MR2 Owners Club Forum banner

MKII Suspension Analysis: Lowering, 91vs93, Graphs, ect

24K views 98 replies 20 participants last post by  janiK 
#1 · (Edited)
12/29/07: Added analysis for the front, see this thread


EDIT DEC 10th: Replaced 93 graphs with graphs using 93 crossmember dimensions supplied by twoina2. Rescaled Y axis on all toe graphs. For refrence, I run roughly .1" of toe in on my 91, so the max toe changes we see in roll for either suspension are roughly 10% of the total toe. Comments related to the reacent edit are shown in blue.


I have been doing a lot of work these last couple of weeks with a suspension analysis program called Optimum K on our Formula SAE car. If anybody else wants to try it out they have a free fully functional demo available, only limitation is you can't save.

I decided to try to analyse the MR2 Suspension, hopefully I can shed some real light on the age old debates about the differences between the 91 and 93 suspensions, the effects of lowering, ect. At this point I have only analysed the rear, and I wouldnt exactly call it a "full" analysis, but I think its a good starting point.

Please note:
1). I got the input data from my own car and a tape measure, so accuracy is +-1/4" maybe. That being said, these results arn't perfect, and shouldnt be used as the final word, especially where the exact values are concerned. However, they should be good enough to spot trends and get a general idea.

2). The 93 is assumed to have a 1/2" lower ride height than the 91. The lowering ammounts are based off of my car with Tien S-techs, and added onto from there to approximate the heights for a more modest spring, the 93 stock springs, and the 91 stock springs.

3). The body roll used is to the right. AKA, the car is in a left hand turn. Therefore, the right wheel is the outside wheel, the really important one. The outside (right) wheel is the dashed line on the graphs. The inside (left) is the solid line.

4). Heave (vertical chassi motion) is defined positive up, negative down.

5). I analyzed 3 different ride heights. Stock (Green), a modest lowering spring such as the Pro Kit or the TRD (Blue), and a more extreme drop such as the S-Tech, the Sportline, or an aggressive coilover drop (Red). This coloring scheme is used throughout.

Now, some results.

First, the 91 suspension in roll. On the lowered graphs you will notice that the curve "freaks out". This is due to extreme roll center migration, to the point that the program doesn't know what to do with it. Basically, ignore the plots after about 1* of body roll. Because of this I think that anyone lowering with performance in mind should STRONGLY consider stiffer sway bars as well, to limit body roll and keep the roll center under the chassis.

One thing to note is that lowering has very little impact on camber curves (which, for what its worth, pretty much suck). Moderate lowering has minimal effect on toe characteristics, however extreme lowering really makes the toe go nuts. I'm not sure if it was the rescale or if I did something different this time, but lowering, even extremely, does not seem to have as big of an effect anymore. If it was a change in my method I believe it to be for the better, as this time I actually setup different models for each state of lowering instead of adjusting the same model over and over. Much less chance for error, hopefully.

Next, the 93 in roll. Again ignore the curves past about 1* of roll. The choppy graph is due to the roll center migrating far outside of the cars wheel base, causing the program to produce errors.


Heres the 91 in heave.


And the 93 in heave.


The heave characteristics are much less interesting, IMHO, although they do show the increased toe in character of the 93.

Some new analysis: Neither rear suspension changes toe MUCH in roll. Both maintain nearly constant toe. The 91 still responds less drastically to lowering, however the change suggested by twoina2 severely reduced the negative impact of lowering on the 93.

It appears now that at stock ride height the 93 toes in on the inner wheel and out on the outer wheel, effectively causing slight oversteer. Severe lowering reverses this, effectively limiting oversteer. The changes are very slight however, maybe not noticeable.

Stock the 91 appears to toe in the outer wheel and toe out the inner wheel (VERY SLIGHTLY). Lowering the 91 has even less impact than on the 93, although in the 91's case extreme lowering should slightly increase the tendency toward oversteer, as it makes the outside wheel toe out slighly more and the inside toe in slightly less than stock.

In conclusion, I might reverse my statement about lowering. The effects on toe characteristics appear to be minor, such that a little bit of static toe in can easily compensate. However, the issue of roll center migration is still present, and should be mitigated by larger sway bars on lowered cars.


SO! Discuss, rip apart my methods, tell me what else to test, tell me what I missed, whatever. I just hope the community can benefit from this in some way. The more thoughtful discussion we have on the topic the better.
 
See less See more
5
#2 ·
So I'm not very familiar with these graphs. So here's my feable attempt to analyze them.

I don't get what the difference of these two graphs are. (graph 1 vs 3 and graph 2 vs 4)



Also, are you just comparing ride heights? How does spring rates factor into this? Also, since you are analyzing the rear, have you taken measurements on a '91 and '93 x-member?

Were you able to do the fronts?
 
#3 ·
Graph 1 and 2 are the 91 suspension at three different ride heights. Graph 3 and 4 are the 93 at three different heights. The 93 is just my best estamate of the 93, based on looking at pictures of the crossmembers. Should be really close. The length of the tie rods is the only difference.

Spring rate and anti roll bar stiffness effect how much the suspension moves, but not what happens when it moves. The actual motion is simply a kinematics problem, which is what this program does. You input 2* of body roll, and it tells you how the suspension responds.

At this point, yes I am just comparing the effects of various ride heights on each style of suspension. I could obviously compare all sorts of things such as relocating suspension pickups, but those are not things we have much control over, so there isnt much point.

I have not gotten arround to doing the front yet, although I plan to at some point.
 
#4 ·
Alex W said:
Graph 1 and 2 are the 91 suspension at three different ride heights. Graph 3 and 4 are the 93 at three different heights. The 93 is just my best estamate of the 93, based on looking at pictures of the crossmembers. Should be really close. The length of the tie rods is the only difference.

Spring rate and anti roll bar stiffness effect how much the suspension moves, but not what happens when it moves. The actual motion is simply a kinematics problem, which is what this program does. You input 2* of body roll, and it tells you how the suspension responds.

At this point, yes I am just comparing the effects of various ride heights on each style of suspension. I could obviously compare all sorts of things such as relocating suspension pickups, but those are not things we have much control over, so there isnt much point.

I have not gotten arround to doing the front yet, although I plan to at some point.
Thanks for the clarification. I will have to pull this up during my lunch break and stare at it for a while!

Is there a way you can just model the entire range of motion (toe and camber) with a single line?

Since spring rates are not a factor and it looks like you can just have a larger range of motion and we are just comparing relative changes in toe and camber, then it may be a bit more easier to read. Take the roll out of the equation for this particular graph. I'm assuming that with each spring, as you take a corner, you will be nearly bottoming out with any of the springs.

Ah, it's too early in the morning. I don't know what I'm talking about. Maybe I can come back later...
 
#5 ·
traffic said:
Thanks for the clarification. I will have to pull this up during my lunch break and stare at it for a while!

Is there a way you can just model the entire range of motion (toe and camber) with a single line?

Since spring rates are not a factor and it looks like you can just have a larger range of motion and we are just comparing relative changes in toe and camber, then it may be a bit more easier to read. Take the roll out of the equation for this particular graph. I'm assuming that with each spring, as you take a corner, you will be nearly bottoming out with any of the springs.

Ah, it's too early in the morning. I don't know what I'm talking about. Maybe I can come back later...
The problem is suspension performance is more complicated than that. I can't take roll out of the equation, because as the graphs show, the suspension behaves very differently in roll than it does in straight heave.

If you look at the graphs you will see that lowering the car has minimum effects on the character of the suspension in heave. Examples of this include going over a rise, hitting a bump mid corner, hitting the brakes mid corner, ect. Anything that causes a plain verticle motion of the suspenion.

Body roll, taking into account moving roll centers and all that, becomes much more complicated, and this is where lowering has an impact on the toe changes.

For example, notice on the 91 roll graph. The stock height causes roughtly equal (and very linear) toe changes with roll. The outside wheel toes in, the inside wheel toes out, effectivly steering the rear toward the inside of the turn, limiting oversteer. Lowering 1" has minimal effect on this. However, when lowered past that the toe in of the outside wheel is much reduced, which should cause more oversteer on a lowered car.

The 93 reacts even more strongly to lowering, with even 1/2" of drop significantly changing the toe curves. However, you will notice that the 93 stock appears to behave oppositly to the 91, with the outside wheel actually toeing OUT slightly. It is possible that this is due to my model not being completly accurate (as mentioned before, it should be close). However, it leads us to question exactly WHY the 93 is more stable / "safer" than the 91, when based on toe changes it should oversteer more. I think to answer that we need to look at the toe change in heave, where we see that lifting of the rear (braking) on the 91 actually causes it to toe out slightly (again due to inacuracies in my model, it is possible the 91 is closer to static toe than to toe out, in fact I think this may be the case), whereas the 93 toes in as the rear of the chassi lifts.
 
#6 ·
I suppose the best thing I can do here is say what I plan to do, based on what I have seen here. First, I had considered swapping the to the 93 crossmember, but now I don't think so. The 91 handles lowering much more gracefully, IMHO, and just in general has characteristics I like better. Second, I will be switching to some springs with a bit less of a drop, and a higher rate, maybe RS*R. And 3rd, getting some sway bars (already planned on that) to limit body roll and prevent that huge roll center migration that I am seeing with the car lowered.
 
#7 ·
Alex,

Nice work. I think your graphs (at least in heave) confirm what we've discussed about the differences. I would stress (as you have suggested) that accuracy in measurements is key to getting meaningful results. But it looks like things are going in the right direction.

The 91 should see very little toe change. The 93 should see toe IN under all conditions other than ride height. It makes sense that if you lower the car 1", it's going to toe out for the first inch of lift until it gets back to stock ride height, then it will begin to toe back in.

Camber should be the same for both cars, and it looks like it is.

One thing you might consider is keeping the magnitude of your Y axis constant. The charts look like we're seeing big swings in some cases, but it's actually just a function of the Y axis units.

Now is when things get fun. Start playing with roll centers. As a starting point, consider moving the outboard pickup for both the LCA and tie rod downward the same amount that you've lowered the car from stock. I'd try both 1" and 2".

Next, move the top of the strut inboard about 2" (the camber plate) and adjust camber back to zero at the strut/knuckle interface. See how this effects the camber curve and roll center migration.

I've done a lot of this modeling, but it's neat to see someone do it using a different program. There are definitely things we can do to make the suspension work when lowered, they just require some creativity and some tricky fabrication.

Nice work.
 
#8 ·
Thanks randy.

You are correct about the scaling. Somehow I didn't notice that, for some reason I thought I had them the same. Basically, toe changes from roll are about an order of magnitude smaller than toe changes from heave. For easier comparison I will rescale those graphs when I get a chance.

A quick look at the 91 heave model / calculations has led me to believe that I did have a slight measurement / entry error that gave the 91 the toe out characteristic in heave. I think it really is supposed to be nearly 0 toe change in heave for the 91. This change might also have an impact on the roll, so I will probably recreate those graphs when I am working on it later.
 
#11 ·
rnoll98 said:
You should have this thread moved to the AEF - Suspension and Chassis forum to isolate it from the "what coilovers should I buy?" threads. That forum could use a little action other than the misplaced "what coilovers should I buy?" threads. :)
well, that is becoming an easy question to answer these days ;)

subscribing....
 
#12 ·
Sorry for the lack of updates and new work on this, I have been absolutly swamped with finals and finishing my senior project and all that. But its 95% over now, so hopefully this weekend I can do some more on this...

I will probably leave it here (as it is MKII based) untill it kinda dies a bit and drops off the front page, then have it moved to the AEF forum (thanks for that sugestion by the way, I had forgotten about those forums).
 
#13 ·
Its always great when another member of the MR-2 community takes a serious look at this stuff.

Looking at my notes, the '93 rear ride height is 0.38" lower per the BGBs -- 7.61" i/o 7.99" as measured at the lower arm inboard pivots. The rear arms are 4" longer, 18.5" i/o 14.5", and the inboard pivot for the rear arm is 1/8" lower relative to the chassis.

Mike R
 
#14 · (Edited)
twoina2 said:
Its always great when another member of the MR-2 community takes a serious look at this stuff.

Looking at my notes, the '93 rear ride height is 0.38" lower per the BGBs -- 7.61" i/o 7.99" as measured at the lower arm inboard pivots. The rear arms are 4" longer, 18.5" i/o 14.5", and the inboard pivot for the rear arm is 1/8" lower relative to the chassis.

Mike R
Excellent. Getting other peoples measurements on this sort of thing should allow me to improve the accuracy of my results.

Edit: A couple of thoughts on this... The stock 93 tire is .2" taller than the 91, so that means with the same spring the 93 is .1" higher in ride height, so add that to the .38 difference and the spring difference becomes almost exactly .5"

Second, my 245/40-17 that was on the car when I measured is 1" taller than the stock 91 tire. I measured my control arm inboard pickup to be 6.75 off the ground. Subtract 1/2 inch for the tire difference, and then add 1.75" (the approximate S-tech lowering ammount) and you are right back at the stock 8" ride height. So that means that all of my previously posted 91 numbers should be fairly accurate. I will recheck the 93 stuff with the new tie-rod measurements. Shouldnt change it much.
 
#16 · (Edited)
Had a chance finally to do some more work on this. First, I tried the new numbers for the 93 rear that twoina2 had posted. Found something interesting.

These are the toe graphs for heave and camber comparing his number, which states that the toe bar pickup is 1/8" lower than it is on the 91 (Green line) with my original guess which was that it was in the same verticle plane (red line). I don't know which is correct, the green line seems to me like a strange way to design a car, but then again I have never measured a 93 crossmember, so I dont know.

Basically, the green line says that the inside wheel toes in and the outside wheel toes out (oversteer) and that it toes out initially on uniform compression.



As always, draw your own conclusions, and let me know what you think. Untill I know otherwise I will be using twoina2's numbers, as I have never personally come anywhere near a 93 crossmember.

First post edited with more / revised info.
 
#17 · (Edited)
At Randy's sugestions, I did a couple of tests on the effects of RCA's (Roll Center Adjusters) and camber plates.

I ran the 1.75" lowering with the addition of a 1.75" RCA. I am not going to post those graphs, because they were exceptionally un-interesting. Essentially, the two curves were directly overlayed. SO!! If we had roll center adjusters for both the control arms and toe rods we could lower without effecting the geometry AT ALL.

Next, the camber plates, moving the strut top in 2" per side. These shows a 91 with 1" lowering with (orange) and without camber plates (blue). As you can see, the effects on camber gain are minimal, if that's your reason for getting them they aren't worth it, IMHO. The toe changes were exaggerated with the camber plates. The roll center migration was improved significantly, but certainly not fixed.





I do now have hard numbers on roll center migration. This graph shows lateral movement of the roll center vs body roll. Again, green is stock, blue is 1" lowering, orange is 1" lowering with camber plates moving the strut top 2" inboard.



The spiking is obviously not real, it is program artifacts caused by the extreme movement. It does however show how quickly the movement ramps up as body roll increases.

WE NEED ROLL CENTER ADJUSTERS!!!
 
#18 ·
Alex,
Show the 1.75" of lowering with and without the RCA's. It gets much more interesting I think. The importance is shown here. Lots of things go south with that much lowering and no RCA.

Also, show the 1.75" of lowering with the RCA's and the camber plates vs. just 1.75" of lowering and stock suspension. What you should see is that both the RCA and the camber plates contribute a little to a better camber curve.

We're working with struts here, so don't expect a lot. You still have to run fairly stiff springs to keep from rolling out of the sweet spot, but if you can make the sweet spot a little wider I think you get some wins. On top of that, the rear suspension is very vertical vs the front. I think the front has a better camber curve to begin with and also can benefit from the RCA and camber plate game. I'll try and dust off my program this week and post some of my numbers.
 
#19 · (Edited)
Ok. First of all, 1.75" lowering, with (black line) and with out (red) roll center adjusters. Notice it looks the same as the 91 stock vs 91 lowered 1.75" curves. The RCA's bring you right back to stock characteristics.


Next the 91 lowered 1.75" with RCA's and camber plates. The difference you see is entirely due to the camber plates, as the RCA's bring the curve back to stock, and then the camber plates improve the camber slightly (in roll), and exaggerate all toe movements slightly. As always, green is stock, red is lowered, and orange is lowered with corrections (RCA's and Camber plates).




I will agree the camber plates help, but not nearly as much as the RCA's, and if I had to chose one it would be an RCA every single time. The camber plates alone dont even come close to fixing the roll center migration.
 
#20 ·
it looks to me like the big problem is the lack of camber under roll.. on paper and in practice.

to get a good handling mr2 you have to run so much camber it destroys the inside of your tire, and straight line suffers. ideally, you'd want much much less camber and have a fantastic amount under roll.. this should also give you better straight line traction and real world faster times.

when i lowered/stiffened my car i didnt think for a second that i lost any particular good handling trait. i have cursed the crappy camber curve many times though. im just wondering if you went through the hassle of correcting all the roll centers and getting that inline.. if you'd genuinely have a faster car.
 
#21 ·
Please define Roll Center Adjusters. Are you talking about the crash bolts at the bottom of the strut? From what I can tell on my 93 na, to get camber plates to be of use, I'd have to change over to coil over size springs as stock diamater springs will limit the amount of neg. camber you can adjust to. Also another reason the 93 handles different then the 91 is the caster is different. If what I read is correct, Toyota changed it to increase stability and reduce snap spin.
 
#22 ·
I think the roll center adjusters move the point that the lca connects to the spindle. Thus moving the roll center closer to the center of gravity. This will help with the car's tendency to roll in a corner. This is a very common theme in Macpherson struts-as you lower, the roll center drops faster than the CofG and so your cars tendency to roll increases. I'm not so familiar with its impact in the rear. But in the front, you will lose front end grip. Now why do you lose grip? Maybe this graph can help explain it.

For a given spring rate, when you take a corner and you have increased your cars tendency to roll, as the graph shows, you may lose camber. Now, you can counter this by increasing your spring rate or thicker sway bars. However, with each step, you've created a potential negative. Sway bars transfer loads (maybe too much, too fast) and high spring rates can be uncomfortable on a street car.

If you have correct CofG/RC relationship (near OEM ride height) even though you are running lower spring rates, your cars tendency to roll is less. So your car rolls less. Providing less degradation of camber and more grip. I think this is easy to feel for the front suspension. Not quite sure about the rear.

Can you please explain how caster may impact the stability? I can understand for the fronts (but you can always adjust to '93 settings) but what is happening in the rear (non-adjustable?) It is my understanding that the lively rear of the '91 is from less dynamic toe-in as the suspension goes through its motion.
 
#23 ·
Alex W said:
Had a chance finally to do some more work on this. First, I tried the new numbers for the 93 rear that twoina2 had posted. Found something interesting.

These are the toe graphs for heave and camber comparing his number, which states that the toe bar pickup is 1/8" lower than it is on the 91 (Green line) with my original guess which was that it was in the same verticle plane (red line). I don't know which is correct, the green line seems to me like a strange way to design a car . . .
Maybe something was lost in s. I will send you a spreadsheet off list with all of the details.
 
#24 ·
Alex W said:
I will agree the camber plates help, but not nearly as much as the RCA's, and if I had to chose one it would be an RCA every single time. The camber plates alone dont even come close to fixing the roll center migration.
It's not an either/or. I'm just trying to throw what I know at you to get the BEST possible camber curve we can without welding A-arms to the chassis.

The simple fact is the further you move the strut top inward, the more camber gain you get as the suspension moves up, because for the most part, it's moving toward that strut top until the LCA starts moving the LBJ inward, thus RCAs. The RCAs make the lower arm point downward rather than up on a lowered car. That means as it swings up it's moving OUT before it moves back in. So we have the bottom of the wheel (lower ball joint) moving out, and the top of the wheel moving in, and we get more camber.

If the LCA is the base of a triangle, and the strut (king pin?) is one side, the smaller the angle created at the LBJ, the more camber you will gain. At 90 degrees, you don't gain any, and larger than 90 degrees, you lose camber. This is why RCAs and camber plates both help camber gain. They both make that angle smaller. Roll center height helps keep this in check so you don't go overboard. Eventually you'd end up with a swing arm :)
 
#25 ·
It's a total compromise. If you create a camber curve so great that you can roll the car 2 degrees and see zero change in camber, you've just hurt straight-line braking/acceleration because as soon as the nose dives or the rear squats you instantly gain a lot of camber and your contact patch starts going bye bye.

I'm gonna try and dig up my files. I recall acheiving something I was content with on the front. I think the rear is tougher because the strut angle isn't as good.
 
#26 ·
Well said Randy. A while back I think we guessed that 1 deg of camber gain with 2 deg of roll would be a reasonable compromise.

The front suspension is a much different beast than the rear. The front struts are just about the shortest struts found on any street car and they are also inclined a fair bit. The result is they don't behave like the typical sedan strut suspension described in the text books.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top