[stretch]

The formula for calculating the resistance of a solid sway bar, according to Fred Puhn, is:

                           500,000 D^4
K (lbs/in) =  -------------------------------------
               (0.4244 x A^2 x B) + (0.2264 x C^3)


               B
       _________________
A|    /                 \  C
 |   /                   \

A - Length of end perpendicular to B (torque arm - inches)
B - Length of center section (inches)
C - Length of end (inches)
D - Diameter bar (inches)

This formula does not take into account the flex of the bushings used to mount the sway bar, which can be significant. I've measured the stock sway bars (roughly, not precisely) at:

Stock Front sway:
A: 5 in
B: 32 in
C: 7 in
D: 0.79 in
hence, K = 467lb/in

Stock Rear sway:
A: 7 in
B: 42 in
C: 7 in
D: 0.79 in
hence, K = 205lb/in

Now, what matters isn't the spring rate of the bar, but the spring rate at the wheels. The motion ratios front and rear in the STI are about 0.75 and 0.875, respectively. To get the wheel rate of the sway bar, we multiply by the spring rate (K from above) by the square of the motion ratio.

WHEEL RATE = K * Mr ^ 2

Thus, the final wheel rates from just your sway bars are:

Front: 267lb/in
Rear: 157lb/in

Add this to your coil spring rates to get your total wheel rate. The stock front springs are 224lb/in with a motion ratio of 0.97 (=210lb/in at the wheels), and the rear springs are 194lb/in with a motion ratio of 0.98 (=190lb/in at the wheels). A stock STI rides on its bump stops up front, which causes these spring rates to hike tremendously, but I'm going to ignore that for this conversation and assume a suspension with adequate suspension travel.

FINAL STOCK WHEEL RATES:

477lb/in front (267lb/in from sway bar, 210lb/in from coil spring)
347lb/in rear (157lb/in from sway bar, 190lb/in from coil spring)

The distribution of these total spring rates match the car's weight distribution almost exactly.

Anybody want to measure the Whiteline sway bars? If they're the same as stock, except for the diameter, you'd get spring rates (not wheel rates) of:

Whiteline Front:
24mm: 955lb/in
27mm: 1877lb/in *this is not accurate because this bar reduces to 24mm prior to the bushings

Whiteline Rear:
24mm: 419lb/in
27mm: 671lb/in *this is not accurate because this bar reduces to 24mm prior to the bushings

You can see that just minor bumps in sway diameter increase resistance quite a bit- it actually goes up by a power of four! So, the question I'm arriving at: is a 24mm sway bar too big? What about a 27mm sway bar?

Javid (of 6gun Racing) has said his car, which uses high spring rates and big Whiteline sway bars, is slower than the competition running without sway bars (but having more body roll). This actually makes a lot of theoretical sense. Quite simply, the car's spring frequency becomes too high for optimal use of the car's tires, which most racing literature suggests takes place around 2.0 to 2.5hz. (Further reading about how to determine your spring frequency based on your wheel rate.)

I think that big sway bars are useful on cars with soft springs to reduce body roll, but on cars with firm springs (400lb/in front, 350lb/in rear or higher), smaller (closer to stock) sway bar sizes should be used for optimal grip. This assumes the car has sufficient camber and enough suspension travel- in other words, no other pre-existing, more significant problems. It also looks to me like running a larger sway bar up front (especially the huge 27mm bar) doesn't make much sense (in theory) for anything other than A-stock autocrosser.

Comments appreciated.

EDIT: I spoke with an engineer who assured me my numbers (from Puhn's book) were wrong in regards to body roll. Since you have twice as much bar flex in a body roll situation (inside and outside wheel move roughly equal amounts in opposite directions), the sway bar rates I calculated should be DOUBLE what I originally said.

In other words, take the number from the original equation and double it.

My original equation only works for a one wheel bump. I thought the doubled force was divided among the two wheels (thus you wouldn't double the rate), but it isn't. That's where I was wrong. The sway bar force is equally applied to both wheels.

So, again- as extreme as the sway bars sounded earlier, double that.

[lackskill]

I wish I had the time/knowledge to be of assistance.

thanks for your continued contributions!

[import111]

I think it is all personal preference. I like a VERY stiff setup. I have 550F/450R spring rates and I have Cobb 25mm hollow swaybars front and rear (rear set to middle). With skinny or non sticky tires this might be a bad setup, but with good tires there is more than enough grip. Stiffer = less body roll = more stable = less camber change mid corner = me happy.

Just my 2 cents

[AnorexicSTI]

That's the theory I used on my 92 Civic... in fact, in that car, because it was so light (less than 1800 lbs) I actually ran NO swaybar and ran quicker... I hardly had any body roll in that car and with double wishbone suspension giving me all the camber I needed, swaybars ended up being deadweight in that car.

I've always gone with wheel rates that are almost matching the weight distribution... so in the STI, it's 58/42... I think my car is now 57/43... what spring rates would u recommend if I'm running COBB front and rear hollow bars (rear currently set in the middle setting).... assuming the weight will be 2900 lbs with driver and fluids.

Big sways are for autocrossers... for track and street, I don't think it gets better than the COBB or Hotchkiss hollow bars.

[Arrow]

Originally Posted by import111

I have 550F/450R spring rates and I have Cobb 25mm hollow swaybars front and rear (rear set to middle).

The hollow 25mm bars are ~equivalent to 22mm solid bars, so your setup conforms to Stretch's ideas about stiff springs and not-so-huge bars.

[stretch]

Well, I think what matters is that your total spring frequency come out to the 2.0 to 2.5hz range. You can get that via springs or sway bars so long as you stop there. The trade-off is this that since sway bars change your spring rate depending on whether one or two wheels hit a bump simultaneously, it's impossible to valve a strut accordingly.

And now we get into that gray area I know little about. Is it possible to properly dampen a car with big sway bars? You're either underdamped in turns or overdamped on the straights.

I'm certain this is why race cars tend to use smaller sway bars: because sways make damper valving a compromise. However, on a street car, we make that sacrifice to reduce body roll while still being able to run soft, comfortable springs for a good highway ride.

If you look at a completely stock car with total wheel rates of 477lb/in and 347lb/in (as calculated above), it already has a spring frequency of about 2.25. (Actually your bump stops increase this while your soft strut tops and sway bushings decrease this, so the actual spring frequency is quite dynamic.) So, you're already pretty high on a stock suspension. Again, that's already in the optimal range for use on racing compound tires.

I know a stock STI feels mushy and has a lot of body roll, but I think that is due to the car's low roll center (also debatably a good thing). Body roll isn't necessarily bad so long as you have enough negative camber, but for sure the STI has more body roll than most similar cars with a multilink rear suspension. Another thing to consider is that a stock car rides on its bump stops, creating horrible corner entry understeer and a jacking effect on turn-in. There are a host of problems on a stock car that get fixed by adding a large sway bar even if it's not technically ideal. Adding a large sway bar can be three steps forward even if it's one step back. In other words, your choice of sway bar will depend on what other problems you haven't yet fixed.

I think Prodrive might have the right idea with their RB320 suspension. They keep the stock front sway bar and bump up the spring rates what I assume is a mild to moderate amount. Once you've increased your stock spring rates, I'm not sure a larger sway is desirable anymore so long as you have enough camber. Prodrive uses a slightly larger rear sway bar to aid rotation, which is understandable given the car's weight distribution.

Personally, I have the 27mm front and 24mm rear bars. This was somewhat against my better judgement, but it was the popular recommendation here so I went with it. I like them but because I don't own camber plates, the reduction in body roll is paramount for me. With camber plates, I think that Whiteline 22mm or Cobb sways (23mm equiv, IIRC, since they're hollow) are probably ideal. I run 315lb/in springs front, 250lb/in springs rear. This would still bump me over the "ideal" spring frequencies, but not by much. For those with even firmer springs, I'd probably keep the stock sway bars or, if anything, install a mildly larger rear for rotation as needed.

I don't mean to say that such-and-such is best, only that it's worth trying. I haven't actually tested all this in the real world. I know opinions are like assholes- everybody has one. However, at least Javid of 6gun racing has written some anecdotal evidence- from actual data- suggesting the theory is correct. A post of his on Nasioc is actually the reason I started looking into this.

AnorexicSTI- don't forget that with your car being so much lighter than everyone else's, your car will have a much higher spring frequency even given the same spring rate. This is important. Also, depending on where you're removed most of your weight, you may actually have a higher center of gravity than your competition, which will cause more body roll given the same spring frequency. (Don't worry, lightweight cars still kick ass from having a low polar moment of inertia.)

[twiSTies]

so what you're saying is that the stock car setup is the best?

[stretch]

Originally Posted by twiSTies

so what you're saying is that the stock car setup is the best?

Well now I'd be pretty foolish to say that, wouldn't I! As I said, larger sway bars can be 3 steps forward and only one step back. I absolutely am not saying they're bad. Heck, my experience still says they're one of the best modifications for an otherwise stock car. (On the other hand, my theory says that modification should be undone once you've prepped the rest of the car.)

It's lots of other engineers and drivers that have come up with the 2.0-2.5 spring frequency rules, and I'm only showing you guys the math to see how your car fares against that norm. I think such comparisons are useful. Calculating wheel rates between cars is the only meaningful way to compare suspension configurations, but as Dennis Grant says, the "optimal" spring frequency range is the same for all cars.

I personally don't have an opinion on what your optimal spring frequency should be since I've never done that research. Your ideal spring frequency (as determined by engineers) has to do with, I believe, the construction and resonant frequencies of typical racing compound tires. I just regurgitate their results, so don't shoot the messenger!

But with some work, I bet the stock spring rates would work well. I mean, mathematically, it should. Off the top of my head, it'd need much longer-travel struts, stiffer bushings, 3-4 degrees of negative camber, slightly larger rear sway for rotation, etc... I bet the car would be fast. Of course, that's a long list of prerequisite modifications.

[Splash]

I'm glad you put "in theory" in there...

I use relatively soft springs (300/280), but I also use WL bars front and rear - set to 29mm and 26mm (their nomenclature, not mine) for use in STU.

In ANY application that requires the use of the stock rear diff, you WILL find a preference towards anything that keeps the inside rear wheel down.

Whether that is stiff springs with medium bar, or any softer springs with big bars, That what you'd see. Even in my case, using street tires.

I don't want to use any more than -3.1deg camber up front and, even with 300lb springs and the 29mm setting on the 27mm bar, it's not enough if the dampers don't droop enough to leave the inside rear down.

The adj 24mm rear bar is set to 26mm (full stiff) with the 280lb rear springs, and I think that is about all it will take before the bar starts helping keep the inside rear up.

I have one of the old Strano bars now, and was gonna fab different endlinks for it and try it with the WL 27mm rear bar and see what that does, Or, I was gonna get something that got me closer to a 1.9 to 2.2Hz frequency spring-wise and bar it until it does what I want, fully expecting the bars to be somewhat less stiff, particularly the rear one.

Now, let me replace the rear and center diffs, and I will hang that rear wheel up all day in the name of retaining grip, but just by a little. I still gotta work within the limits of roll-induced camber loss vs static camber.

For me, the rear bar is the primary tuning tool, everything else satisfies some other aspect of the car's geometry or physics.

Heh, I was reviewing your numbers, and though gee, what would happen if you took the stock spring rates, then the Strano 32mm hollow bar numbers? Most of what I've seen say the Stranobar is 15% stiffer than the WL set on 29mm. Might break 2000lb/in a corner? Maybe then factor in the bumpstops... Hee Hee...

[CGMDan]

All this is also assuming the bars are all made of the same material?

[evilSTi7]

very cool. thanks for taking the time to figure all of this out.

[stretch]

Just to throw a wrench into things, here's a choice I can't explain. TriPoint Motorsports reportedly uses 250lb/in springs in the rear of their Mazda6 SCCA Speed World Series race car. The motion ratio back there is 0.5 for a wheel rate of roughly 63lb/in. So, imagine running 65lb/in springs in the back of your STI (1/3rd stock rates), and that's what TriPoint runs in their Mazda6 race car. I don't have a clue what their front rates are.

But then they use this for a sway bar:


Remember this is the rear of a ~2700lb front-wheel-drive car, but it does show that some people use soft spring, big sway setups for racing- and quite successfully, I might add. I'd love to know how the hell they valved their shocks for this. TriPoint is one of only two authorized Koni rebuilders, so they know their stuff. But I sure don't understand.

I do think the norm for a race car is to run small sway bars and to use them as trim devices, meaning they're only large enough to dial in over or understeer as needed. The above config may only work since FWD cars need whatever rotation they can get (certainly I'm not recommending anyone do that in an STI), but I thought it was kind of funny to post anyway.

[Splash]

That's easy to do on that car.. Make the rear almost immobile, then stiffen the front so that the inside rear comes off the ground just a bit. Since the rear is just along for the ride, they can get away with it.

I'd bet the front bar is medium sized with monster springs though....

[ACHAHockey]

Puhn's book is amazing. I read it on the way to a PCA race when I was 12, and have a copy still. Thanks for posting all this info too man. It's going to help me when I get the Subaru back and auto-xing it. The Porsche has no rear sway, and the front sway is on the stiffest setting. So needless to say going around large sweepers, at about the apex, the car plows. And going through slaloms the front tire of the direction I'm turning will lift about 4 inches off the ground. Helpful info right here, should be stickied.

[4banger]

Originally Posted by stretch

This formula does not take into account the flex of the bushings used to mount the sway bar, which can be significant.

Stretch,
I read this and I remembered seeing an article on the Whiteline webpage talking about some testing they did on the WRX rear sway bar. Due to all of the crazy angles, its rate turns out to be much less than than a 'normally' shaped bar. While looking for this, I found the answer to your question quoted above. I don't think this accounts for the bushings that mount in the middle of the car, just the endlinks. For those interested, here is the link: http://www.whiteline.com.au/articles..._Link_comp.pdf

Here is another interesting Whiteline article, specific to the WRX, which talks about stifness of endlinks: http://www.whiteline.com.au/articles/WL_Alloy_link.pdf

Edited for stupidity