IW STi Forum banner
141 - 156 of 156 Posts

·
Super Moderator
Joined
·
8,957 Posts
I will have a G25-660 on the dyno on Friday. I've had it on the dyno twice, but it ended up having several issues I had to solve. It is now leak free and a 4-port EBCS (couldn't do more than 23 PSI on the 3-port and 10 PSI spring).





Preliminary numbers are the left graph:


I spent a bit of time with this car trying to get the turbo to perform and it just would not. I boost leak tested it myself and swapped it to a 4-port EBCS (which helped), but peak boost was at 5,000 RPM and tapered like 5 PSI by 7,500 RPM. It made 488whp on a DynoMITE dyno, which in my own tests reads a little higher than a DynoJet.

I am pretty sure it was a dud from the factory, especially seeing the pictures/measurements and the story to accompany why this is the second turbo. The first wouldn't even spin by hand. The turbine was touching the housing. Garrett would not warranty it, mind you , it had not even been used! Once the second turbo was received, they took some measurements and they were pretty far off from each other, as well as the housings being visibly different.

The car (not when it failed):


This was scary as it caught fire, which we were able to put out quickly.











I'm will personally stick with EFR's and recommend them. This wouldn't be as much of an issue if they would warranty their parts that have QC issues. Luckily, the motor is fine, but the intercooler may be replaced, which is a custom build.
 

·
Registered
Joined
·
231 Posts
I spent a bit of time with this car trying to get the turbo to perform and it just would not. I boost leak tested it myself and swapped it to a 4-port EBCS (which helped), but peak boost was at 5,000 RPM and tapered like 5 PSI by 7,500 RPM. It made 488whp on a DynoMITE dyno, which in my own tests reads a little higher than a DynoJet.

I am pretty sure it was a dud from the factory, especially seeing the pictures/measurements and the story to accompany why this is the second turbo. The first wouldn't even spin by hand. The turbine was touching the housing. Garrett would not warranty it, mind you , it had not even been used! Once the second turbo was received, they took some measurements and they were pretty far off from each other, as well as the housings being visibly different.

The car (not when it failed):


This was scary as it caught fire, which we were able to put out quickly.











I'm will personally stick with EFR's and recommend them. This wouldn't be as much of an issue if they would warranty their parts that have QC issues. Luckily, the motor is fine, but the intercooler may be replaced, which is a custom build.
Jesus yamaha! Thats a bad failure for a brand new turbski! Usually they just wont make power if the catridge isnt up to snuff or its completely imbalance but that right there is kinda scary.

Im disappointed in how they ruled to not warranty the turbo just to solidify and concrete their policies. The warranty not being a case by case really hurts my image of them. Quality control should always be held accountable for these types of failures.

Hopefully you guys get that sorted out. EFR and Precision has always been good to me. Hell even FP in the early days helped me out on DSM stuff and early ej20 builds for our track car.

Sent from my SM-G965U using Tapatalk
 

·
Super Moderator
Joined
·
8,957 Posts
Turbo #3 went even faster, which was the warranty replacement! I'll post more after Garrett has a look at it, but this one made it like 4 or 5 days. Had it on the dyno to see where it liked boost/timing while making 499whp (could have made more, but didn't want to push it), then turned it down and went to an autoX the next day. It went quickly.

Granted these are the Gen 2 series, but the G-series... But I personally will not use a G-series on anything at this point.

Automotive fuel system Motor vehicle Automotive tire Auto part Vehicle

White Automotive tire Automotive lighting Black Motor vehicle



The Gen 2 turbos have been fine for me. I had one on the dyno the same day (3076 on an S2000), and it made easy power (530whp) and didn't blow up the next day. :)
 

·
Registered
Joined
·
5,745 Posts
IMO you're over-speeding the crap out of it. I think that turbine is waaay too small for the power you're trying to make. If 500whp is your target, I'd go 35R frame size with a mid-size A/R and do it easily. Largest A/R 30R would do it as well assuming primo fuel.

The larger turbines will give power everywhere in the rev range and certainly won't choke flow up top
 

·
Super Moderator
Joined
·
8,957 Posts
We ran it to what it wanted on the dyno. We turned it down afterwards. This turbo should be able to handle 25 PSI on a Subaru without issue. On that dyno, the in-house tuner confirmed that is about he power he sees on that turbo on other platforms, however, it was turned down. I also found several threads on NASIOC where people are running 30-32 PSI on 2.5L Subarus, however, many don't use them like this... So I turned it down knowing the usage and the reference baro.

There wasn't a power goal, that's just what it made. The goal was decent spool, which this was supposed to have, but did not. We were going to try the .71 hot side if it made it through the autox.

Also, I have found that dyno to read a little higher than a DynoJet, so the power you saw is really what the car wanted. If the % trend is roughly the same from a couple cars I had on that dyno for tuning, and then a DynoJet for classing, it was more like in the 450's for DJ numbers. Even if we take those numbers at face value (and they're are corrected at our nearly 6k ft), the turbo is marketed as a 660 bhp turbo (compressor map reaches out to 61-62 lbs/min though), so 500 whp [corrected] should not explode a turbo marketed towards this use case. My own personal STi did 405whp on this dyno, but 367whp on another, so please just understand that I am using it as a tool, and not for numbers.


I have run plenty of turbos harder than they should have and never had a failure, but did not do that here. The thought of over-speeding the turbo was already on my mind and not something I wanted to do, just to avoid the mess of what happened last time. I had boost leak tested the car and fixed all the leaks. The only thing I didn't confirm (but was recently told another tuner tested before me) is the BPV that routes back into the inlet. I never heard any air leak, but didn't pull it apart from the inlet. The other tuner had already did the test I suggested, which was to pull the return back to the inlet and put something of a visual reference during a dyno pull, to see if it was leaking, which they said passed as it should.


One more kicker, the log of the AutoX run showed a max of just under 11 PSI in second gear... Baro that day was 12.1.
 

·
Registered
Joined
·
5,745 Posts
We ran it to what it wanted on the dyno. We turned it down afterwards. This turbo should be able to handle 25 PSI on a Subaru without issue. On that dyno, the in-house tuner confirmed that is about he power he sees on that turbo on other platforms, however, it was turned down. I also found several threads on NASIOC where people are running 30-32 PSI on 2.5L Subarus, however, many don't use them like this... So I turned it down knowing the usage and the reference baro.

There wasn't a power goal, that's just what it made. The goal was decent spool, which this was supposed to have, but did not. We were going to try the .71 hot side if it made it through the autox.

Also, I have found that dyno to read a little higher than a DynoJet, so the power you saw is really what the car wanted. If the % trend is roughly the same from a couple cars I had on that dyno for tuning, and then a DynoJet for classing, it was more like in the 450's for DJ numbers. Even if we take those numbers at face value (and they're are corrected at our nearly 6k ft), the turbo is marketed as a 660 bhp turbo (compressor map reaches out to 61-62 lbs/min though), so 500 whp [corrected] should not explode a turbo marketed towards this use case. My own personal STi did 405whp on this dyno, but 367whp on another, so please just understand that I am using it as a tool, and not for numbers.


I have run plenty of turbos harder than they should have and never had a failure, but did not do that here. The thought of over-speeding the turbo was already on my mind and not something I wanted to do, just to avoid the mess of what happened last time. I had boost leak tested the car and fixed all the leaks. The only thing I didn't confirm (but was recently told another tuner tested before me) is the BPV that routes back into the inlet. I never heard any air leak, but didn't pull it apart from the inlet. The other tuner had already did the test I suggested, which was to pull the return back to the inlet and put something of a visual reference during a dyno pull, to see if it was leaking, which they said passed as it should.


One more kicker, the log of the AutoX run showed a max of just under 11 PSI in second gear... Baro that day was 12.1.
We've run boost into the upper 40s, but the EJ is not like 'other platforms', don't fall victim to that thinking. EJs are horribly inefficient in comparison to a Honda, Evo, and the list goes on. You can't spec the turbo the same, it just won't make good power. As a general rule of thumb with an EJ, run a bigger turbine and smaller compressor. It favors the bore/stroke/displacement and other design nuances of the EJ.

I still stand by that turbo is being overspun, and by a good bit. If the turbine is going to be pushed anywhere near flow limits, you need a wheel speed sensor on them.

I would roughly treat a G25-660 as a slightly smaller GXT3071. Mostly because the turbine side flows less on the G. Looking at the G25-660 compressor map, ballparking good fuel, and decent sized A/R I would expect ~400whp and making good power across the rev range.

Despite the marketing, let the compressor and turbine maps guide your decisions on turbo/turbine selection. If you're going to stick with that specific turbo, don't bother with the .72 A/R, go right to the big one. You'll pickup power everywhere across the rev range and your EGTs will be hating on you a lot less too.

If there's anything I can do to help, please don't hesitate to ask. We've tested crap loads of turbos here (on just Subarus) and if I don't feel comfortable answering your question, I have contacts at Garrett I can lean on as well.
 

·
Super Moderator
Joined
·
8,957 Posts
We've run boost into the upper 40s, but the EJ is not like 'other platforms', don't fall victim to that thinking. EJs are horribly inefficient in comparison to a Honda, Evo, and the list goes on. You can't spec the turbo the same, it just won't make good power. As a general rule of thumb with an EJ, run a bigger turbine and smaller compressor. It favors the bore/stroke/displacement and other design nuances of the EJ.
Eh... I've tuned more EJ's than any other motor. I think I understand them. Also, I am not the builder of the car.

I still stand by that turbo is being overspun, and by a good bit. If the turbine is going to be pushed anywhere near flow limits, you need a wheel speed sensor on them.
That's fine that you think that, just show me my data collection to backup against our use case. Again, it was turned down to 25 PSI tapering to 24 and the log, due to being an AutoX and not getting out of second gear, was 11 PSI, max. Stop thinking the power number given on the dyno are absolute. I've already found it to read higher than a DynoJet and DynoJets are on the higher side of numbers, in general.

I would roughly treat a G25-660 as a slightly smaller GXT3071. Mostly because the turbine side flows less on the G. Looking at the G25-660 compressor map, ballparking good fuel, and decent sized A/R I would expect ~400whp and making good power across the rev range.

Despite the marketing, let the compressor and turbine maps guide your decisions on turbo/turbine selection. If you're going to stick with that specific turbo, don't bother with the .72 A/R, go right to the big one. You'll pickup power everywhere across the rev range and your EGTs will be hating on you a lot less too.

If there's anything I can do to help, please don't hesitate to ask. We've tested crap loads of turbos here (on just Subarus) and if I don't feel comfortable answering your question, I have contacts at Garrett I can lean on as well.
Yes, this is why I used the dyno as a tool and I've also tested my own fair share of turbos. The owner will not bother with the .72 as he's not going with Garrett again. He already had the big hot side and we have less actual exhaust flow up here at 6k+ feet.
 

·
Registered
Joined
·
219 Posts
If there's anything I can do to help, please don't hesitate to ask. We've tested crap loads of turbos here (on just Subarus) and if I don't feel comfortable answering your question, I have contacts at Garrett I can lean on as well.
Hi Chris, nowadays, with all the testing you have done, would you still recommend the gen 2 GTX3576R, or are there better options available now to consider for a 500-550 whp daily? And if so, what size A/R?

What model would the competing EFR turbo be?
 

·
Registered
Joined
·
5,745 Posts
Also, I am not the builder of the car.
Ugh... Aren't those so fun :censored:

not getting out of second gear, was 11 PSI, max. Stop thinking the power number given on the dyno are absolute. I've already found it to read higher than a DynoJet and DynoJets are on the higher side of numbers, in general.
Not thinking of the power, thinking of how choked that hot side was trying to force 25 psi through the engine. No boost at auto-x, that turbo was already on the outs before you left the dyno.

The owner will not bother with the .72 as he's not going with Garrett again.
Roger that.
 

·
Registered
Joined
·
5,745 Posts
Hi Chris, nowadays, with all the testing you have done, would you still recommend the gen 2 GTX3576R, or are there better options available now to consider for a 500-550 whp daily? And if so, what size A/R?

What model would the competing EFR turbo be?
We stopped turbo kit development about a year after the G-Series came out and the early offerings were not great compared to what's available now. Sadly, because we stopped development on the kits, I don't have 3576-to-G Series comparative data, because there was nothing in the G-series lineup at the time. If I were building a new shop mule today, it would have a G-Series. Only because turbo tech gets better with time. NOT like the marketing would lead you to believe, but more in the realm of 5-10% better with each gen. Back to your original question, the GTX3576-R Gen 2 is still my personal favorite for that power goal; middle A/R turbine for 93/Flex and big A/R for track use and/or only E85. It just makes such fat power from ~4K to 7,500 RPMs.
 

·
Registered
Joined
·
287 Posts
We've run boost into the upper 40s, but the EJ is not like 'other platforms', don't fall victim to that thinking. EJs are horribly inefficient in comparison to a Honda, Evo, and the list goes on. You can't spec the turbo the same, it just won't make good power. As a general rule of thumb with an EJ, run a bigger turbine and smaller compressor. It favors the bore/stroke/displacement and other design nuances of the EJ.

I still stand by that turbo is being overspun, and by a good bit. If the turbine is going to be pushed anywhere near flow limits, you need a wheel speed sensor on them.

I would roughly treat a G25-660 as a slightly smaller GXT3071. Mostly because the turbine side flows less on the G. Looking at the G25-660 compressor map, ballparking good fuel, and decent sized A/R I would expect ~400whp and making good power across the rev range.

Despite the marketing, let the compressor and turbine maps guide your decisions on turbo/turbine selection. If you're going to stick with that specific turbo, don't bother with the .72 A/R, go right to the big one. You'll pickup power everywhere across the rev range and your EGTs will be hating on you a lot less too.

If there's anything I can do to help, please don't hesitate to ask. We've tested crap loads of turbos here (on just Subarus) and if I don't feel comfortable answering your question, I have contacts at Garrett I can lean on as well.

I appreciate the feedback, the car being discussed is actually mine. You and I have similar thought processes over the ideas going into this build. As discussed, there was not a peak power goal in mind, the goal was to have a quick spooling and more responsive setup, which none of these G series have done. You mention treating the G25 as a slightly smaller GTX3071, which was exactly my intention as prior to the G25 I had a GTX3071 on this setup without any issues. I wanted quicker spool and better transient response which is what you would expect moving from the 3071 to the G25. Unfortunately, none of the G series I've had have performed, they should not be making peak tq at 6k rpms, but that's what they have done. This is the third one btw and it was on the car for less than a week. My first one didn't even free spin out of the box as the turbine wheel was contacting the compressor housing. The second one was dimensionally and visually different, the turbine depth of the two being different by .06 inches. The third one measured the same externally as the first but had the same measurements of turbine wheel as the second. All three were different from each other in one way or another despite having the same part numbers.

I have also seen tons of evidence of subarus doing better with the larger exhaust housings and making more power everywhere, specifically cars locally at the same altitude as me. This is why I opted for the larger housing off the bat, all of these were using the .92. My goal was to have more power to the left of the curve than the 3071 offered, especially so because I had put a longer second gear in the car. This is an autox car afterall so quick spool and tansient were the goals, not peak power. Ill add as well that the two turbos that separated show zero signs of being overspun either, focusing on the backside of the compressor there's no distortion or orange peel, the logs also back this up. I'm sorry but you're not hitting max shaft speed of 165k at part throttle in second gear at 11psi.

Some build details:

IAG Magnum, fully assembled longblock, fire locks, Stage 3 D casting heads, Kelford 199-Bs ect, rated for 1000hp
Full Cobb Flex Fuel
1700cc IDs
Killer B Holy Header
Tial 44mm WG
Tial BPV
Garrett intercooler core
 

·
Registered
Joined
·
5,745 Posts
I appreciate the feedback, the car being discussed is actually mine. You and I have similar thought processes over the ideas going into this build. As discussed, there was not a peak power goal in mind, the goal was to have a quick spooling and more responsive setup, which none of these G series have done. You mention treating the G25 as a slightly smaller GTX3071, which was exactly my intention as prior to the G25 I had a GTX3071 on this setup without any issues. I wanted quicker spool and better transient response which is what you would expect moving from the 3071 to the G25. Unfortunately, none of the G series I've had have performed, they should not be making peak tq at 6k rpms, but that's what they have done. This is the third one btw and it was on the car for less than a week. My first one didn't even free spin out of the box as the turbine wheel was contacting the compressor housing. The second one was dimensionally and visually different, the turbine depth of the two being different by .06 inches. The third one measured the same externally as the first but had the same measurements of turbine wheel as the second. All three were different from each other in one way or another despite having the same part numbers.

I have also seen tons of evidence of subarus doing better with the larger exhaust housings and making more power everywhere, specifically cars locally at the same altitude as me. This is why I opted for the larger housing off the bat, all of these were using the .92. My goal was to have more power to the left of the curve than the 3071 offered, especially so because I had put a longer second gear in the car. This is an autox car afterall so quick spool and tansient were the goals, not peak power. Ill add as well that the two turbos that separated show zero signs of being overspun either, focusing on the backside of the compressor there's no distortion or orange peel, the logs also back this up. I'm sorry but you're not hitting max shaft speed of 165k at part throttle in second gear at 11psi.
I have to say, stories of Garrett issues are far and few between, but by the sounds of it, you've hit the motherload! Hopefully this is not a new trend and you have better with your next selection.

Are you running a specific Auto-X class, or just for fun?
 

·
Registered
Joined
·
287 Posts
I have to say, stories of Garrett issues are far and few between, but by the sounds of it, you've hit the motherload! Hopefully this is not a new trend and you have better with your next selection.

Are you running a specific Auto-X class, or just for fun?
If I didn't have bad luck, I wouldn't even know what luck is. This car is currently setup to the rules for Street Modified and has several national trophies over the years including its first year in SM at nationals in 2019, and a podium finish at nationals in STU in 2014. Chasing tenths of a second is what at least on paper it made sense to swap from the 3071 to a G25 in hopes of finding more response and a wider tq curve.
 

·
Super Moderator
Joined
·
8,957 Posts
I've been talking with another tuner friend of mine and there is at least one other person that's had this same issue with this turbo. Their fix was upgrading to a turbo with a bigger shaft. I am awaiting more information, however, I am certain that this turbo is not worth the money with other options on the table.
 
141 - 156 of 156 Posts
Top