Whiteline Chassis Talk 8 - Roll center, what is it and why is it important to me?

[Arnie]

Copy of another thread from 2004 tracking a discussion of roll center, roll axis and its relevance to suspension and dynamic setup.

Started by Arnie Medel, WRX and STi specialist based in SoCal and frequent contributor to various forums and boards.

-------------------------------------------------------------------------------------
There is always talk about roll center and its importance in suspension setup. Could you please explain to us, using diagrams if possible:

What roll center is?

Why is it important and how does affect me?

How suspension geometry/components affect it? (do camber plates, height adjustability independent of spring preload type struts, camber bolts, have an effect on roll center?)

The affects of lowering on the roll center, also known as "why is a subterranean roll center bad"? (What is too low? Is there such a thing as being too low in regards to roll center?)

Thanks!

[Whiteline]

G'day everyone,

Arnie, you certainly know what questions to ask :) .

A proper answer to your post is beyond the scope of this thread but will be moved onto the short list of potential "Whiteline Chassis Talk" topics. In the meantime in great shorthand;

- What roll center is?

A vehicles roll centre (RC) is the theoretical point (vertical height relative to road) around which the vehicle will roll, either front or rear. Intersection of these points from rear to front determines roll axis. Calculation is relatively complex and imprecise, particularly on modern multilink systems.

- Why is it important and how does affect me?

The position of the RC(s) is very important in that its height relative to other factors will determine how much and in what way the vehicle will tend to roll in any given circumstances. Its height can vary from below ground (remember it is a theoretical point) to above the CG (centre of gravity). When compared with the CG, we can determine the "roll couple". Imagine an inverted pendulum where the pivot represents the RC and the pendulum weight itself represents the CG. Contemplate this with varying lengths for the same weight, then superimpose a theoretical axle line below the pivot at varying heights.

In general, the lower the RC relative to the CG the more the car will want to roll (at that end) no matter what you do. The lower the RC the greater the propensity to roll.

-How suspension geometry/components affect it? (do camber plates, height adjustability independent of spring preload type struts, camber bolts, have an effect on roll center?)

Though increasing complex to calculate with multilink especially, one reasonably simple example is found with MacPherson strut equiped suspensions as the strut itself forms one of the vectors for calculation of the angles. Hence, relocation of upper mount or lower mount position will change static roll centre. That is, camber adjustable strut top will change roll centre when changing camber. Lengthened or shortened lower arms will also change roll centre where as camber bolts fitted to the hub have no effect.

- The affects of lowering on the roll center, also known as "subterranean roll center bad"? (What is too low? Is there such a thing as being too low in regards to roll center?)

The angle of the lower control arm relative to the ground is also part of the calculation on MacPherson strut equiped vehicles. Excessive lowering will almost always invert this angle leading to more roll, sometimes exponentially if this results in a "subterranean" RC. (Very bad from a handling point of view)

MacPherson strut equiped vehicles are reluctantly used for racing because one fundamental drawback of the setup is that it delivers a "migrating roll centre", generally down. That is, the lower the ride height or the more a wheel is compressed the lower the RC becomes, exacerbating roll which further lowers the RC.... (Note that roll compresses outside wheel suspension further lowering RC as it does).

Playing with vehicle "rake" or front to rear ride height affects the roll axis which is fundamental to the inate balance of the chassis. That is, lowering the front excessively relative to rear (most common mistake with cosmetic lowering as people try to even up tyre to guard gaps) slopes the roll axis further down to the front creating all manner of on-road changes even touching spring or bar rates etc.

Cheers
Jim

Whiteline Automotive

[Arnie]

Originally posted by Whiteline

The position of the RC(s) is very important in that its height relative to other factors will determine how much and in what way the vehicle will tend to roll in any given circumstances. Its height can vary from below ground (remember it is a theoretical point) to above the CG (centre of gravity). When compared with the CG, we can determine the "roll couple". Imagine an inverted pendulum where the pivot represents the RC and the pendulum weight itself represents the CG. Contemplate this with varying lengths for the same weight, then superimpose a theoretical axle line below the pivot at varying heights.

In general, the lower the RC relative to the CG the more the car will want to roll (at that end) no matter what you do. The lower the RC the greater the propensity to roll.


So would it make sense to say that, theoretically, getting the roll center at the same height as the CoG would be ideal?

[Arnie]

Originally posted by Whiteline

-How suspension geometry/components affect it? (do camber plates, height adjustability independent of spring preload type struts, camber bolts, have an effect on roll center?)

Though increasing complex to calculate with multilink especially, one reasonably simple example is found with MacPherson strut equiped suspensions as the strut itself forms one of the vectors for calculation of the angles. Hence, relocation of upper mount or lower mount position will change static roll centre. That is, camber adjustable strut top will change roll centre when changing camber. Lengthened or shortened lower arms will also change roll centre where as camber bolts fitted to the hub have no effect.

Whiteline Automotive

So are you saying that the popular type of adjustable coilovers that have the height adjustment independent from preload, like the high end JIC, TEIN, ZEAL, etc. can actually worsen one's static roll center? That by adjusting the ride height from the secondary "threaded tube" one is affecting the roll center? I can't say I've seen Bilstein or Öhlins utilize this type of ride height adjustment, just the Japanese setups.

In addition using camber plates also affects roll center? It doesn't seem like something as minimal as the adjustment range of a camber plate will really affect roll center to the point that one could see any additional roll. Or is the effect magnified from this point outwards?

So to get more camber up front, running double camber bolts is the best way to go despite the inconvenience compared to using a camber plate?

[Keith Watson]

Good thread, Arnie.

Over on the NASIOC forum John Felstead posted this (http://forums.nasioc.com/forums/showpost.php?p=5200294&postcount=417) about rake:
Ride height is very important on the Impreza, they dont react well to lowering and in fact the lower you drop them, the more they roll. Best handling i have found is using stock TypeRA ride height at the front and have the rear sitting 3mm higher than the front, to give some positive rake, that helps to generate front end bite on turnin.A crew chief friend of mine said an important thing about roll center height is how the amount of roll generated affects the camber gain. I'm still unclear about what happens when.

... slopes the roll axis further down to the front creating all manner of on-road changes even touching spring or bar rates etc.Can you explain further? I'm still unclear about what happens and why. Let's assume the roll axis is relatively close to a balanced chassis. What affect(s) would lowering the rear ride height (roll center) have? If I'm understanding correctly it will cause more roll at the rear, give the inside rear wheel more camber. What happens to chassis balance (adds understeer/oversteer)? Weight transfer?

[DuncanG]

Originally posted by Arnie
So would it make sense to say that, theoretically, getting the roll center at the same height as the CoG would be ideal?

As with most things you can have too much of a good thing. Two problems I'm aware of with too high a RC with Macpherson struts (there may be more):

1) track changes significantly on suspension movement and can increase tyre wear and somewhat reduce grip.

2) more important - if the RC is too high the car will experience a jacking effect on cornering. Its as if its tripping over its own wheels and can cause a significant reduction in grip and inconsistent feel. Its analogous to having too much anti-dive/lift geometry (in the pitch axis) but worse.

A general rule of thumb with strut suspension is to set the height so that the lateral control arms are level. That minimises track and camber change for small deflections. However be aware that the outer ball-joint on the front of a Subaru is offset a good deal above the lateral arm and so the front RC will be much lower than you'd think judging by the level of the arm.

[Jim]

G'day everyone,

Keith, the position of the RC and its interplay with the CG determines the roll couple. All three are important and will affect how the car will roll which has an affect on camber.

In very simple terms, a relatively high RC will tend to promote jacking of the body as it tries to roll over the tyre where as a subterranean RC will see the body roll WITHIN the wheel track. The later will tend to compress the loaded wheels suspension more than the former. Naturally the position of the CG and the resultant roll couple is very important in this calculation but its rare that the CG is lower than the RC.

Lowering the rear ride height may raise or lower the RC depending on the geometry but it must obviously lower the CG. Though it will lower on the rear of a GD WRX, it is not necessarily obvious what would happen with respect to the change in RC. However, all things being equal, a reduction in CG is always helpful from a handling point of view. The problem is that in most systems, lowering CG is almost always done through lowering of ride height which will almost always result in changes to RC. That's one of the reasons why race teams try to get their cars seriously "under weight" so they have a choice where they put back the weight in the form of ballast. Needless to say they will place it as low as possible to lower the CG with out affecting the RC.

Theoretically, a proportionate reduction in the height of RC will result in an increase in roll. The dynamic outcome of that will depend on many things not least of which is the inherent camber and bump steer curve of the affected wheels and the resulting changes to roll couple.

Duncan, it is rare that you will find level control arms from OE as that will almost always lead to a relatively high low static RC. That is, with the stock inward inclination of the strut and the inevitable compression of the loaded wheel in roll, most strut suspensions will see a lowering of the RC further promoting more roll. This must not be confused with race setups where teams can design suspension geometry from scratch allowing a nominally control arm position with minimal suspension travel either side.

If anything they will almost always be inclined up towards the wheel with the GD WRX ball joint centre being around 10-15mm above the lower control arm pickup points (equates to around 5% relative difference). We actually sell roll centre adjusters for Celica's etc for that reason as even a mild 35mm height reduction is enough to bury the RC. Equally, we sell roll couple adjusters for other vehicles as OE engineers increasingly use these positions to artificially change the actual and perceived balance of the vehicle. Double wishbone or upright/wishbone hybrids are a different thing entirely.

The attached table is an extract from an internal document mapping the position of the RC on the GD platform. It shows the migration of the RC relative to ride height and suspension travel. Note how the front RC reaches ground level at around 35 mm lowering height while the rear with the pronounced downward roll axis reaches the ground at around 70 mm compression. Beyond these levels the RC buries deeper below ground which I suspect is largely the reason why John Felstead quoted what he did. Its also a very good reason why we try to avoid excessive lowering on most cars as the WRX is not unique in this respect.

Its effectively our last full day at Whiteline as closed down for the holiday period. We'll be back on Tuesday 4th January and would like to wish everyone the very best for Xmas and the new year.

Looking forward to talking to you all again in 2005

Kind regards
Jim Gurieff

Whiteline Automotive

[Keith Watson]

I had incorrectly been thinking of the roll center as an independent thing. Now I see how the pieces fit together. Fantastic explanation. Thank Jim!

Happy Holidays!

[rbahr]

Hi All,

First thanks for a GREAT thread. OK now on to my questions and thougths.

On a MacPherson strut equipped car, as I understand it, the roll center is determined by a number of things - but the height of the strut is one of the components. If a strut were built with a camber plate so that the TOTAL length of the finished product was the same as the OEM product, AND the lower perch were located in the same position, the RC should NOT change. NOW if we lower or raise the car we will move the roll center, and also add bump steer. How much change are we really talking about here (with the addition of camber plates)?

The reasons I ask are that I track my car a fair amount and need to add more camber in the front (and rear) and have been considering coilovers with camber plates. I REALLY don't like the idea of camber bolts - probably silly but I worry a bit about replacing the 17mm bolts with 14mm bolts and in addition they frequently seem to come loose...

Jim - I have some questions about your products - could you send me an email offline: rbahr@yahoo.com.

Thanks

Ray Bahr



That said,

[Arnie]

Ray, as I understand it from a conversation with Jim, its not the height or thickness of the camber plates its the change of the position of the top of the strut that alters the suspension geometry thus affecting the roll center. How much the roll center is affected by that lateral movement of 50 odd mm inwards and outwards that the camber plate provides is the big question.

[rbahr]

They both play a role - The instantious roll center is described by the intersection of a line at 90 deg to the top of the strut that extends out meeting a line that runs parallel through the lower A arm(s).

The point here is that we are changing this top angle by no more than 2 degrees. I would expect that the bigger component would be the change in height which directly affects bump as well as the RC as well as.... But then we get that when the shock compresses...

Ray

[Wojtek]

Hi All,

I'll try to describe this in a different way.

On a car with a MacPherson strut suspension there are 2 main suspension components (that determine roll centre)-
(1) - lower control arm (LCA)
(2) - Macpherson strut (strut)
There are also 3 pivot points-
A - lower control arm, inner (chassis)
B - lower control arm - to - Macpherson strut (ball joint)
C - MacPherson strut, upper pivot (top mount bearing)

As we all know by now, the roll centre is a theoretical line determined as follows;
1. extend A_C
2. draw a perpendicular line to B_C at C.
3. D = intersection of points 1 and 2 above.
4. draw a line from D to the centre of tyre footprint (on the same side as the strut) E. we now have line D_E.
5. Roll centre is the intersection of line D_E through the centre of the car.

So, points that alter roll centre are A and C. B doesn't change. If a car is lowered, it is point A that changes.
Changing the ride height effects the height of points A and C, and hence point D and roll centre.
Changing the height of top mount bearing (by fitting top mounts of non-factory dimentions) changes point C, and hence point D and roll centre.
Changing the inclination of the strut changes point D, and hence roll centre.

Which change has a bigger effect on the roll centre depends on individual cases.

Regards,
Wojtek.

[Arnie]

wojtek- thanks for the geometry lesson! So i guess the only way to find out how much one's camber plates mess with the roll center is to take careful measurements of all angles and calculate for ourselves?

[Wojtek]

Hi,

Arnie, yes that is what you would need to do. Carefully measure the actual car you are working on, and draw the front suspension.
I believe that there are some software programs that can help you with that - do a search on google.

Cheers,
Wojtek.

[Conecrusher]

Originally posted by Wojtek
Hi All,

I'll try to describe this in a different way.

On a car with a MacPherson strut suspension there are 2 main suspension components (that determine roll centre)-
(1) - lower control arm (LCA)
(2) - Macpherson strut (strut)
There are also 3 pivot points-
A - lower control arm, inner (chassis)
B - lower control arm - to - Macpherson strut (ball joint)
C - MacPherson strut, upper pivot (top mount bearing)

As we all know by now, the roll centre is a theoretical line determined as follows;
1. extend A_C
2. draw a perpendicular line to B_C at C.
3. D = intersection of points 1 and 2 above.
4. draw a line from D to the centre of tyre footprint (on the same side as the strut) E. we now have line D_E.
5. Roll centre is the intersection of line D_E through the centre of the car.

So, points that alter roll centre are A and C. B doesn't change. If a car is lowered, it is point A that changes.
Changing the ride height effects the height of points A and C, and hence point D and roll centre.
Changing the height of top mount bearing (by fitting top mounts of non-factory dimentions) changes point C, and hence point D and roll centre.
Changing the inclination of the strut changes point D, and hence roll centre.

Which change has a bigger effect on the roll centre depends on individual cases.

Regards,
Wojtek.

Wojtek;

Do you mean to define Line 1 as A_C or A_B? I belive A_B (an extension of the lower lateral arm) is what you mean.

Please see this link for an illustration of what I mean. Link - http://www.rqriley.com/images/fig-14.gif

Regards

Mike Barna

[Arnie]

Originally posted by Arnie
So are you saying that the popular type of adjustable coilovers that have the height adjustment independent from preload, like the high end JIC, TEIN, ZEAL, etc. can actually worsen one's static roll center? That by adjusting the ride height from the secondary "threaded tube" one is affecting the roll center? I can't say I've seen Bilstein or Öhlins utilize this type of ride height adjustment, just the Japanese setups.

In addition using camber plates also affects roll center? It doesn't seem like something as minimal as the adjustment range of a camber plate will really affect roll center to the point that one could see any additional roll. Or is the effect magnified from this point outwards?

So to get more camber up front, running double camber bolts is the best way to go despite the inconvenience compared to using a camber plate?

Wojtek or Jim, could you address these points as well?

[Keith Watson]

Hi Arnie,

I was going to take a shot at answering your camber plate question until I started doing my own sketches based on Mike's gif. I thought I had concluded one thing and now I've gone the opposite direction. This is difficult for me to grasp and I'm still shaky on it.

But it looks like moving the top of the strut tower inwards moves the roll center upwards. How much depends on the ride height and the resulting angle of the lower control arm. It appears that the lower the ride height the less the gain in the roll center height.

The angle between the strut and the line from the strut top to the instantaneous center is always 90 degrees. So leaning the top of the strut inwards always brings the distance to the instantaneous center inwards, which always raises the roll center.

Doesn't having the roll center and CG closer together reduce roll? Am I missing something?

Of course I don't have a clue about what might be happening during roll. :eek:

Hmm, maybe the best answer for this specific platform is to get suspension analysis software and get measurements off the car.

Regards,
Keith

[Wojtek]

Hi,

Mike,
Yes, you are correct. I have made a mistake with the letters. It is the extension of the lower control arm, so it is line A_B.

That is a great diagram too. Much easier to explain.

Arnie,
With reference to coil-over kits with adjustable lower brackets, this only effects roll centre IF the height of the top mount is altered.

Mike,
Yes, you are correct. In general, moving the top bearing inboard (effectively increasing negative camber) will raise the roll centre, all else being equal.
The distance between roll centre and CofG is the roll couple. The smaller the roll couple (roll centre and CofG closer) the less roll there will be. But like with everthing in life, there are compromises. Some of the less desirable effects of small roll couple are vehicle jacking, change in wheel track, and less compliant suspension. It is similar to what anti-squat is in pitch.

Regards,
Wojtek.

[Keith Watson]

OK, that makes sense. I figured I was missing something. :)

Thanks for all the great information you provide!

Regards,
Keith

[Arnie]

Ditto, thanks for the great feedback on this topic. I have to say that the Subaru community has come a long way since the days of "throw on a rear swaybar" as the blanket solution for all suspension woes. Thanks for the education.

[KevinM]

Originally posted by Arnie
Ditto, thanks for the great feedback on this topic. I have to say that the Subaru community has come a long way since the days of "throw on a rear swaybar" as the blanket solution for all suspension woes. Thanks for the education.

No kidding!

<--- 20mm STi sway bar was first suspension mod. :P

[Carnut]

Originally posted by Wojtek
Hi,

Mike,
Yes, you are correct. I have made a mistake with the letters. It is the extension of the lower control arm, so it is line A_B.

That is a great diagram too. Much easier to explain.

Arnie,
With reference to coil-over kits with adjustable lower brackets, this only effects roll centre IF the height of the top mount is altered.

Mike,
Yes, you are correct. In general, moving the top bearing inboard (effectively increasing negative camber) will raise the roll centre, all else being equal.
The distance between roll centre and CofG is the roll couple. The smaller the roll couple (roll centre and CofG closer) the less roll there will be. But like with everthing in life, there are compromises. Some of the less desirable effects of small roll couple are vehicle jacking, change in wheel track, and less compliant suspension. It is similar to what anti-squat is in pitch.

Regards,
Wojtek.

Please excuse my ignorance, but I am having trouble with the roll centre calculation for McPherson struts. The diagram posted has a perpendicular from the top of the strut. But we know that instantaneous roll centre postion is independent of link length. The link length only changes the way the roll centre moves with suspension travel. If the perpendicular from the top of the strut determined roll centre (link length influences roll centre), then simply doubling the strut length would raise the roll centre, and I have not seen that (yet).

[Keith Watson]

Originally posted by Carnut
But we know that instantaneous roll centre postion is independent of link length.The intersection of the two lines which determine the instantaneous center are not limited by control arm lengths. But the angles of those control arms are very dependent on control arm lengths.

Lets try a simpler case than doubling the length of the strut.

Increasing the ride height of the car makes the strut longer. Lets assume the angle of the strut remains the same. In the picture what moves? The (lower) control arm angle changes with the outer end being lower, which brings the instantaneous center in and up. So, increasing the strut length brings the instantaneous center in and up and raises the roll center.

[Carnut]

Originally posted by Keith Watson
The intersection of the two lines which determine the instantaneous center are not limited by control arm lengths. But the angles of those control arms are very dependent on control arm lengths.

Lets try a simpler case than doubling the length of the strut.

Increasing the ride height of the car makes the strut longer. Lets assume the angle of the strut remains the same. In the picture what moves? The (lower) control arm angle changes with the outer end being lower, which brings the instantaneous center in and up. So, increasing the strut length brings the instantaneous center in and up and raises the roll center.

Thanks Keith. I can see that changing ride height alters roll centre. What if a longer strut was put in that kept the same ride height because it was mounted higher. Severe and ugly modification of a strut sticking through the hood, but we are testing the theory here. Ride height would be the same, control arm angle the same, but by the diagram the roll centre has changed and I am not sure it has.

[Keith Watson]

I see what you are asking. I wasn't sure what the answer was so I cheated - I printed the diagram and drew some new lines. :D Now that I did that it should have been obvious to me. :rolleyes:

Moving the strut top higher means the line perpendicular to the strut is now higher but still parallel to the original line. Next extend the line through the lower control arm out further to the right. The instantaneous center is now located. Draw a new line from the instantaneous center back to the center of the tire. The roll center has dropped just a little.

In the diagram I drew on I made the strut 50% longer and the roll center barely dropped.

By the way, raising the strut top is sometimes done to lower a car but still keep the same amount of shock travel. A friend did the front of his wife's BMW race car this way.