh, and Neil, the car is to be used for trackdays and the odd amateur hillclimb/autocross event. I'm shooting for a ~650kg dry weight and 350 naturally-aspirated, bent-eight mules!

Your logic is good enough, but it just doesn't turn out that way.

On the rear, you'll be looking for less camber change. Think of a Volkswagen swing axle. That's about the most camber change possible, and very little good was every said about their rear grip, especially on corner entry. You'll want to have fairly long front view swing arm lengths and much less camber gain that what you're planning. If you think that entries could be an issue, at least add an optional mounting point for 20% or so anti-lift. Excessive AL can cause chattering and skipping over bumps, but having some will allow better control of rake with few ill-effects.

Is this for some sort of single seater? If so, then your rear roll center is also fairly high.

Hooking up 350 HP shouldn't be that big of a deal with most tires of reasonable size. If you're running a V-8, you should shoot for at least 500 HP. It's not hard to make an engine put that out with pretty standard hot-rod parts. Hell, some stock 6 cylinders are over 300. If you have a good amount of HP, then hooking it up will start to be an issue, which will mean getting a big contact patch when you're straight. If you have a lot of camber gain, then as you get to the throttle, you'll have less and less contact patch, which will pretty much guarantee exit oversteer and bad powerdown. The more HP you have, the more you have to sacrifice mid-corner to make the thing work on acceleration.

[hydra]

I see your point FB, and its a good one. How much camber change is good for starters? I should mention that I'll be running 7" slicks in a 15" diameter both F & R, which is why I was concerned with the corner entry phase as I don't have the luxury of an extra inch or so in the rear. Remember that we're dealing with an MR configuration with a ~43/57 weight distribution. Another thing worth mentioning is that due to packaging constraints in the rear, we can't make the control arms any longer than they currently are...

Is rear anti-lift achieved by locating the SVSA ahead of the rear wheels and above the ground, i.e. by tilting the front control arm mount down compared to the rear? I assume the resulting caster (and thus trail) change with bump wouldn't have too adverse an effect on lateral-compliance characteristics?


Car is to be a closed-wheel single-seater, sort of like a mini-LMP with a 225cm wheelbase and a 160cm track width (wide, I know...) Reducing the camber gain should automatically lower the rear roll center...

The 350bhp mentioned are to be extracted from a 32V quad-cam all-aluminum 3.6L V8 with a peak torque of no more than 400Nm. For those who are interested the engine in question has an overall length of just 50cm from the crank pulley to the bellhousing face, and should weigh no more than 170-175kg when I'm done with it. Crank centerline is 21cm above the oilpan/lowest point of the bellhousing, and the engine is 56cm high to the top of the valve covers and 72cm wide across the timing belt cover. Any guesses as to which engine we're dealing with here? (Hint: Its cheap and plentiful!)

Back on topic though, my modest power/torque requirements mean that throttle-on corner-exit behavior isn't super critical, even with 7" slicks (or so I'm hoping), so we can focus some of our attention on corner-entry and mid-corner behaviour.

Finally, and I know we've already discussed this, but after all is said and done, can I really do away with the toe-control link?

[NRoshier]

Hydra email me via racemagazine.com.au and I'll send you through some of my articles on my car, we can have a good conversation about this stuff.
I'm still getting over Greg's comment about influencing his view on suspension design, but I might add that camber gain has pro's/con's as you are aware and (IMVHO) what you 'need' will vary depending on the type of motorsport/circuit/driving style etc etc For hillclimbs traction/braking is very very important and locally htere are very few tracks that put a premium on med/high speed cornering. With this in mind a lower camber gain is better. Same could be said for a high speed track that has a majority of low speed corners, here braking and traction out of the corners will again be at a premium. The case may well be reversed for a place like Monza which is nearly all high speed.
You can counter the 'oversteer' in a number of ways, but here I will defer to Greg et al as I know little in comparison, though I would point out that a longer wheelbase is an easy solution.
3.6lt rules out the Lexus V8 as that is 4lt. the Audi V8 is 3.6lt but is far from cheap!
Corner-exit performance may be an issue as I will be running 9" slick/treaded tyres and with 300bhp will expect issues as local tracks are not smooth...I do expect to invest in some very good dampers to help here.

[Fat Boy]

Originally posted by hydra I see your point FB, and its a good one. How much camber change is good for starters? I should mention that I'll be running 7" slicks in a 15" diameter both F & R, which is why I was concerned with the corner entry phase as I don't have the luxury of an extra inch or so in the rear. Remember that we're dealing with an MR configuration with a ~43/57 weight distribution. Another thing worth mentioning is that due to packaging constraints in the rear, we can't make the control arms any longer than they currently are...

I'm not sure what an MR is unless you mean a Toyota MR-2. Sorry. Whatever it is, the weight distribution seems to be a good place to start. I HIGHLY encourage you to put more tire on the rear. In fact, put more on the front, too. Long wishbones are great. We all want them. Having said that, there's really no replacement for contact patch area. Look at the rules and run the max allowed. If there aren't any rules, then run the max that it's easy to get your hands on. I have seen this taken too far with Atlantic tires on a FSAE car, but that's about the only situation. As far as the front/rear split, you need to get more rear tire than front. At the very least, have the tire widths reflect the weight distribution of the car. Throwing tire at a car might seem like a brute force approach, but tire are _so_ critical that I really can't over emphasize their importance. Cavemen were able to drop the wooly mammoth, sometimes brute force works.

As far as wishbone lengths, a lot of sedans have pretty short wishbones. It's the nature of the beast. You can still make it work pretty well.

Originally posted by hydra

Is rear anti-lift achieved by locating the SVSA ahead of the rear wheels and above the ground, i.e. by tilting the front control arm mount down compared to the rear? I assume the resulting caster (and thus trail) change with bump wouldn't have too adverse an effect on lateral-compliance characteristics?

In short, yes, that's what gives you anti-lift. If you tilt the lower one up and the upper one down you can generally get by without hurting much else. Be careful that the anti-squat doesn't start to go up too. I know you've talked about dropping the toe-link. To be honest, I'd leave it there. At the very least, it'll make setting toe an easier chore.

Originally posted by hydra
Car is to be a closed-wheel single-seater, sort of like a mini-LMP with a 225cm wheelbase and a 160cm track width (wide, I know...) Reducing the camber gain should automatically lower the rear roll center...

Back on topic though, my modest power/torque requirements mean that throttle-on corner-exit behavior isn't super critical, even with 7" slicks (or so I'm hoping), so we can focus some of our attention on corner-entry and mid-corner behaviour.

I think you might be selling your car a little short on the throttle side of the corner. It isn't a huge amount of power, but it'll be frisky enough that you'll have your hands full. 7" of tire isn't much. You're probably worried about understeer. Don't be. Understeer is your friend. Yes, people complain about it a lot, but ultimately you can still go pretty damn fast with a car that has a fair chunk of understeer. If a car is loose, you're going to be slow. My rule of thumb is that a car with 10% too much understeer goes about the same speed as a car with 1% too much oversteer.

[hydra]

Neil,
I just emailed you through your website. Looking forward to hearing back from you!
You're 100% right about the design needs stemming directly from how/where the car is to be used. Most corners here are relatively tight-radius low-speed (>100kph) and a lot of them are off-camber. Plus roads here are quite bumpy and slippery due to the aggregates used. My reasoning was that the steering lock needed for the tight radii corners generated a fair bit of steer camber up front. So I figured I'd add camber gain to the rear so I'd end up with similar cambers. In fact, at 2.5 degrees of transient roll, at ~0.3g deceleration going into a 100ft radius turn, the outside rear only loses 0.6 degrees of camber compared with the outside front. However, reducing the rear camber gain by just 25% increases this number to a non-trivial 1.2 degrees... Getting back on the power after said turn gives us ~0.9-1 degree of negative camber change on the outer rear wheel assuming a 1g longitudinal acceleration. Surely this isn't excessive is it? In short I found that having roughly double the camber gain in the rear as the front gives some fair-looking numbers in my case. Whether or not this translates to a well-behaved vehicle is a whole other issue... Of course increasing the wheelbase would make for a less-twitchy setup, with less-pitch during acceleration, but it would also make for more weight and less chassis stiffness.



The 3.6L V8 in question is indeed an Audi, off the Audi V8 quattro of '89-'91 vintage. I don't know about Australia, but these are relatively common in North America and Western Europe - if you know where to look, and typically sell for the equivalent of 200-500USD. In fact I went one better and bought the entire car for a under a grand. My arguments in favor of using the Audi drivetrain are as follows:

- An Audi 016 transaxle bolts right up, so no need to mess with adapter plates and custom flywheels and all that other non$en$e. Tranny is good for 400-450bhp which is good enough for me, and I was lucky enough to find a V8q manual-tranny, complete with flywheel (rare and unique to the V8) and clutch locally

- The Torsen rear diff in the V8 quattro is almost a straight swap into the Audi 016 transaxle. The bearing bore on the differential cover needs to be bored out a little to fit the larger bearing, but that's about it. So basically I get a badass Torsen LSD for very little $$, which cannot be said of any other "junkyard" transaxle.

- Engine is cheap to buy, and top-end/valvetrain is standard 16V VW Golf II GTi, which means that parts are cheap and plentiful locally. Its the later A8 4.2L engine that is expensive...

- Engine is also very compact length-wise (just 8cm longer than a Suby flat-4 actually) , and about as lightweight as junkyard V8s get. Its also very robust, and has been known to hold 20+psi of boost with nothing more than stacked gaskets (ghetto I know) to lower compression.

- I realize that I could have gone with a Suby setup which is both smaller and lighter, with a lower C of G, but I really wanted a V8. Plus turbo subies aren't cheap, and neither are LSD transaxles...

[hydra]

FB,
By MR I meant Mid-Engined & Rear-wheel drive
I realize that tires are by far and away the most important part of the equation. Forget suspension geometry, forget camber gains, roll centers, and all that other nonsense, they're just the icing on the cake. Tires are everything, a brief ride in a racing kart is enough to illustrate this very clearly. I would love to be able to use wider tires, both F&R but I'm limited by what I'm able to get my hands on locally. Importing a set of slicks from abroad would be prohibitively expensive, plus I own two sets of ATS "cookie cutter" Porsche 911/944 wheels in 15x7 which are, at 12lbs each, difficult to better as far as weight goes, not without spending $$$ at least. The best I can do is to minimize weight and make sure I'm making the most of my contact patch to get the best out of my tires. Interestingly enough I should have a similar "tire to weight ratio" as my friend's Pantera GT5, with its 14" wide rear wheels. Besides, wouldn't getting the tires up to temp be a concern if your typical competition event is ~5min long at a time?

The understeer factor can always be tuned by varying ARB stiffness up front. I'll be shooting for a Lateral Load Distribution of 50-51% Front, with ride frequencies of 1.85Hz F and 1.7Hz R at the contact patch. How reasonable is it to expect a 38cm C.G. height with a ride height of 10cm? Or is this asking too much/too little of an open-cockpit racecar?

[Greg Locock]

1.8/1.7 sounds very soft for a circuit car. Good. Coincidentally that's what I am currently estimating for Neil's car. Incidentally his CGZ is 512mm.

[Fat Boy]

Originally posted by hydra
FB,
By MR I meant Mid-Engined & Rear-wheel drive
I realize that tires are by far and away the most important part of the equation. Forget suspension geometry, forget camber gains, roll centers, and all that other nonsense, they're just the icing on the cake. Tires are everything, a brief ride in a racing kart is enough to illustrate this very clearly. I would love to be able to use wider tires, both F&R but I'm limited by what I'm able to get my hands on locally. Importing a set of slicks from abroad would be prohibitively expensive, plus I own two sets of ATS "cookie cutter" Porsche 911/944 wheels in 15x7 which are, at 12lbs each, difficult to better as far as weight goes, not without spending $$$ at least. The best I can do is to minimize weight and make sure I'm making the most of my contact patch to get the best out of my tires. Interestingly enough I should have a similar "tire to weight ratio" as my friend's Pantera GT5, with its 14" wide rear wheels. Besides, wouldn't getting the tires up to temp be a concern if your typical competition event is ~5min long at a time?

The understeer factor can always be tuned by varying ARB stiffness up front. I'll be shooting for a Lateral Load Distribution of 50-51% Front, with ride frequencies of 1.85Hz F and 1.7Hz R at the contact patch. How reasonable is it to expect a 38cm C.G. height with a ride height of 10cm? Or is this asking too much/too little of an open-cockpit racecar?

MR meant mid-engined-rear wheel drive for Toyota, too. I guess I wasn't too far off.

I understand what you're up against for tires. You can't run what you can't get your hands on. Having said that, I'd look around a bit more. I don't know what is commonly raced in your area, but I'd spend more time looking before locking in a design. If wheels are a big deal, then you can have what you own widened. OK, enough about that.

Heating up the tires shouldn't be too big of a deal depending on what compounds you can run. A sticky tire comes up to temp in a 45 second autocross, so 5 minutes is plenty. The first couple corners are sketchy, but you'll get used to it.

Your C.G. height seems optimistic. I've ran a similar sized formula car with about a 350 mm C.G height, but it had a 35 mm ride height. If you were starting with less of a blank slate, you'll inevitably end up higher than what we had.

Make a Front ARB that's big. I mean drain-pipe big. I think you'll need more a lot more FLLT% than you're planning. I wouldn't bother making a rear ARB. It'll just be ballast anyhow. I just took a gander at my last sans-aero car. It was a lot bigger than what you're running, but common ride frequencies were about 2.0 on the front and 1.7 on the rear. It all depends on the track, but as the old man said, "A car is like a primate, it has to squat to go." Body roll and pitch are not necessarily bad things until they hurt the response. With a patient driver that stays ahead of the car, that can be pretty stinkin' soft. That's what it'll take to get good traction in slow corners on a road with very little rubber on it.

Have you thought about a diff, yet?

[NRoshier]

Actually Greg my Vert CoG is 483mm...though I do not know if 30mm makes that much difference?
Re wheels: have a look an enkei racing RP-F1 wheels from tire rack. I will be running 17X9.5 rear wheels and they weigh 15.9lbs. WRT wheels I would suggest too that you have a look at the available tyres in your projected size and then +1-2" if you go larger in width. I chose the tyre width/diameter based on the maximum available tyre choice in both brands and compounds. This way if one tyre does not work very well or goes out of production then I have further choices. For me in my local market 17" is good, 15" and 16" is not (in 'r' spec road tyres and slicks).
The audi engine is an interesting choice and I wish it were possible for me to get and use it. Unfortunately local rules preclude using an old engine such as that as my car has to comply with emissions regulations.

[Greg Locock]

Sorry Neil, you are right, I was reading off the sprung mass CG. I've got 478 for the complete vehicle. I think I can live with a 5mm error!

[NRoshier]

The wheels I ordered will be here by the end of the week. To my delight air freight from the states cost $300US for 8 wheels, so the +35mm offset will be set.
I had a very long conversation with Bill McKenna the local Penske damper agent, who also sells competition tyres in SA. He suggested that a treaded motor sport tyre such as a Bridgestone Re55 should have around 25% increased performance over a Goodyear F1 GS-D3, which I assume you might know a bit about. He has never seen data released on the competition tyres in the past 20 years. The FSAE teams have started a fund raising system to test a batch of tyres, different teams have to put in money and the tyres will be tested, with the data released to the funding teams. Naturally this only relates to their rather small slicks.
Also I found this from germany: Michelin is giving for Formula BMW 2006 following numbers:
Tire size: Front 20/54 - 13; Rear 22/54 - 13, Vert. stiffness: Front 19 kg/mm; rear 18 kg/mm
I assume that this values are for warm pressures around 1.6 bar and camber somwhere near 3°30' recomended by Michelin in their setup sheet.

[hydra]

I know I'll eventually need to calculate this, but what would you say is a reasonable estimate for CG height? I'm starting from a clean sheet of paper, and will be packaging as many parts as I can as low as possible - right on the floorpan. By way of comparison, Goran Malmberg's HemiPantera has a CG height of 415mm @ 75mm ride height, but that's with a Hemi-engine and a roof structure, so I was thinking that 400-420mm should be in the ballpark.. Some rough calcs show that I have roughly 55% of my overall weight at ~320mm above the ground..

What would you say is a good baseline FLLT% to start with? Milliken has a (presumably front mid-engined) example in Chapter 16 that ends up with 44-45% weigh distribution on the front @ race weight, and has equally sized tires front and rear. He then goes to recommend setting the FLLT% as 5% + the Frontal Weight % for initial understeer. Now I went ahead and added another 2-3% on top of that just to be extra safe. Why do you feel that even this is insufficient? Incidentally, this would mean that I'd end up with a front ARB that's 9mm thick.. I've seen drinking straws that are bigger than that!

I'll be running a Torsen/Helical LSD which I may end up modifying to increase the Torque Bias Ratio under acceleration by substituting the needle thrust bearing for a hardened washer on the drive side, as described on the Torsen website.


I also really wanted to thank You, Greg, and Neil for your superb, world-class feedback. This has been a most enjoyable and informative discussion thus far...

[NRoshier]

Hydra, I am happy to discuss my own thinking WRT my own project, as it is somewhat similar to your own and I thank you for your comment, however I would also respectfully point out that I defer utterly to Greg/Fatboy/RDV on questions of engineering/design.
I might suggest that a plate diff is more reliably tunable in most circumstances. I can send you a copy of the TransX catalogue if you like. They have a range of differentials.

[hydra]

Originally posted by NRoshier
Hydra, I am happy to discuss my own thinking WRT my own project, as it is somewhat similar to your own and I thank you for your comment, however I would also respectfully point out that I defer utterly to Greg/Fatboy/RDV on questions of engineering/design.
I might suggest that a plate diff is more reliably tunable in most circumstances. I can send you a copy of the TransX catalogue if you like. They have a range of differentials.

A clutch-type diff may be more tunable, but I already have a torsen-diff (the main reason I bought the donor car actually). Besides, I don't think there are any aftermarket clutch-type LSDs for the Audi 016 transaxle, all the ones I'm familiar with are the helical type.


EDIT: Nevermind, I just remembered that the 944 Turbo (essentially an Audi 016) transaxle uses a clutch-type LSD

[NRoshier]

true! I'd imagine the 944/968 diff may just 'drop in'...but also may not given VAG's ability to change things in a minor way to make porsche parts more expensive. From memory Porsche had a different locking ratio under power and braking to stabilise the 911. I wonder what they did on the cayman/boxster?

[Fat Boy]

Originally posted by hydra
I know I'll eventually need to calculate this, but what would you say is a reasonable estimate for CG height? I'm starting from a clean sheet of paper, and will be packaging as many parts as I can as low as possible - right on the floorpan. By way of comparison, Goran Malmberg's HemiPantera has a CG height of 415mm @ 75mm ride height, but that's with a Hemi-engine and a roof structure, so I was thinking that 400-420mm should be in the ballpark.. Some rough calcs show that I have roughly 55% of my overall weight at ~320mm above the ground..

What would you say is a good baseline FLLT% to start with? Milliken has a (presumably front mid-engined) example in Chapter 16 that ends up with 44-45% weigh distribution on the front @ race weight, and has equally sized tires front and rear. He then goes to recommend setting the FLLT% as 5% + the Frontal Weight % for initial understeer. Now I went ahead and added another 2-3% on top of that just to be extra safe. Why do you feel that even this is insufficient? Incidentally, this would mean that I'd end up with a front ARB that's 9mm thick.. I've seen drinking straws that are bigger than that!

I'll be running a Torsen/Helical LSD which I may end up modifying to increase the Torque Bias Ratio under acceleration by substituting the needle thrust bearing for a hardened washer on the drive side, as described on the Torsen website.


I also really wanted to thank You, Greg, and Neil for your superb, world-class feedback. This has been a most enjoyable and informative discussion thus far...

UGH! You're asking a pretty tough question without much info. Without a doubt, the tires are going to be the biggest factor when it comes to FLLT%. Without having a clue of what they'll give you, my thoughts on the matter aren't worth a whole lot.

I would say that a 9mm FARB is giving you some sort of message. Depending on the span of that bar, it could be within reason or very small. Which do you think? When you're building a car from the ground up, you need to allow for adjustability. I'd recommend building several FARB's. Start out with them in fairly large increments, say a 50-100% increase between bar sizes. Also build them with blade adjusters so you can hope to find a good setting based on track conditions.

As far as giving you an actual FLLT% to shoot for. I'd guess you'll end up anywhere from 55-75%. A huge range, I know. This is about a truthful as I can be, though.

A Torsen diff isn't my favorite. They're pretty tough and don't cause a lot of problems, though, so all in all, maybe not too bad of a choice. With the Quaiffe you can adjust the belleville's in the middle to vary preload. Is this an option on the Torsen you're using? If so, then you might think about throwing a chunk of preload in it to start, maybe 30-60 NM. Most Torsen diffs aren't built to lock very hard. This is one of the things that makes them last a long time and not induce any really ugly traits.

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

Driving style is going to play a huge part in tuning the car. Inexperienced drivers (I don't know your experience level) tend to drive a car in a way that will induce understeer. They tend to brake too lightly, turn in too early or too late (either having to wind in a bunch of wheel at the apex or never actually getting to the apex), and apply the throttle mid-corner--generally in stabs. All of this conspires to make the car understeer. The driver reports this understeer, and then the engineer goes about trying to tune it out. Ultimately, the car might have been well balanced or even an oversteering car when driven by a pro. Don't get caught in this loop. It's one that's very tough to get out of.

If you make a couple changes to the car that _should_ reduce understeer, but it doesn't seem to, then go back to baseline. Does that feel like more or less understeer? Now do something pronounced that should make it understeer. Quite often you'll find as you make the car understeer more, the driver can drive more like he should and the car will actually feel as if it has _less_ understeer even though logically you know it has more. I don't know if this is making any sense, but what I'm trying to get at is that understeer is not necessarily your enemy. It's a tough thing for most racers to grasp because they feel fast when oversteering and sawing on the wheel. It's a shame, but that just doesn't work on the stopwatch.

[hydra]

OK on a slightly different note.

How do you design/tune lift-off oversteer out of a mid-engined car? Even the OEMs get this wrong at times (2nd-gen MR2 and Ferrari 348 being notable examples) What steps does one take to minimize the incidence of this most disconcerting phenomenon? The easiest ways would probably be to lengthen the wheelbase and widen the rear tires, but what do you do if those are fixed?



Btw just for the record it now looks like I'll be running 7s up front and 8s in the back, on 190/570-15 Hankook F200 race slicks all round..

[Greg Locock]

A big red sticker that says don't lift off in corners has always worked for me. It is tattooed on my eyeball.

Neither of those companies are exactly famous for making suspensions work except by following Chapaman's dictum.

What you need to do is work out how the forces are changing and then use them to your advantage. Rear roll steer or compliance steer would seem likely candidates, and you might be able to do something funky with anti squat. Your shocks might need a bit of a rethink. Diff might help.

The basic question is - why are you getting oversteer when you lift off?