[flatthump]

Originally Posted by WolfPlayer

I wish people would stop filling up this thread with wrong information.

Subaru uses the rear O2 for fueling corrections. However, the roll the rear O2 plays is SMALL in comparison to the front O2. Simple as that.

t

I took the information directly out of my streetTUNER manual. I wasnt trying to present some ground breaking discovery..

[WolfPlayer]

Originally Posted by jfitzpat

But, frankly, I don't find your statement any more compelling than flatthump's.

Frankly, I find it rediculous that you make bold statements about how the Subaru ECU works when you don't own a Subaru and you haven't tested/researched this issue with regard to this car.

With the STI if you pull the rear O2 sensor out of the exhaust pipe OR if you disconnect it, then the AFRs will swing to the rich side (as viewed on the wideband). Simple as that. Test it for yourself. Several people in this thread have already tested and verified this. Your turn.

t

[jfitzpat]

Originally Posted by WolfPlayer

Frankly, I find it rediculous that you make bold statements about how the Subaru ECU works when you don't own a Subaru and you haven't tested/researched this issue with regard to this car.

With the STI if you pull the rear O2 sensor out of the exhaust pipe OR if you disconnect it, then the AFRs will swing to the rich side (as viewed on the wideband). Simple as that. Test it for yourself. Several people in this thread have already tested and verified this. Your turn.

t

If it is "rediculous", why engage me at all? Seriously, I'm sorry that you are thin skinned and I'm sorry I don't find your own experiment compelling, (for the reasons that I have already laid out)

But if you insist, let's start with the basics.

1> Is there a cat installed?

If you move a wideband controller from before a cat to after a cat, there is going to be a change in readings. Period. With a PWM measurement like the LC-1, the reading will typically be leaner (which is understandable because the cat removes hydrocarbons). With some lambda meters, the reading will typically be richer. The reason is two fold, first current based meters generally feed 1/lambda (the proportional measurement actually made by the sensor) back into their PID loop (or variant). Second, current based meters are more prone to gas pressure errors on the rich side of peak.

Since you already mentioned that you had been using the same sensor in a very hot location, it would not be unreasonable for gas pressure errors to be exagerated. See the Bosch and NTK sensor specifications for details.

2> If there is NO cat, was a cat originally installed?

This is important because the ECU was programmed for the original install. What the ECU expects is the behavior of a cat. And your laughter notwithstanding, it is a pretty safe bet that the 3 way cat intended for your vehicle obeys the same physics as others.

3> Your points still don't connect.

Even if we ignore the potential problems with your experiement you are making a huge leap. In other words if your experiment turns out to be legitimate, all that you have established is that removing the sensor has an adverse effect.

Jumping a step and assuming that means that a sensor plays a role that is normally not played can be misleading. We know that role is not played on many Subaru's, since the 16 bit ECUs have been extensively disassembled.

On the flip side, there are plenty of other plausible explanations for certain behaviors that don't rely direct use of a sensor that should spend most of its time hovering near 0V. For example, in many HW designs an ADC chip is shared. An open circuit on a sensor might adversely effect switch and settle time on the analog multiplexer. IE, the sensor is not intentionally fed into the circuit, but its absense corrupts the reading of other sensors.

I'm not saying this is the case. I'm just giving a plausible explanation for the symptoms described. Likewise, I'm not saying your readings are absolutely bogus, I'm just giving reasons why they might be suspect.

If you are legitamately interested in getting to the bottom of this, I'm glad to help. I'm a little backed up right now, but would even be willing to look directly at the ROM code. Getting an accurate explanation for what is going on would seem best if someone is interested in properly satisfying the needs of the ECU.

But if an alternate hypothesis is going to continue to send you into a rage, and you think it is a signifcant factor that the water pump I changed today was on a Chevy not a Subaru or that most of my race tuning is on a Lycoming aircraft engine, that's fine too. I just dropped by. You're the one complaining about the accuracy of information on the forum.

-jjf

Edit: P.S. I hate to nitpick, but I'm the only one who has not made a flat statement about how the ECU works. I've only talked about how the sensors and cats work. It is just that, being cursed with sequentially firing neurons, I'd like ECU behavior to be reconciled with those independantly verifiable facts.

[WolfPlayer]

No cat.
Rear O2 installed.
Wideband installed well upstream of 2nd O2.
Car targets 14.7 on the LC1 at idle.
Remove rear O2.
Plug port.
Car now targets lower 14's on the LC1.
Reinstall Rear O2.
Car now targets 14.7 on the LC1 at idle.

Test on somebody else's STI. Same thing.

t

[drees]

The proper way to test the role of the rear O2 sensor is to substitute the appropriate resistance to simulate a lean and a rich signal and see if the ECU leans or richens up the mixture accordingly.

jjf - I know you're on the Enginuity forums, so if you can read ROMs, there are STi ECU dumps there for you to dig into if you want to verify the role of the rear O2 sensor there.

[R4ND0M_AX3]

I've made progress........I printed out the ECU wiring page that shows the connection to the Rear O2. Time for a break.

[jfitzpat]

Originally Posted by WolfPlayer

No cat.
Rear O2 installed.
Wideband installed well upstream of 2nd O2.
Car targets 14.7 on the LC1 at idle.
Remove rear O2.
Plug port.
Car now targets lower 14's on the LC1.
Reinstall Rear O2.
Car now targets 14.7 on the LC1 at idle.

Test on somebody else's STI. Same thing.

t

You didn't answer 2. Was a cat ever part of the design?

If so, then it seems to run against your theory. If the rear O2 is part of the AFR learning, then cat removal should cause a shift, since the rear O2 will suddenly be seeing quite a change.

On the other hand, unterminated wires, open heater circuit, etc. all sound like good prospects for a deceived input circuit or a fault trigger to the ECU.

Also, this does not appear to be your original experiment. You were using a sensor in a position that LC-1 users normally get temp errors and you also reported much richer readings.

Last, I don't think we can generalize this to all STI's. Since I did the free SSM logging thing I've talked to one STI user who is using the rear O2 circuit as a generic 0-1V input for logging. I've also found several threads on this in different forums.

drees - Thanks! You are also correct - though I think you would need to both satisfy the heater circuit (if any) and provide the ecu with a voltage. I'll be interested in looking at the code. The only obvious code application that comes to mind is a sensor fault detection. If the mix is shifted rich, then the cat cannot collect O2 and can't overheat. I've never heard of such protection, but it is at least possible.

R4ND0M_AX3 - Glad to hear you are making progress. Good luck!

-jjf

[WolfPlayer]

Car in a completely stock configuration but with NO cat.
14.7 AFR at idle.
Remove rear O2 from exhaust and leave it plugged in (fully functional and open to the air).
Plug the port port in the exhaust.
Lower 14:1 AFRs at idle.

The only change is the fact that the ECU is now seeing very lean AFRs at the second O2 sensor.

Ironically, I would like to be wrong about this issue.

t

[Merchgod]

I haven't looked at it too closely but I have come across the SSM parameters for 'A/F correction #3' and 'A/F learning #3' which are active on an 06 wrx, so the rear o2 has some authority, albeit probably very small, over closed and open loop fueling.

[jfitzpat]

Wolf Player, repeating yourself is not nec. Again, what does "but with no cat" mean? That you bought your car stock without a cat or that the cat was removed?

If you *are* right, it is important. If a sensor goes from post cat to pre-cat, the readings are going to change. If the sensor is part of the ECU's AFR learning code, then the change could be bad.

Merchgod, the ROM would seem to be the ultimate authority, but I am still at a loss as to how a post cat sensor could be meaningfully factored in. A cat works to remove hydrocarbons, and the sensor requires hydrocarbons to get a proper reading. As I mentioned above, I could understand some sort of sensor failure code, or cat efficiency code, but the normal AFR learning code? That still doesn't make any sense to me.

-jjf

[flatthump]

Maybe the rear 02 sensor acts a monitor for the cat.

I think at some point(when the cat eventually fails) the ECU would be able to tell from a growing variation of AFR readings between front and rear 02 sensor(s).

[WolfPlayer]

Originally Posted by jfitzpat

Again, what does "but with no cat" mean?

"but with no cat" means that there is no catalytic converter on the car. I have an APS turbo back exhaust (TBE). It has a catalytic converter but the converter is hot-swappable (I can install it or leave it out of the car). The APS catalytic converter is not installed. The secondary O2 sensor is still located after the catalytic converter (or, in my case, after the race-pipe).

Again, I WANT to be wrong about this. However, the AFR changes based on a very lean reading from the secondary O2 (or if the O2 is completely removed).

t

[Merchgod]

From what I've read, the rear o2 is given some authority over fuel on most modern cars to account for a bad cat. I'm guessing cats don't all of sudden go bad, but degrade over time.

[jfitzpat]

Originally Posted by Merchgod

From what I've read, the rear o2 is given some authority over fuel on most modern cars to account for a bad cat. I'm guessing cats don't all of sudden go bad, but degrade over time.

As I've said above, that makes some sense to me. Cat efficiency is a pretty narrow band. It is also important that the mix oscillate over stoich, otherwise the cat can never collect O2. But I still don't understand why it would be factored into open loop at all. Once you get past .85-.9 lambda, the cat is almost useless for 2 of the 3 things it converts.

Perhaps a small adjustment to tweak stoich based on perceived cat effiency? I'd love to look through the code, but I've got to finish work and my Enginuity committment first.

flatthump: That is typically all the rear sensor is for. To monitor cat efficiency. However, based on R4ND0M_AX3's tests it seems to me that something is occuring, at least with some ECU's. But I'm still trying to understand what it is.

WolfPlayer: This is starting to sound like a broken record. But I'm going to repeat the basics one more time. I found your initial post non-compelling. Your numbers seemed non credible (13.6) for a closed loop correction circuit at idle. I won't bore you with the math, but the change was well outside the expected precision from the overall system. Also, once I hear words like 'super hot', I don't put much credence in a sensor moved to a cooler, lower gas pressure situation.

However, I did find R4ND0M_AX3's tests more compelling. There appeared to be fewer variables and the change (.007 - .015 lambda) seemed more realistic for an ECU driven correction. Once you have a technical phenomena, you have to get to the bottom of it.

What you seem to be failing to grasp is that whatever the answer, it has to follow the laws of physics. Unless you want to dispute how UEGO sensors work or how cats work, talking about hot swapping cats does nothing to bring us closer to a real answer. You can read about UEGO sensor theory here. If you read carefully, you will also see why cars spend so much time about 14.7 (lambda 1.0 actually). It has nothing to do with economy, nothing to do with power, and everything to do with emissions control.

Two OK references on cat's can be found here:

http://www.fordscorpio.co.uk/cats.htm
http://en.wikipedia.org/wiki/Catalytic_converter

Although they reinforce what I just said (14.7 being an issue of emissions, not power or economy) and give a basic notion of how cats work - you'll still have to connect the dots. UEGO uses hydrocarbons, cat works to remove hydrocarbons.

Last, WHY SHOULD I CARE? Contrary to what you might believe, the point of this is not to make you look 'wrong'. No matter how much you contrive or revise your own original experiment all I am really interested in is reconciling a response in the idle behavior to removal of a sensor that DOES NOT REQUIRE sensors and cats to behave in strange and magical ways. Reconciling behavior with physics is not a matter of making me 'right' about anything. It is about developing a truly effective workaround or correction.

For example, R4ND0M_AX3 can definately program an LC-1 analog output to simulate a narrow band sensor. But if we don't really understand what the ECU is looking for and responding to from that sensor, we are only guessing and the simulation itself may have unintended consequences. Perhaps the answer is simulating a narrowband with a small shift (simulate a working cat to the ECU).

Sure, we can just use guesswork and trial and error, but if Merchgod is correct (and I have no reason to doubt it), there is an impact on open loop behavior. Anything that effects open loop behavior and is not really understood is an impediment to high performance tuning.

-jjf

[WolfPlayer]

Again, I would like to be wrong. Physics - ok. Physics 1 and 2 + advanced Physics Theory in college. Calculus 1, 2, and 3. Differential Equations. Blablabla. I graduated 3.96 in my major. Math is a good thing. I get it.

I'm giving the results of a simple test. I never designed this from the ground up. Fact is - and I don't understand why you don't see this - all that I am saying is that if you take the stock rear O2 sensor and make it see lean conditions then the ECU responds and richens up the mixture. Simple as that.

I am not about to go and design a test around this because, honestly, it just isn't important enough to me. I know that the ECU *can* make some adjustments to AFR based on the rear O2 sensor. That is good enough for me. I don't care about qualifying it, etc. It's just not THAT important. Fact of that matter is that there are people out there who REMOVE their rear O2 sensors and then disable the codes. That is the main point of this thread. The main point is to deter people from removing their rear O2 sensors because that will cause the ECU to richen the mixture. The side point is that if the ECU sees a lean signal on the rear O2 then it will richen things up.

t