28

The Ultimate Guide to Suspension and Handling: It’s All in the Geometry- Part One, The Roll Center

By Mike Kojima

 

 

In the first parts of our suspension series we have covered basic suspension stuff. Now it’s time to bury ourselves in tuning suspension geometry. Geometry tuning is a step above the usual bolt on street parts. Making changes on this fundamental level is what racecar and suspension engineers do for a living, but we’ve found that with the more popular cars in this market there are parts available to help with these mods.  Some of you are also advanced enough to experiment with this as well.  

 

Be sure to check out the rest of the Ultimate Guide to Suspension & Handling series

 

The Roll Center

Roll center is the virtual pivot point in space that a car rotates around when subjected to cornering forces.  The roll center is significant because its location determines how a car will handle and what factors must be considered when tuning its suspension.

 

Finding the instant center is as easy as drawing some lines though the suspension pivots and extending them until they meet in space.  The top example is typically what you would find in a modern multi link car.  The bottom example shows that the instant center can be on the outside of the wheel as well.  The only type of car that I can think of being lake that is a current F1 car.  To find the roll center of a car like this you would have to draw a line bisecting the angle and extend it toward the center of the car.  I don't think any of us are going to design an F1 car anytime soon so we won't worry about this for now.

 

The roll center is located by first finding the front view instant center of each side of a car’s suspension. The instant center is the point in space that the suspension’s links will rotate around.  Drawing lines from the center of the ball joint through the inner pivots of the upper and lower control arms and extending them inward towards the center of the car until they meet will let you locate the instant centers. Now draw a line from the center of the tire’s contact patch to the instant center on both sides of the car.  The point where these two lines intersect is the roll center.  For a car with McPherson strut suspension, the upper line is found by drawing a line 90 degrees from the strut axis starting at the upper mounting point of the strut.

 

The roll center is found by drawing lines from the center of the tire contact patch to the instant centers.  When they intersect at the cars' centerline is the roll center.

 

In the case of extreme angularity of the links, the roll center can be under ground.  Although this isn't a show stopper and many race cars are like this in the front suspension due to a desire to have a lot of negative camber gain in the front suspension, it is not ideal.

 

Finding the roll center for a live rear axle is very different from finding it on an IRS car.  Live axles are common for domestic cars like Mustangs or trucks.  For a leaf spring car, the roll center is where the leaf springs bolt to the axle.  For a 4-link, it is the intersection of the links when viewed from the rear.  For a panhard rod, the instant center is about in the middle of the rod.  For a Watts linkage, it is at the center of the rocker.

 

Page 1 of 5 Next Page
Bookmark and Share
Comments
Series8217
Series8217link
Wednesday, November 28, 2012 12:40 AM
Nice! I like how you posted this at about the same time I finished reading the previous article. Will Part Two be ready in about 10 minutes? ;-)
8695Beaters
8695Beaterslink
Wednesday, November 28, 2012 6:40 AM
MotoIQ: Finding Engineering terms with no names and naming the shit out of them.
Julian ITR
Julian ITRlink
Wednesday, November 28, 2012 8:44 AM
I have been looking forward to this article for some time!

What I don't quite get is, if shortening the roll couple has similar effects on weight transfer as stiffening the suspension, then why are roll center corrections on FWD cars usually made only to the front (by extended ball joints)?
jeffball610
jeffball610link
Wednesday, November 28, 2012 9:20 AM
Since I'm doing so much work to my Datsun 510, I think I'm going to tackle doing some measurements and see where I'm at. The front is a pretty simple strut setup, but the rear has a semi-trailing arm configuration. Would I do similar measurements as the struts for the rears? The concept seems the same as a strut with a fixed point on top and the lower arm doing all the moving.
I'd also like to know how suspension engineers determine what a good spring rate or how to design a sway bar are done on paper. What's the starting point? I know they don't just make some stuff and see if it works. Where do they begin?
HudsonMC
HudsonMClink
Wednesday, November 28, 2012 9:29 AM
@jeffball610: You're lucky in that the modifications to correct 510 geometry in the front and rear and very well documented as the car has a long history with grassroots racers. Check out the510realm.com and search for "slotted rear crossmember" and "bump steer spacers" (actually mis-named roll center adjusters) and "280ZX struts" to get you started. dpracing.co, futofab.com, and datsport.com sell just about everything you would need to get you started. Troy Ermish and Techno Toy tuning also sells some pretty slick coilover setups that would be a great start.
Dan DeRosia
Dan DeRosialink
Wednesday, November 28, 2012 11:04 AM
@jeffball610: Spring rates and sway bars are a complicated topic, especially if you have control of the roll centers and overall suspension geometry too. I'm not a professional, but here goes for a basics version of that.

On a production car, spring rate's more driven by ride frequency more than anything - calculate the unsprung weight on each corner and figure out its natural frequency and that has a fair amount to do with a starting point on how the car feels. Then, knowing the roll centers and CG at each end, you can use that to figure out how much each end is transferring weight, which correlates with over/understeer balance. Sway bars are then used to tune roll stiffness to get the balance where you want it. On race cars, you can run the wheel frequencies higher (to some extent - too stiff only hurts you) and adjust the roll centers, which as Mike is pointing out in this has all sorts of dynamic effects. But that's the basics.

Dynamic effects gets to be where a lot of fun comes in; shocks affect how fast everything transfers weight as does roll center and various other things... and for that matter, influences how the car feels.

Regarding F1 cars pushing... it's interesting that watching practice sessions at Austin, they were actually sometimes deliberately kicking the rear ends out with power. I suspect that with the power/weight ratios and no traction control anymore, there's very little reason to tune for anything other than severe understeer at any speed aero's not in play. But F1 suspension is just goddamn weird; even besides the geometry being driven by aero, the huge tire sidewalls means that there's all sorts of extra gizmos that are applied to damp vibrations there. The mass dampers Renault came up with were pretty neat, and then there's the inerters that everyone has moved to.

On less exotic stuff, I need to dig up the setup book that an ex-GM engineer came up with for the 1st gen RX-7 - the basics of playing with roll centers and geometry is there, and he had some interesting ideas with ackerman, but also some tricks that I've not seen anywhere else.
Nyborg Garage
Nyborg Garagelink
Wednesday, November 28, 2012 11:40 AM
Great stuff, thanks!
Nyborg Garage
Nyborg Garagelink
Wednesday, November 28, 2012 11:50 AM
The mentioned "recommended" roll center height front and rear, would this be a be an "ok" starting point for a car that has to handle somewhat like a RWD world rally car. (Much softer springs than a typical race car, and also a little bit higher CG/ground clearance).

I'm thinking the roll center can be tuned to give the car less body roll than normal so that you could still run soft springs and soft sway bar (so that the car handles the ruff bumps in the track in a good way, but still has acceptable body roll).

I'm thinking of cars like Rhys Millen RWD Hyundai at the time Pikes Peak still had lots of gravel.
Rockwood
Rockwoodlink
Wednesday, November 28, 2012 12:19 PM
@ Julian ITR: playing with the roll centers up front on a FWD car is not only more effective there (more weight), it also means you don't have to run as much spring to counter roll. Less spring (to an extent) means less tire shock from bumps and more mechanical grip.

Basically, running more spring instead of fixing the roll center is a band-aid.

You don't see it out back as much because many FWD racecars are privateer efforts, and those limited dollars are better spent up front.
EB Turbo
EB Turbolink
Wednesday, November 28, 2012 2:06 PM
On the 510's, Is it common to relocate the front radius rods?
malibuguy
malibuguylink
Wednesday, November 28, 2012 8:59 PM
I need some help with my track day car...i'm willing to go custom control arms (i fab for a living) but I'm not sure about adding the distance between the knuckle & the balljoint. I could easily make a spacer...or will trying some form of rod-end joint conversion be better?
Mike Kojima
Mike Kojimalink
Wednesday, November 28, 2012 11:34 PM
Like I was saying, too high of a roll center causes a jacking moment so yes you can overdo it.
dilop
diloplink
Thursday, November 29, 2012 11:49 AM
Great info as always!! learning learning learning here!

Thanks Mike, you make this stuff looks like pre-school math..pretty easy to follow and understand.

HudsonMC
HudsonMClink
Thursday, November 29, 2012 1:48 PM
@EB Turbo: Yes, on severely lowered racecar 510s, folks will commonly relocate both the TC Rod (radius rod) and LCA pivot down an inch or so.
Pilun
Pilunlink
Thursday, November 29, 2012 3:34 PM
Great article on roll axis, but I couldn't help but notice here:

"On most cars the ideal location for the roll center is 2-4 inches above the ground for the front suspension and 4-10 inches above ground for the rear suspension with the rear roll center higher than the front. This is so the car will transfer more weight on to the front of the car due to an increase in geometric anti roll giving a more predictable tendency to understeer at the absolute limit. Most purpose built racecars are like this."

I think there might be a mistake here. If the rear roll center is higher and closer to the c.g., there will be more geometric weight transfer in the rear. This by itself promotes oversteer. My guess is that most people prefer a lower roll center in the front, because of less jacking effect on the steering end, and because the car will yaw slightly away from the turn, which is a more stable feeling.

Some mid-engine setups prefer a lower roll center in the rear to reduce geometric weight transfer in the rear, decreasing oversteer.
EB Turbo
EB Turbolink
Thursday, November 29, 2012 3:47 PM
@HudsonMC, I did a rack and pinion conversion on a 510. I didn't look on the internet to see what was common. I just did what would work.

I essentially took the lower suspension and did a 180°. I relocated the radius rods to the rear of the car with new mounts. This also made the car a front steer. I had more room to put the rack. I also slid the cross member backwards and cut the center out to put the LCA in double sheer.
Mike Kojima
Mike Kojimalink
Thursday, November 29, 2012 5:48 PM
Do your roll center heights like this and it will work.
Supercharged111
Supercharged111link
Sunday, December 02, 2012 7:44 PM
Back to this:

"On most cars the ideal location for the roll center is 2-4 inches above the ground for the front suspension and 4-10 inches above ground for the rear suspension with the rear roll center higher than the front. This is so the car will transfer more weight on to the front of the car due to an increase in geometric anti roll giving a more predictable tendency to understeer at the absolute limit. Most purpose built racecars are like this."

You don't specify which end to shoot for for your particular configuration: i.e. FF, FR, MR, RR. I believe you stated that the moment is smaller in the front of an FF than in the rear. Is this due to the additional mass in the front vs the rear, and is the factory's way of attempting to inject some response to an otherwise undersprung car? I'm dealing with a 1990 Corolla GTS here, wondering what end of the recommended spectrum to shoot for. It's quite front heavy at the moment too with the supercharger transmission, supercharger, intercooler, and all that crap. I'm wanting to make some custom lower control arms for it so I can copy the Sentra's setup while possibly widening the front and perhaps moving the lower ball joint forward a half inch or so. I think I'll first measure my caster and calculate the change to determine how far I move it.
DRIFT_S13
DRIFT_S13link
Sunday, December 02, 2012 8:36 PM
Unfortunately SPL doesn't list the front or rear LCA's on their website. As far as I know they are unavailable.
ZZang
ZZanglink
Monday, December 03, 2012 7:40 AM
page 3: "The converse is true for a short roll couple. If you have short roll couples, engineers say that you have a lot of geometric anti roll which has the same effect as stiffening the suspension, it increases weight transfer to the outside tire. This can increase over or understeer depending on what end of the car you do it to."

"it increases weight transfer to the outside tire" <- this is right??

i think that decrease transfer to the outside tire!!

anybody explain why increase??

my english is poor!!! sorry!!


silverbullet
silverbulletlink
Monday, December 03, 2012 11:04 AM
@ZZang, stiffening the suspension will increase weight transfer on that end. For instance when you put on a thicker anti-sway bar in the rear of a well balanced car, the rear becomes stiffer and has a tendancy to oversteer more. This happens because more weight is being trasnfered to the laden wheel in the rear. This additional weight transfer is causing an over load, causing the car to over steer more.

@8695Beaters, these terms have been around for a long time. The tuner world is just unaware of them, unlike the motorsport world. Mike is exposing this field to the tuner world so there will be less poorly set up vehicles in motorsports with tuner car roots such as time-attack and drifting.


silverbullet
silverbulletlink
Monday, December 03, 2012 11:55 AM
@8695Beaters Ah i spoke too soon, as you were refering to the "mike axis"!
HudsonMC
HudsonMClink
Monday, December 03, 2012 1:36 PM
@EB Turbo: that's pretty slick. I've heard of people doing that as well, but only to convert to rack and pinion. It's similar to how the 240Z of the same era was laid out. MI think the general consensus is that the stock steering geometry of the 510 is excellent, just the weak point is the box. A front steer rack and pinion 510 is relatively rare, especially one with double shear LCAs!
xloki77x
xloki77xlink
Tuesday, December 04, 2012 12:37 PM
2nd picture, 5th page is the perfect example of how NOT to use a rod end. If you've ever heard the term "Rod End In Bending" and wondered what it is, this is it. Rod ends are designed to be used in a tension/compression applications, never to see a bending load. Their use in a bending application, such as an upper or lower ball joint, is considered bad engineering as their axial load capacity is only about 10-15% of their radial load cap. In this picture, the rod end is used as a lower ball joint in a MacPherson application; an even bigger no-no than using it as an LBJ in an SLA config.

The "Mike Axis" is actually just a severely dumbed down version of the "Mass Centroid Axis" which is the exact same thing, but more accurate. Looking at the sideview of a vehicle, divide it into 10 or so sections (the more sections, the better the results), and find the CENTER OF MASS of each section. The axis really has nothing to do with any measurements though. The only thing of relevance is the roll couple front and rear (distance between cog height and roll center height). This is what makes a car "feel". One more thing, center of mass is different than center of gravity and the cog is not necassarily in the center of the car.

Also, you have not even touched polar moments of inertia which are a large part of this conversation. To generalize and say that a car with an upward sloping roll axis from front to rear will oversteer is wrong. There are plenty of mid engined cars that understeer terribly. However, because of a large rear polar moment of inertia they tend to swing out violently once either front grip or more power is available. This is why you dont see too many Porsches drifting.

On top of everything, all of this goes out the window when you start talking about these things happening dynamically and roll/ yaw come into the equation. I know you're trying to help people understand things in the simplest terms, but once you get in this deep simplest terms do not work anymore.

I'm not trying to upset anyone and I hope you guys keep up the great work on the site.

Thanks,
Pete
Rockwood
Rockwoodlink
Tuesday, December 04, 2012 4:38 PM
Correct me if I'm wrong, but unless that spherical binds, it's not going to see very much in the way of bending loads. If the washers capturing the tension rod bushing on an Integra can last hundreds of thousands of miles without issues, then that 3/4" (guesstimate) heim is more than sufficient.

Such a setup seems to work fine in a countless number of off-road cars out there without excessive failure...

http://images.thesamba.com/vw/classifieds/pix/3905665.jpg

And those vehicles see far more abuse than this car does, especially since it's been sitting in a garage undriven (zing! sorry mike ;-p)

Also note that this is part one in a series. Capturing all of the ins and outs of suspension tuning in magazine-ish format is going to take a number of installments. Stay tuned? :)
xloki77x
xloki77xlink
Tuesday, December 04, 2012 5:25 PM
It is not sufficient, especially in a MacStrut LBJ application. It is arguable on an UBJ of an SLA for some time, but should only be used to find the correct geomoetry then replaced with an encapsulated spherical bearing. Using RE's in any LBJ is just bad design, but will see less force in an SLA LBJ than McStrut.
The force acting on the B14's rod end is a bending moment in both braking and acceleration. The argument that "just because its used here" does not make it good design or good engineering. As well, using larger than needed rod ends may work, but its simply a band aid for poor design when lighter components could be used when designed properly.

To summarize:
"Threaded rod ends are designed to be loaded in compression or extension. They are not designed to be used when loaded laterally as this tends to bend the threaded shank and usually applies shear loads across the thread roots. Bending loads tend to distort the eye causing the spherical element to seize. This quickly leads to failure. The shear loads are not only in the vertical plane. Front wheel brake torque is the biggest single force in most cars, and this is reacted through the suspension links.

http://www.formulastudent.de/uploads/RTEmagicC_image007.jpg.jpg

These forces are reacted in a horizontal plane, and if threaded rod ends are used, then a shear load is introduced. Failure of a rod end under hard braking can have catastrophic results."

Please read this (or any) article on rod ends in bending:

http://www.formulastudent.de/academy/pats-corner/advice-details/article/pats-column-rod-ends-in-bending/
xloki77x
xloki77xlink
Tuesday, December 04, 2012 5:28 PM
I also forgot to add that the RE on the B14 is also loaded in bending AND single shear which is the worst possible scenario.
Rockwood
Rockwoodlink
Tuesday, December 04, 2012 6:41 PM
The whole front of that car is bad engineering. Yes, it is not the best design practice, but let's be realistic. This is not an academic exercise, nor is it an FSAE car. Modifying that control arm to accomodate a captured spherical would result in more weight than an oversized RE.

In any case, since you pointed out the problem, time to point out a solution. How would you not only encapsulate the rod end without adding weight over and utilizing the stock control arm (which I believe was for rules compliance), but also double shear it?

There is a point where "good enough" is, in fact, good enough.
Mike Kojima
Mike Kojimalink
Tuesday, December 04, 2012 6:54 PM
xloki77x

Thank you for the insightful comments, I think you might be an engineering student, probably on an FSAE team. I think this because the late badass Carroll Smith used to take points off on student cars with rod ends used like I did on my car. It was one of his pet peeves.

In the real world of Motorsports, there are countless examples of many fast reliable and safe race cars doing this in many forms of racing from World Challenge to off road to Formula Drift.

When designing my lower arms I calculated the loads and used a high quality rod end with a rated strength many times higher than the anticipated load. So far this has seen at least three seasons of racing and quite a few other events including a crash that bent the arm but didn't phase the rod end.

One of my good friends that works for a factory backed world challenge team, that in fact won the championship one year, used a similar rod end with even more offset with twice the moment in a heavier car. I was sketched out by the design and told him so.

In several heavy crashes that the teams cars endured over several seasons, the rod ends never yielded even though control arms, the bolts through the rod ends and the spindles bent and even broke.

In our Formula D car there are several rod ends like this and although we have worn out a few and they get replaced once a season, we have never broken a rod end even though we go through arms fairly regularly.

This is the real world. If I was designing a formula car with a big budget, you bet I would do a bearing with a housing and double shear everything, especially if I was really worried about weight and was designing things closer to the load.

In the real world I design a lot of adjustability in the things I do because I am always learning about the interplay between roll center, caster, KPI and scrub. I have been experimenting a lot on this over the last few years and have learned a lot about the relationships between these and how they affect a car dynamically. What I am learning is pretty interesting and I will probably never write about exactly I have found out because it's one of the things people hire me for.

Secondly you are a smart guy, have a cookie, I did not write this article for people like you. I wrote it for people that are car hobbyists that want to learn more about their stuff to take their hobby to a higher level. I also wrote it to be somewhat entertaining so I would not bore people to death. I did not write it to be an SAE paper up for professional review nor as a primer for engineering students. Their are many detailed and boring things already published in this subject.

Some of your comments were on things outside my intended subject of the story. For things like this, you can write on and on for years. Wait that's what Milliken did. I usually only write with authority on things that I have some practical experience with. Sometimes you can't be sure if the author of your reference material was from a literature, practical or scholarly background. I mostly like to read about practical stuff.

That being said, I like your comments. What is your background? Interested in writing for us?
Dan DeRosia
Dan DeRosialink
Wednesday, December 05, 2012 10:15 PM
This is one of those things where I half think the Formula SAE/Student programs, and engineering schools in general produce bad design practice - I've seen no end of people just sit stymied about solutions because the only practical way to get something done is counter to the theoretical ideal. That's probably a different rant though.

One comment did strike me though, from a Race Car Engineering interview of one of the designers of a Nissan Group C car (can't remember exact names) who was fairly new into that side of things, when asking Eric Broadley of Lola for advice on rear suspension, got something to the effect of "well, the theory's all well and good, but remember to build a lot of adjustability into it."
Mike Kojima
Mike Kojimalink
Thursday, December 06, 2012 12:42 AM
Kenku,

I get you 100%. Sometime I laugh when a student tries to school me on something I happen to be good at. I am impressed with how much knowledge students can assimilate with all the tools they have now days, at least the smart ones. They are a lot smarter than I was as a student!

I look at many FSAE car and I am sometimes impressed with the innovation. Then I am sometimes dismayed at the execution. A lot of FSAE cars have really bad plumbing, wiring and mechanical layout. Since in my experiance at least 80% of your DNF's come from failures of wiring and plumbing, I pay a lot of attention to this in cars I am involved in.

Another thing about FSAE cars is sometimes a lot of thought is wasted in trying to be innovative. In racing well excuted normal stuff will often beat innovation unless there is unlimited money and maybe not even then. But I guess that's why they are students.
Dan DeRosia
Dan DeRosialink
Thursday, December 06, 2012 7:19 AM
Don't get me wrong, I wasn't an exception - one particular idea I had was running a supercharger on a CVT with a detonation resistant head design on E85 (IE, run the restrictor at choke flow as much as possible, ala WRC), fed through a single speed transmission driving all four wheels. Which I think is pretty far towards "innovative for its own sake", though at least I did realize it at the time too. Nowadays I can't help but think that the best innovation a team could do is getting things done ahead of time with roll centers and geometry reasonable, putting data acquisition on it, and learning to tune the sucker.
SGSash
SGSashlink
Thursday, December 06, 2012 9:03 AM
I am 100% with you guys on the FSAE stuff! But as those kids learn, they will find out more about good plumbing and increase their fab skills. We all had a time when we thought NAPA (in my case Canadian Tire) was the only place to buy shit.

I have one interesting observation I wanted to add to your article Mike. While I know dampers can be tuned to adjust transient response, I found that driving cars with a very small rear roll couple makes them very twitchy on quick direction changes (esses, chicanes etc), whereas cars that have longer rear roll couples versus front allow for a very neutral steady-state balance, but that allows you to really throw the car around during direction changes.

Our CTCC Genesis Coupe racecars are not allowed any geometry correction, so as a result the rear roll centre is under the ground. The front is not much better. The car has somewhat lazy response and more roll than I would like, but one good feature is the fact that you can throw the car around really hard in transitions without the risk of spinning or getting into a big slide. The IS300 World Challenge car I drove had roll centres only an inch or two below the COG, and it was VERY twitchy on direction changes. Yes it was super responsive and awesome - but perhaps there is a balance there especially for RWD high power cars.

In other words roll centre can be a tuning tool more for transient balance than for steady state balance - since we have bars and springs for that.

I'm interested in your thoughts Mike!
xloki77x
xloki77xlink
Thursday, December 06, 2012 3:46 PM
Mike,

I totally understand the need to make your articles interesting, informative, and light-hearted while still touching the engineering/theory side of things. I can see how this subject could get kind of dry if you're not familiar with the terms and just spewing info at people is not the way to keep them reading. I also totally understand that the majority of people that come to this site are not engineers and making your articles easy and fun to read is a primary goal.

As you may have suspected, my background is in Mechanical Engineering, but I never participated in FSAE. I'm just a regular 27 year old dude that likes to build stuff in his garage when he gets out of work. The deeper I got into building my project car the more I wanted/needed to know to make it the best it could be. I am in the midst of building a 92 Evo I road race/hillclimb car and I didn't just learn all this stuff from school or reading. I learned by doing, or practical experience as you call it. Yours and Dave Coleman's articles (the only reason I had a subscription to SCC way back when) got me thinking in the right direction, but it was Carrol Smith, Herb Adams, and big Willy Milliken who really taught me suspension design.

And I do have practical experience. I'm in the middle of redesigning the Evo's multi link rear suspension into a fully adjustable SLA configuration by building a tubular 4130 subframe, a-arms, and toe links. The front will stay McStrut for now, but it will also have a 4130 subframe and a-arms. I've already designed and fabricated bump steer kits and, in collaboration with Ground Control, double adjustable Koni coilovers for the Evo I-III.

I do want to thank you for all the work you've put into the site because I'm on here almost daily. You guys are doing a great job and the more poeple you can hit with information like this the more knowledgable everyone can become. If there is any way you would like me to contribute let me know and I'd be happy to give what I can.

Mike Kojima
Mike Kojimalink
Thursday, December 06, 2012 3:59 PM
We would love for you to write about your project!
Mike Kojima
Mike Kojimalink
Thursday, December 06, 2012 4:02 PM
SGSash

I think with proper shock tuning that a lower rear roll couple is actually more stable and easier to control in a slide or at the limit.

Conversely it is way more sensitive to shock adjustment.

What makes are car twitchy in my experience is a can be a combination of a lot of things. Alignment, bushing compliance, suspension geometry and shock tuning can all conspire to make things difficult!
Mike Kojima
Mike Kojimalink
Thursday, December 06, 2012 4:08 PM
Kenku,

How far did you get on the transmission for the blower idea? One of the things I have been thinking about is a transmission for a centrifugal supercharger.
xloki77x
xloki77xlink
Thursday, December 06, 2012 5:25 PM
@Rockwood:

As for a solution to the problem of the rod end in the lower arm:

First you have to figure out where you want your lower, outer pivot point to be. This will affect camber, caster, KPI, strut axis, roll center, etc. Since this example has a rod end in there now you can physically adjust it to your liking, but there are other ways to find the right setup. You can download free software online to play with all these things.

When you've found your setup dimensions its time to start fabricating. You weld a spherical bearing weld cup to your lower control arm; whether its the stock arm or custom one it doesnt matter, its the same process. You can now use a tapered mono ball stud and spacers to adjust your roll center.

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

Rod End vs Spherical Bearing:

Rod End:
Aurora RAB-12T - 3/4" ID / 3/4"-16 thread
Price - $66.50
Weight - .602 lbs (9.632oz)
Ultimate static radial load rating = 27021 lbs
AXIAL LOAD = 3377 lbs (This is the one we care about)
3/4" Jam nut = .125 lbs (2oz)

Total weight = ~11.6 oz

-----

Spherical Bearing:
FK Bearing WSSX8TV - .5000 ID
Price - $17.63
Weight - .100 lbs (1.6 oz)
Ultimate static radial load rating = 21400 lbs
AXIAL LOAD = 4970 lbs!

FK CP8 Weld Cup
Price - $9.16
Weight = .35 lbs (5.6 oz)

Total weight = ~7.2 oz.

In this case, the spherical bearing is cheaper, lighter, and, most importantly, 68% stronger axially. You can even further the weight savings by replacing the 3/4" bolt used with the rod end with a tapered, 1/2" threaded mono ball stud.
Dan DeRosia
Dan DeRosialink
Thursday, December 06, 2012 6:21 PM
Never got anywhere much with the CVT idea for the supercharger; whole thing was back of the envelope thinking and most of that on the head design. The idea was to modify a snowmobile v-belt setup with some sort of electronic control, but any more details than that would take figuring out. Possible you could just do it with proper weights and springs though.
Lorin Mueller
Lorin Muellerlink
Friday, December 14, 2012 9:07 PM
So, where would the roll center be on a twist beam car?

Based on the solid axle/leaf spring, I'd assume center at the level the shock attaches to the axle... Maybe?

How does instant center behave?
stuntman
stuntmanlink
Thursday, January 03, 2013 11:08 PM
In your statement:

"On most cars the ideal location for the roll center is 2-4 inches above the ground for the front suspension and 4-10 inches above ground for the rear suspension with the rear roll center higher than the front. This is so the car will transfer more weight on to the front of the car due to an increase in geometric anti roll giving a more predictable tendency to understeer at the absolute limit. Most purpose built racecars are like this."

Don't you mean that with a lower front & higher rear RC, the REAR will transfer more weight to the outer tire (due to its greater geometric anti-roll and thus increased load transfer) and the lower front will transfer less weight to the outer front -thus reducing understeer in typical front engine cars - ?

0-2" front roll centers can also work quite well. So can RC's below the ground plane -especially on cars with a high ride height that won't bottom out and can afford a vertical load acted on the sprung mass from a lateral force (opposite of jacking). Low front RCs are okay if the body roll dosn't adversely affect your dynamic camber (and you can tolerate a little 'flop' and lack of response) due to reducing load transfer to the outer tire which increases the total tractive effort of the pair.

:)
Mike Kojima
Mike Kojimalink
Thursday, January 03, 2013 11:43 PM
When I am writing it is in generalities, hence the use of the word MOST. For serious development, geometry must be engineered into the overall vehicle system on a case by case basis.

Exceptions to just about any suspension theory can be made to work by balancing out the compromises. There are case studies where successful race cars have had pretty improper geometry among other things but they still won.

Many times as an engineer you can't change things due to rules, budget, etc. The smarter guys figure out good workarounds to these sorts of handicaps. It's part of what makes our job fun.
Mike Kojima
Mike Kojimalink
Thursday, January 03, 2013 11:49 PM
Billy, Thanks for the catch, I could have sworn I edited that out before I Published it. In my original draft I was talking about rear engine characteristics and I screwed it up when I was shortening the story down and simplifying it from what I first wrote.
stuntman
stuntmanlink
Thursday, January 03, 2013 11:58 PM
You don't call me anymore like the old days where we enjoyed hanging out and discussing this stuff... :(

In our mid-engine FXMD car, lowering the rear (CG and RC) was a HUGE improvement in rear grip and stability which helped break the BW 13 record. -Going toward what you were saying how it can be the opposite in a mid/rear engine car. And like you said, setup is a compromise of many different things. The right course of action depends on what you're dealing with.
Mike Kojima
Mike Kojimalink
Friday, January 04, 2013 12:15 AM
If you would quit messing around living in the deep south and racing NASCAR I would see you more :)

I wrote the story in the middle of the night after working on a car all day. Thanks to you I went back and looked at what I wrote it and damn it if it make any sense. Sometimes no one here can edit my tech except me and I was so dog tired I just wanted to crank it out and get some sleep.

Experimenting with roll center location made a huge difference with our FD stuff and is one of the reasons why we can generate so many longitudinal G's while sliding and sometimes nearly yank both front wheels off the ground like a WOO sprinter but still be stable and drivable.
Jonathan Spiegel
Jonathan Spiegellink
Tuesday, January 22, 2013 12:36 PM
The Mike Axis - related to another SCC term we named in honor of Dave Coleman - the Dave Point. Refers to the imaginary point where the steering axis intersects the ground.
kyoo
kyoolink
Thursday, February 07, 2013 10:06 PM
for me personally i thought the level of the material was perfect, though i'm not an engineer and probably more of the target audience for this kind of material. my general takeaways from this article are:

- that you don't want the roll center to be too high or too low
- you want the roll couple (distance between the cg and the roll center) to be as short as possible
- thus if you lower the car's cg (good thing), due to the roll center going even lower (bad thing), you want to find a way to shorten the roll couple again, either by fabricating a suspension component or aftermarket if it exists.

is this a correct summation of the ideas?

for evo 8/9s, its usually considered a bad idea to lower the car more than an inch and a half, because it does put the roll center beneath the ground. then the guys who can use the WL roll center adjuster to bring it back up
S15Stephanie
S15Stephanielink
Thursday, April 04, 2013 7:29 PM
So are there monoball pins made for toyota? I am trying to keep the machining I need done to a minimum.
KevinK2
KevinK2link
Monday, August 12, 2013 2:57 PM
@xloki77x
"The "Mike Axis" is actually just a severely dumbed down version of the "Mass Centroid Axis" which is the exact same thing, but more accurate."

I believe this has nothing to do with the 3 principle axis of the sprung mass. It is more of a construction line used to estimate where the mass center (or weight center) is. Mike actually confused his readers by introducing this as a significant axis, as well as using the slope of it to predict weight transfer.

I subscribe to Fred Puhn's (How to Make Your Car Handle) method of chassis analysis. Fred uses this line to link the two roll centers, and assumes the CG will be on this line, with it's fore aft position established by the front and rear centers of mass ( those based on load on F&R axles ).

Then the vertical distance from the CG to the roll axis ( which Mike correctly described ) becomes the lever arm for the total roll moment on the sprung weight.

To determine weight tranfer bias, you use the front and rear roll stiffnesses. "Mike axis" slopes, relative to the roll center, is irrelevant to weight tansfer at each end of the car. Example, take any popular street car, put a 4" dia front sway on it with aluminum pivot blocks. 99% of the sprung mass weight transfer will happen at the front wheels.

I tried to correct Mike back in his SCC days, and it was like a Spinaltap sceen ... " but this knob goes to 11"

KevinK2
KevinK2link
Tuesday, August 13, 2013 2:04 PM
xloci77x: "The only thing of relevance is the roll couple front and rear (distance between cog height and roll center height). This is what makes a car "feel". One more thing, center of mass is different than center of gravity and the cog is not necassarily in the center of the car."

Have to disagree here. Your only likely to deal with front and rear roll couples when working on a cheap go-cart, with a flexible frame.

For a car with a chassis stiffness at least 10x the total roll stiffness, you NEVER consider F&R roll couples. You calculate the for/aft CG location on a line between F&R mass centers, and then assume it is located on the center width of the car. Then you compute the single roll couple by dropping a line vertically from the CG location down 'till it intersects the roll axis. That vertical distance is your roll couple lever arm.

Also, the center of mass of the sprung weight is the SAME as the center of gravity. The same integration method is used to determine the location, the G term drops out, and in both cases, the CG is at the same place as the CM.

W=mG



Mike Kojima
Mike Kojimalink
Saturday, August 17, 2013 1:55 AM
Of course the roll axis is the mass centroid, if i called it that I would loose 80% of the readers so I made up a funny term for it. The roll center location and roll moment have everything to do with with weight transfer distribution of the sprung mass, I am not sure what you are talking about when you state that it doesn't. Geometric antiroll has an effect on weight transfer percentage just as any of the other elastic methods of increasing roll stiffness.

Maybe I don't understand some of the points of your comments, I think we agree on a lot of the same things perhaps but your semantics are different.

In practical terms on real cars, roll couple and the slope of the roll axis has a lot to do with how the car points and takes a set in a corner and how it feels at the edge of and exceeding the limit of adhesion. The roll axis slope and direction has a lot to do with how a car feels to the driver. It's the difference between a car that is easy to drive and twitchy.

Try it some time on a car that is highly adjustable. An easy experiment that will teach you a lot is to restore proper relationships on a sedan based race car that has been lowered a lot. Assuming that this is the only variable (typically it wont because a lowered car will have a whole bunch wrong with it) it will feel much better.
Crackers
Crackerslink
Wednesday, August 28, 2013 9:03 AM
@Mike- I read this about a week or so ago and its given me a great deal of food for thought, but the lack of discussion about the F1 setup really has a monkey on my back. Given the extreme low CG on an F1 car, do you think, or know, if the RC is intentionally raised to invert the roll couple? This would result in body roll exhibiting as a lifting of the outer corners respectively would it not? (Assuming the body moved enough to measure, lol) If that were the case how would that effect weight transfer relative to traditional roll couple orientation? Could it actually create downforce on the inside corners through the sway bars?

I realize I'm more likely to be killed by the abominable snowman than ever have a practical use for the information, it's just trying to figure out why it works is bugging me.
Mike Kojima
Mike Kojimalink
Thursday, August 29, 2013 1:55 PM
I am pretty sure that since the suspension on an F1 car doesn't do a whole lot, they give up some in the geometry department for better aero.
Crackers
Crackerslink
Tuesday, September 17, 2013 10:58 AM
Thanks Mike, I'm sure a lot of us appreciate you taking time to answer questions from complete strangers. When do we get to see the next installment?
KevinK2
KevinK2link
Saturday, November 23, 2013 6:19 PM
Mike,

First I want to thank you for all you're great articles on suspension, both here and back when you were with SCC.

But as shown by my questions, I have a problem with your neglecting to include the effect of springs & anti roll bars, in the creation of front and rear roll stiffnesses, which when added together represent the total roll stiffness. This is part of Puhn's method of suspension handling analysis.

In the past I said the front and rear roll couples don't matter, but I was slightly wrong. What is used in the Puhn analysis is the weight center vertical locations over front and rear axles, as discussed later.

Along the "Mike axis" , you determine the COG of the total sprung weight, from the front and rear weights over the axles, and then calculate the location along the "Mike axis" of the singular COG.

Knowing the COG of the total sprung weight, and it's distance along the roll axis, you get the total moment TM acting on the sprung weight using the lever arm defined by the vertical distance between the roll axis and the "Mike Axis" .

For a long sweeping high speed corner at constant speed, this TM will be resisted by the Total Roll Stiffness, ie the sum of Front and Rear roll stiffnesses. This will determine the angle of roll, and you can calculate the weight transfer at the front and rear tires due to body roll.

Note to calculate the front and rear roll stiffnesses takes some work. You need to know things like Moment Ratios (MR) related to spring or sway bar deflection vs wheel deflection.

Now here is where the front and rear weight centers are needed. We calculated reactions due to the Total Moment and the sprung body roll. As far as weight transfer at front and rear tires, you also need to include the front and rear moments due to the unsprung weights. The lateral forces at the F & R weights (end points of the "Mike Axis") due to cornering, and the vertical distances from there to the ground, when multiplied provide front and rear moments from the unsprung weights ( wheel, tire, brake caliper and disc, etc ).

So looking back at your article, page 2, yes, lowering the rear roll center will increase the rear lever length and tend to increase roll. But per the Puhn method, you CAN'T look at this in isolation at one end of the car. You see how it changes the roll axis, and the lever length at the COG, and then the weight transfer at that end of the car. If the roll stiffness is high at that end, the change in roll will be small. So you can't make all the conclusions you made on page 2, as being true in all cases.

Mike Kojima
Mike Kojimalink
Saturday, November 23, 2013 7:08 PM
KevinK2, it is all very simple. I am only talking about the geometric aspects of roll resistance here. This article is intended to explain that simply. It is not an article on how to analytically one dimensionally model your car. It should be obvious even from the title that it isn't.

I could write about this but it would be beyond the scope of most MotoIQ readers, be more like a book you would pay for and I am reluctant to teach the world my particular modeling methods for free.

The way I do things works well for me and frankly I get paid a lot to do this sort of work for customers and I don't really want to teach everyone in the world what I do specifically.
KevinK2
KevinK2link
Sunday, November 24, 2013 12:21 AM
Mike, thanks for the reply, but recall the title of the series:

"The Ultimate Guide to Suspension and Handling"

I guess I was assuming this was an "Ultimate" guide for making a car handle, and I should have re-read part 3, Balancing the Chassis, where springs and bars are lightly discussed.

Thanks for clearing this up. By the way Fred Puhn was/is a Mechanical Engineer like myself. The equations he uses do not represent a "one dimensional" model of the car. It's a mathematical model of the suspension, described by a group of equations of 1st and 2nd order. The model has 2 degrees of freedom, vertical displacement at the wheels and rotation about the roll axis.
Mike Kojima
Mike Kojimalink
Sunday, November 24, 2013 4:01 AM
What book is this?
KevinK2
KevinK2link
Sunday, November 24, 2013 1:38 PM
"HOW TO MAKE YOUR CAR HANDLE"

by Fred Puhn

cw 1976

Still available on Amazon.com

He included an equation for the stiffness (lb/in) of a sway bar (I think he created it). I derived the same exact equation, and found that it represents a "bench test" stiffness, where it is mounted to the bench, pinned at the bushing locations and at one end. The load and deflection is measured at the other end. Because of this "bench test" stiffness rating, you must DOUBLE the calculated rate, when the bar is in a steady corner, like at 1g . Many people that discuss how the suspension works miss this point.

I believe Fred raced, but was also known for setting up race cars back then.

He had his own company, Puhn Engineering
Mike Kojima
Mike Kojimalink
Sunday, November 24, 2013 8:19 PM
I think I used to have this book, I think someone borrowed it many years ago and never returned it.
Racing Aspirations
Racing Aspirationslink
Wednesday, July 16, 2014 4:35 AM
Interesting article... Now you're ready to start using http://www.racingaspirations.com/apps/suspension-geometry-calculator
Post Comment Login or register to post a comment.

MotoIQ Proudly Presents Our Sponsors:

© 2014 MotoIQ.com