Jump to content

Header Design Theory & Turbo EJ Headers


Recommended Posts

After really thinking about the OEM header I decided that I really dislike the design and wanted a better designed aftermarket header. As with just about any major purchase I start with reading then building a spreadsheet to compare one part vs the other. I've been trying to get as much info about primary sizes, designs, weights, etc., but it gets pretty hard to find info on some headers (with Tomei being the most generous with header information).



OEM UEL Manifold Design Overview

USDM OEM manifold has 2.625" long primaries on the driver side and 2" long primaries on the passenger side. This is a typical design of a log manifold, it's just a very stretched log manifold. What's worse about this design is it's pairing the two adjacently firing cylinders together with a short divider. With only two inch primaries the exhaust gasses don't get very far before getting sucked back into the neighboring cylinder causing reversion.


With the small stock turbos manifold efficiency under the curve doesn't really matter, because you'll be building boost and overcoming back pressure in no time. Since the primaries and overall manifold is fairly short it doesn't become too big of a restriction up top either, hence OEM manifolds actually perform pretty well up top. With a bigger lagy turbo stock manifold's shortfalls really start showing though. This is what forced me to start looking for a new manifold, city driving on a bigger turbo lugs the motor a lot easier, requiring more downshifting, which leads to reduced fuel economy.




No worries on reading them all as I'm going to be quoting specific points from these articles throughout the post. It's really hard to find good turbo manifold/header design articles, thus I'll be referencing NA designs concepts too, since they can apply to turbo applications too.


NA Header design articles:

Header-Exhaust Design Effects on Engine Power

Understanding The Different Types Of Headers On The Market – Streetable Rods

Mazda: SKYACTIV-G's 4-2-1 Exhaust System


Turbo Manifold/Header design articles:

Warpspeed's Post in "Exhaust scavenging on a turbo header" thread

Turbo Exhaust Theory



More reading material to help set the stage

Intake & Exhaust Tuning - One of the biggest misconceptions is that ram tuning is not important on a turbo-charged engine. Through extensive testing, I've found that a turbocharged engine responds to a tuned inlet and exhaust system and large-port cylinder heads as much as a naturally aspirated engine. The engine also needs much more cam than most people think. Don't strangle the engine!


One of the best examples of this was a 2.0-liter BMW that Gary Knudsen at McLarne Engines developed for Can-Am racing. The engine produced 540 HP on a gasoline with Mack air-to-air intercoolers and a simple log-type exhaust manifold.


Gary replaced the log manifold with tuned headers and picked up 60 HP. When he adjusted the cam specs to take advantage of the better exhaust, the engine produced 640 HP on the dyno and 600 on track! When Gary installed the tuned headers throttle response also improved tremendously.


This exert has a lot of gems in it; it reassures the importance of header tuning on turbocharged engines, it shows the tremendous gains from going away from log manifolds (what our stock UEL manifolds/headers are), and finally it shows the cam tuning opportunities that a better flowing header provides.



Primary Diameter


The diameter of the primaries determines the how responsive the engine will be and how peaky the torque curve will be.



Bigger diameter shifts peak torque to a higher rpm compared to a smaller diameter.


The bigger the diameter, the more cross-sectional area. Exhaust flow must overcome this extra tube cross-sectional area and therefore the flow travels slower . It takes the rpms to climb to a higher rpm before the speed of 240 ft/sec (and therefore, peak torque) is reached. So increasing diameter shifts when 240 ft/sec and peak torque is achieved to a higher or later rpm, because it takes lon for the air flow speed to reach 240 ft/sec.


In addition, a bigger diameter will increase the actual peak torque number (i.e. not only does diameter change the location, it also increases torque).


The diameter of the primary pipes directly affects the speed (or flow velocity) of the exhaust mixture traveling through them. Simply put' date=' gas moves faster through a small tube than a large one. Change the pipe diameter and you change that speed. You also change the rpm where the torque peak occurs. The lower resistance of a large-diameter header moves the torque peak into higher rpm; a smaller-diameter pipe moves the torque peak to a lower rpm. Bolt on a set of headers that are just too big and your torque peak will move so far up the rpm range, you’ll never see it again-and you’ll wonder why your new headers screwed up your motor.[/quote']



N/A cars: As most of you know, the design of turbo exhaust systems runs counter to exhaust design for n/a vehicles. N/A cars utilize exhaust velocity (not backpressure) in the collector to aid in scavenging other cylinders during the blowdown process. It just so happens that to get the appropriate velocity, you have to squeeze down the diameter of the discharge of the collector (aka the exhaust), which also induces backpressure. The backpressure is an undesirable byproduct of the desire to have a certain degree of exhaust velocity. Go too big, and you lose velocity and its associated beneficial scavenging effect. Too small and the backpressure skyrockets, more than offsetting any gain made by scavenging. There is a happy medium here.


For turbo cars, you throw all that out the window. You want the exhaust velocity to be high upstream of the turbine (i.e. in the header). You'll notice that primaries of turbo headers are smaller diameter than those of an n/a car of two-thirds the horsepower. The idea is to get the exhaust velocity up quickly, to get the turbo spooling as early as possible. Here, getting the boost up early is a much more effective way to torque than playing with tuned primary lengths and scavenging. The scavenging effects are small compared to what you'd get if you just got boost sooner instead. You have a turbo; you want boost. Just don't go so small on the header's primary diameter that you choke off the high end.




Smaller Primaries

  • Faster response & reduced lag due to higher velocity
  • Max torque shifted to the left


Bigger Primaries

  • More power mid-top end where smaller primaries would choke the engine
  • Max Torque shifted to the right



EJ Headers

This is not a definite list and might contain errors, finding this information took a lot of web searching.


Primary Internal Diameters

Exhaust Port Sizes: 39mm

USDM OEM: 39mm

Full Race: ~35.3mm (Educated guess based on tube thickness of 3.68mm, from Geoff's forum posts, and guessing 1.5" OD SCH10 pipe.)

Invidia: 39mm

Killer Bee Holy: 36.56mm

Perrin E-4 Regular: ~36.6mm

Perrin E-4 Big Tube: ~39mm

Tomei EL & UEL: 39mm

MAPerformance EL 4-1: 42.98mm


Secondary Internal Diameters

In this section I will list the secondary and tertiary (4-2-1 headers) diameters that I've found, calculated, or measured. While secondary diameters are less important, the do also contribute to overall responsiveness. Only two companies listed their secondary diameters, rest are based on my best judgement


OEM (4-2-1): 40.7mm Secondary, 41.7mm Tertiary

Full Race (4-2-1): Same as Primary

Invidia (4-2-1)- About same as Primary

Killer Bee Holy (4-1): Unknown: Best guess: ~50mm

Perrin E-4 Regular (4-2-1): About same as Primary

Perrin E-4 Big Tube (4-2-1): About same as primary

Tomei EL (4-2-1): 42.5mm Secondary, 57.5mm Tertiary

Tomei UEL (4-1): 57.5mm Secondary

MAPerformance EL (4-1): 55.68mm


Based on this information, Killer Bee and Perrin regular (Full Race too if it is 35mm) would have the least amount of lag and higher low end torque but at the cost of top end. This was also proven by Perrin's side by side dyno of the regular header vs the big tube. I personally was worried about going with bigger 39mm diameter header, until I found out that stock port sizes are 39mm and stock header is also 39mm.





Primary Length



Longer tubes will create more torque at the rpms before peak torque.


How do they do this?


Longer tubes will speed up air flow velocity. The flow velocity of 240 ft/sec and peak torque will occur at an earlier rpm compared to a shorter tube. Changing the length of the header primary tubes does not increase the value of peak torque like diameter does. Instead length changes the behaviour of the torque around peak torque along the rpm band.


If you imagine the torque vs r curve from a dyno to be like a see-saw: then, on a see-saw there is a point where the plank sits to allow it to rock up and down. This is usually in the middle of the see saw and is also called the fulcrum. On our torque vs rpm curve, imagine the peak torque to be the fulcrum, although this fulcrum doesn't necessarily have to be in the middle like the see-saw...it can be moved. Changing length "rocks" the torque curve about the peak torque.


If you have a longer primary header tube, the torque curve will "rock" in such a way that the left side is higher than the right side. There is higher torque at earlier rpms before peak torque. There is less torque at later rpms after peak torque.


If you shorten the length of the primary tube, the torque curve will will have the see-saw with the right side higher than the left. So there is more torque at later rpms after peak torque.


Mazda took advantage of long primary headers for another reason, it has reduced reversion, which reduced heat and mixture dilution thus allowed them to increase the compression ratio to 14:1.

4-2-1 Exhaust System



One option to significantly reduce residual gas is the adoption of a 4-2-1 exhaust system. As shown in Fig.3, when the exhaust manifold is short, the high pressure wave from the gas emerging immediately after cylinder No. 3’s exhaust valves open, for example, arrives at cylinder No.1 as it finishes its exhaust stroke and enters its intake stroke. As a result, exhaust gas which has just moved out of the cylinder is forced back inside the combustion chamber, increasing the amount of hot residual gas. With a short exhaust manifold, the high pressure wave arrives at the next cylinder within a short amount of time, causing this adverse effect to continue from low to high engine speeds. However, with a long 4-2-1 exhaust system, since it takes time for the high pressure wave to reach the next cylinder, the effect mentioned is limited to extra-low engine speeds, making the reduction of residual gas at almost all engine speeds possible.



On a small turbo street application where exhaust back pressure is considerably higher than boost pressure, the best approach seems to be the shortest length/minimum volume manifold that does not actually restrict flow.This is going to give best turbo response, but this is for a tractable street setup, not a max power application.


Exhaust reversion is going to always be a problem because of the very high static exhaust manifold pressure, and there is nothing you can do in the way of manifold design that is going to really help much.


Now consider a more healthy engine, with a much larger exhaust turbine, where exhaust back pressure might be about the same as boost pressure. Here there might be some gains to be had by thinking a bit more about the exhaust manifold design.


One factor to consider is exhaust cam duration and how many cylinders feed (each?) turbo.


For instance on a six cylinder engine with three cylinders feeding one turbo, the firing interval would be 240 degrees between each cylinder. If exhaust cam duration is 240 degrees or less, each exhaust valve will completely shut before another opens. In this case you can combine all three cylinders very close to the head and build a short minimum volume exhaust manifold, again for best turbo response.


The identical setup with a 290 degree duration exhaust cam will have periods where two exhaust valves are open together. One cylinder will be right at the beginning of the high pressure blow-down phase, while its neighbor will be at the sensitive valve overlap period. One cylinder will blow straight into the other.


You can fix this by using long individual runners from each port to the turbo flange. The pulse has to travel two runner lengths before it can blow into an adjacent cylinder. So above any reasonable mid range RPM, you can effectively isolate the exhaust ports.


On a big cam, big turbo engine, low RPM turbo response is not an issue, so the extra pipe volume does not hurt.


Pipe tuning on an n/a engine works by having the end of the pipe discharging into a low pressure. This creates a negative return wave that can be timed to coincide with valve overlap.


A turbo manifold terminates in a restriction, because the turbine scroll is always the most restrictive part of the whole exhaust system. So there can be no return negative wave to tune.


Most of the individual runner exhaust branches that you will see, are usually about twelve to fifteen inches long, which is sufficient to separate cylinders, but far too short for exhaust pipe tuning in the usual sense.


While we, turbo guys, are not worried about velocities or scavenging, other factors like reversion and even resonance tuning with free flowing enough turbos. Longer primaries can help reduce reversion since the exhaust pulse will have to travel further before it can travel back into another cylinder, by that point exhaust valves should be closed on that cylinder. Another huge bonus that longer runner buys us is, ability to have equal length primaries.



Longer Primaries

  • Fatter power curve below peak torque
  • Can be Equal Length
  • Can become a restriction up top, I think a bigger diameter pipe should help resolve this.
  • Lower EGT's due to pulse having to travel a longer time before a merge with another cylinder, giving exhaust cams more time to close before mixtures are diluted. This means longer runners will have less reversion with longer duration cams and/or higher AVCS advances.

Shorter Primaries

  • Faster Response time due to less time needed to reach the turbine
  • Reduced lag due to hotter exhaust gasses hitting the turbine
  • Less likely to become a restriction at higher RPM's = more power up top
  • Harder to make equal length
  • Higher EGT's due reversion



EJ Headers

This one is even harder to find information about then diameters, this information is based on my best judgement



UEL 4-2-1: USDM OEM (2"-2.75" primaries), Litespeed (Discontinued), WBR UEL



EL 4-1: Killer Bee Holy, MAPerformance

EL 4-2-1's: JDM Style clones: Perrin, Tomei EL, Full Race, Invidia



UEL 4-1: Tomei UEL & Similar



Most aftermarket headers seem to really shine in the mid ranges and at peak torque times, this is probably due to their longer runner designs. Also, long primary headers should be able to run a lot more AVCS advance then stock like headers, AVCS controls overlap and can be used to increase the Volumetric Efficiency (VE) of the engine.


It looks like typical UEL 4-1 headers have the longest primaries if you average out the lengths between all cylindres. Two of the primaries are short while the other two are very long. Depending on how much longer the furthest two cylinders are, it's possible that the two shorter cylinders will fire again before the furthest cylinders finish evacuating, that collision creates that lovely rumble. This also stacks the pulses, requiring a bigger exhaust then an evenly spaced pulses would need, more on this later in cylinder pairing section.


Aftermarket EL vs Aftermarket UEL

There isn't a good controlled test comparing EL and UEL headers from same manufacturer. Best we have is this numberless graph from Tomei. Now I've found some forum posts saying that it UEL was about 3-4whp less then EL. Most people say that EL's have much better response, which is not measurable with a dyno.




Cylinder Pairing & Merge Collector


There are couple ways to pair the primaries: 4-2-1 sequentially, 4-2-1 nonsequentially, 4-1 sequential (circular firing order in collector), and 4-1 nonsequential (diagonal firing order in collector).


NA 4-2-1 Pairing Sequentially vs Non-sequentially (keep in mind this is form an inline 4 article):

Here's Larry Widmer's (of Endyn) take on sequentially pairing the header primaries (i.e. 90 crankshaft degrees apart from one another instead of 180 crankshaft degrees):

The reason sequentially pairing of header primaries works is due to the energy imparted to the exhaust charge. If you just do 180 degree timing on the exhaust side, the exhaust pulses are evenly spaced, and they do permit a certain amount of "tuning", as opposed to just dumping everything into one collector.


When you space the tubes so there are more sequential pulses, the energy from one tube will have a much greater impact on the cylinder it's paired with, and the combined energy will have a much greater effect on the other tube it merges with.


Even (non-sequential) spacing (i.e. pairing header primaries from cylinders 1 with cylinder 4 and pairing cylinders 2 with 3) is nice and smooth, but pairing sequential pulses provides more energy to work with.


It's similar to the use of two single cylinder 2-stroke engines. If you want long running and smooth operation, connect the engines where they fire at 180 degrees to each other. If you want ball-busting acceleration, fire them together. It's all energy.


You get the same amount either way, but the combination you pick will allow you to properly select the energy spread.




  • Stacks the pulses to provide a stronger push
  • Can create stronger vacuum to help evacuate the adjacent cylinder
  • Can cause reversion on shorter primaries
  • Can require larger secondaries & exhaust to avoid chocking the engine



  • Evenly stacked pulses provide a more linear/consistent power delivery
  • Paired cylinders are 180* of crank rotation apart, reducing the chance of getting reversion
  • Can utilize smaller diameter secondaries & exhaust better


I believe sequential pairing applies to turbo manifolds/headers too. It would partially explain why OEM header and other UEL headers have more of a punch instead of more linear power that EL headers provide.


Our sequential headers are also unequal length, thus they stack pulses behind and on top of each other, you would need bigger diameter header & exhaust to not choke the engine which shifts the power curve more to the right. With equal length 4-1 and 4-2-1 non-sequential headers pulses should be evenly spaced thus bigger diameters are not required, smaller diameters end up aiding low end performance too.


EJ Headers

It's important to know the firing order and cylinder layout before you can determine how each header is paired.


EJ Firing is Order: 1-3-2-4

EJ Cylinder layout:

3 4

1 2



Sequential 4-2-1 (1-3, 2-4): USDM OEM UEL, Litespeed(Discountinued), WBR UEL

Nonsequential 4-2-1 (1-2, 3-4): Full Race, Invidia, Perrin, Tomei EL

Sequential 4-1: Killer Bee, MAPerformance, Tomei UEL



OEM headers are technically sequential, but they still perform poorly, this is probably because of very short primaries and flow restrictions. Litespeed header, which is OEM style, was able to put out fairly impressive numbers (they claimed 20whp/40wtq tuned). I think this shows the best case scenario gains for OEM manifold that has been ported and with a better cross pipe. High cylinder 4 EGT's would still be an issue with this manifold design.


WBR did make a slightly longer primary sequential 4-1 header, but LittleBlueGT lost power on it, though it might have been because of small primaries and not the sequential design.


Metal & Tubing Thickness

Tubing thickness is important for heat retention and longevity of the header. To my surprise, according to Corky Bell, thinner headers will actually retain heat much better then thicker ones.


The wall thickness of a particular material will strongly influence the heat transfer, in that the thicker the material, the faster heat will travel through it. This seems contrary to logic at first thought, but consider how fast heat would be drawn out of a high-conductivity, infinitely thick aluminum manifold, as op¬posed to a very thin piece of stainless surrounded with a nice insulator like air. Heat transfer is directly proportional to surface area. It is therefore reasonable to give considerable thought to keeping the exposed surface area of the exhaust manifold to an absolute minimum.



EJ Headers:

USDM OEM: Uknown Thickness Very Thick Cast Iron, Steel Crossover Pipe

Full Race - 3.683mm (~10gauge) 304 Stainless

Invidia - 2mm (~13.5gauge) 304 Stainless

Killer Bee Holy - 1.65mm (~15gauge) 321 Stainless

Perrin E-4 Regular - ~1.5mm (16gauge) 304 Stainless

Perrin E-4 Big Tube - ~1.5mm (16gauge) 304 Stainless

Tomei EL & UEL - 2.5mm (~12Gauge) 304 Stainless

MAPerformance EL 4-1 - ~1.5mm (16gauge) 304 Stainless


Most of the headers that we have are way too thin, this is why a lot of them will crack over time. 304 SS is just not durable enough at high temperatures and with thin tubing. This is why Killer Bee decided to make their header with thin 321 SS.


Full Race's 10 gauge 304 SS header is actually slightly thicker then typical SCH10 piping, I see SCH10 304SS turbo manifolds last a long time in the SR20 world.


Out of the bunch, of thin walled headers, Tomei sticks out the most with 2.5mm piping. You also don't hear of Tomei headers cracking like the rest of them do.


-- Please let me know if there are technical or grammatical errors in this post! --

05 LGT 16G 14psi 290whp/30mpg

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

Link to comment
Share on other sites

  • Replies 90
  • Created
  • Last Reply



This thread was born because I couldn't decide if going equal length was all that much better then aftermarket unequal length headers, especially since you loose the iconic rumble. Then to help me decide if 4-2-1 or 4-1 would be better for a street car.


EL vs UEL?

After much thought I personally decided on an equal length header, mainly because I wanted all of my cylinders to have consistent EGT's and I was willing to loose the rumble in order to have more reliability.


4-2-1 vs 4-1?

These articles are usually talking about full throttle operation, where the header is under pretty high pressure (2-3x intake manifold pressure), at part throttle there shouldn't be a lot of pressure in the header, thus I believe overall header design becomes important. 4-1 headers are generally more peaky, while 4-2-1 should have more midrange. For a daily driver I decided on a 4-2-1, but with diameter similar to OEM size. I feel like this is the best compromise between low end and top end power.



My requirement was for a thicker header, to avoid cracking. While Full Race is amazing, it was way out of my price range, thus I went with the runner up in thickness Tomei. And since it comes with an up pipe, that further sweetened the deal.



TO DO: Add some pictures of various headers.

05 LGT 16G 14psi 290whp/30mpg

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

Link to comment
Share on other sites

The Tomei header is the same as the kinugawa header. I've had both in my hands at the same time.


Don't put to much thought into the 4-2-1 vs 4-1. In theory they're different, but in every application I've looked at, there's not enough of a difference to matter. Sometimes the 4-1 outperforms the 4-2-1 across the entire RPM range.


Do of the other headers have the expansion joint in them? Engines do change shape as the heat up.

Link to comment
Share on other sites

The Tomei header is the same as the kinugawa header. I've had both in my hands at the same time.


Kinugawa is probably an SSI clone, which is what all other Tomei like headers are (which all EL's clones of the JDM headers)


Don't put to much thought into the 4-2-1 vs 4-1. In theory they're different, but in every application I've looked at, there's not enough of a difference to matter. Sometimes the 4-1 outperforms the 4-2-1 across the entire RPM range.


I think the reason the two designs are about the same is due to other contributing factors. We don't have a long primary 4-1 EL header (Killer B is short primary EL 4-1), I think that would be an interesting design to compare to the EL 4-2-1 design.



Do of the other headers have the expansion joint in them? Engines do change shape as the heat up.


Most of them don't actually, it doesn't look like even the Full Race one does anymore (found some old pictures where it did before).


I'm not a fan of slip joints that Tomei and clones have, but it's better then not having one. Optimally I would rather see high quality flex pipes.

05 LGT 16G 14psi 290whp/30mpg

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

Link to comment
Share on other sites

When the little gremlins get rid of the parts legacy in my garage then replace my clutch and break it in, so sometime next year :lol:



I got the Tomei header in today and it's really nice. It's actually only 16.4lbs, up pipe is 3.8lbs. Up pipe is a thick 3mm while header is around 2.5mm. I really wish I could install it sooner, but with a slipping clutch I can't do meaningful dyno runs :lol:

05 LGT 16G 14psi 290whp/30mpg

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

Link to comment
Share on other sites

Like many OEM manifolds Subaru's USDM OEM UEL turbo header is designed with a lot of compromises, it has to have universal fitment, has to last the life of the car, have decent performance, be easily mass producible, while being as cheap as possible. It's no surprise that aftermarket headers outperform stock headers.


One thing the OEM header really has going for it is, it has one of the shortest runners out of all headers, yet it's the heaviest coming in at 25lbs without the uppipe. Cast iron requires thickness to be durable.



Here is what the driver side and passenger side primaries look like.




As you can probably tell they extremely short, I measured them at 2.625" on driver side and 2" on passenger side. You can see the light bleeding into the other cylinder.




Primaries that short causes the exhaust gasses to bleed into the adjacent cylinder, to makes things worse the two adjacent cylinders fire sequentially (Right-Right-Left-Left, Firing order: 1-3-2-4). If the firing order was non-sequential (Right-Left-Right-Left 1-2-3-4), I don't think this header would be nearly as bad, you wouldn't be trying to cram two pulses into the same small diameter pipe.


Primary diameters are oval to say at best, this is probably cast's fault. I measured each cylinder to be between 38-39mm in diameter.



Like all cast manifolds there are packaging issues, to allow you to install it with decent sized bolts/nuts they usually will have a dimple that reduces the internal diameter of the pipe. Extreme case was cylinder 4 with 36.4mm diameter dimple



Cylinder 2 & 4's manifold outlet is at 38.9mm. I feel like this is very undersized for two sequential pulses, I think it should have been 45mm easily.



Now we are at the crossover pipe (x-pipe) for cylinder 2 & 4, Crossover pipes are taunted as the worst offenders for flow, after measuring everything I tend to agree, the biggest issue with this design comes form cheap manufacturing. Crossover pipe has the flex section welded on, from what I've read it has a lot of flow restrictions too. It's not wise not to have a flex section though, since aluminum heads and steel/cast iron headers will expand and different rates.



Crossover's inlet flange has a 44.7mm diameter, then there is about 5mm before the actual pipe begins, this in itself is not bad because it can help reduce reversion. Then finally the actual pipe is 40.7mm, which is presented as a step up thus creating a flow wall, if it was a smooth transition from the flange I wouldn't have as big of a problem with it.



Crossover's outlet flange has similar characteristics as the inlet. Flange measured in at 41.7mm and the actual pipe necked down to 39.4mm



At this point the crossover pipe merges with the second manifold from cylinder 1 & 3. The merge here is not bad too, it's a bigger 39.7mm, which is good.



Now we are at Cylinder 1 & 3, as you can see these runners are much shorter then 2 & 4's, here is where the unequalness of the OEM manifold comes from. Since 1 & 3 has this sharp 180* turn to take, it's the same as adding a few inches to the overall travel distance (gasses will slow down to the same speed as if they were to travel a longer pipe).



As you can see the O2 sensor is actually in a decent spot to receive mixtures from all 4 cylinders.


Finally the header outlet, it measured at 41.7mm. With sequential cylinder pairing I once again believe this should be much larger, 50mm minimum.



Another thing to keep in mind is how much space the O2 sensor uses. This reduces the effective diameter by a good amount.




From here the OEM Catted Uppipe is at 38.8mm at the inlet, since this steps up it creates a wall for the exhaust gasses to hit, not good.



Outlet for the OEM catted is at 41.4mm



STI/Catless Uppipe's inlet is much better with a ~43mm opening, since this is bigger then the header outlet the step down doesn't create a "flow wall". Unfortunately I don't have a Catless up pipe to measure the outlet with.



Finally here is the stock VF40 turbo inlet, like most other cast things it is oval too, I measured it to be between 44-45mm. I picked the smaller number to display, since your exhaust is as big as the smallest pipe.






















05 LGT 16G 14psi 290whp/30mpg

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

Link to comment
Share on other sites

Tomei Equal Length (EL) 193105 header is an overall well designed and engineered header. It sports ~12gauge 304 stainless tubing, with fairly minimal flow restrictions with overall excellent craftsmanship. The biggest drawback of this header is the uppipe flanges are on the thinner side.





My bathroom scale gave it 20.2lbs with up pipe, which is not far from advertised weight.



Header weight without the up pipe is 16.4lbs, this is so that it can be accurately compared to other up-pipeless headers.



The head to header inlet ports are pretty well port matched, though I probably will clean it up some.



Primaries measured at 38.9mm ID and 42.7mm OD in couple different spots




There are slip joints on 2 & 4 cylinder primaries. I was always curious on how they look/work, you can also see how thick the pipes are.




Since the primary to secondary collector is cast, it usually has to be thicker to have the same strength. There is a lot of room for error when dealing with different diameters. In this case secondaries are socket welded to the collector, this means one of the pipes has to be bigger then the other, this opens you up to more flow mistakes like pipe walls and turbulence zones. Tomei made the part of the collector where the secondary slips in be slightly bigger and then cast the collector to be the same inner diameter as the secondary pipes, which is good, but there is still room flow walls in this design. Tomei handled it pretty well for a mass produced item, there are only slight protrusions in both collectors, though I still plan on grinding these down.



Final merge collector cast is bigger then the secondaries that's why you can see the pipe ends. This is fine because it functions as a anti-reversion step.



One thing I'm not a fan of is the oxygen sensor location it could be a little more forward of the collector. In the EL header, it will make adjustments based on readings from only 2 cylinders (1 & 2). This is still better then the Tomei UEL header, which has the bung before the single merge collector, thus making adjustments based on only one cylinder.


Header outlet is 57.1mm



Uppipe inlet is very smooth and clean, is 57.1mm just like the header outlet.




Uppipe outlet is necked down to 52.4mm by the turbo flange. I'm a little torn on weather I like this or not. I don't like it because it is a very short neck down, but it could be worse and could be just a wall. Now, I like it because you can bore it out for bigger turbos.




Now if your going to run this with the stock turbo (VF40) keep in mind the turbo inlet is at 44mm, which would create a fairly big step up from Tomei's 52.4mm. Port matching the turbo would be beneficial to reducing flow restrictions.



I wish Tomei didn't include the useless EGT gauge port, plus the bung does protrude in.



Flanges thickness

Head flanges are 12mm around the gasket and 11.3mm around the edges. Header to Uppipe flange is on a thinner side at 10.4mm around the gasket, is thinner around the edges though.



Uppipe to header is also 10.3mm thick, same as header side. Uppipe to turbo is also 10.3mm where the gasket sits and 6.3mm at thinnest points (where supports are bolted). Once again I wish these were all a little thicker, 12mm, to avoid warping overtime.





















05 LGT 16G 14psi 290whp/30mpg

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

Link to comment
Share on other sites

The one thing that is hard to explain to people when you change from unequal length headers to equal length headers is how much smoother the engine feels. It was like a night and day change.


You should get your hands on a set of oem GD Twinscroll headers , you will be impressed with how good they are as Tomei barely gained any power by switching to their header in the GD EJ207.


Also the sound is quite wonderful

[ame=http://www.youtube.com/watch?v=AzxTSp6SnEs]S204 EJ207 Highway pull - YouTube[/ame]

Link to comment
Share on other sites

I used to run a log manifold on my SR20VE, it sounded great (similar to the rumble actually), and ran great on low cams (0* overlap). But when I would engage the bigger cams it would actually loose power (~40* of overlap). Basically it would have major reversion. I then switched to an oem cast manifold (I doubt it was equal length from factory), and it was night and day difference too, much less detonation on high cams too.


I wanted to try out JDM EL headers, but I couldn't find them for less then $1k (they did have other goodies though). If I could score one for $300 I would have been all over it and would have made a custom up pipe. But I got the Tomei's on sale, on black friday so I'm happy.

05 LGT 16G 14psi 290whp/30mpg

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

Link to comment
Share on other sites

Here's some measurements from my JDM LGT twinscroll header/UP. I had taken some measurements of my JDM STI twinscroll header/UP but did not take pics, as it matched the numbers reported on the Tomei site.


Just measured. The ports for the block/header flange of are 40mm, but immediately neck down to roughly 32mm as Unclemat suggested. It was too deep for the jaws of my caliper so I couldn't measure exactly, but 32 seems really close. The secondaries have 34mm ports.


The ports on the UP are all 35mm. This matches up nicely with the inlet ports on the turbo, shown on the previous page.






Link to comment
Share on other sites

That's really small overall diameters, probably because it's meant for a 2.0L which would have smaller exhaust pulses. It's probably also fine with running smaller piping because pulses are properly timed (because EL) and doesn't try to cramp two pulses into the same pipe.


What metal is it made out of? Doesn't look like cast iron.

05 LGT 16G 14psi 290whp/30mpg

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

Link to comment
Share on other sites

I didn't see STI data on tomei's page just the stock numbers (wrx?) which are the same same for big header and smaller one.


As I mentioned in my build thread, based on this picture looks like JDM header is stainless. But the pipe bends are horrible and the merge collector is mediocre. Overall Tomei is a much better designed header even if you go with the small diameter one.



Do you by any chance have the outer diameter for JDM header? Tomei says that it's 1.5mm (~16gauge). If OEM Subaru is only 1.5mm thick I wonder why they are not cracking?


Also looks like Tomei JDM header is only 2.0mm, while USDM single scroll is 2.5mm (I'll take it thicker any day).

05 LGT 16G 14psi 290whp/30mpg

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

Link to comment
Share on other sites

The info listed is for the GDB STI twinscroll header. The listed weight for the stock header changed on the large pipe page, but I can't account for that.


STOCK JDM STI ------ 9.0mm 4-2-2 Φ42.7 Φ42.7 Φ42.7 1.5mm 12kg/13.1kg

TOMEI #193081 SS304 8.0mm 4-2-2 Φ38.0 Φ42.7 Φ42.7 2.0mm 8.5kg

TOMEI #414001 SS304 8.0mm 4-2-2 Φ42.7 Φ45.0 Φ45.0 2.0mm 9.7kg


Agreed that thicker is preferred, but keep in mind that the OEM stuff comes with insulation and heat shielding for retention.


At one point Tomei had released a technical report regarding their development of the twinscroll manifold. Essentially they pointed out that while the bends and collector could be improved, it yielded little improvement due to other systemic restrictions. I'll keep digging and see if I can find it.

Link to comment
Share on other sites

Stay stock stay happy. Just get a stage 1 tune to smooth everything out and give you a little more torque. Don't change any hard parts. Don't floor it, stay out of boost. The XT / GT is the wrong car to care about fuel economy.


Once you start changing out hard parts there is no general consensus. Sure switching out the downpipe will give you more torque, but drivability is the same. Fuel economy is worse (because it's fun to floor it).


Depending on who you talk to there is no reason to change out your header until you get to a larger than stock turbo.


If you read covertrussian's thread, you can see he made changes to the ECU map and his turbo (to delay boost) to help him with fuel economy, but how long will it take to recover those costs? Probably a couple years.


Your best mod is your own feet.

Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

  • Create New...

Important Information

Terms of Use