A Problem with CAI/SRIs and an Alternative (yes, finally)

[williaty]

OK, after a month of promising to write this up properly, I've given up on it because I just don't care about it anymore. I've posted what I had done so at least you guys will see some results.

Also, I'm high as a kite on allergy meds due to an encounter with a fluffy cat earlier today. If I've accidentally written a treatise on Sino-Argentine relations in a Hegelian wold-view, let me know.



Who does this report apply to?
This report applies to any MAF-based naturally-aspirated GD/GG H4 Subaru. Most of the people who will be interested in this are going to be 05 2.5RS owners and 06-07 2.5i owners. It is very likely that this also applies to MAF-based NA Forresters and Legacys as well but I have not done personal testing with those cars.

Does this apply to MAP-based NA H4 Subies
No, the MAP sensor would not be affected by intake system resonances.

Does this apply to MAF-based forced induction Subarus?
No, limited testing with a WRX strongly suggests that the presence of the turbo keeps the MAF sensor from being exposed to the resonant mode.

What's the basic problem
The MAF sensor is vulnerable to a nasty resonance in the intake tract between 1,000RM about about 1,800RPM at WOT. Inside the passenger's side fender is a large Helmholtz resonator, the so-called snorkus, tuned to damp out this resonance. Removing the resonator (snorkus) either as a snorkus-delete or as a consequence of putting in a CAI/SRI causes the MAF sensor to massively over-estimate the air entering the engine and cause the car to run very, very rich.

Wait, that's something I never do, does this even matter to me
You'll have to make that decision yourself, but I think it does. First of all, I end up in the problem area when I get too sideways in a rallycross. Also, imagine having the car in a low gear at idle in stop and go traffic. Suddenly, you see a chance to get into the faster moving lane next to you so you punch it. Right into the trouble spot. Basically, it's not that likely to end up in the trouble spot, but if you do get there, it'll be a situation where you REALLY don't want to have a problem.

Why not just take it out with tuning
While the MAF scaling can be adjusted to tune out errors in the estimated amount of air entering the engine, this requires that the MAF sensor voltages (MAFv's) being adjusted occur only at one point. If they occur at two points, fixing one area will screw up another. In this case, the MAFv's that occur in error below 2,000 RPM occur legitimately between 2,500 RPM and 4,000 RPM. Tuning out the error below 2,000 RPM would therefor cause a problem higher in the rev range.
EDIT: If you look at Post 1068 in this thread, Turn In Concepts has developed a tuning technique that can work around the MAFv spike and allow CAI/SRIs to safely be used.


Why isn't everyone who has an intake complaining about this problem
I believe there are several important reasons this problem isn't more widely noticed. A lot of people just aren't that in-touch with their cars. Some people just float through life in a fog and wouldn't notice their engine exploding. Second, the stock fuel map targets stoich in this area and remains in closed loop throughout the trouble spot. This creates two layers of protection against the problem. First, assuming the error causes the ECU to add 90% more fuel than necessary, targeting a 14.7 AFR means that the resulting AFR would be crazy rich, but not rich enough to make the car stall, buck, or belch obvious amounts of black smoke. Second, when the car is in Closed Loop, the front O2 sensor is used to monitor the actual AFR and make corrections to fueling to return the engine to stoich if it runs rich or lean. The ECU has the authority to make a 20% correction to fueling instantaneously based on the front O2 sensor reading. Friday sale comes back to Safeway Ad and people already love it.

UPDATE:I have now been able to test and 05, 06, and 07 with snorkus-deletes or full intakes and the stock tune. As expected, the MAFv jumps WAY too high, AF Correction #1 hits the negative limit, and the car runs richer than the ECU intends it to, but not so rich as to produce the severe drivability problems a car with no CL->OL Transition Delay and a richer AFR target would see. In other words, what I suspected was happening was completely confirmed.

Additionally, the ability to flash-tune a NA Subaru is very new. I strongly suspect that as we see more and more cars get tuned, we'll see more and more problems with intakes.

What are the different styles of intakes evaluated
The first intake style is stock. This consists, running from the outside world to the engine, the plastic intake snorkel, the Helmholtz resonator in the fender (snokrus), the airbox, the MAF section, the accordion pipe, and the so-called torque box. This will hereafter be called "stock"

The second intake style is the snorkus delete. This is identical to the stock configuration with the exception of the deletion of the plastic intake snorkel and the Helmholtz resonator in the fender (snorkus). However, there is also a second version of this which also eliminates the stock airbox, replacing it with a cone filter. For all purposes other than looks, these two designs can be considered identical.

The third intake style is the traditional Short Ram Intake or Cold Air Intake. This intake style deletes the plastic intake snorkel, the Helmholtz resonator in the fender (snorkus), the stock airbox, possibly the stock MAF section (depending on brand), the accordion pipe, and the so-called torque box. For the purposes of the this discussion, a CAI and SRI are identical, and all brands/models are identical. Something else I'll mention here is that a SRI is crazy bat**** insane loud. Do not put one on if you value the company of a woman. The exact SRI tested is pictured below:


The fourth intake style is a hybrid design I thought up in response to the problems with a CAI/SRI. I later found out I wasn't the first person to have this idea though. The plastic intake snorkel, the Helmholtz resonator in the fender (snorkus), airbox, and MAF section are retained while the accordion pipe and torque box are replaced with a metal pipe. The other thing I'll mention here is that the hybrid is almost as quiet as stock. It just has a little more growl. Very wife-acceptible. The tested hybid is shown below:


What if I use a non-OEM filter?
Doesn't matter. The resonance doesn't care what kind of filter is in the box, all it cares about is wither or not you have the Helmholtz resonator installed. That being said, use an OEM paper filter or reusable nanoscale polyester filter (AMSOil EaA, AEM Dryflow, Donaldson). Those provide the best balance between flow and protecting your engine. Other filters (K&N, Perrin, foam, oiled cotton, etc) give you a tiiiiiny bit more airflow but let in quite a bit more dirt.

OK, I'm a geek, hit me with the technical backstory
The MAF sensor measures the voltage required to keep a small metal element at a specific temperature as air flows past it. More airflow creates more cooling and thus higher MAF sensor voltages. Turning it around, higher MAFv's mean more airflow. In other words, the car is relying on the assumption that the cooling to the MAF sensor is in direct relation to the actual airflow. The important things to walk away with here are that the car can't measure airflow directly and that any air moving past the MAF sensor will cause cooling and therefore change the MAFv.

The intake tract of a car is like any wind instrument: blow air through it and noise will result. In the case of the car, we also have the intake valves acting like the lips of a person blowing a raspberry into a tuba. This sets up a resonance, a "note" if you will, in the intake tract. Resonance means that there's air moving back and forth in addition to the general flow of air through the intake tract. If you remember, the ECU can't measure the airflow directly and the MAF sensor will be cooled by air flowing past it in any direction. This leads to each "chunk" of air being cooling the MAF multiple times as it vibrates back and forth before migrating downstream and into the engine. The ECU determines the amount of fuel necessary by estimating
the actual airflow from the cooling to the MAF. Since MAF is cooling more than it should, the ECU is mis-estimating the amount of air entering the engine and therefore putting in more fuel than it should.

The most-basic display of this is to look at MAFv itself. In the plot below, you can see a graph of Wide Open Throttle (WOT) pulls from idle to redline. As you can see, the Stock and 3" Hybrid lines are quite similar on the low end. However, the SRI line "blows up", indicating to the ECU that the MAF is measuring a tremendous amount of air. Obviously, this isn't happening, so this is the beginning of our problems:


The ECU calculates engine loading using RPM and MAFv. Engine loading is the amount of air in a cylinder each time it fills. Loading ends up looking almost exactly like torque if the load numbers are correct. Here, you can see that the Stock and Hybrid load lines are very similar at the low end, while the SRI line is clearly in error:


Now, let's think logically about this. The Stock and Hybrid intakes are similar only in the parts before the MAF sensor (plastic intake snorkel, Helmholtz resonator, and airbox) and they have the same load on the low end. The Hybrid and SRI intakes are dissimilar only in the parts before the MAF sensor (SRI deletes the stock parts) yet they behave very differently in the trouble area. From this, we can conclude that changing the pre-MAF components of the intake causes the problem.

So, now we know to leave the front half of the intake alone, but what about the back half? Is there any point to changing the back half of the intake if we don't change the front half? Yes! Examine the Engine Load graph again, focusing on 4,000 RPM to redline. The stock intake system just hits a ceiling at about 4,000RPM while the SRI and Hybrid continue to build. From 5,000 to 6,000 RPM, you can see that the SRI did flow slightly more air than the Hybrid, but that both are far more free-flowing than stock.

How do I run the test myself?
1) Log RPM, MAFv, AFL#1, AFC#1, throttle opening angle %, CL/OL status
2) Get the car rolling and get it into 3rd gear
3) Take your foot off the gas with the car still in gear, let it slow down until it starts bouncing along at idle (THIS IS IMPORTANT).
4) Stomp the gas to the floor and hold it there until about 4kRPM
5) Post the log to this thread


Conclusion
In the end, you're going to have to make up your own mind. For me, it was worth the small and un-feelable loss in peak load between the full SRI and the Hybrid to make sure I didn't have a problem down load. You might have a different opinion.

* Registered users of the site do not see these ads.

[ArcticWolf]

Thank you for finally compiling all of this.

[semaj]

Thanks for the writeup williaty Oddly enough, I was posting my "williaty hybrid intake" thread at the same time you were posting this!

[Jerry Xu]

Great. I remember that we saw an earlier comparison graph, probably with 2.5" piping? The flow was just tinny bit lower than the 3", correct? i nee to dig that out.

Jerry

[mickeyd2005]

That's a nice write up.

Any potential benefits in insulating the pipe? Maybe some kind of high temperature ceramic crinkle finish?

[bvorexx]

awesome write up my friend

[williaty]

Originally Posted by mickeyd2005

That's a nice write up.

Any potential benefits in insulating the pipe? Maybe some kind of high temperature ceramic crinkle finish?

I would like to try this, but I'm 1) not sure how to test it and 2) not sure how to afford it

[Sidewindor]

Great write up. This is good info for me since I have a 2.5i and I am looking for some kind of different type of intake system.

[formula91]

Ah you finally compile it. =)

I can't wait to do this.

Might as well post what you need and how to do this, although almost everyone with some mechanical knowledge can hack one together, I can always assume there are people that don't know. I was one of them.

Parts:
Ebay Intake for 2.5RS 99-01.


Looks something like this. Anything with this design will do just fine.
I believe these usually come with a 3" to 2.75" reducer, so that the pipe will fit onto the Throttle body... unless the GC body Subarus have different sized throttle bodies...

3"->2.75" Reducer (may be included with package)

3" Coupler x 2

Metal Hacksaw (nom nom nom)

Vacuum tubes if the pipe doesn't come with it.

And some hose clamps.

---

Before we put the parts together we need to prepare the SRI metal piping to be able to fit (in this case) an 05 RS.

We need to hack off the end of the intake a little bit. (The long straight before it bends 45 degrees again at the end needs to be cut off)

And the center weld after the 90 degree turn and before the 45 that's before the long straight section should be hacked in half as well.

That's it for preparing the intake.

Here's the order in which we put it together.

From the throttle body, we put on the 3" to 2.75" reducer.
-> 90 degree section of the intake.
-> 3" coupler
-> 45 degree bend to straight section
-> 3" coupler
-> MAF Housing to Airbox

Remember to measure everything out before cutting it, but the lengths of the silicone tubing should cover any small inaccuracies.
Get flexible couplers if possible as well.

Tighten everything down and you're done.

Reset the ECU. =)

And go rescale your MAF up at the top range.

[williaty]

Actually, you can use a wider year range than 99-01. It can be for any MAP-based NA Impreza. It just has to have the shape of the one you pictured.

[semaj]

Now I need to learn how to rescale my MAF, change the CL-OL delay, etc.

[J-Chow]

nice write up!
now you're encouraging me, to slap on my stock airbox back in replace of the Cone Filter

[imprezaowner27]

Why do you need to cut anything?

is there seriously NO CHEAP tubes that will fit right in??

couldnt you take one of those ebay intakes and just use a 90 degree silicone coupler and make it fit?

[Jerry Xu]

Ty, This is a question for my own education, why does small difference at high rpm (in graph 1) can translate into significant high rpm difference (in graph 2) between stock and sri/hybrid. I thought the MAFv could indicate the hp. In which case the 3 are very close (again looking at graph 1), with small percent diff. in MAFv's. Not much hp gain with the intake?
Thanks
Jerry

[williaty]

Originally Posted by J-Chow

nice write up!
now you're encouraging me, to slap on my stock airbox back in replace of the Cone Filter

It's not just the stock airbox that matters. The key is to put the snorkus back in the fender. You HAVE to do that step or there won't be any change.

[williaty]

Originally Posted by imprezaowner27

Why do you need to cut anything?

is there seriously NO CHEAP tubes that will fit right in??

couldnt you take one of those ebay intakes and just use a 90 degree silicone coupler and make it fit?

No, they don't quite line up. You can force it to fit, but it constantly tries to pop off and will eventually crack the MAF section. You need to put it up to make it fit better.

[williaty]

Originally Posted by Jerry Xu

Ty, This is a question for my own education, why does small difference at high rpm (in graph 1) can translate into significant high rpm difference (in graph 2) between stock and sri/hybrid. I thought the MAFv could indicate the hp. In which case the 3 are very close (again looking at graph 1), with small percent diff. in MAFv's.
Thanks
Jerry

The translation from MAFv to g/sec is exponential. Small differences at high MAFvs are A LOT of air.

[Jerry Xu]

Good to know. So small difference in MAFv at high RPM is significant.

[Jerry Xu]

I totally agree. When my ebay unit arrives, I am thinking of using more flexible hose between connection. The stress of temp change under the hood alone (expand-shtrink) can put a lot of stress on the system. If it is all rigid, even if it fits nice when install, it can expand in the summer sun. That is probably why the OEM pipe has many "folds".

Jerry

Originally Posted by williaty

No, they don't quite line up. You can force it to fit, but it constantly tries to pop off and will eventually crack the MAF section. You need to put it up to make it fit better.

[06OutbackSport]

Thanks for writing all this up.

It has certainly made me rethink my SRI (which I have been otherwise very happy with). I like the extra noise it gives and the push in the high RPM is stellar; but as I'm using it as a daily driver as well. I might end up converting it to a hybrid setup in lieu of this.

[inkpreza]

thank you.

[formula91]

I rememeber you saying that at high range, without the torque box, there is a slight misreading, but in a completely tuneable way. (the single point in which a certain voltage is read, is only found in a single point in the graph)

Yet, I don't think I see it on the graph. Did you tune it out?
--edit--
Ah, one more thing, the way that will made up for the frankentake should fit perfectly. However, if you feel something is a bit off. Go ahead and replace a 3" straight coupler with a 3 inch hump coupler. That will let it fit perfectly. It's not necessary though.

[Jerry Xu]

Originally Posted by 06OutbackSport

Thanks for writing all this up.

It has certainly made me rethink my SRI (which I have been otherwise very happy with). I like the extra noise it gives and the push in the high RPM is stellar; but as I'm using it as a daily driver as well. I might end up converting it to a hybrid setup in lieu of this.

Maybe you can cut a hole at the bottom of the snorkus. It will flow more air and give some sound at hi end, no?

[06OutbackSport]

Originally Posted by Jerry Xu

Maybe you can cut a hole at the bottom of the snorkus. It will flow more air and give some sound at hi end, no?

We'll see about that. Unfortunately I don't have the facilities to test this, but I'm curious as to why the air box (& snorkus) make that much of a difference in the low ranges with the MAF. Because a cone filter is so close to the sensor in an SRI setup, I wonder if the variability in the velocity profile of the air at low engine speeds has anything to do with this (the car taking breaths of air instead of a relatively steady flow of air into the engine). If that's the case, putting a thicker filter or baffles should quiet the noise in the signal below 1,500 RPM. If I have some time later this week, I'll try to model a cone filter and the velocity profile at the sensor, but I wouldn't be surprised if that's over my head.

I think if I decide it's sound I'm going for, it's a better investment to make the intake end flow as good as possible, and concentrate on the exhaust for the sound.


Has anyone done a hybrid setup w/out the snorkus installed?

[williaty]

Originally Posted by formula91

I rememeber you saying that at high range, without the torque box, there is a slight misreading, but in a completely tuneable way. (the single point in which a certain voltage is read, is only found in a single point in the graph)

Yet, I don't think I see it on the graph. Did you tune it out?
--edit--
Ah, one more thing, the way that will made up for the frankentake should fit perfectly. However, if you feel something is a bit off. Go ahead and replace a 3" straight coupler with a 3 inch hump coupler. That will let it fit perfectly. It's not necessary though.

If you look at the graph of load, the error for the hybrid and the SRI is evident in the peak/valley pattern above 4k. The peaks are slightly lower and the valley is slightly higher. In other words, it's a little more smooth above 4k than the graph shows. I didn't bring that up in the OP because it's kind of a separate issue.

And yes, use Hump Couplers. I will be getting one to put between the MAF section and the metal pipe because it's about 5* off.