Variable geometry turbo "VGT" implementation on an sti.


New member
I'm starting this tread to compile information to generate knowledge and understanding of the requirements to implement a VGT turbo setup on an sti or a EJ25 equiped subaru for that matter. Ill start with a brief explanation.

Brief Explanation:
A variable geometry or variable nozzle turbo is equipped with a special exhaust housing that allows change in the volumetric size of the turbine area threw an actuating mechanism that is proprietary to each turbo manufacturer. The system works by moving a set of vanes that will enlarge or contract, open or close, to allow a change in the exhaust gas flow thus affecting turbine impeller velocity. This is the main way to control boost and turbo spool times in this turbo chargers.

As some of you might have noticed, vgt is only used on the most rare of oem setups. Take for example the very exotic Porsche vgt turbo using materials that are state of the art. You will mostly find Vgt setups on Diesel operated engines. This turbos offer the capability to become an effective engine brake when couple with other parts in the exhaust system. For reference on engine brakes please fallow this link :>

Design issues:
If you have connections and deep pockets you might be able to find a Porsche vgt turbo to work with your setup. Sadly the rest of us wont. This will leave you with 2 options: Use a garret vnt turbo or use a diesel engine vgt turbo. As any one that works with a diesel engine will tell you, Egt vary from one fuel source to an other. Diesels for the most time burn at colder egt than a gas operated engine. This means that the structural and operating components in the turbo will be subject to grater thermal stress. To put it simple, something can or will melt if Egt get out of control on a normal diesel designed turbo. On the garret side of things, their vane control is flawed. Their exists to much slack in the vanes to allow them to move when the materials undergo thermal expansion, this in turns allows build up of carbon deposits on an area that cannot be self cleaned. What happens is that the vanes slowly begin to get stuck, regular inspection of the vanes is required. The system uses a mechanical actuator that leaves some users wanting more control.

Introducing the Holset Vgt:
The holset turbocharger model HE351 uses a different Technic to control and actuate the vanes. The holset only uses about 3 points of movement to control the vane actuation. This system is more robust than any other in the market at the current time. From here on after this thread will concentrate on this turbocharger. If any other comes to my knowledge I will make changes accordingly, if it can be implemented.


Here is a cut out of the turbo "notice that this turbo presents a pneumatic actuator for the vgt vanes, the HE351 uses an electronic actuator":


The HE351 was designed above and beyond the capabilities of a normal ranges of operation for a diesel engine turbo. It's complete construction relays on the fact that its going to see high load scenarios for sustained periods of time. Holset claims that on the HE line of turbos, they used cast aluminum compressor wheels and a inconel alloy for the turbine. I my self will confirm this on a future time when I'm able to use a metal analyzer to see what alloy comes up in the scanner. As of now materials remain unconfirmed by the end user.

The turbo impeller shaft takes a different approach to the conventional design. The shaft is larger towards the exhaust side, this design offers a more robust design on critical control points. No other turbo has this type of design, it has proven it self on the holset HX and HE family of turbochargers.

This turbo is water and oil cooled, and the 2 can be used on an STI. I plan on implementing them to control turbine temperatures. If temperatures become a persistent problem, a small cooling unit can be attached at the inlet line. I don't think that this will be a problem but I'm prepared to make the adjustments.

The turbo vgt vanes are actuated by an electronic microprocessor that controls a brush less bipolar motor that is equipped with hall effect sensors to measure the motors shaft current position and thus relaying information of the vanes position to the processor. The turbo also has a bigger Speed sensor "also hall effect" that measures shaft speed. This is used by the ecu to control turbo speed by opening or closing of the vanes. The controller itself has enough electronics inside of it to act independently of any vehicle ecu, it just needs the inputs of some reference sensors to work.

The problem is that the controller uses Holset by Cummings proprietary commands that some have tried to crack and or talk to the unit with out success.

This rises the issue of finding a method to control the vgt vanes.

Vane control:
The biggest issue with using this turbo is to find a way to control the vgt system successfully. This will be a problem depending on the level of control that you require/want. My research has lead me to varius methods of control that have the up's and downs.

Current ideas and methods of control:
  • Wastegate actuator- There are setups out there that strip away the electronic controller to expose the vgt actuator shaft. Fabricating a bracket, you install a wastegate actuator with the psi that you want the vanes to react to. Mot people use an actuator that will start its movement at about 12 to 14 psi. This is to allow the turbo to enter a fast spooling stage and make the transition to a slower spool up top. The draw backs of using this method is that the system will be static, you will not be able to make any adjustments based on electronic inputs. The switch point will always be the same. It will be predictable but it lacks control options.

  • Electronically controlled motor: By far this would be my preferred method of vane control. There are many setups that I can think off to control this, I am only limited by my ic programing knowledge "0" I can build a circuit using a schematic and have basic electronics knowledge. Still if you can source a pair of helping hands this can be an easier task.

    The HE351 comes equipped with an array of hall effect sensors to monitor the brushless motor, this is the main way to reference the vanes position on the stock setup. You could use a variable potentiometer attached to the motor shaft and use that to reference the motor position if you are unable to tap and read the sensors. You can actuate the motor by using the input signal from a map sensor, tps, maf etc. You could program the vane position to open gradually in response to this signal. This simple setup would work by adjusting the motor constantly depending on the potentiometer or hall sensors. Problems whit this simple setup, if the vanes get pushed back by the back pressure in the exhaust system your motor adjustment will be wrong to begin with. The Ic will see the current vane position but if you don't implement a correction logic in the code, it simply wont work. This means that for an electronic controller to work you will need to implement a PID correction method to be able to keep the vanes in check.

    This is easier said than done and requires extensive code programing and knowledge. If you have someone with the capability I have an even better sensor data reference input for vane control, the hall sensor monitoring the the turbine shaft speed. It is theorized that the most efficient impeller speed for the HE351 is about 111,400 rpm. If you can monitor the shaft speed with the hall effect sensor you can program your vane control method to keep the turbine speeds in this area, staying at the efficiency islands of the compressor map. The HE is capable of making 50psi of boost and still be in an efficient range of the compressor map.

Turbo Size:

The HE351 is not for the faint of heart. This turbine is capable of supporting 600+whp on a subaru with out braking a sweet. It has a 63lb/min compressor capability. The exhaust side sports an ar of .35 fully closed to about 2.01 fully open. This is about 4cm^2 to about 25cm^2.

A turbo of this size would be spooling at about 5000 rpm on a normal setup. On an inline 2.0 engine this turbo is capable of making 15 psi at about 3,000rpm and full boost at about 4,000rpm. It spools considerably fast thanks to the vgt. This means that you could have that fast spooling, top end monster that most dream off. This is what I'm aiming for.

Here is a the only compressor map that I have found available on the web:

I'm going to try the next big thing, contacting Holset to see if they are willing to publish or confirm this is the real compressor map.

Keeping up with updates.
Last edited:


New member
Here a side cut of the Vane actuation mechanisms.

Here you can see the difference in moving parts from this turbo to a garret vnt. Notice that the garret moves all the vanes independently of each other, meaning all of the vanes have a linkage system to the main movement mechanism. Holset moves a ring where the vanes are fixed in place.



Adaptation and installation:

Given the gigantic size of this turbo, you will have to fight with it to install it. The exhaust flanges are T3 for the inlet side and 3" v-band for the exhaust. The turbo it self weights about 45 pounds or 50 with the stock tail housing. This is a very heavy turbo, you will need to fabricate a support bracket to the base of the turbo. So much unsupported weight will put stress on all of the exhaust tubing welds and will eventually brake something with the added vibrations if it is unsupported.

You have 2 ways to adapt this turbo to an sti. You can either make an up-pipe to mate to your header and make it long enough to place the turbo at the rotated position or you could fabricate an up-pipe adapter to rotate the turbo to the side and lift it about 3 or 4 inches.

Heres a Picture of the up-pipe adapter that needs to be fabricated to adapt the turbo to the stock location up-pipe: "This picture is not mine, Evil06sti started a holset project but after a long time he stopped updating it because of some life problems. Still mad props to him to venture off into the unknown and doing an excellent job!"


Notice that the pipe has a bend to place the turbo a little further from the block and the the flange is rotated to clear the strut tower and intake manifold. I prefer this method because in case something goes wrong you may place the stock location turbo back in there. Also look at the added support bracket, this is an important part of this setup. In case that you are wandering, greddy also uses a similar method of turbo placement for their t67 turbo kit but the parts are different.

Here is th holset placement in the engine bay, if you get it right:
Notice that the compressor housing outlet is parallel with the inlet, which means that if you have an txs or ssac intercooler setup, the piping will not be a problem for you.


As you see, the turbo outlet mates almost directly to the intercooler pipe. I'm keeping this compressor housing just for this reason. I will however make anti surge ports and a port and polish job on the compressor housing.

On the exhaust side I will use ceramic coating to minimize the radiant heat transfer into the hood. The clearance is about 2 inches from the hood, if you don't do anything to control the temperature it will probably do some paint damage. After the ceramic coating I will also, eventually use a turbo blanket tailor made for this turbo.

The downpipe will need to be costume made, the bend coming out of the turbo is somewhat steep but its the only way to get it to fit right with out touching any other components in the engine bay. I will also do ceramic on the downpipe.
Last edited:


New member
Boost control with a VGT setup.[/B

This in my opinion is a delicate topic to discus on this type of turbos.

Lets start by theory. Variable vanes means that you can enlarge the exhaust housing to the point that the exhaust flow isn't enough to sustain boost or to continue the spool up. When the vanes are fully open, the space is enough for the gas to flow over and transfer less energy. This means that you could negate the use of a wastegate to control boost.

This is in paper, In real life you need a really fast response from a controller to achieve this, a 3D map reference to either rpm and tps or Vehicle Speed Sensor. It would use the inputs to manage boost like a solenoid does with a ecu. In fact this has given me an idea that ill go further into briefly. Still this would be extremely difficult to achieve and in the case of failure you would need a mechanism that would make the housing fail in the fully open position to avoid problems if the housing remains fully closed.

Side note: As Fujiwara stated, a closed exhaust housing will increase back pressure but a closed holset could turn it self into a partial engine brake... this will be bad as the egt will rise and melt something off.

Alternative to an external vane control system "using a programmable Ems":

In my case or any Engine management system that has a configurable boost control system "Aem, Hydra, link etc" you can control boost with a PWM output. In the case of the Aem Ems "my current engine management" you can choose the PWM frequency of operation from a wide range of selections. If I'm understanding this, I can use the PWM signal that drives the boost control solenoid to drive a servo motor with the % of duty cycle. I'm still a little confused, I don't know if the PWM duration interval changes with %duty cycle. If the duration does not change then this method of control is dead to start whit. Sad thing is that I don't have an oscilloscope to analyze the duration of the pulse.

If indeed the duration of the pulse changes then I can tune the vanes opening angle to the rpm & tps references. This still isn't optimal but it is a huge improvement over controlling it with a wategate actuator.
Last edited:


New member
Sent an Email to holset asking for a confirmed compressor map and inquiring about failure temp ratings. Lets see if they replay.


New member
Also, since this is a generic VGT implementation thread, I'll dump some more stuff off the top of my head that I've found out.

-The turbo used by porsche in the 2007+ 911 TT 997 is known as the BV50
-The mass flow of the BV50 is less than a 3076
-the BV50 is not on the public market as of yet, but is on the Borg Warner catalog for this upcoming year of products. The price will be between $4-5k for just one turbo, not both.

Wow for 5k I can retrofit billet blades into a holset and modify it to hold the heat lol... That is steep, yet again materials are crazy in that turbo.

I'm going to cover the exhaust components later tomorrow. Still I didn't know the exact size for the v-band flange on the downpipe side. Nice to know since I'm ordering the flanges soon.


New member
Ems pwm control

Modified the second post to include more info on vane control and other stuff.

Fuji, care to elaborate on my idea of vane control using pwm from the boost control solenoid to move a dc servo? I'm really exited to know if this will work since in my case I can work with the ems. For stock ecu ppl, this is another story since the stock ecu can't change the pwm frequency at the solenoid.


New member
I Need to revise the flanges. For some reason various ebay vendors say that the inlet flange is not t3 or t4 but a hybrid of some sorts. I will need to wait to have the turbo on hand to see this for my self.


New member

Found this simple 555 circuit. I don't understand if the control voltage means that if I input from 0 to 10 volts it will output the pwm signal from 1 to 2 ms or so. That is what it want.

If I don't manage to do this Ill have to go with the archaic solution of the wastegate actuator. I don't really want it to be permanent but will test it to see the transient response difference from wastegate to electrical control.

I'm satisfied with with it operating based on boost reference.


System Operator
Staff member
Seems like alot of work for minimal benefit compared to some of the new techs out there. Like the EFR turbos from full-race.


New member
The only hard part here is adapting a control method which isn't that hard for someone with the right knowledge. After the control scheme is dealt with, we can have a vgt standard setup on a sti with less than 1200 to invest on turbo wise. I bet that the EFR turbos are not going to be below 3,000. Sure for a full blow race car, heck Ill go with them and after reading they sound amazing. Still I'm into this more as a passion than to go with what everyone else does. No offence to any one but I don't want to be the zombie that goes out and buys everything because its ready made to work. I like to work with the kinks in things, the feeling of fabricating something with your own hands and seeing it work is just amazing... religious if you might say.

That said I welcome all points of view in this thread, as it also helps analyze the pros and cons of this project.


System Operator
Staff member $2500 list for the 550 hp turbo.

I love the VGT concept it's an old one (newer to gas engines), it's smart but the packaging/quirks can create problems. Now if you were to make a VGT that integrated alot of the "extras" sorta like the EFR turbos do you could have a very sweet setup.

I admire the desire to fabricate.

I guess it really comes down to results on EFR vs VGT, if we can see huge gains in both spool and top end it maybe worth it. Have you thought of doing a bottom mount turbo like the new Legacy runs? It may help with some of the packaging issues and would reduce exhaust piping.

I really like the simplicity of the EFRs which is one of my biggest gripes with VGT (complexity). I will be following this thread to watch the progress. :)


New member
this is all very interesting. I would have to agree that simplicity probably wins over complexity in my book.


New member
I just checked that link and it says to call for price.

Still I think that I can implement a billet compressor blade in the holset. If you see the comparisons of the turbo with a evo16g you will notice that blade geometry is similar, wich means that we could yam a better performing wheel in there if it comes to that. We will see soon enough. After this is done, It will look simple with a set of instructions and parts list for fitments.


New member
With a Flux capacitor installed, anything is possible. Lol

Fuji hows your research on amplification circuits going? Need some help? I think that I can source a bit of guidance from my electrical engineering department at my college.

Side note, the compressor housing seams quite big enough to support a daily driver of 500whp. If you want to go further than this then sure have a whack at it. Maybe the ppl from blouch can set something up. I can source the info and see if they can do it. "Billet blade swaps"


New member

I need to convert this circuit to output a pwm signal based from 0 to 5 volts. Looking to have various methods of control to test out. I like the 555 timer circuit since it doesn't require any programing or expensive parts, just to have it setup right. I want to reference this one to a boost sensor signal. The circuits that I found reference themselves to 0-10v and that is just to much for this application.


New member
I was looking for servo options. I'm still thinking about torque capacity. I have found hobby servos capable of 15 pounds of torque but I think that the back pressure on the turbo will make much more than that, even if the vector wont be applied towards the vanes it self it will put some force, pushing back into the vanes current position and making the servo correct it self. I have seen back pressures of 35 psi on diesel engines running about 10 psi boost on the lab.

I liked the Homemadeturbo setup because it uses the stock motor which has the needed torque to control the vanes but in my case, i cant figure out how to program the board or the code itself. I liked the 555 timer idea because it requires no code, but then you need to swap the motor unit for a servo that will hold up temp and have the torque to operate it.


New member
Well Holset replaid to my email.

"Unfortunately we are unable to provide efficiency islands as these are for internal use only, also our turbochargers are designed for the commercial diesel sector and therefore my not meet the requirements of a petrol (gas) application."

Weird that they consider a compressor map as delicate "internal information".


New member
Well, I just had a blow to my wallet thanks to my pos stock radiator. The top cover blew up causing my car to over heat, luckily I saw it on time and shut off the engine before any damage was done. Still I needed to get the car up and running in the same day and no shop had an aluminum radiator in stock... so I had to buy a stock unit for 230.00... so there goes my savings for the holset. Lets see if I can get them back by mid December. I think I need to hit the casino soon...

On other news, I'm still stumped on the electronic control methods. Fuji, hope that you have better success than me.