Okay suspension geometry gurus (ahem, Austin)... need your help inside!

[sperry]

It's time to talk about roll/bumpsteer on the SVX!

First, if you haven't been following, here's the background on the modifications I made to the SVX in order to run some big StopTech brakes:

http://www.subaru-svx.net/forum/showthread.php?t=53570

The short version: I adapted some STi front knuckles to replace the SVX's knuckles. This involved fabricating a ball-joint adapter (both cars are coilover McStrut with an LCA and swaybar). The result is that the knuckle rides about half an inch closer to the centerline of the car than before, and I think because of it, I've got a rollsteer issue.

The car drives fine enough on the street, but at the track, it could use some help. Here's my summary of the handling at the track from the other thread:

Originally Posted by sperry

I'm not super happy with the car on track. First, with those wheel spacers pushing the front wheels outwards and the super soft stock suspension, I'm actually rubbing the shoulder of the tire on the rubber fender liner under hard turning. It's actually minor enough that I can probably just pull off the liner strip... but if I'm going to do that, I'll probably also put a really minor roll on the front fenders as well just to be sure.

Second, the car pushes like a plow... I mean sure, it's not any worse than the car was before the new front knuckles/brakes, but I was really hoping the additional camber up front was going to help a little... but with a 70/30 weight bias, I don't think the SVX is ever going to be super fun on track... it's probably always going to be a bit gruesome in the slow corners.

Third, and this is the most disconcerting, I've got a weird shimmy in the steering on corner exit. I don't know how to explain it better than to say that if feels like a speed-wobble on a motorcycle. Basically, as you feed in power exiting the corner, if you hit even a minor bump, the steering starts to shudder back and forth. Just backing out of the throttle a tiny bit stops it, but leaving your foot on the floor allows the shudder to worsen. My best guess is that the car has a bumpsteer issue that coupled with my 231,000 mile old struts can put a wobble in the steering. Hit a bump with the car's weight shifted off the wheel, and the bumpsteer pushes the steering one way, then the strut starts to oscillate on the one corner and as the wheel travels up and down it pushes the steering back and forth. And I think the -2deg of camber probably allows the shudder to self perpetuate, as the wheel turns back and forth just a little bit, the camber keeps the strut from ever damping the vibration out. The only thing that stops it is lifting a little and shifting some weight back on the strut.

Now, I'm not super educated on the details of bump/rollsteer. I understand what causes it, but I don't know enough to figure out how to fix it. I'm not even sure I can fix it, simply due to the limitations of the space and parts I have under the car.

I did a little drawing in Visio just to see if I can understand what moving the knuckle towards the centerline of the car might do rollsteer-wise. If the lower balljoint moves inwards, then the dynamic camber change of the wheel should be more dramatic. Because the length of the steering rod has only been shortened enough to give me zero toe at static right height, I'm guessing that there will be excessive toe-in on the outside tire when cornering.

Here's my guess at the geometry:



The blue diagram is the stock SVX geometry, the black is the new STi geometry. The lower arc in each diagram is the LCA range of motion (from static height, pushing up under compression), the upper arc in each diagram is the steering rod range of motor through the same compression.

Assuming the blue diagram represents "proper" rollsteer, i.e. it's just right to compensate for the change in camber, at a delta of 3.6 deg, you can see that a shorter LCA and more static camber results in a greater change of rollsteer angle under compression, 4.5 deg.

Now, the diagrams above have no real basis in reality, as they're not from any measurements off the car... they're just me attempting to see what would happen to the rollsteer under the conditions I'm guessing are occurring on the SVX. How far off base am I? Is this totally incorrect or am I on the right track? I think that without accounting for the strut angle itself, I'm probably missing a big part of how this works.

But if this is the cause of my issue, what can I do to address it? Narrower steering rack? Push the knuckle back outwards and come up with some means of extending the upper strut mount (which is the reason the knuckle is closer to the centerline to begin with)? Neither of those options are very likely... but I can probably raise or lower the ball joint adapter to change the angle between the LCA and the steering arm... but I don't know how that would help/hurt the geometry.

People smarter than me... throw me a bone!

[AtomicLabMonkey]

If you changed knuckles without modifying the rack/tie-rod assembly, then you almost certainly have different bumpsteer now than stock. I can't say what it is though without exact parts measurements or measuring the steer itself.

It's all easy to visualize once you've moved things around in a 3D suspension model, but I'll try a little explanation. The tie-rod axis must intersect with the instant center of the control arms (or strut) to have zero bumpsteer (illustrated in pic). If you look at the movement of the assembly as it goes through bump/droop, it becomes clear that the length of the tie-rod as well as its initial angle determines dynamic angle, and consequent steer effect. To modify the initial tie-rod angle, you need some kind of bumpsteer kit which will let you adjust the height of the outer tie-rod pivot. Alternately, you can change the height of the rack (usually more difficult). To modify the length of the tie-rod, you obviously need to either change the width of the rack, or the steering arm on the knuckle (neither is usually practical). There are pics that illustrate all this movement in my RCVD text, but I don't have a scanner here.

Bumpsteer is fairly easily measured, although it takes some time to get the car set up to take the data. Basically you put the car on stands, pull the springs & swaybar endlinks out, attach a bumpsteer gauge to the hub, and just measure the amount of steer you get with dial indicators as you run the suspension through its range of bump/droop. If you have a means of adjusting like a bumpsteer kit, then you change spacers & test again. Change spacers & test again. Etc.

http://www.maximummotorsports.com/st...roducts_id=193

The simplest thing to do is just adjust the spacer stack to minimize any steer, although you can tune it for some desired handling response if you want. OEMs use roll & bushing compliance steer effects to tune understeer into their vehicles. If you're getting dynamic toe-in during roll at the front of your SVX, it could account for some of the 'plow' you're feeling. I'd advise measuring/adjusting the bumpsteer for this reason alone, cause you can definitely feel these effects on track.

Now, that being said - I doubt bumpsteer is responsible for power-on steering shudder. I'd look at all of your suspension bushings, steering rack mounts, and the ages-old strut dampers as a cause first. Usually when there is any kind of shudder it's something to do with the combination of all the rubber and damping (or lack thereof) in the system, from the tires up.

Also.. 70/30? Really?

[sperry]

Dude, thanks for all the info! I'm on my phone reading this so I'll have to go back over it later when I can think about it more clearly. I think I understand the relationship between the steering rod angle and the instant cente of the control arms, but I don't know that I understand what the instant center looks like on a mcstrut suspension. Plus, it's going to be a bit of work to properly measure the svx in order to really diagram what my issue is since I've never had to measure suspension geometry before. I'm sure my margin of error will be pretty big.

Just from driving the car though, I'm willing to bet the car is toe in under bump. If it wasn't, I'd expect it to be darty under braking what with the wide tires and bumpy braking zones at rfr. Since it's really stable under hard braking, plus all the understeer, toe in sounds most likely.

And yeah, the weight distribution is terrible. Probably more like 65/35 in reality, but it's bad enough that the rear suspension goes light when you put just the front of the car on jackstands.

[AtomicLabMonkey]

Yeah, measuring existing suspension geometry is tricky and you can't really do it accurately enough to realistically predict steer effects, because you need the exact location of every pivot center. When you're adjusting bumpsteer, changing the height of the tie-rod end by very small amounts affects your steer response, I mean like .050" or less is enough. That kind of measurement accuracy is not really possible with ball joints & tie-rod pivots since they're encased in their housings. For that exact info you have to either section an existing part to measure it, or have an engineering drawing for it.

You can measure all the geometry to construct a model and see how things generally work, it just won't be a predictive tool for bumpsteer.

McStrut geometry:



Basically the same as dual A-arms, but the UCA is replaced with the strut. The terminology in this pic is a little different; "Virtual reaction point" is the instant center, "Pivot radius" is scrub radius, etc. Also the tie-rod and LCA are co-axial in this front view which makes them hard to distinguish. But basically you draw a line from the upper strut bearing, perpendicular to the strut axis; the intersection of that line with the LCA axis is the instant center.

[AtomicLabMonkey]

Originally Posted by AtomicLabMonkey

There are pics that illustrate all this movement in my RCVD text, but I don't have a scanner here.

Pics attached. My 'scanner' consisted of taking a pic with the digital camera.

[sybir]

Originally Posted by AtomicLabMonkey

Pics attached. My 'scanner' consisted of taking a pic with the digital camera.

Cackalacky Tech™

(I do the same thing, lol.)

[knucklesplitter]

Good info in this thread. Thanks.

Originally Posted by sybir

Cackalacky Tech™

(I do the same thing, lol.)

Me too... and not 'cause I lived in South Cackalacky for 10 years .