Update 1/05/11 Dyno Results This is what you have all been waiting for! Finally the "Proven Power" side of things. But first a few questions surrounding the GTX3076R and other GTX turbos.
- The Question, "Is the GTX turbo Laggier then the GT turbo?" Answer, "NO"
- The Question, "Is the GTX turbo going to make more power?" Answer, "YES"
- The Question, "Is the GTX turbo better?" Answer, "See above answers"
- The Question, "Is the GTX worth the extra money?" Answer, "YES"
- The Question, "When can I get a PERRIN Kit with one of these?" Answer, "NOW if you ask nice"
The SetupSo my STI was used as the guinea pig for these turbos (along with a bunch of others recently) and its setup exactly like a normal PERRIN GT turbo kit car should be.Parts installed: PERRIN Headers, PERRIN GT Turbo kit, PERRIN FMIC, PERRIN Catback, and our PWI-2 Methanol Injection kit, PERRIN EBCS Pro, Deatchwerks 850cc injectors and Walbro 255lph pump. The engine is built with Cosworth Rods, Pistons and bearings, and ARP Studs. Things to take special notes on are: Stock heads, Stock Cams, OEM head gaskets, OEM TGV housings in place and functioning and stock fuel rails with PERRIN Stumble fix installed.We use 92 octane fuel here in Oregon, but like any big power car running on pump gas, we run 50/50 methanol. The kit is setup to flow about 800cc's per minute of methanol/water. Tuning duties on this beast are controlled by a Cosworth ECPro standalone ECU. An Innovate Wide Band is integrated into the ECU as well as a gauge so i can visually see AFR as well as log it accurately. That along with a Omnipower 4 bar plug and play MAP sensor, so i am able to run the car to higher boost pressures. This part is a must for any turbo kit customer! Setup: Dyno we are using is our awesome Dynapack Dyno. Car is setup to run in 4th gear, starting at 2000 RPM and ending at 6800RPM. We use a settle time of 4 seconds, which means, the dyno holds the RPMS at 2000RPM for 4 seconds before the run starts. During this 4 seconds, we have the car at 100% throttle for about 2-3 seconds. The fans we use are not super crazy fans that flow 100MPH wind, but each fan flows exactly 28MPH (measured with our wind meter) and both are aimed at (not on) the intercooler). This doesn't provide perfect airflow to the intercooler compared to to driving on the street, but it does provide very accurate and repeatable results and if anything lower HP that you may see on the road. All runs were done with the ECU reading about 30C intake temps, and this was very consistent. The dyno temp measured 58-60F. The dyno runs were done with about the same amount of time between them as well as coolant temps were always at 90C. The idea is we did it consistently to eliminate as many variables as possible. Now with all that out of the way its time for the good stuff. Garrett GT vs. GTX If you do not know, this whole test is the GT3076R with a .82ar housing compared to the GTX3076R with the .82ar housing. We will leave out the other GTX turbos for now as well as the .63ar housings. I have really good results on the GT3076r.63 comparing it to a .82 that was swapped out and dynoed in about 1 hour. Its about 300RPM quicker spooling and that should be good enough to give you an idea how the GTX with .63 will perform in the spool department. Car was previously tuned at 1.2bar, 1.5bar, and 1.7bar of boost with the GT3076R w/.82. The tune was not pushed to the limits by any means, meaning AFR target was .80-.79 Lambda (11.76-11.6AFR) running 50/50 meth. I did do runs previously at .81- .82 Lamda 11.9-12.0 and it was very happy, but added fuel for safety. Timing could have been bumped up about 2 degrees more as I found it could run that also. Either way, we re-dynoed the car with the previous tune (super safe) and did NO tuning between runs. I ran, 1.2, 1.5bar, and then 1.7 bar with just twisting the SI drive button. AFR's were perfect, engine noise low, and WGDC was right where I left it. I did do a 1.9bar run and AFR's were the same .79-.80 lambdo with no knock. I did this because i knew that some guys would be asking for a big power run. Below are the results overlaid with the GTX results. Next up was time to swap the turbo. As you can see from the NASIOC posts I made with my iPhone, we started the swap and in less than 1hour it was back up and running. I did have help doing this to make it go as fast as possible. 1.2Bar of boost... First run I was crossing my fingers, as I left the same tune in place. At 1.2 bar (17.4psi) of boost I was impressed with the response as it was almost exactly the same! I did a few hot and cold runs(Heat in the header is what I am referring to) and after comparing them to the GT3076R runs we just did, the GTX was maybe 50RPM laggier. No one will notice this! I am pretty excited as even at this lower boost levels, it gained power! 15WHP more makes it 370WHP which is very good for low boost! Now at these boost levels, the GTX compressor wheel isn't really being utilized as it will be at higher PR's. The GTX wheel, according to Garrett, is more efficient by about 6% at redline, which means cooler charge temps. Typically this means denser air, and means if not accounted for that the AFR will be leaner. I only found the AFR to be leaner by a microscopic amount. This is something only visible using some of the Cosworth toolbox tools to see the average change. The AFR was leaner by .01 Lambda(.15AFR) showing its is more efficient! The Cosworth ECU has closed loop boost control so it hits my target boost every run. This keeps it very consistent and eliminates some variables. The GTX on all runs needed about 2% more Wastegate Duty Cycle(WGDC). This shows the turbine needs a tiny bit more engery to spool the same. The Cosworth ECU also has a great visual way to see engine noise and using this to compare both turbos, its almost exactly the same. For the next few examples, I will make sure we cover WGDC, AFR, engine noise and of course power. Again, keep in mind, ZERO tuning was done from the GT to the GTX at this boost level. 1.5Bar of boost.... With the twist of the I-drive we are now running 1.5 bar (21.75psi). Just like we saw at the 1.2 bar level, we gained power! Another 15WHP over the GT at this level is pretty good, again for no tuning. This is where I feel that 75% of our turbo kit customers fall into. It's that boost range that works ok with 91 octane and makes plenty of power to get you in trouble! No one would be disappointed with 430 WHP and 430 ft-lbs of TQ! I also think that this is what you would see on pump fuel setups. During these runs the AFR was again leaner compared to the GT turbo, this time it was just under .01Lambda (.15AFR), but at redline it was just over .01 Lambda. Indicating cooler denser air and more efficient turbo as expected. On the dyno the AFR changes were not something we could see at all. This could be due to the wideband 02 sensor being located at the tailpipe versus my Innovate sensor is in the downpipe where its more accurate and faster. Engine noise this time around was slightly quieter. Here is a quick shot the Pi ToolBox software that is what the Cosworth ECU uses. By far this is the most powerful datalogging software out there. I can overlay two runs by offsetting the time (or a value in the data) and compare runs exactly. You can see the engine noise graph at the bottom and start to see the GTX run (in green) and how the engine noise is quieter. This is just one of the many veiws i have setup to tune with. 1.7bar of boost.... Now we are starting to get to the sweet spot of the turbo. This time we are seeing 15-20WHP gain over the GT turbo. Again NO tuning is done but we see the same thing where, AFR is .01Lamba leaner, engine noise this time is about the same average. Making almost 470WHP and 470ft-lbs of TQ is very impressive for what this is. While this may only represent 20% of our turbo kit customers out there, who wouldn't want to take their existing 440WHP car to 470 with a simple bolt on part. Ok maybe not simple, but its not that hard. 1.9 bar of boost. Ok so this is not normal and doesn't represent what customers normally do, but I had to push just a little. The 1.9bar (27.55psi) run on the GT3076R made about 450WHP. This is what we tell customers the cap is for this turbo. I have see more but its when customers really push to 30-32psi of boost(which tapers down at redline). But this is a lot of stress on a Suby engine! This is the only time I did tuning on the engine, and it was only fuel related. This higher boost level is where the the turbo starts to kick butt. Initial runs were showing 20WHP gain, but with only a minor tweak to the AFR curve, we saw 480WHP! While this was kind of a glory run, the engine noise was quieter (just barely), and the engine seemed very happy. Realistically customers that have any kind of wheel HP over 450, it becomes somewhat of a blur as the car starts to become scary fast. But either way, if we can now say that 480WHP is the cap for this turbo......sweet! From all this it goes to show that the GTX turbo works and if tuned for, there is at least 20WHP to be gained. Why does it work, simple its more efficient, and it will do higher PR's. The PR(pressure ratio) is not really a Subaru thing as they typically don't get pushed past 30psi very ofter, but that efficiency thing is what makes this a great option. Here is the GT turbo and GTX Turbo compressor map overlaid with air flow from a 2.5L engine. The graphs show air flow for a 2.5L engine plotted similar to the boost response found on the GT or GTX 3076R turbo. As you can see the plots really runs off the map on the GT compressor map at higher boost/airflow levels. The GTX really fits this engine better! Could I have made more power? For sure! I think the GTX could have made more power for sure had I turned up the boost a bit more and really tuned the AFR and Timing, I could have hit 500WHP. But pushing 30psi to do so is a lot of stress on a Subaru engine (big bore/short rod issue). This is where a GT or GTX3582R would come into play. It can make 500WHP with a few PSI less, and blow way past the 500WHP barrier. I choose not to as this car is driven everyday and 30psi on the street is just stupid! I know, so is 25psi, but still..... With speculation of this turbo acting like a GT3582R, why DIDN'T it make tons more power, well I think that is a simple answer. While the GTX3076R flows like a GT35R, the GT35R has a much larger exhaust wheel and there for has much less back pressure as well as more potential to flow more air. So while the GTX3076R compressor wheel flows enough for tons of power, the GT35R turbine wheel needs to be in place to push past the 500WHP mark. So the question is, "Is the additional cost worth it?" Looking at the standard retail price for a Garrett GT3076R vs a GTX3076R, the GTX is only $230-$250 more. So in this case, I say yes! People spend that on crank pulley plus installation! In regards to the GT3582R vs GTX3582R the retail price difference is $700-ish. Without any proof (yet) the additional $700 is at least worth 20WHP over the other. Again, look at TGV housings and installation cost.Someone could easily spend $700 for housings/modification and installation. So with out proof that it makes any power, I say yes! Question, "When can I get a PERRIN kit with a GTX turbo?". Answer, they are available now, so.....now! Final comments: Your welcome Garrett...... One of the features Garret proclaims is quieter operation. After my first minute driving the car, i would absolutley agree! Its about half as loud as the old one. There is no way cops will here your turbo screaming from a couple blocks away! I expect tons of questions, start asking! Email me, IM me, call me, i am here!
Back on the dyno again finally. I was pretty excited to see what kind of power this GTX3582R was going to make. I installed it with the .63AR housing as this has been becoming the more popular choice lately. My initial impressions on the street were that it's a little laggier, but still pretty snappy. Also I found that on the street it "felt" like it made more power. So when I put it on the dyno my plans were to do it a little different and start with GTX3582R w/.63 first then go to the GT3582R w/.63. Then do the same boost levels, AFR levels, timing and same 7000 redline.To recap, car has the exact same parts as before, all our normal parts, plus our Meth injection kit, Cosworth ECU, and stock TGV housings. Goal is to run 1.2bar, 1.5bar, 1.7bar, and 1.9bar all at .79-ish lambda.First run at 1.2 bar was exactly the same power, exactly the same everything except laggier. At these boost levels it was consistently 400RPM laggier. Here is a graph of some runs. This shows the 1.5bar levels of the GTX3076R and the GTX3582R. you can see that on a typical run its 500RPM slower, but when quick runs are done the split is more like 350RPM.Moving on to 1.5 bar then 1.7 bar I found the exact same thing! Only difference was i added a bit more fuel (no power change). I was really wondering when this larger more efficient compressor wheel was going to start coming into play. I felt cheated a bit so I started tuning the 1.7 bar range. I added some timing, leaned it out, and basically the changes were not worth anything, maybe 5WHP. I put the changes back and hoped that 1.9 bar would show some real gains. So one last ditch effort, I ran it at at 1.9 bar and guess what! NOTHING!! It made the exact same power as the GTX3076R in all instances and since it was laggier, it made less peak TQ. You can see here the raw data runs and how they were making the same power. At these boost levels i also did some tuning and found that timing was much more on the edge. Adding a degree or half neted in a large change to engine noise. Still no knock, but again, the 5HP wasn't really worth it to make it that much on the edge. I did see evidence that the compressor was more efficient as i was needing to add more fuel, but not tons. Here is a nother few runs, showing the GTX3076R at 1.5bar of boost and the GTX3582R at 1.5bar of boost. Its easy to see what turbo is going to be the next popular turbo from this. The question is why? I am sure there is a bit of power to be gained with the TGV's being deleted (Proving this soon as well), and potentially there is some power to be gained with bigger cams, but realistically it's the turbine housing. In the past with GT turbos, we never really found the flow limit of the turbine housings. In this case we did. Now looking at the Turbine housing flow charts we can see why we hit this limit. You can see how a 30R.82 flows about 23lbs/min under those conditions and as the turbine pressure raises it hit a wall. Basically adding more exhaust pressure will not do anything for the ultimate flow of the engine. On the 35R turbine map the .63AR housing hits a similar wall at 23lbs/min. This is why at the same boost levels as the 30R the engine wasn't flowing more air making more HP. But if we step up to the .82 there is an instant 5lbw/min of air flow gained from the turbine wheel. 5lbs/min is 50 engine HP worth of airflow so this should net close to 50 more Wheel HP. But of course at the expense of lag. In past tests .82 GT3582R would spool around 4500 RPM and with the GTX i think it will be the same. In the past we would say that when pushed the GT30R w/.63 made about, 400WHP, then .82 added 50WHP to that number, then a GT35R w/.63 added another 50 ish and then the .82 added even more. But with the case of the GTX3076R it really does fit a broader range of HP without sacrificing spool. The GTX3076R w/.82 AR now covers 2 turbos In the end your choice of turbos are going to be GTX3076R w/.63, or GTX3076r w/.82, or GTX3582r w/82. From this I can 100% conclude that the GTX3582R with a .63 housing is NOT the turbo to buy. A much better choice is the GTX3076R with .82 housing. Its cheaper and spools quicker. From this I can also conclude that a GT3582R with .63 housing is also not a good choice as the GTX3076R is better, but it does cost a bit more. The only thing left on the table is GTX3582R with .82 ar exhaust housing test. If there was one on the shelf i might have stolen it for this test, but i think its going to have to wait a week. Sorry! If people are still wondering are the GTX turbos worse in any way performance wise. I am pretty confident in saying not at all. The only issue they have is they cost more. But overall when looking at a complete turbo kit or a complete car build, its not that much overall. Onto the fun stuff. Here is the GTX3582R w/.82 housing at different boost levels. It does make more power than the GTX3076R but not tons and tons more. I think that if i had reved the engine 8K we would see the GTX3582R start to show its worth. At 30psi, the 2.5L is needing about 70lbs-min of air and going up to 8000 adds 10lbs-min more. That is where the GTX3076 falls off and the GTX3582R would really show some gains. The question is, "Is it worth going to the .82 over the .63 GTX3582r?" I would say for sure, if you are wanting more power and are going to rev the engine high. Here is a graph showing the .63 runs versus the .82 runs. Like all other runs, i did no tuning to increase power, just to bring the AFRs back in check. What i found was i needed more fuel at the higher RPM's which is expected with a more efficient turbo. What i notice on the road is the car is laggy, then BAM, power, and once on boost, its a very linear feel all the way to redline. There is no drop in power and it really makes me want to rev the car higher and higher, but i am not.... My personal choice after all is said and done... GTX3076R with .82ar. Check out the difference. And what you can't see on the graphs is the off boost power. Personally if i was limiting my redline to 7000 RPM, then i would loose 30WHP to get the better response and fun factor of the smaller turbo. Here is another comparison of some turbos. I couldn't overlay the GTX3582R w/.82 without doing some free hand drawing. But check out at 4000RPM and where the TQ is on all these as well as the above graphs. Keep in mind these are not all GTX turbos, but what i am showing is the response and TQ from the GT or GTX 3076R and the GT or GTX3582R with different housings. At 3500RPM........ -3076R w/.63 hits 340ft-lbs of TQ -3076R w/.82 hits 280ft-lbs of TQ -3582R w/.63 hits 240ft-lbs of TQ -3582R w/.82 hits 220ft-lbs of TQ AT 4000RPM...... -3076R w/.63 hits 420ft-lbs of TQ -3076R w/.82 hits 420ft-lbs of TQ -3582R w/.63 hits 320ft-lbs of TQ -3582R w/.82 hits 270ft-lbs of TQThat is all you ever really needed to know about the GTX turbo and how it compares to the GT stuff. Overall for a couple hundred extra bucks you get more HP, and or more potential for more HP. That seem cheap compared to some parts you buy that only give 10WHP.