Roll Center Height/Roll Resistance

[johnc]

From Mark Ortiz's chassis newsletter.

 

ADJUST ROLL RESISTANCE DISTRIBUTION WITH GEOMETRY OR SPRINGS/BARS?

My newly finished track day car, based on a Porsche 914 chassis, has a front RC height adjustment separate from the ride height adjustment. Track testing is showing me that the car wants a more rearward biased roll stiffness distribution than I had anticipated when selecting the road springs and ARB’s. I could add roll stiffness to the rear but I am considering lowering the front RC as an alternative. It was during that thought process I remembered that you had addressed the pros and cons of that relationship. Is my memory correct?

 

I have discussed various things that relate.

 

One in particular is the possibility of roll center height changes having unintended or unanticipated effects in strut suspensions, especially in cars that have strut suspension only at one end. The 914 has strut suspension in front and semi-trailing arm suspension in the rear.

 

To lower the front roll center in such a system, without changing the ride height, one has to either raise the ball joints or lower the inboard pickup points for the control arm. Either of these increases the length of the front view projected swing arm, and reduces camber recovery in roll.

 

In general, strut suspension forces us to have less camber recovery in roll than we would like, just to keep roll center height and steering axis inclination within reason. Adjustable ball joint or pickup point height is useful primarily to correct for the effects of lowering the car for competition, and it does provide some limited freedom in choice of roll center height. However, it cannot provide an escape from the fundamental limitations of the design.

 

Lowering the front roll center will also increase the amount of roll, absent other changes. If the front suspension has less camber recovery in roll than the rear, and we increase the amount of roll, the poorer cornering camber at the front will tend to dilute the understeer reduction from the change in roll resistance distribution. It may even happen that the effect from camber may outweigh the effect from load transfer distribution, and we may get more understeer rather than less.

 

A theory that one sometimes hears is that the geometric roll resistance distribution has a particularly large influence on entry characteristics. This is claimed to be due to the fact that the sprung mass takes time to roll, which delays the effect of elastic load transfer, whereas geometric anti-roll moments are present as soon as there is lateral ground plane force at the tires.

 

Let’s examine that. Two things make the roll displacement take some time: friction and inertia.

 

The friction consists of (largely unintentional) Coulomb friction in the pivots and sliding elements of the suspension system, and (intentional) viscous friction from action of the fluid in the dampers. The friction force is anti-roll (resists roll) when roll is increasing, does not resist roll when roll velocity is zero, and is pro-roll (acts to maintain roll) when roll is decreasing. The frictional forces will add or reduce wedge according to their front/rear distribution.

 

The inertia is the sprung mass roll inertia. It acts in opposition to roll acceleration. It can become highly significant in tall, softly sprung, lightly damped vehicles subjected to abrupt inputs. If the car rolls rapidly to some angle, holds that roll angle for a time, then de-rolls, roll velocity:

1. starts at zero,

2. increases to some value outward with respect to the turn,

3. then decreases to zero,

4. stays at zero for a time,

5. then increases inward with respect to the turn,

6. finally decreases again to zero.

 

Consequently, roll acceleration:

1. starts at zero,

2. is outward with respect to the turn for a while, first increasing and then decreasing,

3. passes through zero and then becomes inward with respect to the turn, again increasing and then decreasing (roll velocity is outward but decreasing, so roll acceleration is inward),

4. stays at zero for a time,

5. is inward with respect to the turn, first increasing and then decreasing,

6. passes through zero and then becomes outward with respect to the turn, again increasing and then decreasing, finally to zero at the conclusion of the maneuver.

 

During phases 2 and 3 above, the frictional forces act against roll. Since they occur within the suspension system, they must react through the tire contact patches, and they contribute to lateral load transfer in proportion to their front/rear distribution. More rear damping dynamically de-wedges the car and adds oversteer. The frictional forces act in parallel with the front and rear elastic and geometric anti-roll components. Therefore, their influence depends on their comparative magnitude, relative to the elastic and geometric moments.

 

During phases 5 and 6 above, the frictional forces act against de-roll: they act to maintain the roll displacement, and are thus pro-roll forces. More rear adds dynamic wedge, and adds understeer.

 

During phase 2, roll inertia is anti-roll in direction. During phase 3, it’s pro-roll. During phase 5, it’s pro-roll (anti-de-roll). During phase 6, it’s anti-roll (pro-de-roll).

 

The contention that geometric anti-roll has a disproportionate effect on entry behavior is based on the idea that the geometric forces are directly related to ground plane lateral force and therefore are not diminished by the anti-roll effect of roll inertia during phase 2.

 

I think that makes some sense, qualitatively. But then what about the effects of roll inertia in phases 3, 5, and 6? And how big is the effect really, in race cars as actually driven on road courses or ovals?

 

In a tall, softly sprung, lightly damped sedan or truck, undergoing a J-turn test at the test track, we might see the effect of roll inertia as a reduction of roll in phase 2, followed by an exaggeration of roll as the vehicle approaches steady-state cornering. There should in fact be a degree of roll overshoot and subsequent de-roll as the vehicle approaches steady state. With really light damping, the vehicle might exhibit roll oscillation as the driver tries to get the vehicle to steady-state cornering.

 

 

 

But do we see roll overshoot during late entry in race cars, ordinarily? Do we see noticeably higher roll velocities during late entry than during early entry? Not usually.

 

Race cars have relatively small roll displacements and velocities, and relatively stiff damping. The main thing that slows roll displacement during entry in a race car is damping, not roll inertia.

 

So, what of the effect of lowering the front roll center, versus adding rear anti-roll bar? For identical steady-state understeer gradient, will either option produce noticeably freer entry in a road race car? Probably not – but the lower front roll center will produce larger roll displacements and velocities, and will therefore make the car more responsive to damper settings than adding rear bar.

 

And, returning to the first question, if greater front percentage is accompanied by more geometric roll resistance at the rear (e.g. higher Panhard bar), will the car turn in more readily? Maybe, if the rear is damped less in roll than the front, which is common with beam axle rear ends.

[HowlerMonkey]

In 914s we used to use high rate torsion bars and only an adjustable front bar to fine tune the roll stiffness distribution.....no rear bar at all and some guys experimented with running no sway bars.

[Leon]

Thanks again, John. His calculus threw me off (the acceleration 1-6 bit is wrong) but the rest was an interesting read as usual.

[letitsnow]

Is it actually reasonable that increasing the roll stiffness can add grip by reducing camber loss? That is, assuming a somewhat softly sprung car to begin with.

[HowlerMonkey]

914 tech doesn't transfer well to cars with a different configuration.

 

I drove a 914/6 with steel GT flares and 8 inch fuchs and went through many torsion bar/sway bar combinations.

 

Getting rid of the rear one once spring rates went up helped.

[johnc]

The question was about a 914 but the answer is generic and applies to all cars.

 

Is it actually reasonable that increasing the roll stiffness can add grip by reducing camber loss? That is, assuming a somewhat softly sprung car to begin with.

 

Yes. The Golf GTIs racing in SSC in the early 1990s are perfect examples of that.

[randy 77zt]

On my old z (I sold it) I raised the front roll center by relocating the control arm mounting hole on the cross member up 3/4" and increased the track width by 3/4" with the relocation mod.I welded a thick washer on both sides of the new hole for more strength.I didnt use those aftermarket bump steer spacers.I heated the outer tie rod ends to get rid the bend on the end and make the tie rods run straight.When the car was on its wheels the front control arms ran parallel to the ground.The tie rods ran parallel to the lower control arms.Actually at normal ride height my front lower control arms were lower at the ball joint end very slightly.I came to this idea by reading the old book called how to make your car handle.And after lowering a few fox body mustangs and finding they looked cool but still handled like dog doo.After playing monkey-see-monkey-do it took me 3 years of track days at 3-4 days a year of parts swapping to be able to run decent lap times and to like the car setup.I know changing the roll center is not allowed by scca and nasa for the improved touring classes.But I was building a track day/street car and didnt care.

[Snailed]

I heated the outer tie rod ends to get rid the bend on the end and make the tie rods run straight.

 

Why? That makes no sense to me.

[letitsnow]

Yep, the pivot points are what's important, the tie rod could have a loop in it and act the same way if the pivots are in the same place.

[heavy85]

Is it actually reasonable that increasing the roll stiffness can add grip by reducing camber loss? That is, assuming a somewhat softly sprung car to begin with.

 

Dont know anymore but many moons ago the hot stock fwd autox setup ran huge front bars even though they were front heavy. Rules limited spring & rear bar changes so front was all they had to work with. Theory was the improved camber control due to limiting roll was the dominant factor and produced the best lap times.

 

I'm starting to think my car (240Z) is limited by camber control more than anything. It's running 425F / 375R springs, stock front bar, no rear bar, front inner pivot up 5/8", front 3/4" 'bumpsteer' spacers. Went to stock front bar this year as the big front bar had braking issues where it would want to lock the inner front unless braking perfectly straight and it tended to push center off. 3.5 deg camber up front and 2 deg rear on RS-3 street tires and I'm still wearing off the out edges and while some better it is still pushing corner off more than I would like. This is on road courses too not tight autox where one tends to abuse the edges more. It's still quite fast but lays over more than I think it should to produce highest possible grip. This is actually my biggest goal for the winter off-season to figure out and build whatever I need to make some gains here. I'm thinking more track width and more caster (from just over stock now) is the starting point but am still very much in the early planning stages and open for discussions.

 

Cameron

 

PS - finally getting around to re-read Carroll Smith's 'Tune to Win' again (last time was ~15 years ago) and highly recommend it as a very readable, insightful book that's still relevant 30 years after it was written.

[johnc]

I'm thinking more track width and more caster (from just over stock now) is the starting point but am still very much in the early planning stages and open for discussions.

 

That's a good direction to go. Also, try removing the bumpsteer spacers and correct the bumpsteer in a more traditional way. If you increase the front track and get more caster you can soften the rear springs by 25 to 50 lb. in.

[heavy85]

That's a good direction to go. Also, try removing the bumpsteer spacers and correct the bumpsteer in a more traditional way. If you increase the front track and get more caster you can soften the rear springs by 25 to 50 lb. in.

 

I have the 'bumpsteer' spacers installed to raise roll center not for bumpsteer control. Rod end with shims on the tie rods are used to set bumpsteer. Why would you remove the spacers as that just lowers the roll center? How much track increase is meaningful - I'm thinking 1/2" to 3/4" per side but thats just a swag I pulled out of the air? Caster at ~7 deg?

 

Cameron

[johnc]

Bumpsteer spacers negatively affect camber gain in bump. If you want to get a better negative camber gain as the suspension compresses then remove the bumpsteer spacers. Since you're relocated the inner LCA mounting point and you plan on more caster, you probably don't need the bumpsteer spacers. Its worth experimenting at the track.

 

A total track increase of 1" (1/2 per side) is a good place to start. I would run +5 caster on a track car, depending on how wide your tires are. The problem with increasing caster and widening the track is that you're increasing scrub, which slows the car down - so you have to compromise on track width and caster. Scrub is an issue for a 200hp 240Z. Not an issue for a 400hp 240z.

[heavy85]

Bumpsteer spacers negatively affect camber gain in bump. If you want to get a better negative camber gain as the suspension compresses then remove the bumpsteer spacers.

 

Now I'm confused. I always thought but never tested the opposite. So I just measured it at ride height, +1", and +2" of bump travel.

 

w/ 3/4" 'bumpsteer' spacers camber changed -1 deg (-1" ride height) then - another 5/8 deg (-2" ride height) - Note 1/8 deg tolerance. LCA started 5 deg pointing down and ended 4 deg up - NOTE 0.5 deg tolerance.

 

W/out 'bumpsteer' spacers camber changed -7/8 deg (-1" ride height) then - another 5/8 deg (-2" ride height) - Note 1/8 deg tolerance. LCA started 1.5 deg pointing down and ended 8 deg up - NOTE 0.5 deg tolerance.

 

From my quick and dirty measurements camber gain is within tolerance the same with and without the spacers. So what's the real value of the spacers in adjusting roll center? My theory is I need as high roll center as I can get because with 25" tires need all the lowering I can get. Is there something I'm missing? To be honest I've never tested without them but want to make calculated changes.

 

The problem with increasing caster and widening the track is that you're increasing scrub, which slows the car down - so you have to compromise on track width and caster. ...Not an issue for a 400hp 240z.

 

I am planning on longer LCA instead of wheel spacers to avoid adding scrub. Yeah I'm closer to the latter (365 whp) so scrub is not a big concern in terms of losing momentum.

 

Cameron

Edited October 26, 2012 by heavy85

[JMortensen]

It's pretty linear according to Dan McGrath as well:

 

 

From my quick and dirty measurements camber gain is within tolerance the same with and without the spacers. So what's the real value of the spacers in adjusting roll center? My theory is I need as high roll center as I can get because with 25" tires need all the lowering I can get. Is there something I'm missing? To be honest I've never tested without them but want to make calculated changes.

If your control arm is pretty flat, I'd keep them. Underground roll center will roll more, and I still think you might be undersprung.

 

If you're still wearing the outside of the tire, I would say more neg camber or stiffer springs. Getting the tires flat midcorner is the name of the game. Are you taking tire temps?

 

 

 

[johnc]

Now I'm confused. I always thought but never tested the opposite. So I just measured it at ride height, +1", and +2" of bump travel.

 

w/ 3/4" 'bumpsteer' spacers camber changed -1 deg (-1" ride height) then - another 5/8 deg (-2" ride height) - Note 1/8 deg tolerance. LCA started 5 deg pointing down and ended 4 deg up - NOTE 0.5 deg tolerance.

 

W/out 'bumpsteer' spacers camber changed -7/8 deg (-1" ride height) then - another 5/8 deg (-2" ride height) - Note 1/8 deg tolerance. LCA started 1.5 deg pointing down and ended 8 deg up - NOTE 0.5 deg tolerance.

 

From my quick and dirty measurements camber gain is within tolerance the same with and without the spacers. So what's the real value of the spacers in adjusting roll center? My theory is I need as high roll center as I can get because with 25" tires need all the lowering I can get. Is there something I'm missing? To be honest I've never tested without them but want to make calculated changes.

 

 

 

I am planning on longer LCA instead of wheel spacers to avoid adding scrub. Yeah I'm closer to the latter (365 whp) so scrub is not a big concern in terms of losing momentum.

 

Cameron

 

OK, your numbers are different then what I got years ago - although I think I checked the numbers on a car with the stock position LCA inner mount. On your car I would less included to remove the bumpsteer spacers.

[heavy85]

and I still think you might be undersprung.

 

If you're still wearing the outside of the tire, I would say more neg camber or stiffer springs. Getting the tires flat midcorner is the name of the game. Are you taking tire temps?

 

Im not too concerned with the tire wear in and of itself but its telling me that it may be camber control thats needed to help the center off push. Agree with flat mid corner. Already at 3.5 deg camber and 425 lb on street tires. Sure seems like geometric camber control is where I need to focus. So instead of springrate add track width and instead of more static camber add caster. I used tire temps when I first started running these tires but that was two years ago and I didnt take any temps this year. Need to take them again as I have already made ARB and aero changes since I last measured. When measuring camber curves I also noticed I only have 2" of bump travel before starting to kiss the bumpstops (Koni foam ones) so I definitely need to cut then down another inch.

 

Cameron

[JMortensen]

Yeah, I got focused on the camber curve there for a minute. More caster is definitely a good answer (caster was one of the biggest changes I made to my car) and wider track will help too.

 

You might look at using different bumpstops and maybe not focus so much on not hitting them. I always thought hitting the bumpstops meant that you didn't have enough spring or you had too much rebound damping, but after getting into Miatas I'm finding that from the factory they are riding on the stops A LOT, and even with the higher spring rate coilover setups the bumpstop is considered much more of an active part of the suspension. You might try FatCat Motorsports and see what they suggest for bumpstops. Or to restate: if you cut them down and still hit them the effect of hitting them might be more dramatic and hurt handling more than having a thicker, better stop.

[clarkspeed]

I say invest in a geometry simulation program. I've been playing around with one for a few months now along with re-reading many books. You can change that spacer, lengths of CA, inner pivot points, etc. and see the results in 2 minutes. Some things have a big effect, others do not. It will point you in the right direction.

[heavy85]

I say invest in a geometry simulation program. I've been playing around with one for a few months now along with re-reading many books. You can change that spacer, lengths of CA, inner pivot points, etc. and see the results in 2 minutes. Some things have a big effect, others do not. It will point you in the right direction.

 

What are you using?