Roll Steer, Roll Axis and Roll Center and How They're Related

[The WidowMaker]

i have a few questions for rrc, ra, and roll steer.

first, how in a c4l is roll axis related to roll steer. ive put my propeller hat on, but its not spinning. ive heard that the rear roll axis is related to the roll steer, but my assumption was that the angle of the lca determined roll steer and nothing else. and since the uca in a c4l determines the roll axis (??? see below), the ra can be changed without touching the lca's. the only way roll steer can exist is if one bar is shorter or longer than the other. this creates toe in on one wheel and toe out on the other which steers the rear. but, if they are level at ride height and a turn is entered, one bar moves up and one moves down an equal amount (this must be where my error is) and therefore the angles cancel each other out.

next question; rear roll axis is used to calculate the rear roll center, but the whole cars roll axis is drawn through the front and rear roll centers, correct?

what points are used to calculate rear roll axis on a c4l? my reading tells me its the convergence of the fronts of the rlca for one pt, but the information goes extinct from there. is it the uca housing mount? the convergence of the uca's in the trunk area? and if its either, how would it be any different? if i drew a pt from the ic through the uca housing mount, i would still end up at the convergence of the uca's, wouldnt i? now is the roll center a point vertical to the axle centerline, or is it that area in the trunk?

thanks for your help, its late and im going to bed now, cause my wife says!!!!

Tim

[Norm Peterson]

i have a few questions for rrc, ra, and roll steer.

first, how in a c4l is roll axis related to roll steer. ive put my propeller hat on, but its not spinning. ive heard that the rear roll axis is related to the roll steer, but my assumption was that the angle of the lca determined roll steer and nothing else. and since the uca in a c4l determines the roll axis (??? see below), the ra can be changed without touching the lca's. the only way roll steer can exist is if one bar is shorter or longer than the other. this creates toe in on one wheel and toe out on the other which steers the rear. but, if they are level at ride height and a turn is entered, one bar moves up and one moves down an equal amount (this must be where my error is) and therefore the angles cancel each other out.
Rear axle roll steer is the amount of axle roll about its own roll axis that shows up as steer, IOW axle angular change as seen in plan view. Any axis is a line, which requires precisely two points to define, and in this case the two points must represent points where lateral motion is effectively restrained.

In the case of the C4L (or any of its "relatives" such as a Satchell link), one point on the axle's roll axis is the virtual convergence point of the lowers, and the other is the virtual convergence point of the uppers. Where the axle's roll axis crosses the axle centerline in side view is the geometric (aka kinematic) roll center. The slope of that line is the amount of roll steer - if it is inclined at 5°, the slope is approximately 0.087, or 8.7%. Moving any of the pivots on either the uppers or the lowers whether up/down/sideways/forward, or rearward will affect the axle's roll axis and roll center, but moving some points in some directions has greater effects than other pivot point relocations. Actually, moving the axle side pivots of the UCA's upward causes greater changes in RCH per inch of relocation than any other pivot point move.

It's not the easiest construction to visualize once a little roll is introduced, since most people's minds "fight" the idea that the one plane containing the LCAs and the axle and the other plane containing the UCAs and the axle must necessarily warp as the car rolls, and that under roll neither the UCA nor the LCA constructions remain converged to a single point (they go "skew", separating vertically, and midpoints between the construction lines at what look like convergences in plan view seem to be points as good as any to use).


next question; rear roll axis is used to calculate the rear roll center, but the whole cars roll axis is drawn through the front and rear roll centers, correct?
Yes. But don't attach too much importance to the vehicle roll axis - the car does not actually rotate about that line except by either momentary coincidence or truly unusual suspension design.


what points are used to calculate rear roll axis on a c4l? my reading tells me its the convergence of the fronts of the rlca for one pt, but the information goes extinct from there. is it the uca housing mount? the convergence of the uca's in the trunk area? and if its either, how would it be any different? if i drew a pt from the ic through the uca housing mount, i would still end up at the convergence of the uca's, wouldnt i? now is the roll center a point vertical to the axle centerline, or is it that area in the trunk?
See item "first". Do not confuse the IC as used more for drag racing setups with the virtual convergence of the LCAs. They are not the same thing. For the IC, you're looking at the convergence of UCA and LCA projections in side view. For roll axis, it's the convergence of the LCAs in plan view and convergence of the UCAs in plan view, which you then link together in side view.


Norm

[The WidowMaker]

thanks norm, i was hoping you would reply.

i guess i still dont understand how the roll axis determines the roll steer. if the lca control the fore/aft movement of the housing, and they are of equal length throughout the entire movement of the housing (again assuming 0* to ground at ride height), how does the housing turn? any pictures, or sites with pics?

in order to decrease the angle of the roll axis and acquire a neutral steer (assuming pts form an angle down towards front), the front point needs to go up, the rear point needs to go down or the points need to stay on the same horizontal plane but move further apart. one of those solutions would be to move the lca front pt up. lets say i move it up so the arm was on a 3.5* up angle. by doing so, in a turn i could have 7* of differnce btwn the arms. this would lead to .219* of axle oversteer (assuming 22.5" arm length and 44" lca to lca housing mount), but my roll axis line would have gotten flatter therefore telling me i would have

caculating the roll axis makes a little more sense now. im not drawing point through the arms until the meet (uca meeting lca) but rather finding the point where one lca meets the other and where one uca meets the other and then connecting the pts. i thought it would be the same, but now i see that its not.

[Norm Peterson]

Take a pen or pencil that has a pocket clip on it. That'll be a suspension roll axis.

Slip a small thin ruler under the pocket clip - anything about 6" long and thin enough to get under the clip and rigid enough to hold its shape will do - so that it is perpendicular to the pen/pencil. That's your axle.

Hold the pencil over a piece of paper on your table (this is to make things more readily visible) at some angle to the paper (like you were going to write something) and look straight down at it. The angle you use doesn't matter for purposes of illustration.

Now roll the pencil without lifting or dropping its "high end" and observe how the ruler "steers". If you draw a line and line up the ruler with it (sighting down on it) before you roll the pencil, you can get a little better idea of how much steer you get with different roll axis (pencil) inclinations.

Be sure that you are keeping these thoughts which have to do with the axle's/suspension's roll axis separate and distinct from the vehicle's roll axis (which is a different animal entirely and not nearly as useful).


Norm

[jerome]

I don't think it was explicitly stated so,

If rear roll axis points up (towards front of car), that is roll oversteer.
If rear roll axis points down (towards front of car), that is roll understeer.

You will understand it better once you can think it through to get the right answer to the above two points. I understand it better if I imagine the car body level, and the rearend/ground rolling in a turn.

[The WidowMaker]

i see what you mean by the example, i just dont understand how it can do that with the lower arms in control of the fore aft movement. if the lca's are the same length at a given roll (whether one is + and one - or both the same), i dont see how one side of the housing can be any further forward. what am i missing?

[Norm Peterson]

Because the four links determine axle rotation in a plane that's perpendicular to the axle/suspension roll axis, which is not necessarily the same as a plane parallel to the ground (it usually is not). The links control the movement, and steer is a consequence of that movement. That's different from assuming that only two of the four links control steer and whatever overall axle movement that results is simply a consequence.


Norm

[jerome]

I'll assume the C4L you're talking about has angled upper arms and parallel lower arms.

Imagine you can "roll" the rearend in space and simulating chassis roll side to side. Keeping the the center of the rearend at the same point in space, you see that the location of the top of the rearend is determined by the upper arms and similar for the bottom of the rearend.

For simplicity, assume that the upper links both start at the same point on the rearend housing and extend outwards (from plan view to the chassis). In this example, the upper links only connect to the rearend at a single point, and are thus unable to determine the "steer of the rearend.

Now imagine the lower links in side view. Assume the lower links are parallel to the ground. To roll the rearend (with chassis pickup points constant), one arm must angle downwards and one arm must angle upwards. The arm that angles down pulls that side of the rearend not only lower, but also forward. The arm that angles up pulls that side of the rearend up and forward (the same amount as the other arm).

This scenario does not "steer" the rearend. However, things change when you assume that the parallel lower links are angled upwards from the rearend. One link moves down and pulls that side of the rearend forward. The other link moves up and pushes that side of the rearend backwards. (This is only true until the link moving upwards reaches horizontal. At that point it begins pulling the rearend forward again). The lower side steers forward and the upper side steers backwards. The lower side is the side that is on the inside of the turn (because car chassis rolls towards outside of turn). With the rearend angled towards the outside of the turn, the front and rear steer in opposite directions causing oversteer.

Best way to visualize is to draw one line for neutral position of both link, draw one line same length originating from same point, angled lower, and one angled higher. You'll see that the x-coordinate of the lower one is more forward and the x-coordinate of the higher one is more back (as long as its lower than horizontal)

[Norm Peterson]

However, things change when you assume that the parallel lower links are angled upwards from the rearend. One link moves down and pulls that side of the rearend forward. The other link moves up and pushes that side of the rearend backwards. (This is only true until the link moving upwards reaches horizontal. At that point it begins pulling the rearend forward again). The lower side steers forward and the upper side steers backwards. The lower side is the side that is on the inside of the turn (because car chassis rolls towards outside of turn). With the rearend angled towards the outside of the turn, the front and rear steer in opposite directions causing oversteer.
I agree with the [vehicle] oversteer conclusion for uphill parallel LCAs, but not those intermediate steps. If the LCAs are running uphill toward the chassis, it's the outer LCA that will drop at its chassis end due to roll, and at least initially push the outer wheel rearward. The inner LCA pulls its wheel forward as its chassis end rises.


Norm

[The WidowMaker]

jerome, your explanation of the lowers in side view is how i have always assumed it worked. any angle other than parallel to the ground makes the arm shorter. it doenst matter if one is +1 and the other -1 since they will still be the same length. but if both arms start at -1 and in a turn one could be -2 with the other being 0. this would cause a rear steer condition and this is how ive always calculated it expressed in degrees of toe per degree of body roll. but the roll axis is a concept i still cant wrap my head around. my only assumption at this point is that although the links themselves stay the same length, that because of axle roll the lca rear mount may move in relation to the axle centerline. if this happened, one side could become longer than the other.

either way, i have to do some more research and just take your info for fact at this point.


thanks, Tim

[The WidowMaker]

norm - that example is what ive known as the reason for roll oversteer. the outer gets longer and the inner gets shorter which points the rear towards the outside of the turn. if they were the same length in the turn because the angles cancelled each other out, then it would have neutral steer. again, im going to have to do some more research to figure out how roll axis controls it.

[jerome]

Left turn, right side of rearend comes closer to chassis (rotates upward) and pushes rearend backwards. Left side of rearend moves further from chassis (rotates downwards) and pulls rearend forward. Rearend drives toward right, and thus oversteer.

My example was unclear. We are saying the same thing, but I am imagining the rearend rolling while you are imagining the chassis rolling. My way is less intuitive (but somehow easier for me to understand), and your way is the way it actually happens.

[The WidowMaker]

that makes sense

here are my new numbers;

50% AS assuming 20" cog at camshaft center
90" IC
uca 6.38* down angle
lca .818* down angle
18.5" Roll Center
9.14% Roll Axis figured off a 116.5" lca convergence @ 7.88 above the ground and a 8.1" uca convergence @ 19.3 above the ground

my fear is that the AS is too small and the RC is waaaaayyyy too high. but my plan is for either a watts or a phb. will either of these work since there is such a big difference btwn where the rc is now and where i want it (12-15)? will there be too much non compliance? the plan is to swap out to rubber bushings in the housing.

this is with the stock style c4l on my 70 velle, with the front of the lca moved 1" up and the front of the uca moved 2" up and running a 28.5" rear tire. the car is lowered a ton, but i dont know where stock was so i dont know exactally where its at now.

thanks, Tim

[jerome]

The goals you have pretty much require something other than the C4L.

Lower bars aiming down is usually a no-no. It creates weird roll-steer effects.

With a C4L, it is impossible to get 100% antisquat and neutral steer at the same time. 100% antisquat forces lower bars to aim up, and that is roll oversteer.

What you want to do requires a 3link or a satchell link.

You can also get a better mix of SVSA length and roll center height.

[The WidowMaker]

well i could get the lca to point up slightly which would raise my lca convergence pt, decrease the angle of my roll axis, raise my AS and lower my IC length. but im still unsure of how this would affect roll steer. my original thought was that the bars pointing up would cause roll oversteer, but my roll axis is pointing down which would cause roll understeer (correct???). so until i figure that out, i guess im stuck.

dont get me wrong about my desired performance. i want it better than stock, but this is still going to be a street car. i really just want it to be predictable (read: roll understeer vs oversteer) and i dont want snap oversteer at 75+ in a turn. i am also going through this because i want to understand it. i dont want somebody to tell me what to do just cause, but i want to understand what is happening and why it is happening. that way i can begin to formulate my own opinions and realistic expectations.

so some more questions;

would you raise the AS to something closer to 100. maybe in the 75ish range?

any compliance issues btwn the high RC and the phb?

im going to swap the front sphericals on the uca to johnny joints. this shouldnt change the articulation capabilities, but it will help them last longer and theyll be a little quieter. now is the time to do it since i have to fab up some new stuff to run the front of the uppers higher.

thanks again for the help,
Tim

[The WidowMaker]

i guess im also sticking with my comment above

"my only assumption at this point is that although the links themselves stay the same length, that because of axle roll the lca rear mount may move in relation to the axle centerline. if this happened, one side could become longer than the other."

because the mount is 4" below the axle centerline (thats just a guess), if the housing was rotated 90* backwards so the pinion pointed up, the links would be 4" longer. this is again the only way i can see how you could get steer when the lengths are at the same or same but opposite angles to the ground.

any truth or do i need to keep looking?

tim

[Norm Peterson]

I think what happens instead is that the actual roll center (not necessarily the geometric/kinematic one that you construct) rises to suit the fixed length requirements of the links. If there is any rubber or poly anywhere, it distorts. And the axle itself might shift laterally, although I haven't tried to work out any sort of geometric proof of that.

As far as LCA inclination being the sole thing that determines roll steer, that only happens with lowers that do not converge in plan view. And the convergence does not have to occur ahead of the rear axle either. The S197 Mustang's LCA convergence is over 20 feet behind the rear axle.

What converging the lowers buys you is specifically the ability to dial in some anti-squat without driving the axle steer into vehicle roll oversteer or requiring steeply inclined uppers (resulting in a short side view swing arm). Converging them LETS you run them uphill toward the chassis.

IIRC, somewhere in the 50% - 65% antisquat range is preferred for cars being tweaked for road course use or similar driving. Part of the reason is that while anti-squat is a beneficial effect at the drag strip and can be used to affect forward bite coming off the corners (particularly in oval track racing), it has an evil siamese twin in anti-lift. Want more of one, you get more of the other. Too much anti-lift will make it much more likely that you will experience "brake hop". Not a nice effect when you're rapidly approaching a corner at speed and need to slow down RIGHT NOW. Admittedly, you'd have to be running pretty hard to experience this, but it does happen in some cars even at auto-X. 4th Gen LS1 F-bodies are especially prone, though in that particular case other factors are involved.


Norm

[jerome]

If you run a 3link, you can use a dual action upper link (actually two links, one has effect in compression, one has effect in tension) so that you get 100% anti squat without ill effects of brake hop, because the lower angle of the "braking" link makes a long SVSA


I have 100% antisquat on a 3 link with 50+inches SVSA, so it is possible to get a reasonably long SVSA length to reduce chance of brake hop, even without the dual action upper link.

For roll axis on a 3link, the roll axis is drawn between the PHB height and the point of convergence of the lower arms in side view. So you can have uphill lower bars for antisquat, but as long as they converge below the PHB height, you also have neutral or roll understeer. I have my PHB (and roll center) at about 13" with neutral steer.

Sorry if I sound like a **** pushing the 3link, but it's just my opinion that if you are doing so much work and calculations, you might as well do it all out.

Jerome