Is Suspension Tuning Backwards? - A Test with RX-8

[///M-Spec]

I'm not projecting hostility on the laws of tuning, i'm projecting some well-founded criticism and amazement that the posters who wrote such articles as "The Pocketguide to GT4 Tuning" evidently NEVER tested it in the actual game. As if they just assumed it worked.

Maturin, I'd like to point out a few passages at the begining of my sticky at the top of forum.

This Guide was written for GT3. However, it will serve as a great guide for understanding cornering forces and suspension terms. We will update it as required as we learn more about GT4*

* emphasis mine.

I actually don't know which of us wrote that, since it wasn't me who set up the this forum (I'm just a Mod, only Admins can create forums). But needless to say, the guide tells you right off the bat where it's coming from. Ask yourself: how could the sticky be seriously tested if GT4 has been out all of three weeks in North America and just now making to the shelves in Europe?

The tips in this guide are taken from real life tuning principles, but its important to remember that the GT3 suspension model is abstracted and grossly simplified compared to a real racecar. I also claim no special knowledge of the inner workings of GT3's physics modeling: I'm simply applying what I know about real cars to what you can do in GT3.

In addition, I've also striven to ensure the accuracy of the information contained within, but obviously can make no guarantees as to the effectiveness of any proposed solution. In other words: I hope this information is useful, but Your Mileage (literally) May Vary.

...and of course, this passage which I did write tells you exactly where I'm coming from.

And to answer your earlier question, yes I play GT4. No, I'm not very far in the game yet, since a) I'm very busy these days and only get to play a few hours a week and b) most of my time is spent on license tests at the moment, since I don't move on much until I gold them all.

You are, of course, welcome to disagree with any thoughts on tuning I've put forth, but let me stress something: everything I've written, I've taken from real life principles and simplified them for use in GT. If something does not appear work the way it's supposed to in GT4, I would be just as disappointed as the next guy. However, I would refain from jumping to any conclusions about the accuracy of the sticked guides, since they do, in fact, concur with information taken from the real world. And since GT is a self-proclaimed simulator, I would except it to reflect real world properties... such as spring rate settings. Wouldn't you?

That said, I spent a few minutes (not long, just 20 or so) playing around in GT4 last night. I don't have an RX-8, but instead picked another 50/50 weight dist, FR car: the M3.

I tried numerous combinations of bone stock, default, moderately stiff front, moderately stiff rear, super stiff front, super stiff rear while leaving all other bar, camber, toe and height settings default. Tires were default Sport Mediums that came with the car. Test course was Grand Valley Normal Practice Mode.

Results: moderate understeer in steady state turns using ALL configurations. Strong understeer under throttle using ALL configurations. Neutral under braking in ALL configurations, except for super-stiff front.

The only variation I observed was with the super-stiff FRONT setup (which is supposed to promote understeer), I got plenty of LIFT-THROTTLE oversteer. But under throttle, the front end continued to plow forward. Interesting.

Not sure what to make of the results yet. The test was not in-depth or very scientific. But my initial impression is spring/shock settings.. on their own, do not seem to affect the handling properties of a typical balanced, FR car in GT4. The lift-throttle oversteer was unexpected, however. I'll do more testing this weekend. I've been meaning to buy an RX-8 anyway.

A question for those of you who have tested thus far: WHERE are you getting oversteer/understeer. Are you getting them part, open under throttle or steady state coasting? What tires? What track?


M

 

[eggsOverEasy]

Greyout;

Since i have seen people being understanding of you as you seek to grasp your knowlege of settings several times in this forum, i would ask that you have the courtesy to do the same. You tell people to add their two cents, and then tell people who don't understand not to post. Don't be so eager to dismiss someone who doesn't understand, as long as they are interested in tuning the same as you are.

That being said, i am having trouble understanding when you say

it has been established that in real life, stiff rear / soft front causes oversteer.

When you say soft front do you mean the spring rate, and/or the bound and/or rebound?

I had done alot of reading in december mainly from the GT3 forum. Someone posted a link from a racing site that had a list of real world adjustments in order of how much affect they had on over/understeer. From that and discussion on that forum, i gathered that bound and re-bound have the most effect. So i thought perhaps this was overshadowing the adjustments he made to spring rate. (which i think is what you are reffering to when talking about the front/back being soft/hard)

Cheers,
eggs.

 

[bfifteenv]

One of three things is the case:

1) someone in charge of something at PD royally ****ed up, and the slider on the spring stiffness scale is backwards - as the slider increases, the spring rate is actually decreasing. The car DID seem to exibit a bit more body roll up front with my superstiff springs, but I may have just been imagining it - but its POSSIBLE this is the case. If this IS the case, the question is, what about the damper slider? or the sway bar? We know the camber slider is correct at least...

2) The physics model created by PD has a serious, UNINTENTIAL flaw (again, read the descriptions in the game!).

Either one of these two is equally possible.

Is it NOT equally possible that the new physics model is just much more complex than in previous installments and that the general rules of thumb for tuning a car in the game no longer apply in all cases?

Yes if everything else is sorted, lowering your spring rate should have the result of increased grip at that end, but when sorting out the suspension on a race car the first thing professionals do is never change spring rates, it's usually alignment adjustments. I have, on just about all of my cars in the game, been able to make great improvements in the handling with minor camber and toe adjustment. If those don't do the trick, then I move to damper rate and anit-roll, then move onto spring rate. And in every case the modifications have had the effects that I have expected.

Adding rear toe helped the car turn in better
Adding a slight amount of camber helped increase ultimate grip at that end in a corner, adding too much caused the oposite affect
Decreasing damper bound helped the car remain more stable through bumpy corners
Increasing rear anti-roll helped an understeering car remain nuetral through corners
And every time I have messed with the spring rates (granted only a few times because I usually get better results through other means) it has required more tuning in other areas to see the full benefits of a spring rate change.

I would submit that PD knew that the new physics model was going to cause trouble for people and that's why they included the new "Original" suspension option in the tuner shops as in my experience these come closer to fully sorted than the other suspension systems do.

it has been established that in real life, stiff rear / soft front causes oversteer. If you disagree with that, please do some more reading before you post in this thread.

I've read a few books in preparation for being in charge of suspension design and setup on my Univeristy's Formula SAE car my third year on the team, and when the car was built and had an oversteer problem, the first thing I did was NOT swap out springs.

 

[eggsOverEasy]

Yikes! Earlier i said

I'm not sure i see why the settings are backwards.

In going from setup 1 to setup 2, you are reducing the bound and rebound in the rear....

And i went on to explain why i thought this would indeed cause understeer. The only problem is maturin was NOT reducing bound and re-bound in the rear! He reduced it in the front, which would of course cause the opposite.

So now i am us mystified as everyone, else. Plus alot more embarrassed. I appologise for adding confusion to an already difficult subject. I can see why greyout was frustrated!

 

[sukerkin]

Not a problem Eggs - I for one just assumed that you were a long-standing member of the "Lower Is Stiffer, Higher Is Softer" school of Damper tuning .

It's nice to have another engineering brain on the case, so don't be put off contributing if things get rough in a thread. I'm an engineer too, unfortunately not a Mechanical one (electrical and software), altho' I have spent the past three years or so studying suspension dynamics in an effort to improve my GT3 tuning :blush:.

Also, never be afraid to make a mistake and confess it later - I've had to do it recently myself too when what I typed wasn't what my mind was thinking (... mutters to self "Negative Toe numbers means Toe OUT in the game, eejut!")

 

[eggsOverEasy]

Thanks for the encouragement!

Not a problem Eggs - I for one just assumed that you were a long-standing member of the "Lower Is Stiffer, Higher Is Softer" school of Damper tuning .

Yes, i remember that debate! I think both camps had hoped that PD would do a better job of indicating what the in-game setting numbers reffered to exactly when they came out with GT4! Sadly, i think there is even less in-game info on the settings, and even the cars themselves. Perhaps this will appear in the soon to be released guide.

 

[Maturin]

Message removed by poster, final four test done with ASM on, due to using the tri-set buttons at the top.

 

[Maturin]

message removed by poster.

 

[///M-Spec]

Thanks for sharing your test data with us, Maturin. I presume you did this again at Trial Mountain?

Would you mind trying this on a fairly flat track with many constant radius turns, such as Toyko, Seoul or New York? Without cambers on the turns to upset balance, I think it would make for somewhat more scientific testing. We could always throw in some big cambers later like High Speed Ring.

If I have time tonight, I will try this too. I'd like to get to the bottom of this as much as anyone else.

Off the top of my head, I recall the RX-8 has roughly 50/50 weight distribution. Does anyone know if the Japanese S spec car has same width tires front and rear?


M

 

[Maturin]

Thanks for sharing your test data with us, Maturin. I presume you did this again at Trial Mountain?

Would you mind trying this on a fairly flat track with many constant radius turns, such as Toyko, Seoul or New York?

I did do it at Trial Mountain because it's my "test track" (bumpy, curvy with two straights). I'll go to Tokyo since I know that one better (I went to New York )

 

[Maturin]

message removed by poster.

 

[Canadiandrifter]

EDIT...

Alrighty then...

 

[JaeTea]

Default/Set A/Set B/Set C/Set D

Springs: f10/r10 -- 10/10 -- 10/12 -- 10/13.5 -- 10/17
Height: 120/120
Bound: f8/r8 -- 7/8 -- 7/8 -- 7/9 --7/9
Rebound: f8/r8
Camber: 2.0/1.0
Stabilz: 4/4


Results:

Default: Very slight oversteer, or essentially neutral
Set A: Slight oversteer
Set B: Moderate oversteer
Set C: More than Moderate oversteer, almost driftable
Set D: Noticeable understeer, hit wall coming out of rotary

Crrrazzzy!

I gotta try this tonite.

 

[Maturin]

so somewhere between 13.5 and 17 is where it crosses the imaginary line with that setting..

Can anyone make up a calculation out of that?..There has to be some numerical formula for determining it?

message removed by poster.

 

[Maturin]

We still have the original problem. When I did the tests, I used the "Set" buttons at the top, but forgot that it doesn't migrate over the ASM settings. So the four tests done after default that I did, were using the ASM and the results are useless.

I tried replicating my results with ASM off, and I could not do it. The game still oversteers when you soften the rear.

 

[Greyout]

first: I apologize for my harshness before. I just didn't want to see a bunch of posts explaining the "physics" behind how a softly sprung rear end will loose traction over a stiffly sprung front.

anyway, moving on.

here is a quick test I did before I ran off to work again

Car: Honda prelude

1st test:
Springs: at 10/10
Shocks: at 7/7
ride height: ALL THE WAY UP (I'll explain why in just a second)

I then went around Tsukuba

results: slight oversteer when off the throttle, understeer when on the throttle. This was my base line.

I then LOWERED the spring rates ALL the way, to like 3.5 or whatever it is, and left the dampers the same.

Same track... now the reason I had the ride height all the way up, was in hopes of being able to have the most suspension travel available as possible, to make sure my findings were accurate.

All though the balance, and total grip of the car, did not change too much (wouldn't expect it to...) the body leaned hard in all directions, and pitched foward and backward much more. Was harder to control in transitions.

I then went full stiff front and rear springs, (18/18), again leaving the dampers the same.

The car cornered much flatter, with much less pitch and squat.

my conclusion from this basic test is that the slider & label on the spring rates is infact CORRECT, and that when the sliders move to the right, the car does behave as if it has stiffer springs, at least when it comes to body roll & pitch.

At this point, I hadn't introduced any different spring rates between front and rear, and although this seems like a useless test, it was a slight possibility that they had the slider backwards. Well, for what its worth, its not.

my next test was the sway bars.

with my FULL SOFT spring setup, front and rear, I moved the sway bars from 1 to 7 at each end.

the sway bars don't have a huge effect. I always wondered what 7 really means - slightly thicker then stock, or a veritable solid steel, tent-pole thick bar. Well, its more the former. The car exibited SLIGHTLY less body roll.

I hadn't considered that the rebound and bound might be reversed as someone had mentioned earlier in this thread, and that is a new point I hadn't considered, but I would doubt it (anything is possible I guess).

my next test involved putting the rear springs at their minimum (3.x) and leaving the fronts at 18. Dampers were left at 7/7.

The oversteered LIKE CRAZY. The idea of a FWD car, at FULL THROTTLE, in second gear, drifting completely sideways around a 180 degree corner with the tires pointed straight ahead, or even slightly countersteered, is quite hard to wrap ones head around.

I didn't mess with the dampers much. Dampers have a big effect on the balance of a vehicle in any kind of transition, for sure. however, I limited all my observations to steady-state corning, ignoring any turn-in push or drift, as well as any trends when jumping back onto the gas.

3.5kg/mm is about the spring stiffness of a completely stock prelude. The theory that the default settings match the stock settings, at least with spring rate, is fale. 18kg/mm is a very common relm of race spring run in the rear by track-prepped FWD cars (well, they actually go beyond that frequently...). My ride height was insuring that no bumpstops, bottoming out, or lack of travel was throwing the results.

I did not do the same test for sway bars, and did not play with shocks.

moderate levels of the same test resulted in similar findings

F= 15
R= 10

(same everything else)

-good amount oversteer, but nothing quite as insane as before.-

so, spring stiffness works backwards in this test. I can not imagine the in-game physics that could rationalize it, but I suppose there is more tests to be done.

the next question is: Does stiff front / soft rear sway bar do the same thing, or does it behave as it should in real life?

edit: those 4 test results are very, very, strange - especially how the rear spring rate crosses some barrier and becomes ineffective (that maximum rear spring rate you tested is by no means an unreasonable spring rate)

 

[Maturin]

edit: those 4 test results are very, very, strange - especially how the rear spring rate crosses some barrier and becomes ineffective (that maximum rear spring rate you tested is by no means an unreasonable spring rate)

It's because I forgot to turn ASM off for the last three tests. The tests are useless. Although the "understeer" cross-over is true, must be a quirk of the ASM (or it stopped correcting after a point).

 

[Greyout]

oh ok - n/m...

I guess I'll play with it some more tonight...

 

[awdrifter2]

Deleted.

 

[KAuss]

With a NSX Type S, the stock top suspension mod settings are

13 / 10
111 / 111
8 / 8
8 / 8
2.0 / 1.0
0 / 0
5 / 5

The NSX has really mild oversteer and no rebound to straighten out the car... Note this was with a fully tuned NSX Type S with 426 hp... I tested this in Trial Mountain and in practice mode to do the flat S turn only... The car would turn in with mild oversteer, and hold the slip angle with no rebound until ultimately it spins out or head into the wall...

What I did to counter this was out of this world and made no sense to me...

13 / 12
111 / 111
4 / 4
4 / 4
2.0 /1.0
0 / 0
5 / 3

Now, to counter the NO rebound, I soften the bound rate and lower the rebound rate... This made no sense to me at all... However it seemed to work... I also lowered the rear stabalizers to hope the rears will catch while at slip angle... To which it did... However the car would catch TOO much and end up oversteering on throttle... So i tried to SOFTEN the springs and hope to get it to catch more pavement when I floor it and it just got worst... So what I did then is set the rear spring rate from 8 (the lowered #) to 12 and the car is absolutely nuetralized itself... Only on full throttle and body roll will the car over steer...

At this point, the game made no sense to me... Also, I have no idea why the car will NOT turn in under trail braking... You'll generally have to brake during the straight and let off the brake to coast into a corner... Mild understeer if you plan on WOT out of a corner with no body roll...

I've given up on tuning it any more because the physics just didn't match what my intentions were... I didn't dare to say the game was messed up, but now I'm reading this thread and it's reassuring my findings...

 

[RoxxorJoorSoxxo]

You are understeering because you are turning in way too hard and braking, more than likely

 

[Kremithefrog]

To me it makes sense that stiffer rear springs remove some oversteer. Keep the tires planted better.

I think when you lowered the bound rate it lessened the need for rebound coupled with your other settings it made a neutral car. I doubt those settings are THE best, but if it improves your times then they're definitely not bad.

 

[Mr Whippy]

LOL, GT4 sounds totally fooked up beyond belief...

I'm really gonna have to test this later tonight.

I have a feeling that dampers are a big problem here. I've done some work with a PC sim (Racer) and built an M3 recently. I've been implementing all the camber change with bump and bump steer and toe changes with bump etc to get it perfect. The tyre model is all good too.

The dampers made so much difference to the handling. Too much bump at the front makes a car understeer like crazy, so softening the front springs leaves the car over-damped, so as softening the rear springs leaves them horribly overdamped.

Having the rear over-damped by a large degree makes the car ride on the dampers. A damper creates a reaction force to the velocity of the damping shaft. If the spring is very soft in comparison to the damper, like you may now be using, then the shaft velocity and hence damping force is very large in that small instant at turn in. This in effect works like a very stiff spring. There is very rapid weight transfer as the suspension does not deflect, the damper just transmits the force to the tyre quickly.

I never liked GT series for dampers. What on earth does 1-10 mean? They should give N/m/s ratings for the dampers, so we can really understand what data we are putting in.
For example, are the dampers absolute, 1-10 for the softest cars to the stiffest race cars. 1 in this case would be like 1000n/m/s, and hardest might be 10,000n/m.s for a very heavy stiffly setup SLR Merc for example.

If this is the case, most road car wheel rates, like 50,000n/m say at the front of an m3, want around 2000n/m/s bump, and 4000n/m/s rebound.

Again, are bump and rebound relative to each other? Are the force curves linear, digressive, sqrt*v, who knows? I can only guess again that perhaps the fact we use roughly 1/3 to 1/2 bump values over rebound is taken into acount?

Ie, irl, an M3 might use 1500 bump, 4500 rebound, but in GT4 what do we use? I'd guess we'd use maybe 4 bump, 4 rebound. 4 rebound gets us maybe 4000n/m/s damper rate, and then the 4 bump is an appropriate relative amount for the given rebound rate.

Either way, for this testing, I'd always set damping to 0 or 1, dunno if you can totally remove the damping. This leaves it all on the springs for weight transfer.

If it's still oversteers with a softer rear roll rate, then I assume that the front inside should be lifting off the floor?

Is it possible to soften the front and make the rear super stiff and make the rear inside wheel lift?

I think either it's the dampers, or GT4 is seriously wrong?

I hope someone can work this out, as if they can't, it seriously throws into question the quality of GT4... lol... what a bug if it is one

Dave

 

[Greyout]

I have not been able to lift either inside wheel, regardless of how extreme the spring settings have been.

it would be quite comical to see a FWD lift the inside front, as it oversteers drastically. It just doesn't make any sense.

I'm starting to think more and more that there is a limit in the game that will not allow a wheel to leave the ground on its own from body roll - i.e., the body rolls, the suspension is in droop, and runs out of travel. Kind of like a reverse bump stop - when you reach this limit, normal physics is thrown out the window.

I agree that the testing should be done with the dampers at the lowest setting, and on the flatest, smoothest track.

I don't think Trial mountain is a good track for this. Lots of elevation changes, for one.

I'd suggest Motegi (either short course), Tsukuba, motorland II, or even the driving park.

edit: also, the tests listed above are difficult to take info from, because you are adjusting several variables at once. When the setups are as confusing as they are here, I strongly suggest adjusting only ONE item at a time.

 

[Mr Whippy]

Hmmm, has anyone checked to see if front/rear are labelled right...

Try putting ride to like 200mm front, 10mm rear, and see if it's going around on it's back end to check your actually changing the front suspension when altering the front!

Is this on PAL and NTSC, or just one? Could be a translation error if above is true?

If it is, then I'm guessing it's pretty bad...

Is this with ALL the suspensions? Has anyone tested with the others?

Hmmmm

PS, just ordered my DFP

Dave

 

[Greyout]

I have a drag-race-only prelude in the game just for run. The front is at 98mm and the back is at 140, or something like that.

the car is indeed slammed in the front and raised in the rear. the height adjustments are correct.

that doesn't mean the spring ones aren't backwards though.... that would be interesting...

I'm actually hoping it turns out that they are just labelled backwards - I can deal with that without thinking twice, but I can't deal with a completely screwed up physics model.

 

[sukerkin]

Hi Mr Whippy

Nice to see that such threads still attract informative posts and questioning approaches 👍

It's reassuring to hear someone else asking the questions "out-loud" so to speak that a small group of us have been discoursing in detail over the past few years .

I've always likened suspension tuning in the GT series as somewhat akin to plaiting fog and it's only by thousands of hours of testing that we've arrived at a general consensus of what works and what doesn't. It's good to know that we weren't the only ones who decided that it was not possible to mathematically model what's going on 'under the bonnet' of the GT games :sigh: .

 

[Greyout]

I guess one way to test backwards labelled theory would be to go full stiff in the front, full soft in the rear, and see how the pitch & dive under acceleration and decelleration changes when you reverse the settings (with dampers at 1/1...)

with the soft front / hard rear, the car should pitch foward heavily under braking, but not squat very much at all under acceleration.

I'm at work right now (geee, imagine that) - anyone at home able to test this?

 

[Mr Whippy]

Will be testing extensively on my M3 CSL when I get home...

1 damping, no camber or toe etc... Use ballast to get perfect 50/50... N1 or N2 tyres (most realistic response to weight transfer imho)...

Then I'll set front rates to 20,000N/m and rear to 20,000N/m and no arb

See if I can then up the front to say 100,000 and notice it oversteering (which is what people seem to be getting)... also I'll try note if the front end gets stiffer, or if it's the rear end getting stiffer! Hmmm, that could well be the problem!

Then I'll go back to 20k/20k, and then work on the arb's... again, see if they are working right or labelled right...

PS to others, oversteer is effected by loads of stuff, if, a steady state turn in gets you oversteer, reducing rear roll stiffness gets you more understeer. You can still hold a slide well with low rear-roll stiffness. Drifting/sliding is not oversteer, oversteer is the phase when we end up sideways, once sideways, we are technically oversteering, but it's not effected by the setup as load transfers are stable now, not changing as much!

Steady state turning on a skidpan will reveal the results... I'm sure I can get to the bottom of this tonight

Dave

 

[Greyout]

stupid work