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

[Greyout]

I will try the "small adjustment" thing again tonight I guess...

And guys... there REALLY isn't more weight transfer at the end of the car with softer springs

 

[Bad760]

Ok here are my observations from an hour or so of tinkering last night.
Brace yourself it's going to be a long one

Now I didn't do super scientific testing, I just did a whole lot of hot laps and looked at the handling of the car from a racers point of view.

Controller - Dual Shock 2
Track - Grand Valley East section
Car - M Coupe
I have all stability control & TC off always
Shocks for all setups - F/R - 7-8/8-7 ( bound and rebound)

First with R1 tyres-

My race setup (well the parts that matter to this discussion)
Springs - F/R - 9.5/9.0

I did half a dozen laps with this setup for a start to get settled in
-----
Changed set up
Springs - 9.5/8.5 (softer rear)

Now with the slightly softer rear tyres I was immediately slower, I could not carry as much speed into the corner I had to slow it down more or use more lock (which automatically scrubs speed) to get it to turn in the same and could not get on the power as early as before ie. it had less corner exit steering

Conclusion - the car had less steering everywhere, which means the rear had more grip, which was predictable
-----
Changed setup
Springs - 9.5/9.5 (rear springs harder than original setup)

I went faster than my original time 2nd lap! I could carry speed way deeper into a corner than my original setup and get on the gas earlier, the car just rotated better in the corners.

Conclusion - the car had more steering everywhere, which means the rear had less grip, again predictable
-----
Changed setup
Springs - 9.5/15 (full hard in rear)

Was slower again, the car had a lot more corner exit steering, I could light the rear tyres up relatively easily, but surprisingly it developed some understeer on corner entry in the mid to highspeed corners.
Overall the car seemed very unbalanced and would switch from oversteer to understeer seemingly at will

Conclusion - car had less rear grip everywhere except for where I highlighted, it seemed like the settings were so out of whack that the physics engine got confused.
-----
Changed setup
Springs - 9.5/12.5 (rears inbetween full hard and my normal setting)

Was still slower, now this setup behaved more like I thought it would, not much rear traction on corner exit which meant that I had to go easy on the gas coming out of corners, on corner entry the rear was a little bit more skittish than my normal setup but not dramatically so, the only exception was on the very highest speed corner it still under steered a tad.

Conclusion - car had less rear grip, but the car felt more balanced and consistant than the full hard rear spring setup, it didn't switch from over to understeer at will
-----
Changed setup
Springs - 9.5/4 (fullsoft rear)

Was still slower, now the car had heaps of turn-in off power steering, infact violently so but as soon as I touched the brakes and slowed down into the middle of the corner it switched to massive understeer also with massive understeer on corner exit.
Now I could get the car into a nice drift if I was violent with the steering and brake/throttle so as to keep the rear of the car unstuck from the initial violent oversteer, but if I let the rear settle down the understeer reared it's head again.

Conclusion - car had way more corner exit rear grip, but less high and midspeed rear grip - same sort of comment as the full hard rear spring setup but reversed, again I'm not sure how to explain the oversteer, but it seems to correlate wit hmy real-world findings
-----
Changed setup
Springs - 9.5/7.5 (rear in between full soft and my normal setup)

Was still slower, the car had a little less corner exit rear grip than the full soft settings and but still had the oversteer in the highspeed corner throttle liftoff case

Conclusion - see my middle stiffness rear spring conclusion but just reverse it

Now with N3 tyres

The same observations apply as for the above cases although I was 5 sec a lap slower and the differnces in handling where increased with the less grip available, I was however more easily able to control the attitude of the car with the throttle, being able to make it wheelspin almost at will in any corner.

-----
At the end I went back to my new slightly altered race setup (changed the back springs to 9.5 from 9.0) and went almost 1sec quicker than I had 3/4 of an hr ago with the same setup.

-----
So the conclusion that I take from all of that is that possibly the physics engine in GT4 is very lifelike, it is alot more consistant to the engine in GPL which I play a lot and is widely recognised as a very good sim, the problem that I think we may be seeing is that because it is so real not many people would have run a real car with such hugely different spring rates and so they would not know that it is not a strictly linear thing, if you go to soft the you will get some oversteer, and so setup changes made in GT3-2-1 may not apply for the full adjustment range

This is just my opinion from what I have observed myself

 

[Maturin]

-----
Changed setup
Springs - 9.5/4 (fullsoft rear)


Conclusion - car had way more corner exit rear grip, but less high and midspeed rear grip - same sort of comment as the full hard rear spring setup but reversed, again I'm not sure how to explain the oversteer, but it seems to correlate wit hmy real-world findings

er...I don't think so. That car goes sideways with that set-up, pure oversteer. Is that what you found in your "real-world findings"?

Also, your test isn't any good without your shock settings. And "less steering everywhere" doesn't exactly convey "understeer" or "oversteer"

 

[Bad760]

er...I don't think so. That car goes sideways with that set-up, pure oversteer. Is that what you found in your "real-world findings"?

Also, your test isn't any good without your shock settings. And "less steering everywhere" doesn't exactly convey "understeer" or "oversteer"

My real world findings show the same off power turn in oversteer that I found in GT4

Now looking at a real world example, I race 1/10th scale RC touring cars at a national level.

Now I can tell you that when I put the hardest springs on the front and softest on the rear I most definitely get oversteer turning into and mid corner as the rear of the car wobbles around with such soft spring rates, but bad understeer out of the corner as the car squats and loads up the rears. I can't easily explain that as my understanding of roll and pitch moments and moving roll centres and roll axis isn't 100%.

If you had bothered to read my post on the previous page you would have seen this example

Perhaps I should have explained myself more clearly, when I said less highspeed rear grip I meant purely when coming off the throttle.

I disagree that "it doesn't mean anything without shock settings", we are testing spring rates here not shocks as long as the shocks are the same for each test they will have no big impact, for the record though I have edited my post to show shock values.

Now I don't know how else you would describe "less steering everywhere" apart from understeer. Understeer basically means that the front of the car is going in a wider arc than you would expect for the steering input that you input.

And i did say at the end of my post

Now I could get the car into a nice drift if I was violent with the steering and brake/throttle so as to keep the rear of the car unstuck from the initial violent oversteer, but if I let the rear settle down the understeer reared it's head again.

Now I wouldn't call that an oversteering car I would call that a very un-balanced car (and I did) switching from oversteer to understeer rapidly

Contribute to this discussion(that you started) by all means but please don't dismiss other peoples findings and opinions because they don't support your own.

I think that I have proven that what you have found is true and also can happen in the real world

 

[Shoes_98ls]

this is what im starting to think as well. I believe that PD may have used a pendulum style phycics engine...

why do i say this?

well it all has to do with that damn G-meter at the bottom of the screen, which i might add is TOTAL inaccurate compared to RL. If you go into a corner, lets say a right turn for the sake of discussion, the ball in the meter will swing wildly to the left of the screen... ok thats true to real life, other than the fact that you can pull 1+Gs on almost any stock car with N1-3 tires.

I THINK what im starting to see is the car acts like a little weight on the end of a string. you turn the car in, the "string" will swing to one side, and stay there regardless of how fast or slow you are going.... however, once you get on the gas to rotate some thru the corner the "string" will stay in the same spot.. why it does this im not sure, and it seems to do it more with rwd, vs fwd, for some reason.

really the only way to consistantly get the rear end out on a car is to feint into a corner... regardless of entry speed, or the angle of the turn.

this, to me, shows the pendulum style physics engine. Think of the "string's" movement, agian in a right turn, if you were to feint into the corner. It would swing to the right (towards the apex of the corner) and then accelerate all the way thru the neutral (strait up and down) and left most position it would reach. perhaps this is how the decided to emulate weight transfer?

agian, im not to sure of this myself, so im going to go test it out some more...

please feel free to call me an idiot

disregard all of that garbage....

after playing a tad more, and actually paying close attention to my input vs the reaction of the car... i realized that my "theory" was, well, just dumb. this might be partially due to my tire selection.

ill play around with it more tomorrow.

 

[Bad760]

hehe

 

[awdrifter2]

Deleted.

 

[sukerkin]

Excellent contributions chaps - Bad760's write-up has pretty much convinced me that the physics engine in GT4 is okay for the application of real world techniques. At first, cursory reading, his results are exactly what I'd expect. Colour me a happy engineer .

Many thanks for the kind words too Pontiac - it's good to know that the scribblings of myself and my compatriots have added to the enjoyment of the game for someone outside of our own group 👍.

Greyout, I'm not sure but I think what it is that you're missing, regarding the contrary behaviour with settings that are too soft at one end, is that it is not weight transfer per se but rather grip loss due to not enough weight and a mangling of the suspension geometry and CoG.

This is a reasonably recognised phenomenon in automotive circles and the fact that GT4 replicates it is actually a hopeful sign for the game mechanics.

I'm afraid that practical input from me on this will be limited for a while as I've, stupidly, dropped a TV on my right hand and haven't really got a lot of use from my index and middle fingers {or accelerator and brake in game terms }. It'll be okay shortly but I think we're nearing the end of this investigation anyhow and, altho' more data is always good, I may be in danger of performing redundant testing by the time I check out some spannering details.

 

[rk]

I hope this remains an ongoing investigation which these observations are intended to help, I anxiously await a conclusion. I spent approximately 10 to 12 hours with the CLK-GTR on the same track in practice, changing one thing, one or two clicks, then driving it; I may post a tuning algorithim based on the experience. The car seems to react to suspension changes consistent with expected results. Usually one extra click on the rear spring would make the car more responsive, one lesser click would make it more stable, the same, in reverse, held for the front wheels. (That is what we are supposed to expect, isnt it?). Something I encountered during this test that surprised me: even with max front and minimum rear downforce- and minimum lsd (asm disabled)- I could elicit no more than a sluggish throttle oversteer, which seemed odd for around 1000 hp.

 

[awdrifter2]

Deleted.

 

[rk]

I have a set of settings of the CLK-GTR that can kick the tail out easily. I will post them later when I get home.

Thank you for the offer and I will gladly compare your values to mine, however, the point I was making was that it seems odd that any combination of suspension settings could result in all but preventing throttle oversteer when the aerodynamics are set to such extreme; remember the driving aids were mostly disabled and lsd was 5/5/5 if I remember correctly.

 

[awdrifter2]

Deleted.

 

[DrBiggly]

In testing the settings in GT4, I've found settings to be consistent with what Greyout has experienced.

In real life I compete in a modified '02 WRX for SCCA Solo2 in STX class. After using two different types of coilover setups, multiple swaybars, more alignments than I can count, and springrates of the following:

7k front, 5k rear
8k front, 5k rear
7k front, 8k rear
8k front, 10k rear

I can attest from firsthand experience that a higher rear springrate indeed causes more oversteer.

My personal opinion is that while the update physics engine seems updated and more realistic in ways, a serious flaw seems to be that the effect of the front and rear springrates are backwards with respect to understeer and oversteer, but not for any longitudinal weight transfer issues.

I can setup a car similar to my own in the game (including camber numbers, which currently is about -3.2 in the front and -1.1 in the rear) and the result is a vehicle with handling dynamics absolutely nothing like the real thing.

As another test, I had a friend of mine who runs a driving school (http://www.seat-time.com actually) and has not only raced on a multitude of tracks, but has done development work (as a driver) for an LMP team as well try out a CRX in the game. He has one that he also runs in SCCA Solo2. We set the car up exactly like his...and nothing worked. I tell him to swap the springrates from front to rear. The car handled a lot more realistically (but still not quite the same as in real life.)

Just thought I'd add in some real world experience and my own game experiences to the debate.

(Pics of my racecar can be provided upon request for the doubters.)

 

[sukerkin]

Aw ... just when I thought that everything was probably going to be okay DrBiggly comes along and raises the eerie spectre of doubt once more .

 

[rk]

Aw ... just when I thought that everything was probably going to be okay DrBiggly comes along and raises the eerie spectre of doubt once more .

Wait, there is yet hope, and no offense DrBiggly, thanks for your excellent comparitive analysis- do I HAVE to doubt you to see your racecar, btw? This is an important consideration when comparing data or measuring systems. All values must be of the same unit measurement to derive any useful meaning from comparing data. It is concievable that real world tuners can use thier hard learned numbers for ride height and spring rates, two very important factor in suspension dynamics. But is anyone outside of the Polyphony crowd able to define a shock compression value of 5? And is that 5 half of 10 or, like on the Richter scale, could it be half of 6? And are these values absolute- 1 has a specific number in any application, on any car, or are they relative- 1 is the lowest of 10 values you could use specific to this particular application (car). Then there are the sway bars; a value of 3 would be, um 3 inches? Perhaps 3 mm? Or it is measured in strength, so 3 would be 3 times stronger than 1...or maybe it is 3 arm wrestles stronger than 1...3 horsepower?
You can, of course, see how these "nuances" might affect MrBigglys game WRX. For every unnasigned value he must take his real world number and convert it into every possible unit of measurement and try each one and each combination of "re"assigned numbers while he is at it. In other words, if a number applies to a torque value, he must convert it to lbs/inch, lbs/ft, newtons, joules/kg, etc. Until he has done this he can learn nothing aside from the fact that the programmers use a different (and perhaps alien) measuring scale.
Personally, I prefer the empircal school of tuning, real world theory is an good basis to understand what the programmers were trying to accomplish, but there is no way they could have factored every variable into their equations. In the final distillation, when you use real world values, you are comparing mathematics to mechanics and you mustn't get lost in translation.

 

[sukerkin]

An excellent series of points, Aarque and some that as both an avid GT fan and a software engineer I should have arrived at and posted myself .

My excuse is that I'm at work and rather harried today - I'm shipping a control system and had the customers in for final sign-off testing and approval {and unlike PD I'm not allowed any mistakes or omissions }.

Can't stop to type more I'm afraid, much paperwork to do ... :glum:

 

[Domejuddv10]

Quote:
Originally Posted by Bad760

Now looking at a real world example, I race 1/10th scale RC touring cars at a national level.

Now I can tell you that when I put the hardest springs on the front and softest on the rear I most definitely get oversteer turning into and mid corner as the rear of the car wobbles around with such soft spring rates, but bad understeer out of the corner as the car squats and loads up the rears. I can't easily explain that as my understanding of roll and pitch moments and moving roll centres and roll axis isn't 100%.

I’ve been racing 1/10th Touring cars for many years and I agree with the above, you can also take into account the diff settings, rollbars.....which effects everything else. What I found is not to adjust everything in one go and not by huge amounts.
3mm adjustment in ride height on a say Audi R8 in real world, can make a big difference in the cars characteristic in and out of corner etc.

Love the game, only problem is that I have to work and sleep!

 

[///M-Spec]

An excellent series of points, Aarque and some that as both an avid GT fan and a software engineer I should have arrived at and posted myself .

My excuse is that I'm at work and rather harried today - I'm shipping a control system and had the customers in for final sign-off testing and approval {and unlike PD I'm not allowed any mistakes or omissions }.

Can't stop to type more I'm afraid, much paperwork to do ... :glum:

I'm also very happy to see this level of participation in the thread.

Work?! phhftt! That's no excuse.


M

 

[DrBiggly]

Aw ... just when I thought that everything was probably going to be okay DrBiggly comes along and raises the eerie spectre of doubt once more .

Well I admit to only reading through pages 1,2, and this current page and firing off a post without reading the number of posts in between to arrive at this. Terrible thing to do, just to toss one's two cents onto the pile.

aarque,
Your argument makes perfect sense. However, on a real coilover the amount of bound and rebound for example (or if a single adjustment knob exists like on most Japanese coilovers it will effect both while some companies like Koni only adjust rebound on a single adjustable shock vs. double adjustable) can be measured on a shock dyno and given in real values. Often times though, they are just rather vague "clicks" on the adjustment knob itself. The effect in change of performance from one click to the next is a function of how the coilover is valved and shimmed internally. Thing is, a coilover that is valved for say an 8kg springrate typically can really only properly damp anything that is in the range of +/- 2kg from that rate.

Thus leading me to my question of how PD is dealing with this in respect to their own shock values when one changes the springrate. Is it "automagically" revalved for the purposes of game simplicity and thus the bound and rebound are then handled as properly mated to the spring and adjusting up or down on their scale is a function of simply tweaking, or a one-size-fits-most perspective where more on their sliding scale is needed to match for a higher springrate and thus they would be very gross adjustments?

Leaving all other values alone in the "tuning" section on most of the game I only wanted to comment that the springrates themselves are quite counter to real world data and even PD's statement that scrolls across the bottom of the screen. I was quite befuddled when upping the rear springrate on vehicles that wouldn't rotate in a manner I thought acceptable didn't produce better results. Only from dumb luck was I able to discover that their game physics (alien math as you suggested?) were almost basically opposite albeit in that one particular characteristic only.

One thing I noticed that was right was where I would up the front and rear springrate and longitudinal weight transfer dropped accordingly. This makes sense.

The only thing I have a problem with are the over and understeering characteristics of altering the springrates. I'll have to give it a shot by changing the damping rates to go along with the springrates when I get a little while; perhaps tonight. (Although if the weather is warmer I really need to change the oil in my vehicles, racecar included. 💡 )

If you guys have any Solo2 (also known as autocross or autox) questions I'd be quite happy to answer them as you all seem to be an intelligent bunch for the most part. I also encourage anyone who wants to go out and try their hand at a tiny bit of real racing to at least try an autox; it's quite a bit more difficult than it looks and gives you a fantastic perspective and more insight into what your vehicle is doing and why. Not to mention it's a great way to further oneself as a driver.

Here are a couple of photos since it was asked. First off was my first serious attempt at autox in a stock-class '88 Jetta. Then my first WRX, which was sold a few months later to buy a fully prepped WRX (for my particular class.) (Truth be told I liked the red car better as far as looks, but the one I have now is just faster.)

For those not bandwidth impaired (56k go get coffee) one can click here for a larger version of the blue car. Current Racecar


-Biggly

 

[rk]

The only thing I have a problem with are the over and understeering characteristics of altering the springrates. I'll have to give it a shot by changing the damping rates to go along with the springrates when I get a little while; perhaps tonight. (Although if the weather is warmer I really need to change the oil in my vehicles, racecar included. 💡 )

-Biggly

I am liking to think (hoping) that your car is riding on the dampers, at least initially, when you just soften springs. I have found that lowering dampers 2 clicks all around doesn't impair handling to much to allow me to feel comfortable plinking and plucking springs and relying on the responses, when I get them where I like them I go to work on the dampers (usually).

 

[awdrifter2]

Deleted.

 

[Warp3]

In testing the settings in GT4, I've found settings to be consistent with what Greyout has experienced.

I was just about to go post this in the NASIOC SE forum to get some of the autocross/road race guys to chime in about this topic, but I see at least one has already made it here. (Which reminds me, we need to ask Gary if he has tried to duplicate his USTCC WRX setup on this game and see how it behaves. I don't suppose the USTCC goes to Laguna Seca or Infineon, do they? )

I haven't done any thorough testing in the game yet, but I do know that my '89 Skyline GT-R which was understeering far more than it should be, suddenly changed its tune when I swapped the spring rates on it (thus ending up with softer springs in the back)...and yes this is completely contrary to everything you'd expect from established (real life) suspension theory. Sure if you get a spring TOO soft it then the weight transfer effects will overpower the outside wheel and you get it to slide, but it doesn't seem like that is what is happening here. If that was the case then leaving the front rate stock and setting the rear to either maximum OR minimum should have a similar effect (i.e. oversteer tendencies), but it doesn't seem to work that way.

The more I read this thread the more it seems like the explanation that fits how it behaves (at least in my GT4 tuning experience thus far), is that the front/rear weight transfer is correct for given spring rates (thus the dive/squat is the same), but it's almost like the formula for the left/right weight transfer is reversed as it applies to spring rates. So you still get the same pitch as you'd expect from a spring setup, but a different cornering balance than you expect.

I, too, had hoped this was just a "front and rear springs labeled backward" deal, but the aforementioned tests for front/rear pitch seem to have disputed that theory.

 

[DrBiggly]

I do feel more squat in the rear when I soften the rear springs. So I'm not sure it they are actually reversed. Maybe they reversed the code just in the physics engine? Spring break has finally begun, (well, unofficially anyways), so I will have time to test things out some more.

That's my thought exactly. The only thing wrong that I can find is the correlation between over and understeer and the front and rear springrates. Everything else is right, including longitudinal weight transfer for the springs. I.e. up the rate and the weight transfer under braking and acceleration are correct. Nothing else seems to be amiss.

 

[turbotim3]

the stiffer the suspension the harder it is to turn. i think you all don't understand the suspension settings and what they mean.

 

[Maturin]

the stiffer the suspension the harder it is to turn. i think you all don't understand the suspension settings and what they mean.

Somewhere, far away, a village is missing its idiot.

 

[DrBiggly]

Somewhere, far away, a village is missing its idiot.

Actually, he's got a point but it has nothing to do with suspension. But his statement applies only to super-stiff (i.e. *very* limited-slip) differentials. I've experienced this in real life on an STi and seen it in rally cars where one can hear the diff binding and clunking as simple (very low speed) turns are trying to be made; typically like when trying to maneuver to park and such. It doesn't inhibit turning, it just makes it more difficult. Same with a 4x4 truck and locked differentials (such as if one were to lock the transfer case for snow and then hit a spot of dry pavement.) Quite a resistance to turning.

But I can guarantee that my car (stock about or 3kg/2kg springs but now with 8 and 10kg springs) turns quite well. You're welcome to ask Warp3 above, whom I know in real life and also autocross with on a somewhat regular basis. He's seen my car turn.

Our friend above is quite misguided...perhaps flamebait or troll? I suppose by his own statement he's admitted to never playing GT4 as the cars that "turn" the best (i.e. LM cars and such) typically have the highest springrates by game default. Good job there pal! 👍

 

[rk]

Well I will tell you what I have been learning, so everyone can have fun dessicating my pat theory. At this point I take a freshly modified car, turn off the driving aids, add exactly 5kgt/mm to each pair of springs (nothing to full race cars), twiddle my brake balance where I like it ( I favor rear heavy oversteer braking) and drive it. When I return to the tuning screen, the difference in spring tensions exactly correlates to my impression of degree of oversteer or understeer. I raise one a bit or a lot (depending on how hard the ride felt) and lower the other spring until they have the same (or almost) value; and now the car needs only a fine tune to get what I consider perfect. Dial in the dampers, fine tune the lsd, if even needed, to adjust the accelerating and decelerating turning radii and this car is tight. This has been my same formula for the past 20 cars...Now I know what works for me is only what works for me; but-and this is the whole point of this post- this is a test any of you can easily perform yourselves. As far as I can tell it will completely remove any doubt as to whether the springs work as they are intended and labled IN NORMAL DRIVING ADJUSTMENT RANGES.
(bows)

 

[DrBiggly]

aarque,
While that sounds quite reasonable adding the same amount to each springrate. However, most vehicle manufacturers dial quite a bit of understeer into their vehicles by design. It's safer for the average motorist especially for most folks' panic reactions and tendencies.

Adding an equal amount of springrate boost to each end of the car changes the ratio of springrate difference between front and rear and thus brings any vehicle closer to a neutral ratio, despite what starting springrates might be. Therefore your test will always make a vehicle more neutral in the game or reality.

I don't understand what you mean by "a freshly modified car" though. Do you mean one that you have completed all available power upgrades to? I currently can't conduct any experiments as I'm running some longer races otherwise I'd try what you've suggested here.

 

[rk]

I don't understand what you mean by "a freshly modified car" though. Do you mean one that you have completed all available power upgrades to? I currently can't conduct any experiments as I'm running some longer races otherwise I'd try what you've suggested here.

I can try to clarify; as an example I would take your basic Subaru WRX and after my basic set change it would have, as you have alluded to, a mild understeer I would bring the two spring rates to within about 2 or four clicks of each other (I too am engaged in a series so I am "remembering") and as you have also stated the steering is neutral, but now has a nice throttle oversteer. Additionally my controller provides a low frequency rumble, in cadence to the rising and falling engine note on acceleration that indicates to me the wheels are following every irregularity on the pavement- it kind of feels like the tires are spitting little ribbons of pavement behind me, actually. I know that is very subjective and superlative; however, another example would be the Skyline Concept car I had been having trouble with until I developed this algorithm. With 800+hp (built) and 4wd to lay it down, the only thing lacking was the ability to apply air effects to rein it all in. Today I set everything to default and applied my formula. I found that the intense oversteer could easily be converted to that same controller throbbing traction by bringing the high rear spring down a little, then raising the front one to almost match it. Dial in the dampers, which are surprisingly close to the original 8's, click out the brake chatter -heavy front on this one, opposite my usual pattern, personalize the lsd and camber and it pulls 10 seconds on the aforementioned Sube at Deep Forest.

 

[Maturin]

add exactly 5kgt/mm to each pair of springs ... When I return to the tuning screen, the difference in spring tensions exactly correlates to my impression of degree of oversteer or understeer.

Could you explain this again? If you're adding the SAME amount to both springs, shouldn't it do exactly as it did before, but just, you know, stiffer?