Lithium

Haha that's actually pretty damn cool, I have to admit I was always a huge fan of the old Japanese drag cars running 3240s etc - they had an epic sound to them and looked choice under the bonnet and I'm sure this will look and sound unique, albeit I don't think twins are actually functionally that great a prospect on RBs and while I hope it works... especially with this turbo combo there's a good chance it may prove to be quite a headache trying to make it work as well as the 102+mm singles you're effectively trying to compete with here. VERY keen to see how it goes, though! Any more details on the engine spec?

It depends on what you mean by underperforming? I've seen a few unrealistic (imho) expectations being set, people seem to keep choosing similar or LARGER G-series turbos vs GTX equivalent assuming that size for size the G-series will spool better. I know some sellers/Garrett fanbois will disagree here, but that is the opposite of what I both expected and have seen so far. Size for size a G-series will be slightly laggier in terms of boost threshold than a GTX, but their transient response stays similar to be fair. A lot of them are fairly ambitious in terms of turbine flow versus power they're meant to support as well, and I think Garrett haven't helped themselves with turbine offerings which aren't always that well spaced - your only options may be a turbine which gives good response but chokes the turbine, then the next size up allows the turbine to breathe easily well enough to support the power you might be looking for but it also becomes significantly lazier.... when in fact the option right in between would have been perfect, ie. no .8x a/r turbine housing for G25s. It's worth trying to understand and check turbine flow maps to ensure you're not going wrong here with *any* turbo you go for if you can. The next thing is turbo failures/under performing power wise etc. Another thing I saw coming before these were even used, they flow big numbers for their size and people automatically assume that it means they need lots of boost to do it. They are infact NOT that suited to high boost, and at their best between 16-24psi at sea-level in most small/mid frame cases. After that their efficiency starts dropping off hard, and to add to it Garrett gave frankly bullshit hp claims for their turbos. What predictably happened here is you get a bunch of turbo illiterate people going "I want 900hp from my Evo", (which needs 40psi+ to make that kind of power) so they buy a G30 or G35 900 (because it's rated to 900hp) and put it all together and give it to a tuner with instructions to push it until it makes 900hp. If you actually read the compressor map data and make conservative assumptions that to make the power the owner would be content with requires the turbo to be able to move 90lb+/min of air at pressure ratio 3.7+. On the compressor map for this turbo it's pretty clear the turbo can barely squeeze out over 65lb/min at PR3.7 without overspeeding, and it gets WAY worse at any more boost than that. What makes this worse is that the turbine is actually a pretty efficient little high-speed beast, so while exhaust back pressure will start taking off as you go off the map it will continue to push the compressor harder and harder - meaning you can SERIOUSLY overspeed these turbos pretty easily by running them on a mismatched setup. Soooo, yeah. A lot of the issues IMHO so far come from poor matching. I am not saying that these should be the greatest or anything like that - I for one still for the most part like the performance of the EFR range the most, but overall anything you can do with a GTX I suspect you can do better with a G-series *if* the option is there that appropriately matches a setup. And this is the clincher, a good turbo match is often worth more performance wise than the extra 5-10% or whatever performance you may get from a much fancier turbo. So, TL;DR version: There may be legit failures but I'm not aware of any beyond where people have tried pushing a setup I could have told them would cause issues before they started. If you're going T4 twin scroll then I don't think Garrett even have a native exhaust housing option for the GTX range, and I personally wouldn't hesitate to use a G-series on an RB25 because the boost range you'd be looking at fit right in the "happy boost zone" of the G-series range. Not sure of your power targets, but a G30-770 with a divided hotside has potential to be quite hilarious

Still heaps of stuff going on out there, just a lot of people who do the investigation and development themselves no longer or at least rarely post on forums. I can't speak for everyone, but I lost a lot of interest in sharing things in here as it reached a point where the main people interested in responding were the ones who were more interested in talking shxt than actually helping people learn and get better results and tech discussions consistently went downhill - so whether or not I have anything of value to share, it didn't seem like there was much point bringing it here. One of the builds I have been involved with had an update recently, more modest than the big dollar builds but definitely fitting what you describe in your post. This car is fully developed by the owner ( @Looney_Head ) with only help from his mates, no throwing big cash at the top brands or workshops to get things done. Just lots of yarns, rums, headscratching and knuckle tearing in his garage. There are plenty more of these around NZ, even in our group of friends and I am sure there is plenty of it going on in Oz too.... just maybe largely off the radar.

let us know how you get on

Of course not, but knock becomes a lot more likely if you introduce boost to a setup which has erratic timing

Have you upgraded your CAS setup? There are various options but an aging RB25 and knock on boost has a definite trigger error feel to it

I got a bit of momentum writing a ramble about this but then decided that it was probably a waste of time, the short of it is that you should be trusting your tuner to make the call - there are WAY too many variables to ask a bunch of randoms on a forum about. Your tuner should have all the variables and the wisdom to choose what best suits your interests and setup, if you don't trust him to make that call then you should probably have a different tuner. In regards to power, a mate just tuned an RB26 with a 7675 Gen2 on NZ BP98 last week and stopped at this point (it will be wound up more on E85, but it's a flex fuel setup): That would still be >700whp on a typical Oz rolling road, so realistically with a 7685 on an RB32 you'd think 600-700whp should be quite doable if the right combo has been put together in the right way to allow it but there are plenty of things which could get in the way of that being achieveable.

Let us know what you decide on, and how it goes

To be fair, "after 4500rpm" is pretty open ended haha. The power versus response on the "6662" I would expect to be a fair bit better, but would expect it to be a reasonable amount lazier in the lower rpm than -5s - yours seems laggy, even for those. Are you actually looking for more response, more power, or a good combo? I have in my head you're Kiwi/NZ based? I'd consider talking to @infomotive as he is a NZ based Pulsar turbo dealer, and already has a fair bit of direct experience with running these turbos on RBs so can probably give you some realistic expectations on the G2 GTX3582R spec thing in terms of power/response or other suggestions.

Wow, that's lazy - definitely not right for that kind of setup. I personally am not a fan of the ATP T4 housing and actually have no idea how you have a T4 open housing G30 - Garrett don't have a native T4 open housing so I'd wonder what that housing spec and quality is and how much it could be contributing to that result. Otaku intake manifolds seem to have a pretty dodgy reputation at least from what I've heard, I'd make sure there has been no sneaky cracks for boost to leak out of. I know nothing of the CXRacing manifold either, but the smorgasbord of cost cut spec parts and a sub-par result definitely make it feel like this could turn into a "the poor man pays twice" kind of story. I would at least be going through a diagnostic process to rule out the typical "something ain't working right" process. A genuine Garrett G30 with a .83a/r exhaust housing on a stock cam RB25 with VVT working, and quality intake and exhaust manifolds with no leaks should NOT be that lazy. G30s are a little later in boost threshold than other turbos with the same wheel sizes, but no way near that later - twin scroll or otherwise. I know of an G35 1050 running an open housing on an RB25 and it's significantly less lazy than this. That altitude wouldn't help, but I wouldn't expect it to hurt THAT much. Something that is catching my eye is how linear the spool is - it doesn't seem like it's getting into a proper "ramp" like you'd normally expect. Just to rule out something very obvious that I've definitely seen other people miss, and end up with this kind of lag (if not actually being outright unable to build more than a few psi of boost) - did you definitely install the valve seat/fire ring/whatever you want to call it in the wastegate? Without it you effectively end up with the effect of the wastegate being constantly slightly cracked, the exhaust will sound fairly pitchy/angry the whole time and it makes the turbo a lot laggier.

This response may be controversial, but I'll get the ball rolling with it as at least it is an answer/some conversation This is basically a pretty solid copy of a Garrett Gen2 GTX3582R, and basically the same copy is sold under different branding as a Pulsar GTX3582R Gen2 - funnily enough. As such I'd consider using G2 GTX3582R information as a general guide of what you can expect flow wise from these (they're solid!) but also spool wise. Realistically it's a pretty "big" turbo, definitely laggier than the previous 3582s you could get. Depends really on what you are looking for power wise, how much lag you're willing to live with (realistically all in after 4500rpm on a RB26) and that general kind of thing.

That's fair, though I did make it clear it was a copy and not the real thing - a lot of people consider the Pulsar turbos in this part of the world so in the interest of general data sharing I thought I'd pass it on and let people make of it or disregard it as they will. Probably should have added that to the Pulsar thread. Interesting that HKS are more or less sticking to their "old" airflow to PS estimation method, instead of using Garrett's ultra generous power claims that they moved to from the release of the Gen2 GTX. Either way, nice modern Garrett turbos feels way more "HKS" to me than the strange Mitsi JB period they went through recentl

Me, too. It would be pretty fascinating stuff - I don't know what we should expect, I can say "for the supported power" the G-series seem really good transient response. The only first hand experience I've had is with a Mitsi Evo which I tune which previously had a TD05-20G-esque thing, kinda really almost the equivalent of the original FP Green but using a 7cm TD05 hotside instead of the 8cm TD06 that FP used to use, and a non-Mitsi 50lb/min compressor wheel. It had a fairly decent boost threshold, we were seeing 20psi by around 3700rpm in a 3rd gear pull but it was ultra surgey mid throttle load, and its boost response was atrocious. You could literally feel it having to re-spool in gear shifts... a very strong case for the whole "dyno plots don't tell the story" thing, as you could expect it to be a response monster from how it looked on the dyno. Another thing about on the dyno is it was HEAVILY knock limited by 17psi/18psi, enough that I was not happy to continue. The owner swapped to ethanol to try and get around that and we were able to get a bit more boost into it, but when we got past 20psi I (not exactly surprisingly) noticed that VE was plummeting at higher MAP levels. EMAP was clearly getting up there, quickly. In this case the owned decided to give a Pulsar (Garrett copy) G25-660 with a .72 hotside a crack, basically with the hope of trying to sort the choking issues with minimal sacrifice to response while also having headroom for more power when he got a built bottom end. Also went back to pump gas as ethanol is shockingly expensive and hard to get in NZ again now, so man cams/big turbos and pump gas tunes are back in season in here. So far have only road tuned it, but the fuel map needed massive changes from 17psi and up and the old timing map which was still knock limited - not a peep so far, have gone past 20psi and called it there for until we get to the dyno. Unfortunately we're in lockdown atm so the dyno booking we had for tomorrow has gone up in smoke, no idea when it will happen now. What I can say is that, as expected - the turbo feels a lot lazier than you'd expect if you were comparing the wheel sizes with the likes of a GTX series Garrett turbo. In GT series nomenclature it would be a "G2567" which paints the picture of something that would be in much earlier than a "20G" type sized turbo, but it was the opposite by a bit. Noticeably doughier up until around 3500-4000rpm, but from there it's ramping up quick - probably seeing 20psi marginally after 4000rpm in 3rd gear. What is different however, is that it "feels" much snappier anywhere but the basement... basically once over 3500rpm it takes less provocation than the old turbo to start "winding up", just that it can't wind up to quite the same boost level the old one could until up to around 4000rpm. Once over those rpm it is almost NA-like, WAY more snappy than the old turbo. Like the car feels much more alive pound for pound, and "back into throttle". This is kinda what I predicted before they were used, but it was nice to experience it - basically, the boost threshold of something half way between what you'd expect from a GTX2867R and a GTX3071R, while providing the flow of something halfway between a GTX3071R and a GTX3076R, but the transient response of a GTX2867R... or even slightly better than that, which I guess makes sense given its a lower blade count and similar dimensions (so less inertia) than the old turbine. Something I feel compelled to add here, I've experienced a twin scroll EFR7163 on a 2litre Toyota engine and it makes both other turbos on this particular engine feel lazy <4000rpm BUT I doubt I'd be able to tell the difference between the G25-660 and the EFR7163 once above 4500rpm. The G25 660 in this comparison is open housing. I would guess that this kind of behaviour would translate through the rest of the range to a degree, both the EFR8474 and the G35 1050 are likely to have way better transient response than the boost threshold shows when keeping in mind the best of the "2000s" ball bearing turbos. It's quite interesting to see how they've both done it, both turbos have exactly the same compressor wheel measurements but the EFR has the 74/68mm 11-blade TiAL turbine wheel versus the 68/62mm 9-blade Mar-M turbine wheel to result in different ways of reducing inertia while allowing enough turbine efficiency to give a reasonable performance level. They both will have very low inertia for things making the kind of power they're making, and will probably be significantly snappier under foot than pretty much anything else capable of 750+whp - other than perhaps the Xona Rotor UHF range, but that's a whole other kettle of fish. Beyond those, if I were a betting man I think from this result if anyone tried EITHER of these two turbo specs on an RB they would probably smash the transient response over any of the other typical Garrett/Precision/whatever combos you normally see for this power level on RBs.

Ditto, I wouldn't touch them.

Yeah sounds like you're going a good way, have yarned heaps with Brian from NDT and he is on to it and a GC. Be good to see how it comes out

Ok, I'm no engineer so I'm not going to claim my understanding is 100% correct but this is my general understanding on the difference between the two major approaches. Constant pressure This approach essentially is about smoothing out spikes and drops in pressure and making the gas flow through to the turbine run at a more linear rate. Basically they need to have a nice even smooth collection of the pulses from all cylinders, and rely on a reasonable volume to ensure the pulses dissipate and even out. The idea here is instead of each pulse driving the turbine individually, instead the drive pressure gradually builds up and then keeps a constant amount of force driving the turbine. This means that there is a natural transient lag when exhaust pulses get stronger as the drive to the turbine doesn't really increase until the whole manifold has increased in pressure to match the higher "load" the engine is under, but on the flipside it means that once the drive pressure has built up to a useful level the turbine is kept operating quite efficiently. They're at their relative happiest at full boost/higher rpm situations, they are at their saddest going from low rpm/low load to high load. This is a textbook constant pressure manifold, x2. 3x even length runners of reasonable length and diameter feeding into a CNC collector with all runners facing the turbine at a similar angle. The volume that each set of 3 cylinders feed into are significantly higher than the volume of the cylinders that are feeding the manifold, so when exhaust blow down is finished for a cylinder the runners are fairly well far from saturated so instead of all the energy from that pulse driving the turbine... much of it basically expands out in that side of the manifold, at least until the pressure has reached a point where the turbine is still effectively being driven by residual drive pressure (remember that there are gaps in the pulses when you have just the front or rear 3 feeding one side). Pulse Convertor This is basically what twin scroll exhaust housings were designed to work with, and they swing the other way. Basically the idea is the front and rear 3 cylinders from an I6 engine feed alternating scrolls on a common turbo. The general idea is keep the volume down and the path from each cylinder to the turbine as efficient as possible, meaning the least messy path and the least opportunity to expand along the way as possible. Each pulse should bring the pressure of it's side of the turbine up rapidly, then die down again - with the other side of the turbine getting drive by the following pulse on the other side of the engine. Each side of the manifold should have fairly dramatic spikes in pressure and it will drop (relatively speaking) for when the next cylinder blow down event is due in that side of the manifold. You don't run a collector, or even length runners - you make sure that each pulse is able to drive the turbine as effectively as possible. If the manifold takes less time to pressurise, it also takes less time to empty - there will not be direct pulse collision as there is a gap in the blow down events between the sides of the manifold to allow it to clear a bit before the next cylinder gets a go. This is a more of a nice "split pulse" type manifold: The design isn't trying to set up for the pressure to build up evenly so much as trying to set up as there it's just a pair of exhaust energy sources alternately blasting at the turbine. The idea is that the drive pressure builds up FAST which results in more spool from low rpm and generally a much faster reaction at the turbine to an increase in engine load. Basically turbine efficiency is MUCH higher than with a constant pressure turbo configuration (merge collector style) going from low/mid rpm or lower load to higher load, but then when the pulses become fast enough the constant pressure method becomes more efficient - and they start coming into their own when you start trying to really push the limits with a given size turbo, or generally turn high rpm and make max power. Seeing what happens with cars which often run pulse collector manifolds stock or aftermarket, realistically if someone made them for RBs I'd suspect they'd be ideal for what a lot of the folks here would be normally talking about. Anecdotally speaking I'd pick them as awesome in the responsive/torquey 400-750hp @ hubs area with mid sized turbos, but not what you'd use for roll/drag racing or very high power circuit/time attack cars. TL;DR: Basically all manifolds designed for RBs are a design which is optimal for steady state and max power, pulse convertor manifolds (stock single turbo and many aftermarket BMW manifolds) are optimal for quick spool and transient response. It CLEARLY doesn't mean one type doesn't perform well where the other one performs best, just more pointing out what each are "optimal" at. An amusing example of what I consider a "missing the point" of how pulse convertor manifolds work is people freaking out when B58 Supras came out with "two runner" manifolds, thinking it was way too small for a 3litre... not realising that those two pipes are just the last pit of a pulse convertor manifold pairing cast into the head, and they are not actually the kind of choke point it would seem because they're not operating in the same way. If you've ever seen a dyno plot for a A90 Supra making solid power with a highflow stock turbo, you'd know that they are not laggy at all and they CAN make 700+whp/do 9s reliably with this configuration. Please don't confuse this kind of design with a generic log manifold which has just been made to be small and cheap to make. There are some rubbish ones out there, which do NOT use the exhaust energy well at all and just result in restriction and sometimes even lag despite the small volume. There are obviously draw backs, I'm just explaining because I was asked... but they 100% have their place imho. I would argue that the T4 divided tubular manifold above is not a split pulse manifold in the sense that it combines the pulses to provide a steady drive pressure, as opposed to separating the pulses to the turbine.

Nice choice! The best responding RB26 I've ever been in was a stock bottom end one running an EFR8374 - though it DID have V-Cam as well. Head and shoulders above any other RB26 I've experienced in terms of torque/response, and was also making ~460awkw. Great option for having a nice road setup with plenty of headroom. Who is doing the build?

Here's a thing I discovered in my travels which may be of interest in here, this is the kind of comparison I've been hoping to see but not expecting to, finally... This is a dyno plot of a 1.05a/r divided EFR8474 (dotted line) and a 1.05a/r divided G35-1050 (solid line) with otherwise identical specs, including boost curve on an Evo X. Both hit target 2.15bar at 4500rpm and creep out to 2.35 at peak, which is as far as they were willing to push the engine at this point. Apparently the EFR8474 had stopped giving gains at this point however the G35 still seemed to have more to give, which they said they may try sometime if they get brave. So yeah, seems like the G-series range get a bit more interesting now that the divided housings are starting to seep into circulation! Worth noting that "km" stands for koń mechaniczny, or "horse power" in Polish

That's because most manifolds are not true split pulse - I don't think I've ever seen a "proper" pulse convertor manifold ever used on an RB. Every aftermarket twin scroll manifold I've seen for RBs are constant pressure (ie, designed to work with an open housing) but divided into two - which means they are effectively just operating as two single scroll turbos, but at least have the advantage of having no high pressure collisions and having a slightly smaller volume than a single 6>1 collector manifold would have... all which is good stuff, but not really stuff that is properly focussed on exploiting a divided turbine housing.

Seems like you have pretty realistic hopes by my estimation - of course depending on what you mean by "full boost". I'd expect there will be "good boost" by around 4000rpm, especially with VCT... should be a really nice setup

Most recent I've seen, .83a/r T3 open IWG EFR7670 on a stock manifold: I've been hoping to go for a blat in this car, so far my limited exposure to 7670s make me suspect the 8374 is better "all round" for most RBs. The 7670s don't have an amazing boost threshold (seems to be a few hundred rpm later at worst than what the 8374s offer) but fall quite well short of what 8374 makes

Only thing I could imagine is that it suggests there is something wrong with how the EMAP is reading - that seems way too low, unless it means 23psi of excess EMAP versus IMAP (like, EMAP-IMAP = 23psi) Epic build, either way - awesome to see it at this stage!

I'm with you here. Love the concept and the performance of a lot of the EFR range, but I recommended them to a couple of people who both received EFRs in the last couple years with worryingly shitty casting and I was straight up embarrassed about pushing a >$3,000NZ product to someone because of the quality and performance promised for something that crap to show up and I'm not going to do it again. There are options which will tolerate more punishment, are nicer made, easier to package and potentially may compete as well... at least the balance of the other variables make them more appealing.

Thought I'd randomly bump this thread because of a strange unfolding of events. As I implied here, a Holset HX35 was my "first choice" for my R33 GTS25t but at the time it was not so easy or cost effective to get an aftermarket exhaust manifold which would suit my needs and I was kind of interested in seeing what I could do with the IWG GT3076R and had quite a bit of fun with that. When I decided to move on from the car I sold the GT30R setup to a mate and reverted the car to a stock turbo before selling. Some time down the track the new owner of my car asked about it and was thinking of replicating what I had done, and I mentioned that I probably wouldn't do the same now as I hated the IWG setup - and after a mix of "things falling into place" etc he ended up doing the setup I would have done in the first place. There are aspects of this setup which are better than what I had (ID1000s instead of Nismo 555cc, G4+ Link controlling everything instead of V5 Link and Apexi AVCR, GReddy intake manifold and Sinco exhaust manifold instead of stock manifolds, and a NZ Wiring cam trigger setup) but other than the ECU and injectors everything else would have been doable at the time I did it back in 2006. We didn't push it especially hard, it runs the smaller T3 undivided exhaust housing and I was starting to see hints of climbing EMAP - the aim was a responsive 300kw @ hubs on 98 so we called it a day at just over 310kw on 20psi. The wastegate opens around 3300-3400rpm, and the thing basically drives like a stock turbo. I've mapped target boost against throttle angle and if you're over 3500rpm then the target boost and actual boost track almost like it's an NA style log, the thing would have murdered my old GT3076R in terms of response and made a good 20kw @ hubs more. Kinda makes me wish I did go the Holset in the first place, and really I think even though it is using an old "truck" turbo the result still stands up pretty well for things making this kind of power on 98.

9280. Powertune were corrected and never updated the post.