posted on March 23, 2010 22:36
Project STI- Generation 3, an Introduction
By Mike Kojima, photos by Jeff Naeyaert
Introduced in 2008, the latest Subaru STI is the third generation of its model type. Designated GR, the new STI is an all new platform. The new GR chassis has a 3.3 inch longer wheelbase but a 2.8 inch shorter overall length for less overhangs and a lower polar moment of inertia. The new chassis is 2.2 inches wider and 2.2 inches higher and has gained about 70 lbs for a curb weight of around 3,375 lbs.
Subaru states that the new chassis is stiffer in bending and torsion but we were hard pressed to find data on exactly how much. Considering that the GD was known to be an exceptionally stout chassis, that data would be interesting to review. The big difference between the new GR chassis and the second generation GD chassis is the suspension.
|Look how much more simple the GD's rear suspension is. OK, its not an STI but this is what we could come up with.
Here you can see the GR's rear suspension from the side. The trailing arms job is mainly to absorb braking and engine torque loads. Its bushings are designed to resist fore and aft motion and allow side to side movement so the short toe link can toe the suspension in. You can see the upper links position here as well. Note the bosses for a disk brake caliper in this drum brake equipped car.
The GR’s suspension is a multilink vs the old McPherson struts. Due to their potential ability to keep the tire's tread flatter on the ground for more grip under roll when cornering, multilinks are usually coveted as performance increasers over more mundane and common sedan suspension systems like the economical and pedestrian McPherson strut. Frankly, we are puzzled why Subaru has opted for a multilink in the back of the STI. Usually nose-heavy, understeer prone AWD cars are hurting for front grip. Putting a multilink in the back seems a little counterproductive for performance as it would tend to increase dork-safe but racer-annoying understeer. A multilink also has more parts and is inherently much more complex and expensive to produce than a McPherson Strut system, hence Honda doing away with it on the front of the new Civic. Engineering additional cost into a mid-priced sedan is a very unusual move in this day of brutal competition and shrinking margins in this competitive market.
Our take is that the main reason why Subaru has opted for the multilink is to offer a car with a flatter trunk or rear hatch area for more utility. With no intrusion of the McPherson Struts shock towers, the trunk or in the case of the soon to be released WRX, rear hatch area can be one vast plain of flatness, better to store your luggage, golf clubs or bodies in. Doing away with shock towers also means that rear seat room can be improved and there are perhaps some safety reasons as it is possible to design more crush from side and rear impacts with the multilink. Perhaps the reason can be as simple as Subaru is trying to emulate its arch rival, the Mitsubishi EVO which also has Mac struts in the front and a multilink in the rear.
Multilinks are cool but if Subaru was only intending to use it on one end of the car, we would prefer it in the front instead, ala the old Infiniti G20 or Audi A4. We would take less understeer and more front grip over a flat trunk floor any day.
|The present Impreza’s rear suspension is a good ole McPherson strut. The toe link is nearly the same length as the lower link which means that toe in under roll is minimal. As you can see, this is a very simple and straightforward rear suspension.
Yes, it is a real multilink. A lower link, a trailing link and a toe link all in a pretty close to the same plane line act like a virtual lower control arm. The shorter upper link swings in a tighter arc than the lower link, assuring a gain in negative camber under roll. The short toe link gives passive toe in under roll.
The Impreza’s multilink has its geometry arranged so that the suspension will gain negative camber under roll at a greater rate than the old cars McPherson struts. This keeps the tires tread flatter on the ground rather than allowing it to tilt up onto its outside edge. With its footprint maintained, the tires grip is improved under cornering over the old struts. Some multilinks like the one found in the back of the Scion TC do not have much negative camber gain but this suspension has a good amount. With shorter front lower links the suspension will also passively toe in the rear wheels under roll. Both of these measures will tend to increase understeer and provide stability when cornering.
|A simple and straightforward trapezoidal parallel link with trailing arm with a McPherson strut, the exact same stuff you will find in the rear of an old Nissan Sentra! Its probably fine for the rear of a front heavy AWD car.
|We wish Subaru had put a cool multilink with a fast camber curve up front where the car needs all the front grip it could get ala Audi or the Infiniti G20 but the wide boxer engine would probably make this a nightmare.
The GR’s Multilink rear suspension has more antisquat geometry than the GD’s. This will reduce the amount of annoying engine torque induced squat under acceleration. Anyone who has driven a healthy modded GD STI with street suspension knows what I am talking about. This also means that the suspension will become more bound up under torque load increasing the suspensions resistance to movement under load. More antisquat can mean that the car can have a greater tendency to oversteer on the throttle because the bind will act like an increase in suspension stiffness when the throttle is applied. Perhaps this may cancel out some of the understeer inducing effects of greater negative camber gain and passive toe in later in a turn.
Another observation is that Subaru used small rubber bushings at the pivot points. This is good because it reduces overall compliance and will help keep the tires pointed where the Subaru engineers intended under load. We have been rather dismayed at the huge squishy rubber bon bons that many late model suspension systems are seemingly sprouting lately in an attempt to reduce noise and harshness.
Thursday, March 25, 2010 6:48 AM
There's nothing fragile about the EJ257. It can be tuned fine to a reliable ~420whp by anyone who actually knows how to tune safely/properly -- which is about 1 in 50 people who say they know how to calibrate an EJ* ECU for more power.
So for every 49 jackholes with shattered ring lands from detonation posting in forums...
Thursday, March 25, 2010 8:15 AM
The Honda battlecry! With the right tuner I can do 1,000,000hp in a stock block reliably all day! :)
Thursday, March 25, 2010 8:21 AM
The EJ is fragile, the ring lands on the stock pistons break easily and block integrity becomes questionable past the mid 500's. The crank becomes iffy at much past 7500 rpm. There are oiling issues as well. Compare that to a 4G63.
Thursday, March 25, 2010 11:12 AM
Maybe the Subaru engineers fall into the bad end of the 1 in 50 people who know how to tune the EJ... or so some Subie forum members will say. :)
I'm excited to hear what you guys have in store for the car! Cool break down of the suspension changes. Curious ab the history of your particular car... it has an odd grill and the front suspension has green top plates (Tein?) in the last picture, so I'm guessing it isn't fully stock anymore.
Thursday, March 25, 2010 12:46 PM
The car has Tein stuff on it now, but were are probably going to put something else on there.
Thursday, March 25, 2010 10:40 PM
I thought that place looked familiar. Dope pictures. I know Jeff was there!
A stock EJ257 isn't fragile, but that's kind of like saying that a stock 13B-REW isn't fragile either (it is for you young guys). The moral of story is that it's FAR easier to break than other engines. What does that make it? Fragile.
Glass isn't fragile unless you drop it. And glass breaks far easier than say a piece of steel.
Friday, March 26, 2010 12:55 PM
When you say it is fragile, you should say compared to what. Compared to the iron block closed deck design of the 4g63, ANY other 4 cylinder turbo engine is fragile. LOL.
Seriously, it takes a lot of doing to blow up an engine while tuning it. Most of the engines blown by tuners is the result of carelessness on the part of the tuner. They are tuning for a living anf they want to do the job as quickly as possible and do it just enough to get your car off the dyno and get another car on the dyno. That is the MAIN problem, IMO.
Another problem is that they do not understnad the tuning strategy on the OEM tune. For example, Subaru tunes their OEM turbo cars to run close to stoich at WOT right up to 4000 rpm. It is known as the Closed Loop/Open Loop Delay. The main reason for this is emssions. Subaru wants to certify their cars as LEV. So they do this dangerous practice.
The FIRST thing that you have to do when tuning a Subie is to ZERO the CL/OL delay. If you do not, and you up the boost, then you are going to blow up the engine for sure. Imagine what will happen when you up the boost to 18-19 psi from stock and your still running stoich AFR. OUCH!!!
I have written about this many times. Here is one of them:
Another issue with these cars is intake tuning. When install an intake on a WRX with a hot wire MAF, the fuel trims tend to go negative because the car is running rich and the ECU is trying to compensate by pulling fuel. The problem happens when you go WOT with negative trims. Say your car was tuned to 11:1 AFR at WOT. The fuel trims were contained because the intake was just installed on the car. The car was tuned and all was fine. Then you drive around for a few day the crusie trim become -10%. when going WOT that -10% is applied to WOT AFR. So 11*10%=1.1+11AFR=12.1AFR. So your nice and steady 11:1 AFR has just become 12.1:1 AFR. That is why before tuning AFR the tuner MUST re-scale the MAF to prevent the trims from going negative. Otherwise, the engine is going to knock and blow.
One last issue with these cars is Subaru's timing startegy. Subaru uses dynamic timing. It has multiple timing tables that interpolate and advance the timing under certain conditions. We do not know all the conditions under which the timing gets advanced. So the best strategy is to make the BASE timing maps the same, zero others, and make the additive maps the same. That prevents the sudden onset of timing advance that might make the engine go BOOM.
I have tuned over 160 Evos and Subies and I have NOT blown one engine. If the tuner takes his time and is not in a rush to make a living it is impossible to blow up a car.
Friday, March 26, 2010 2:25 PM
GD and GR Brake rotors are the same size...
Friday, March 26, 2010 3:54 PM
Are you sure? Edmonds and other sources say 12.7 vs 13"?
Naji, our particular car blew up on the dyno bone stock, with no tune or aftermakret parts at all. I would say that is an issue with being fragile, besides the early production cars having a poor tune. Early GD'd had a habbit of blowing up bone stock as well. At SCC we had a test car and it detonated like crazy in stock tune. The ring lands on the Subaru are pretty high and cannot take much abuse.
I think with a proper tune these car are a lot less likely to blow up as long as you don't exceed what the engine can take which is under 20 psi with stock internals and pump gas.
I was at Superlap Battle and I think at least half of the Subies, maybe more blew up (Duck).
I really like the EJ's power delivery, in fact I like driving a properly modded GD alot, maybe more than an EVO but I hate working on them! Try changing cams, ugh. Helping some of my Subie friends work on cars makes me glad I own an EVO!
Friday, March 26, 2010 11:46 PM
Jeff, while 420whp is nothing to sneeze at, that's not even breaking a sweat on Evos, 4g63 or 4b11t. A number of drag Evos with 4g63 have pushed over 1000whp. The 4b11t has been pushed to over 700whp so far; granted, not much info/time on the engine at that power level.
However, Ryan Gates has about 10k of TRACK miles on his Evo X/4b11t with 500-550whp range, STOCK bottom end! The EJ25 doesn't have a chance of surviving at that level. Crawford is destorying EJ25 engine blocks left and right, and I think he's only at 700-750whp. They barely last one track weekend.
Saturday, March 27, 2010 3:50 AM
Saturday, March 27, 2010 5:05 AM
At least on all aftermarket rotors that I've seen (DBA,Powerslot, etc) the GD and GR use the same part numbers for the front. Haven't tried checking the OE part numbers though.
Sunday, March 28, 2010 5:03 PM
Hi Mike...Man do I miss talking to you.
Compared to the 4g63 and the 4b11 the EJ 2.5 liter engine tends to be weaker. The EJ 2.0 liter is a stronger engine. I read somewhere that the boxer engine and the 0.5 liter increase in displacement makes the engine less balanced and more likely to not handle high hp levels. I was able to confirm that by looking at the per cylinder timing maps on some STIs. I found that in #4 Subaru runs timing advance whiel it does not do that in 1, 2, and 3. Subrau tries to balance this engine with timing changes. Very interesting.
You are right, the early production cars have had issues. It started with the 07 STI and carried over to the 08-09 STI. Some have speculated that Subaru changed piston suppliers to cut costs and in the process weakened the pistons on the 07+ STIs. Add to that the PISS poor tuning strategy from the factory and you got many blown stock engines. I suspect the tuning culprit is Subaru's continued use of CL/OL delay for emissions reason coupled with using small orifice pills in 07+ STIs. While a small orifice pill tends to up the boost, it also makes it more likely to spike. The best stock boost pills for the VF series turbo is the 04-06 pills. This is the pill that I use when tuning WRXs that do a turbo swap from the TD04 to the VF series.
I am glad you said 20 psi, because that is the maximum boost I will take the GD/GR engine. I just tuned an 09 STI this Saturday and set the boost to hit no higher than 20 psi. I usually set the boost at 18 psi in 3rd-4th gear anticipating that it will be higher in 5th/6th gear due to the bleed type solenoid inability to react fast enough. I have seen tuners boost these engines higher than that, but I think it is the reason why other tuner blow up these cars. There are so many inrresponsible practices in the tuning industry, it is down right scary. From block tuning the timing and fuel maps, to maxing out the boost and letting the turbo efficiency limit dictate that boost, to tuning w/o a wideband, etc....
There is nothing nicer than a tuned EJ at power delivery. I still remember the first STI I tuned. It was a stock 06 STI. Stock STI boost peaks late at around 4500-5000 rpm. It is tuned that way so Subaru can run near stoich AFR and sell the cars as LEV cars. What I do is disable CL/OL and make the boost come early at 3300 rpm. That makes the car a torque monster. The low end gains on these cars is unreal. After tuning this car, I discovered that there was no need to use 1st gear. You can take off in 2nd gear very easily even on an incline. I loved daily driving the car so much, that I used to borrow it for a drive whenever I met the owner.
Working on a boxer engine is a PITA. Changing plugs is hard, getting a compression and leak down test is hard and as you mentioned a cam change is ultra hard.
Monday, March 29, 2010 12:39 PM
The comments posted here about crappy stock tunes and pain on the leak down test (to ensure the motor isnt shot) have worn a bit of the luster off looking for a used STI. Is the non-STI impreza ecu just as bad?
Perhaps its like Eric said. Just buy an EVO X and be done.
Monday, March 29, 2010 4:39 PM
I agree. The STI is one of my favorite cars, but after reading all the problems about the 2.5L engine, it doesn't make it seem as good. Kind of makes me wish Subaru used the JDM 2.0L when it brought the STi over to the US. Seems odd that Mitsu can also attain a LEV rating, have similar power output, AND have smaller displacement but not have the problems Subaru has. Diff timing on cyl 4?!? Yikes!
But surely, Mitsu has a bad rep for a reason... the Evo must have SOME reliability issues, right?
Wednesday, March 31, 2010 3:20 PM
I'm not so sure I agree with the GDB vs GRB STi front suspension comparison. For starters, the underbelly pics you included are that of base model imprezas and they are not the same as that of an STI.
The STI has a much stouter front suspension with solid aluminum front control arms unlike the hollow ones pictured in this article. There are also as well as a front crossmember support plate which is mounted to the front crossmember and control arm inner mount. I don't think Subaru engineers would take a step backward and manufacture the top of the line STi model with weaker front suspension.
Wednesday, March 31, 2010 3:43 PM
Do you have a picture of the support plate?
Thursday, April 01, 2010 7:47 PM
Jeez some of the comments in here about the STI's reliability are a bit overblown. Issues with the stock tune didn't really arise until 2007+, and even those are not as common as some people would have you think. My 2004 has 88,000 miles on it, granted I am fastidious with maintenance, but it was my DD for a long time, the car sees track, autocross, and drag duty too. Intake, header, downpipe, exhaust, custom tuned, 60whp over factory. Never let me down and runs great to this day.
Monday, April 05, 2010 4:24 PM
here's that support plate:
and why do you keep calling the GD a "GB" in the article?
Monday, April 05, 2010 6:29 PM
I have no idea why I kept calling it a GB, I see I used both GB and GD in the story, I must have spaced out. Thanks for catching that and bringing that to my attention.
Thanks for the picture. Although the STI bridge plate is a well designed piece and greatly beefs up the area of the front lower control arm mount, I don't think the front end of the car is built as stiff as the heavy, large cross section perimeter frame crossmember of the GD, especially if it has an aftermarket bridge brace.
MotoIQ Proudly Presents Our Partners: