IW STi Forum banner

1 - 20 of 26 Posts

2,392 Posts
Discussion Starter #1 (Edited)
A brief history:

I bought a used 2005 STI with low miles from an old guy. I'm not sure what this guy's deal was, but he owned his own business and CarFax showed he registered it as a business car. It was Aspen white, but with pinstripes and door ding protectors- how dorky. He claims to have driven it regularly for a few months before switching back to his BMW 7-series as his preferred highway cruiser.

The previous owner took the car to a tuner shop, and they put an intake and catless turbo-back exhaust on it with no engine management. It had horrible boost creep and was obnoxiously loud. The car had Nexus gauges ($$$) that were improperly wired and reading 22-24lbs of boost. All four tires were at a low 27psi. Since it was a great deal and I saw no signs of engine damage, I agreed to buy it anyway. It was a gamble.

The car was two hours away, so I bought a one-way airline ticket to pick up the car. A few days before my flight, the previous owner called and asked where I was. He had been waiting for me all day for me at the airport on the wrong day, was pissed, and hung up on me after a few choice words. Then, he checked his e-mail, found that I did give him the correct date and that it was his secretary's mistake, and called to apologize. Things definitely weren't going well with this guy.

A few days later, I picked up the car. In hindsight, I can't believe that I even went through with the deal, and I remember agonizing over it until my flight. But, it gets better- when I picked up the car, there was a hand gun in the center console. I was checking the car over just prior to saying goodbye, and, "Whoaaaaa!" I thought this rotten, suspicious deal was going to end with me being robbed.

It didn't. I drove home safely and began reversing all the crap this guy had done to the car.

I've since put the stock intake and exhaust back on, and switch to a quieter Cobb downpipe with a proper Cobb stage 2 tune. After an oil change and the proper tuning, the idle of the car became extremely smooth. It burns almost no oil. I gambled, and I'm glad this car has so far turned out to be a gem.

My previous car was a 4cyl Mazda6 with no options- deliberate to keep weight down. It was a great car. I put an aftermarket suspension on it and raced it in DSP autocross. I tuned the car to oversteer as much as possible. It was great- the car would oversteer in a turn at anything less than full throttle. It proved to be very fast and taught (forced) me to brake early and late apex. I never once lost to another front-drive vehicle, although the BMW's in my class usually won by a large margin. I'm done with front wheel drive.

Now, I plan to race the STI.

I've read several books on suspension tuning and a few books on turbo engine tuning. I enjoy suspension stuff more since I've always been one to enjoy tight roads. My favorite books include:


2,392 Posts

2,392 Posts
Discussion Starter #3 (Edited)
Goal: create an STI that handles well enough to be competitive in regional STU autocross while being perfectly comfortable as a daily driver (and to do it as inexpensively as possible)

Current engine modifications:
Cobb downpipe (stock catback)
Cobb Accessport (version 1)
Cobb Street Tuner (...untouched)

Current suspension modifications:
Turn-In Concepts single-adjustable coilovers (6k/5k rates)
Group-N strut tops
GT Spec anti-lift kit
"Free caster mod"
Racecomp 25mm front & rear sway bars
CarLab X-Brace
2007 STI OE fender braces

Other products I've tried:
Cobb 25mm hollow front sway bar
Whiteline 24mm rear sway bar
Kartboy rear endlinks
Prodrive bump stops (stock front struts)
Whiteline camber plates
eBay Top Mount Intercooler
Solid aluminum fender braces
Ground Control coilover sleeves (stock struts)
Ground Control custom built Koni coilovers (front)
Koni inserts into WRX strut housings (rear)
Strano 31mm front sway bar
Whiteline 27mm front sway bar
AEM cold air intake
HKS catback exhaust

2,392 Posts
Discussion Starter #5 (Edited)
Parts I recommend:

I get asked for recommendations quite a bit. Here they are. If the part isn't listed here, I don't recommend it or I'm ignorant to it.

Springs / Coilovers / Struts:
I do not recommend any lowering springs for the stock struts because the stock struts have no bump travel to spare.

I do not recommend any low-end coilovers because the quality control (and often design) on them is junk. Some are better than others, but none are good. There are very few suspension companies on par with, let alone better than, KYB, who makes the stock struts.

The stock struts are good. They match the stock springs decently. However, they run out of lubrication with use, and the stock bump stops are a bit firm initially. I recommend repacking the struts with synthetic grease, which is extremely easy to do, and replacing the front bump stops with Prodrive/RCE parts. If you do not do this, the stock struts will deteriorate until they make noise and/or ride uncomfortably.

The cheapest way to lower the car a small amount, plus get a modest boost in performance, is to buy RCE's "lowering" camber plates. This will lower your car 3/8" up front only, to even out the fender gap, give you a more aggressive alignment, and get rid of the very soft rubber front strut mounts. They lower the car by moving the mounting point of the strut upwards, not by changing the resting point of the strut so that it is further compressed, as lowering springs do. Thus, suspension travel is identical to stock- there is no loss.

Ground Control:

A much better option is to buy Ground Control's camber plates along with Ground Control's coilover sleeves. These only work with 2.5" diameter coilover springs, but the smaller diameter spring means the camber plate can be designed much differently. These lower the car a full inch without sacrificing suspension travel, plus allow for more caster and camber adjustment than the RCE plates. For the stock struts, run a 250lb, 8" long Hypercoil front spring. Since you gain an inch of bump travel with camber plates, and you will now have an adjustable-height coilover sleeve, I recommend you lower the car only 3/4". This will net you 1/4" more bump travel than stock even though the car has been lowered 3/4".

I do not recommend Ground Control's coilover sleeves for the rear without running camber plates in the rear, otherwise you will have a coil bind problem when the car is loaded. You need a 10" spring to have sufficient travel at 200lb/in rates, and with Ground Control's mount for the stock strut tops, that'll raise the car about 1.5 inches. Instead, you must run 8" springs, which have 5 inches of travel when you need 6 inches or more. So, to save money, keep the stock rear springs and mounts.

Helper springs are recommended, but not required, for spring rates over stock (~200lb/in). I again recommend Hypercoil and their "zero rate" parts, others are much too long at full compression. The stock struts can safely handle roughly 300lb/in front springs and 275lb/in rear springs, but I recommend staying below this threshold because the car gets a little floaty there.

Because Ground Control's kit for the rear is a little more complicated, due to spring travel issues, an ideal setup would be Ground Control's front "coilover" kit with 250-275lb/in front springs and traditional "lowering spring" rear springs, such as those from Swift or RCE. I do not believe vendors sell front and rear springs separately, however, so doing so combining parts would be expensive! That puts you into coilover price territory.

Speaking of, when your struts wear out, you can easily replace them with Ground Control's custom Koni struts. The single-adjustable Ground Control struts are not decent, but not great. I find their modifications to the default Koni valving to be a step backwards, although Ground Control's double-adjustable might be better. Still, it's nothing I wouldn't (and haven't previously) put on my car, and it represents an easy upgrade path.

KW Variant 1 / RCE Tarmac 0 / Eibach Pro-Street Coilovers:
TireRack sells these under the Eibach name for cheap. They are made by KW and are, basically, cheaper KW Variant 1's. Variant 1's are not perfect, however, this is the cheapest decent way to replace the stock struts if they've gone bad. If you want performance, go to the Variant 3's, on which KW clearly spent more time doing R&D.

Why do I not recommend Ohlins struts, the only direct-replacement strut for 2005+ STI's? Because they are bound by the same suspension-travel issues of the stock struts. The Ohlins can handle stiffer lowering springs, technically, but all lowering springs are bound by very limited suspension travel and progressive rates. The struts themselves may be good, but the total package remains flawed, unless you insist on staying at the stock ride height. For that, they are the best and only way to go, unless you find a rare-but-superior Prodrive RB320 setup, which vendors like Racecomp can import.

Also worth noting is that 2004 STI's have a variety of decent strut options, such as Tokico D-Specs and Koni "Yellow" inserts in WRX strut housings. Both are good for the price.

Turn-In Concepts SST:
These are outstanding- the best I've felt, in fact. Their drawback is price, since TiC insists on using only the best Swift springs and some other custom parts to improve upon a somewhat flawed AST base design. TiC's coilovers differ from traditional AST dampers by offering drastically more suspension travel and different valving derived from AST's high-end 3-way adjustable racing coilovers.

I recommend them with Vorshlag rear camber plates to optimize the fitment of the rear spring and allow more tire clearance. Front camber plates are not required because the strut mounts are slotted for a ton of camber on the factory bolt, however the stock strut mounts are soft and not optimal.

These coilovers have, for some reason, been marketed towards daily drivers. That's a total misnomer. They're comfortable, yes, but that's because they're versatile, not because they weren't designed for racing use.

The valving is brilliant, especially for higher spring rates. Turn-in Concepts sells these coilovers near with springs at the very lower limits of what works well with the designed valving. They work even better with firmer springs, if your application demands firmer springs. I wouldn't recommend firmer springs for a daily driver, only a track car on R-compound tires. On R-compound tires, 9k front, 7k rear seems to be a good choice.

KW Variant 3's / RCE Tarmac 2's:
These appear to be great too, although I have substantially less experience with them. These are the only other coilovers that I have been able to verify have good valving.

For a daily driver, KW's OE-sized rear spring is superior to RCE's design as it uses the stock mount and allows for plenty of travel. The downside is that this rear spring is slightly progressive, a necessity by design (like all lowering springs), and thus can lead to a minor stability issues while cornering. This drawback is largely negated by a large sway bar.

Like all other coilovers with narrow-diameter rear springs, they work best with camber plates in the rear. The only difference between KW and RCE's coilovers, besides the rear spring (and color), is a minor adjustment to the rebound valving for use with firmer springs.

These coilovers have a reputation for being track coilovers, but that too is a misnomer. They also work well on the street because the valving is good. I see more similarities in KW and TiC's design goals (not their construction, but in damper length and how the valving is tuned) than I see differences.

KW's product allows bump and rebound damping to be adjusted separately. However, I find adjustability to be unnecessary for a product with proper valving, so the extra knobs only create extra work in finding that setting. Racecomp does tell you exactly what settings to run in their manual, thankfully, although I'd recommend sticking closer to their "street" setting to keep rebound forces a little lower. I believe too much rebound resistance in the rear is what causes the wheel lift at an apex, which some folks have encountered. (It is NOT a droop travel issue.)

Zzyzx Coilovers EM Sport:
These coilovers are full of great ideas. Unfortunately, I'm not sure if they're still available.

On paper, the valving of the double-adjustable Koni model looks much better than the single-adjustable. Zzyzx seemed to deal almost exclusively with track/race cars, but I believe these coilovers would work very well with soft spring rates for daily drivers.​

Sway bars:
To make a stock STI a whole lot more fun, add a 22mm rear sway bar. I recommend Cusco for one reason, and that is it uses rubber bushings. Every other manufacturer uses urethane, which will require periodic lubrication (every few years, if you're lucky, twice per year if you're not).

Sway bars are largely personal preference. They are one of the most profound changes that can be made to any car because they, not your main springs, are what primarily reduce body roll. Adding a stiffer rear bar will get the car to rotate more, but adding a stiffer front bar is what really cuts down on body roll, making steering inputs feel more immediate.

I recommend Cusco's 22mm front and rear bars for daily drivers, with the rear on stiff. I recommend Whiteline's 24mm bars, with the rear on full stiff, for folks wanting to get a bit more aggressive. I recommend Racecomp's 25mm bars with the rear on soft for those wanting a great street/track option, and Whiteline's 27mm bars for dedicated track cars. The more you lower your car, the stiffer the bars you want, but be aware they substantially cut down on suspension independence and will reduce grip, particularly braking grip, on bumpy roads.

If your roads are not smooth, do not run 27mm bars anywhere, and strongly consider sticking to 24mm or lower. Focus on suspension geometry improvements to compensate for your inevitable body roll because overall, that's your best route.

I have found nothing but understeer when running a larger front bar than rear, even with lots of front camber. On softly sprung cars, particularly those without a lot of negative camber, running a larger rear bar results in a very tossible car with neutral handling.

On R-compound tires, rear wheel lift may be an issue with equal-sized rear bars. Consider lowering the car more as a first response to this, if possible to do safely, and make sure your rear dampers aren't too stiff for rebound. If nothing else works, because your car has lots of grip and is transferring a lot of weight, you may need to downsize the rear sway bar. Your goal should be maximum weight transfer of the rear tires (rear tire touching the ground, but no weight on it) in a steady-state turn, meaning it will lift only on corner entry and stay planted the moment you hit the gas pedal.​

I prefer Hotchkis endlinks for the rear. They rust, badly, but only on the exterior, not on the bearing. They are the only sealed, spherical endlink for the STI I know of, and that makes them the only great choice. Kartboy endlinks will never fail, but require periodic lubrication, and their bushing design deflects a bit.​

There are substantial improvements to be made here and are not to be underestimated. Search this site for explanations as I will just give a brief overview of parts I recommend.

Camber plates:
These are covered a bit in the coilover talk as they are quite integral to some designs. For the stock struts with stock springs, I recommend RCE "lowering" camber plates. For lowered cars on stock struts, I recommend Ground Control's package. For all coilovers, I recommend Vorshalg plates, front and rear. They use a second bearing to allow rotation of the spring, which you'll appreciate as reduced noise from your springs. The GTWorx camber plates are a decent, cheaper alternative without that second bearing.​

The "free caster mod":
The rod that holds the front control arm to the rear bushing can be reversed 180 degrees, and since it's off-center, it changes the size of the control arm. Result: your wheels move forward, giving you more caster and an improved weight distribution. This results in heavier steering and more front grip.​

An Anti-lift kit:
Get GT-Spec or Whiteline and laugh at those who buy Perrin's kit, as Perrin clearly has no idea what their own product does. GT-Spec uses a rubber bushing while Whiteline uses Urethane. Rubber is a better NVH isolator, but Whiteline does offer a reasonably soft urethane version. It's still not as soft as the GT-Spec rubber. Primary effect: reduction of terminal understeer on corner entry / braking.​

Whiteline Roll Center Adjuster:
Also called a ball joint extender, because that's how it adjusts the roll center. It changes the angle of the front control arm at your ride height by a very small amount. Consider this a free pass for lowering your car 10mm without the geometry-related drawbacks. No, that's not much. There are alternate kits which raise the front roll center much more- too much, in my opinion, unless your car is lowered drastically (over two inches) and you are also running a custom rear subframe to raise the rear roll center. In other words, stick to Whiteline unless you have a very custom, purpose-built race car.​

There are a few bushings in the car that appear to be worth replacing, and I do not wish to cover them all. The most important, in my opinion, are removing the stock front strut tops (replace with camber plates) and those in the rear lateral links. The rear lateral link bushings are quite soft, and their deflection results directly in delays to tire input.

Additionally, the rear sway bar connects through the rear lateral link, so deflection of the bushing results in a delay in your sway bar engagement. Your front sway bar does not have this delay, so this creates an imbalance in how quickly your tires load up. This results in a strange dynamic since your proportional roll stiffness, front to rear, changes depending on how much body roll the car has at that very moment.

However, I do wish to mention that the soft factory bushings are there to prevent road harshness from entering the cabin as noise. The STI has very little sound insulation, especially at the rear of the car. It is thus prone to increased noise with stiffer bushings in the rear, especially things like rear camber plates. This can be minimized by quiet tires, since tires are the very first layer of insulation of road noise.​

90% or more of bracing on the market is cosmetic, or the designer saw an exposed bolt and chose to brace that point because it was easy.

A brace is generally only useful if it rigidly triangulates one or more highly-stressed points. So few products do this that I can only think of three: a replacement front subframe, front fender braces, or rear x-brace. Of those, I've tried two fender braces and the rear x-brace, and I've only noticed obvious results with one.

CarLab X-Brace:
is the only brace I've ever felt did anything, in any way, to any car I've worked on. It's effect to handling is minimal, and the improvements to grip are nowhere near the 0.1g claimed by the manufacturer. This brace becomes increasingly important as spring rates go up, however, including those running stiff rear sway bars. This brace also reduces NVH particularly if you have rear window rattles.​

I won't say anything specific here, just that tires are the most important part of the suspension. They are the first thing you consider when choosing your suspension. Do not run stiff springs/bars on soft tires. Do not run soft springs/bars on stiff tires. You pick tires based on what you want to do, then pick a suspension that works well with your tires. There's a deep science to this, but as the end user, you're left out of it. Just remember not to go too stiff on soft tires nor too soft on stiff tires.​

2,392 Posts
Discussion Starter #9 (Edited)
On Coilover Development and Tuning:

This is what I've read, and honestly they're quite simple rules:
- Run a higher natural frequency in the rear
- Use twice as much rebound resistance as bump resistance
- Aim for roughly 65% critically damped (average between bump and rebound) below 3in/second, less thereafter

I'm coming up with slightly different ideas although they're in the same ballpark. I haven't tested them all, yet:
- Run a higher natural frequency in the rear
- 50-100% critically damped in the very low speed is OK so long as there is strong digression
- It is OK for digression to happen as late as 6in/sec, does not have to be at 3in/sec
- Use only slightly higher bump valving than rebound up front
- Use a lower damping ratio in the rear than the front for bump valving (thus requiring a higher damping ratio for rebound) to compensate for a higher rear spring frequency
- Run as little high-speed bump as possible, not much high-speed rebound either

Some things I've learned:
- Firmer springs make your average ride quality worse since the car will respond to bumps more abruptly
- Firmer dampers make your peak ride quality worse since it'll resist the suspension motion required to absorb large bumps

2,392 Posts
Discussion Starter #10 (Edited)
On choosing spring rates:

Manufacturers rate their springs in kg/mm or lb/in, but neither is how you should be making your spring rate decision. (For reference, though, one kg/mm is equal to 56lbs/in.)

No, instead you should be looking at the natural frequency of the spring. This magical number takes into account the suspension geometry, weight, and weight distribution of the car- which just so happens to mean that natural frequencies can be fairly compared between cars. Want your car to ride like a Cadillac? Copy its spring frequencies, not its spring rates.

The STI's stock spring frequencies are roughly 1.6hz front, 1.8hz rear. This is actually pretty aggressive. For comparison, I've read the following:
1990 Mazda Miata ~ 1.15hz front, 1.01hz rear
Mitsubishi Evo VIII ~ 1.3hz front, 1.2hz rear
Honda S2000 ~ 1.3hz front, 1.4hz rear
Lotus Elise ~ 1.8hz front and rear
C4 Corvette ~ 2.0hz front and rear

Most experts and textbooks would suggest use around 1.6hz front, 1.8hz rear for a sports car and 2.0hz front, 2.25hz rear for a race car. Of course, there's some wiggle room in those numbers, and we see coilovers for the STI extending all the way to roughly 2.3hz front, 2.5hz rear. The important thing here is that you do not go too stiff for the road you are driving on or you will lose grip. Remember that on a perfectly flat surface, you wouldn't need a suspension- the suspension is there for the bumps, so optimize for them accordingly.

It's interesting to me that most of the cars listed above use softer or equal spring rates in the rear because I believe that is the wrong thing to do. Subaru's stock spring rates are great by comparison. Let me show you why.

Let's say our front spring frequency is 1.8hz. If we make our rear frequency 1.8hz too, our chassis motion will look somewhat like what you see in the following graph:

The two lines represent chassis movement at the front and rear axle. Any time they are different, the chassis is pitching fore and aft, creating an unpleasant rocking sensation. (And the speed at which this happens determines ride harshness.)

You can see that equal spring frequencies and (with dampers matched identically to the springs) will produce two bumps that feel of roughly equal harshness, but the car will see substantial pitching due the front and rear of the car not settling into a rhythm with one another. We can fix this by running a higher spring frequency in the rear, which will cause the rear of the car to oscillate faster and catch up to the front oscillation:

Notice how the front and rear of the spring settle together, very nice! However, now the front and rear bumps are of different magnitudes due to the firmer rear springs. While there is less chassis motion in this simulator, the change in direction is more abrupt and that's what determines ride harshness: rate of acceleration. We can soften the dampers to allow more of the bump to be absorbed by the spring, and in doing so we get this:

In this graph I've reduced the rear damping to 36% critically damped, down from 50% critically damped from the previous graph. Notice the bump is now of equal magnitude front and rear, except we clearly need more damping for the rear! This is what being severely underdamped looks like- lots of body motion. We don't want this. So, what we want is that same low bump resistance, but higher rebound resistance. This graph was made in Photoshop, but the end result is intended to look somewhat like this (the particular program I was using did not support variable damping rates):

What you have there is an ideal spring and damper combination for stability. Stability leads to increased grip from the tires, too. You can achieve this reduced pitching regardless of how soft or firm your springs are (to an extent). Pretty neat, eh? This is why I recommend firmer springs in the rear with less bump resistance relative to the front.

2,392 Posts
Discussion Starter #11 (Edited)
I've been working hard on simulating suspensions using Microsoft Excel. I'm quite proud of how far I've gotten doing this. Here are some pictures from my virtual 4-post shaker which I'll make available publicly some day.

Before I go any further, a HUGE thanks goes out to Wes (4banger on these forums) and Turn-In Concepts for helping get and analyze the data that has led to this program.

Each one of these screenshots below will show you three graphs. The first is chassis movement at each corner of the car. This shows you how much overall chassis movement you are getting. The bottom graph is the height of the asphalt the car is riding over. You'll notice that while the simulated bumps are very abrupt, the chassis motions are much smoother. While the least amount of body motion might at first seem desirable, what we really want is the height of the lines in that graph to change directions as slowly as possible. This is the acceleration of the chassis, which is also shown in the middle graph.

Consider this: a one inch change in asphalt will always result in a one inch change in chassis height- eventually. What matters is how long it takes for the chassis to make that change and settle. The quicker the chassis height changes, the more uncomfortable the ride will feel. However, a chassis that settles with little excess motion will feel more composed. What you want, then, is low chassis acceleration that settles again with little excess movement. This also results in high grip since this same formula keeps the most consistent load on your tires.

That middle graph is plotted as G's. 1 "G" is simply the force of gravity. 2 G's is twice the force of gravity, meaning that if you are experiencing 2 G's, your body will feel twice as heavy as it normally does. This creates discomfort. I've also provided RMS (root-means-squared) chassis acceleration numbers, too. This number is sort of an average deviation from 1G. This is an industry standard way of mathematically measuring ride quality. Lower means less harshness.

With those things explained (hopefully in an easy-to-understand method), here are some graphs I've produced.

The difference in ride quality between 7k and 5k springs (the simulated damper had changes primarily in low-speed valving only).

Here are some valving revisions I've been working on for my own front dampers:

Based on my simulated results and LOTS of reading, I've come up with this experiment in damper valving. Turn-in Concepts has agreed to try it out- we're both going to have a set of front dampers valved this way.

Compared to Ohlins Sportlines, the TiC dampers look much better. (But this is no surprise, the current valving is too.)

Note: in that last graph, the TiC coilovers look much softer than the Ohlins- HOWEVER, what you do NOT see is that they ALSO have more resistance in the critical "car control" region (measured at 50mm/second damper speed). Thus, you're gaining comfort but- in theory- also gaining performance. My virtual shaker rig isn't yet a 7-post simulator though, just a 4-post, so I can't simulate that just yet.

1,198 Posts
Stretch, the graphs are X'd dude!!!! :(

946 Posts
Stretch, your reviews and write ups are immaculately put together and quite informative. Thank you so much for sharing your knowledge on the forums.

Ever think about putting a book together?

Keep up the awesome work, sir!

2,392 Posts
Discussion Starter #17
I'll see if I can dig up the graphs and make the images work. If I accidentally deleted them, I can redo the work.

MikeO, something is in the works. :)


1 Posts
Stretch, Great info. I just wished all the links to the photos and graphs were not broken.

I look forward to reading more of your notes.
1 - 20 of 26 Posts