Here's where we start to geek out on the suspension a little. First, some definitions for those who aren't 100% up on this stuff:

Roll center: An imaginary point at which a suspension rolls about. Old cars had quite low roll centers in the front, more modern cars have higher roll centers to control roll without stiff springs or sways. The roll center is complicated to describe, but if you draw a line between the center of the tire's contact patch through the instant center of the suspension, for each side of the car, the roll center will be where those lines cross each other. It can be above or below the ground, and it can be to the left or to the right of the center of the car. Ideally it shouldn't move around too much in relation to the body's center of gravity.

Roll couple: The distance between the roll center and the body's center of gravity, creating a roll moment. Increase cornering force or weight, or increase the distance between the center of gravity and the roll center, and the car will generate more roll moment.

Roll axis: A line drawn between the front and rear roll centers. Front suspension roll centers are usually much lower than those in the rear on live axle cars, a bit more level on independent suspension cars.

Camber gain: The tendency for a suspension to gain (or lose) negative camber upon compression. A good front suspension will have this feature tailored to keep the tire planted at all times.

Instant center: This is an imaginary point in space where lines drawn through the upper and lower suspension pivot points meet. Extend a line through the upper balljoint and upper inner pivot, and another line through the lower balljoint and lower inner pivot, and where those lines meet is the instant center.

Swing arm length: In a 1:1 relationship with camber gain, the swing arm length is the distance between the instant center and the wheel. Shorter swing arm lengths mean more camber gain, but too short can also cause handling issues.

Front Lateral Load Distribution: The FLLD is analogous to the relative roll stiffness of the front suspension, compared to the total roll stiffness of the car. This takes into account roll couples at each end plus the stiffness of the front & rear springs & sways.

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

Stock, at rest, the front suspension roll center of a 122 is at ground level due to the arms being parallel and horizontal. Since the arms are parallel, camber gain is zero and the swing arm is infinite. When you lower the car on lowering springs, the roll center drops below ground by approximately the same distance that the car's been lowered: 1.5" lowered car means the RC is 1.5" below grade, and this is good because it maintains the same roll couple as at stock height. The net result is that the center of gravity is lowered so it won't put as much load on the outside tire, but with the same roll couple the car will roll as much as it did before (minus the effect of stiffer springs, if present). Since the control arms on a 122 are so long (relative to the width of the car) there isn't much camber gain, and all the roll you get in corners (no matter the height of the body) puts more positive camber into the outside tire, which ruins handling. There IS slightly more camber gain at this new lowered position, but not much.

The rear roll center is where the panhard rod crosses the body centerline - that's the way it is with beam axles - and when you lower the car 1" the roll center drops 1/2". The roll couple in the rear reduces a little, so the rear end doesn't generate as much roll moment. Since the front roll center has dropped more than the rear, the roll axis gets steeper. This means that the car is rolling less on a horizontal axis, if you can picture it, and this causes some handling traits that could be considered "old fashioned".

So what does this all mean in my car? Well, the low front roll center and the stiff springs & stiff front sway combine to give me an FLLD of 70%, according to the suspension analysis program I'm running. In theory the FLLD should be about 5%, numerically, than the front weight distribution. On a car with 52%/48% F/R distribution, AND with everything else squared away nicely, the FLLD will be in the neighborhood of 57%. 70% suggests that I've got a TON of understeer, and this is exactly what I found in steady state cornering. Additionally, I've got a compromised camber condition: I know the car rolls more than 1.5 degrees, so in hard cornering my outside tire has lost all of it's static negative camber and is going into positive camber, and the high tire pressure in the front (needed to try and help the edges of the tire survive) reduces grip even further due to reduced rubber on the road.

Analyzing the front suspension, and building on past experience, I believe I will try lowering the front inner pivots (the "dogbones") by 1" this winter. Doing this does two things: first of all, it raises the roll center from 1.5" below grade to 1" above grade and will reduce the roll couple by quite a margin. This will make the FLLD a little worse, though, but changing the dogbone position also gives me a lot more camber gain so it will help keep the outside tire vertical and should give me more front end grip right there. My program tells me that the camber gain is currently 0.38 degrees per inch of travel, and it will change to 1 degree per inch with the new setup. If I keep the current 1.5 degrees of static camber, the car will be able to roll more than 2.5 degrees before the tire goes positive camber. That's good. I would also like to get the FLLD a little closer to correct, which means I'll have to put the rear sway bar in and possibly replace the IPD front bar with a stock one. With the reduced roll couple that much bar might not be needed anymore to control the roll.

(by "past experience" I mean on Dale's car, where we lowered the dogbone by 2" and that created some other issues. The swing arm length got really short, and at his lowered stance the negative camber was pretty extreme. There were some bump steer issues too. The car did go around corners well, though!)

The loose transition could be down to a couple things. First, the tires. They're sooper squishy, with full depth all season treads, and are tuned by Continental more for all season performance and good ride/quietness rather than sharp handling. I chose them for these qualities, but I think the tradeoff is that I'll have to drive it a little more carefully to let the loads build up more gradually than when I spasmodically hack away at the wheel in the slaloms. Another problem could be the shocks: They're Bilstiens from VPD, but I don't believe they're the revalved VPD version. The rears are mighty stiff compared to the fronts, and could be resisting a lot of roll on the transition (but not in steady state cornering like a sway bar would). I have some KYB rears that I will try, to see what they do. I will also consider taking out the two polyurethane donut bushings under the axle and replace them with rubber: the stiffness of the poly in this location WILL affect the articulation of the axle. I will also remove the limiting straps, as they're no longer needed to keep the car from falling over and on transitions they might be stretching out and yanking a wheel off the ground, temporarily upsetting the car. We ran Dale's car without straps and it didn't explode, so I think I'll be fine.

For what it's worth, my best, worst, and average times were about 2 seconds slower than Craig's, but unlike at Streetwheelers it was on a 60+ second run so it seems like I'm catching up. I know that I was losing a TON of time on the two 270 degree pinwheels, I was loose on entry and very very tight on exit. I still couldn't power out of them either, so I know a limited slip would help and I might install one this winter too.

Again, the goals must be maintained: the car must handle well but must not be brutal or uncomfortable, and I'm not in it to win it, I just want to improve myself and my car.

I measured up some stock stuff today:

Front coils
.572" wire diameter
4.75" OD
6.5 active coils
11.875" free length
325 lb/in spring rate
91 lb/in wheel rate

Rear coils
.468" wire diameter
4.5" OD
8.5 active coils
15.25" free length
125 lb/in spring and wheel rates

Currently I am running VPD springs all around. The fronts are said to be 250-500 lb/in variable and the rears are 145 lb/in linear. At VPD they claim the actual stock front spring rate is closer to 250 lb/in. I suspect that measurement is taken from the free length position, which uses all 8 or so coils: I only counted coils that showed no signs of having rubbed up against another coil, i.e. 'active' coils. With that in mind, the VPD lowering spring might actually be stiffer than 250 lb/in going down the road, it's tough to say. I'm not sure how much they lower the car but it's 1-1.5" in the rear and around 2-2.5" in the front, but Craig had cut them a little to get the stance he wanted. The fronts are 5" diameter which are also said to be the same as some R-sport springs that were available way back in the day. It's tough to calculate the wheel rate from this at ride height, but the motion ratio is 53% so the wheel rates are 70-140 lb/in if the published spring rates are accurate.

I have on order a pair of 5" x 10.5" 400 lb/in springs and a pair of 5" x 9.5" 550 lb/in springs. The 400 lb-in springs should bring the front up about 1" from where it is now, the 550s should bring it up about 1/2". The 400s give 112 lb/in wheel rate and the 550s will give 154 lb/in.

More Data:

(sorry, dead photo)

These numbers are calculated with the assumption that my car weighs roughly 2400 lbs, with a 52/48 balance, and with 110 lbs unsprung on each front wheel & 139 lbs unsprung on each rear wheel. I've made fairly educated guesses in most of these departments, combining actual measurements with some common knowledge.

And more info:

Ride Frequencies are a subjective thing, but the guidelines here are pretty simple. For comfort, you want a ride frequency of about 1hz, maybe slightly less, but less than about 0.8hz is a pretty wallowy ride. A very light, very sporty car will have a ride frequency of about 2hz. More than this is considered too harsh for street use (unless you're 21 years old or have <6% body fat). To achieve 2hz or higher you need limited wheel travel or to have some other means of keeping the springs from being unseated at high amplitude wheel travel. For instance, my 550 lb-in spring will compress less than 2" from full free length, any higher spring rate than this and I risk having them flop around at full droop.

So why am I going through all this bother? Simply put, I don't like that much variability in the spring rate. It might be a compromise between ride and handling, but I think both parts suffer more than either part gains. The greatest benefit to a spring like this is that you can lower the car more and not risk bottoming out as easily. Since I don't need/want the car LOW, I would prefer a linear rate spring.

A spring will compress a certain amount with a given load. Remove a given load, the spring will extend. A stiffer spring will compress or extend less for a given load difference than a softer spring. It's pretty easy to see this happening when you load a car up: add weight, the suspension compresses; but you don't often think about what happens when you remove load. The softer spring will extend further, the car will gain more height.

In a corner, load transfers from the inner wheel to the outer wheel due to the height of the center of gravity and the track width. With a CG at around 18" and a track width of 52", a 0.8G corner will take about 315 lbs of load from the inside wheels of my car and put them onto the outside wheels. The outside suspension must absorb this extra load and will compress, while the inside suspension will extend upon release of this load. Stiff springs travel less, so the outside wheel compresses less and the inside wheel extends less.

But a variable rate spring will do something funny: With added load it will compress a certain amount but will then stiffen up and stop moving. When relieved of the same amount of load it will extend further than it will compress when that load is added. So when going around a corner, the outside wheel will compress a little while the inside wheel will extend a lot. This means that in addition to the car rolling "more" (more than it would with a pair of linear springs equal to the stiff part of the variable springs), the car rises up, which raises the center of gravity, which adds load transfer, which adds load, which adds roll, etc.

Similar issues occur when it hits a bump. Imagine a 2" bump: The wheels compressing the suspension increases the load on the springs. This load is then released, which converts into body momentum travelling upwards. If the compressed position puts the springs into their "stiff" zone, they will store and release much more energy in a short distance, and will release it more quickly. This moves the body faster, and as it goes back into the lightly sprung range the body will keep moving further because the lighter springs lose load more slowly. A stiff linear spring would compress the same amount and would store the same energy, but would lose it more quickly and would "bounce" not as high as the variable rate spring. On small bumps the ride is similar to stock, but that makes the big bumps all the more surprising when the nose heads to the sky.

So what you really get with variable rate springs is a spring that uses both heavy coils on big city street bumps, but only one heavy coil in a corner. I'd rather have linear springs that produce the same effect in corners but ride better on the street, or springs that ride the same on the street but work better in the corners. Savvy?

The 400# spring is stiffer in proportion to stock in the same general amount as the VPD rear springs are to stock, which was my first choice of things to try. The idea with the 550# springs is to try something about as stiff as the "stiff" position of the VPD variables and to bring the ride rates and balance more into the sports car range. For reference, a similarly "old" guy as myself just installed 300 lb/in springs in the front of his 242 and that gives a ride frequency in the 2.0 hz range, and he doesn't think it's TOO harsh, so my 550s might be just fine.

More to report when the springs arrive.

Just read the whole thread - awesome as usual.

Any updates on spring arrival or how the car is doing as a DD?

I also need to rekindle the working in the garage fire - seeing all this progress in such a short time was just so damn impressive!

Rabin

I just got the 400 lb/in front springs installed tonight. I may have miscalculated something... the front end came up nearly 2", or pretty much back to stock ride height. I expected it to come up about half that, maybe a little more. Also, it turns out that 5" diameter springs really aren't what we're looking for: the stockers are closer to 4.75-4.875". The 5" springs don't quite get all the way down into the pocket of the lower control arm, but they might make their way down there given enough time and road hammering. Maybe they need to be chamfered with a grinder before installation, I'm not sure.

That's the bad news.

The good news is that the ride has improved, in my opinion. It's stiffer on small bumps but softer on the big ones, it goes over speed bumps without that sudden jolt at full compression, and the balance feels pretty much like stock (just stiffer). I should really put the Bilstiens back in the rear, or KYBs in the front, because now it feels a bit soft back there. I would consider these to be an excellent rough-road spring when matched up with something equally stiff and tall in the rear (maybe an overload of some sort).

I ran the VPD and these 400 lb/in springs through my calculator and came up with much higher numbers than what I expected, so either my calculator is messed up or I'm not using it right. In order to get these 400 lb/in springs to actually calculate at 400 lb/in, I need to claim one more "active" coil than I would expect. Adjusting the VPD and stock springs down to match, the stockers and the VPDs both approached the claims posted on the VPD website, which frankly, I believe to be true. So ignore my 325 lb/in "stock" spring measurement, it's likely closer to 225 lb/in.

I'll run these for the next couple weeks and will probably pull the car off the road before I absolutely need to, so that I can start hitting this long list I have for the winter. The 550s will go in when I reassemble the car, I just wanted to see what the 400s would be like before I put it to bed.

In all honesty...that was one of the worst aspects of these springs. They'd give you that coil bind jolt and after a week or two you wouldn't be loving it anymore. So I'm glad that's gone. I think if I'd known more back when I had them, I would have swapped them out and done something about it. On the track, they were fine as the surface is smooth and in the corners it felt stable and was really much better than on the street.

I was about to post a while ago and question your statements on roll centres, but felt unqualified.
Your writings on progressive springs helped me understand why they aren't OE on... pretty much anything.
I'd never read a good..."rebuttal" of some claims for them. I'd gotten as far in my understanding as to say the lack of rate in rebound situations was a bit of a brain teaser (what does that "do"?, etc)

Anyway, why I post: this morning on my way to work, I happened to run into some very thought provoking statements on a Locost forum (no plans, per se, but a source for from scratch builds, which I am working towards) WRT roll centers. I reread them just now having left the screen open at here at home.
Thought I'd repost them.


&


A bit subtle a difference from "geometric calculation of roll centres", and not exactly worthy of his claim those are wrong, but it helped my brain... "get it".

Prior to that I read up a lot on anything Ron Sutton wrote online (I know Canuck has cited him here), and played with V-susp a moderate amount. GregK on Locost USA posts sparingly, but I would cite him as probably #2 in my "accidental tutors", as happens online.

I'll take this on, as a point of discussion. I understand you were only posting it up for provoking thoughts, here's why I think this guy is only making boldly incorrect claims:

We all agree that the measure of a car start at the contact patch, right? Everything to do with the performance, ride, and handling of a car happens here. Every force is transferred from the ground to the car at these points.

In a corner, the force generated at the contact patch acts through the suspension arms. The suspension arms develop an instant center, the two instant centers and the two contact patches create the roll center. The roll center is therefore impossible to ignore in the development of roll, as it's the link between the body and the contact patch. With a low roll center, the car might roll the same amount (2 degrees, for instance) as a car with a high roll center based on the spring/ARB combination, but the body will move further side-to-side because the chassis pivots around the roll center, WHEN YOU CONSIDER THE CONTACT PATCH AS A REFERENCE POINT. I highlight that statement because the contact patch is truly everything.

If you play with a toy car that has suspension (like a remote control car, I have several here) this can be easily seen. The guy saying that roll center doesn't matter will almost certainly say that a toy car proves nothing, but that guy ISN'T saying at what point the car IS rolling around, only that it's not rolling around the point that pretty much every textbook on the planet says so.

Another way of looking at it is that a car with a high roll center will have some of the cornering forces go towards lifting the car. Force vector diagrams will show this. A car that lifts up as it rolls can be seen to roll from a higher point than a car that sinks down when it rolls, I think.

In steady state cornering, the car with a high roll couple and high roll resistance will have the CG push further to the outside tire, developing less grip at that end. In a transition, the car with a high roll couple and high roll resistance will take longer to set into a roll because the body has to move further sideways before the roll action can be completed. A car with high roll resistance rides worse in single-wheel bump. A car with low roll center doesn't generate as much camber gain so requires more aggressive static camber and roll resistance. You can see why modern car manufacturers have gone with higher roll centers in recent years...

On the other hand, a car with a high roll axis and low roll resistance might feel tippy even though it rolls the same amount as the other car. In total the passengers will move up and down as much for a given amount of roll, but on a car with a high roll axis the passenger on the inside of the turn will feel his seat "rise up" much more than the guy in the same seat of the car with a low roll axis. The guy on the outside of the turn will drop further in the car with a low roll axis.

Mission accomplished: that just triggered a thought. A car with progressive rate springs will compress less on the outside suspension and the inside suspension will decompress more, when compared to a car with linear rate springs. A car with a low roll center will compress the outside suspension more than it will decompress the inside suspension, when compared to a car with higher roll center. Therefore a car with a low roll center might react much better in a corner to a progressive rate spring than a car with a higher roll center would.

The other quote statement that the roll axis can be ignored simply because the roll resistances of the F/R suspensions are different is complete nonsense. Geometry cannot be negated by roll resistance.

Craig put me on to Ron Sutton too, and I picked up a version of Circle Track Analyzer on his recommendation. It's a pretty powerful tool but I may spring for an upgrade in the future.

Sorry for the delay, I've been out in the garage. I don't have much to add to what Matt said, other than I think the quote was trying to get at the roll of a chassis. I don't think he got it right, but he got part of it right. The roll axis (determined by geometry) is the axis of rotation of a chassis (all relative to contact patches as Matt has said) and the distance between the roll axis and the centre of gravity forms the roll moment. The greater this distance, the easier it is to move the chassis from side to side. It's a lever. I think what Matt's saying with his model car analogy, is take the springs/bars off your car and suspend it by some magical force and if it moves, it will roll based on the geometry via the links and relationships of parts.

I think the original author has to read up on roll couple...maybe this is where he was going? You can't really alter how the roll happens for a given geometry, but you can counter act the forces with springs/shocks. I believe my testing on the 242 has provided a reasonable demonstration of this as I've worked on it this year. Roll attitude is a response to a force. I can offer not much more than this.

Mission accomplished again, more thoughts provoked: I retract all of what I said earlier. I just did some sketches and found the flaw in my slaw, which fully backs up what your guy claimed, Ian.

If you take an extreme case and say that the outboard spring is solid, any roll in the body will occur around the midpoint of the outboard tire regardless of the geometric roll center.

Bringing it back from that extreme, install progressive rate springs in a car. The outboard side won't compress as much as the inboard side will extend, so the apparent roll point relative to the outboard contact patch has been altered from the geometric roll center.

Pretty much destroys all arguments to the contrary, right there.

Edit - I retract my retraction, see below.

Aren't you just changing the resistance to the force and not the point of rotation? I've always thought that things like the CofG and such are a little too simple related to what actually goes on dynamically. But I don't think that geometry has nothing to do with it.

I may not have a complete handle on it just yet, I want to study the sketches a bit more before I make any further comments, but if you run the outside spring in coil bind the body will HAVE to roll around the outside contact patch, no? There's your point of rotation. I think that may be an extreme case, but this moment of "clarity" came from my sketches, and without springs there was nowhere to set the body height in roll. I can roll it around ANY point I like. I can prove my point by rolling the body around the roll center but I can prove this other guy's point by rolling it around anything else.

I'm still working on it, though. There are a lot of things about suspension movement I don't understand yet.

Just acknowledging I know know there were replies. Skimmed what you wrote, and wasn't sure I got it, will return.

The guy I quoted does do this for a living, in England. All biases aside, that made me give him some credibility.

Been wondering how much difference in feel Kaplehenke's (sp?) front roll steer corrector would give, now that theoretically the 245 is mine (in my garage, but no paperwork to say it's mine). I drove it and it's not as bad as I remembered.
I'm kind of opposed to high roll stiffness on the street, and adamantly opposed to fast steering ratio, so I'm not his intended market...but I'm curious. I bought Paul Curran's extended front LCA's... I figure if nothing else that will allow me to put some caster in (strut top settings can now be focused on getting caster because I'll have reasonable camber regardless of the top lateral setting.). Anyway sorry to side track.
If you guys have a handle on geometry anyone want to say why Paul C found the steering odd/objectionable out at lock? I assume that means the ackerman is put off, but that might be an incorrect guess...and I can't see why it would be put off.

We're getting pretty far off topic for the "build" thread, so I'll likely move some posts out to somewhere. Ian, would you like me to start a new thread so you can ask your question about the roll correctors?

Here's some further analysis on the roll center versus roll origin discussion:

I modeled up two simple suspension systems in Autocad. The first one closely represents a stock 122, the second one is make believe. The first one (we'll call it "LOW RC") and the second one ("HIGH RC") have the same CG height, the same track width, the same kingpin angle, the same scrub radius, and the same lower balljoint position. "HIGH RC" has the upper balljoints and both inner pivots raised 2", with slight changes to the arm lengths. I've shown the geometric roll center as a circle below the chassis.

(sorry, dead photos)

Then I rolled them around the midpoint between the two lower control arm pivot points, by 5 degrees, and adjusted the angles of the links to get everything connected together again, without adjusting the lengths of the links. This is an important point to note: nothing has stretched, bent, or deflected once I started moving things around. The two models might be different, but the components of each model stayed the same throughout the experiment. As a final touch, I added a white line perpendicular to the chassis and hanging below the midpoint, to show where it crosses the vertical line positioned between the two contact patches.

(more dead photos)

The roll center was recalculated and the ground/center reference point was moved to the middle of the two contact patches, as we determined earlier that "contact patch is king". The astute observer will note that because the chassis' were rotated between the lower control arm pivots, and the height of the lower control arm pivot in the "High RC" model is 2" above the other model, this means the High RC model is rolling around a point 2" higher which could cause an error. To address this I re-ran the experiment, rolling this time from the midpoint of the chassis at grade. Now both chassis are rolling from the same point (right where the engine drips on the pavement).

(even more dead photos)

And then I overlaid the three conditions of each chassis, using the midpoint between the contact patches as an alignment tool. Are you ready for this?

(sorry, I lost all of these photos over time)

Things to observe: Though I rolled both chassis from different locations, once I referenced them back to the contact patches the overall shape is the same. You can't tell, but there are two white lines hanging below the chassis on each of the combined models, they're THAT CLOSE to a perfect overlap. Notice too that the white lines are pointing very close to the location of the original roll center: The car IS rolling around the roll center, when referencing the contact patches. As long as the springs are linear rate, and the car is considered to be rolling from somewhere along the vertical centerline of the chassis, the roll center is very close to the point at which the car rolls.

When the spring rates are progressive, I believe that complicates matters. The closer the car gets to having a solid outside spring, the more it deviates from the geometric roll center.

Thoughts?

I wasn't satisfied with all of that, so I ran two more tests. This time I rolled it around the inside top corner of the outside tire. This simulates a super-stiff spring on the outboard edge, which lifts the chassis in roll and only compresses the outboard suspension fractionally. Guess what happened?

(more lost photos, you'll have to take my word on this stuff)

The pivot point didn't come up too much at all.

Frankly, I'm amazed. I originally expected the pivot point to match the average of the roll center as it moves around during the increase of roll, but I begun to second guess that when thinking about assymetric spring stiffness. The only apparent difference in the pivot points are coming from the attitude of the body: if it's higher due to jacking or whatever other forces, the pivot point is too but the path of the roll center is similarly altered.

With all of that being said, look at the offset of the CG symbol. The one with the higher CG and higher pivot point keeps the CG closer to midway between the contact patches. Also the roll center moves around a whole lot less on the higher CG model (on the low CG model, one RC point is way off screen, to the left). Add to that the reduced roll couple and the reduced need for suspension stiffness (better grip) and it's looking like a pretty convincing argument to pay close attention to roll centers.

Again read but not fully comprehended...

If you have the mental energy, here is a short, diverse, & inconclusive thread that touches on many of the topics we're touching on here.

http://www.locostusa.com/forums/viewtop ... 18#p204518