Catless and the imfamous CEL

[engineermike]

Do you mind sharing a bit of info on how you tested this?

Yes, a close associate and a bit of an OCD data scientist had spent lots of time perfecting his MAF transfer function and his fuel trims were very close to zero. He then did a cat delete. He did anti-foulers because he didn't want the not-ready status associated with disabling things in the calibration. His fuel trims were suddenly erroneous by 10-15%. As a test, he disabled FAOSC and trims went right back to 0 +/- a few %. That led us both down the path of researching how FAOSC works and the explanation became obvious.

We haven't even talked in this thread about the potential effects of blow-through adding more error to the WBO2 readings when doing cat deletes, though this is generally only associated with supercharged setups.

When talking about hurting motors by running lean, it may not be as dangerous as most of us thought. I was checking around and both the GT and GT500 (along with a bunch of GM engines) actually run full load/WOT at 1.0 lambda. The GT500 pushes the throttle wide open at 90% pedal and power-enrichment doesn't come in until the same 90%, so at 89% pedal you're at full boost with no enrichment. If you didn't botch the lambda borderline correction table, then the timing accounts for any additional tendancy to knock.

Sponsored

 

[ZXMustang]

Yes, a close associate and a bit of an OCD data scientist had spent lots of time perfecting his MAF transfer function and his fuel trims were very close to zero. He then did a cat delete. He did anti-foulers because he didn't want the not-ready status associated with disabling things in the calibration. His fuel trims were suddenly erroneous by 10-15%. As a test, he disabled FAOSC and trims went right back to 0 +/- a few %. That led us both down the path of researching how FAOSC works and the explanation became obvious.

We haven't even talked in this thread about the potential effects of blow-through adding more error to the WBO2 readings when doing cat deletes, though this is generally only associated with supercharged setups.

When talking about hurting motors by running lean, it may not be as dangerous as most of us thought. I was checking around and both the GT and GT500 (along with a bunch of GM engines) actually run full load/WOT at 1.0 lambda. The GT500 pushes the throttle wide open at 90% pedal and power-enrichment doesn't come in until the same 90%, so at 89% pedal you're at full boost with no enrichment. If you didn't botch the lambda borderline correction table, then the timing accounts for any additional tendancy to knock.

you’re seeing that lean condition because the LTFTs are big time rich. I’ve seen that too and when you disable LTFTs you will see the lambda correct since it’s all being controlled in realtime. I had this happen on a few pulls seeing .85+ lambda at wot when I was commanding .77-.78. The LTFTs were showing +15 and the shorties were at zero.
Case in point running lean at wot is very bad. And what you are seeing is a false lambda since the long term fuel trims are doing the work and it’s not reflected in the fuel enrichment logging for lambda. Scared the shit out of me. But after disabling the long term trims, all went back to normal and the car runs consistently at .77-.78 at Wot. Whipple car. That’s why so rich. Coyotes like to be rich too.

 

[K4fxd]

And what you are seeing is a false lambda since the long term fuel trims are doing the work and it’s not reflected in the fuel enrichment logging for lambda.

Except from reading his post @engineermike he said they turned off FAOSC. This would require a re-flash thus disabling, or zeroing out the LT's.

I would agree with your statement if the car was driven for 25 or more miles before the pull.

 

[engineermike]

Zx it sounds like you’re logging commanded lambda (lambse) not the wbo2 reading. The lambse has the fuel trim baked into it but the lambda does not.

 

[ZXMustang]

Zx it sounds like you’re logging commanded lambda (lambse) not the wbo2 reading. The lambse has the fuel trim baked into it but the lambda does not.

No Im logging it correctly.

 

[junits15]

Yes, a close associate and a bit of an OCD data scientist had spent lots of time perfecting his MAF transfer function and his fuel trims were very close to zero. He then did a cat delete. He did anti-foulers because he didn't want the not-ready status associated with disabling things in the calibration. His fuel trims were suddenly erroneous by 10-15%. As a test, he disabled FAOSC and trims went right back to 0 +/- a few %. That led us both down the path of researching how FAOSC works and the explanation became obvious.

We haven't even talked in this thread about the potential effects of blow-through adding more error to the WBO2 readings when doing cat deletes, though this is generally only associated with supercharged setups.

When talking about hurting motors by running lean, it may not be as dangerous as most of us thought. I was checking around and both the GT and GT500 (along with a bunch of GM engines) actually run full load/WOT at 1.0 lambda. The GT500 pushes the throttle wide open at 90% pedal and power-enrichment doesn't come in until the same 90%, so at 89% pedal you're at full boost with no enrichment. If you didn't botch the lambda borderline correction table, then the timing accounts for any additional tendancy to knock.

Interesting! Appreciate the response!
So the belief is that if this individual had kept running the car the trims would have gone back to 0 as the bias was applied to the 02 sensor reading and the error was obfuscated into the lambda sensor bias?

I think where I'm getting hung up on is if the bias is applied to the lambda sensor data directly I would think (all speculation) that we shouldn't be able to detect that with fuel trims. If the lambda sensor is being biased such that 1 lambda looks like .8 lambda to the ECM and we are commanding 1 lambda, in real life the exhaust stream will actually be 1.2 lambda. I would think (again I'm not totally well versed in this) that in that scenario the fuel trims will be near 0, as the ECM is not aware of the fuel mismatch.

If the bias is applied to the fueling, it would make sense that we will see the issue in the trims. Assuming the same .2 lambda bias as in the previous paragraph, if its biasing fuel that would push the base fueling lean by .2, which would be detected by the lambda sensors. The feedback loop would correct and generate that +20% fuel trim.

I've been running with FAOSC on for a while now, aftermarket cats but no non-foulers. I just switched it off, and I'm going to compare actual fuel flow from before and after I disabled FAOSC. I haven't had any weird fuel trim issues, which would point to the bias being applied directly to the lambda sensor. If that is the case I should see different fuel flows in the same situations, I'd expect to see more fuel now.

I'm naturally aspirated, so thankfully I can ignore blowthrough for the time being.

I wonder if this is why people claim to get better fuel economy with headers, when they're actually just running slightly lean


Very interesting stuff for sure! Love this deep dive level investigation

 

[WildHorse]

When talking about hurting motors by running lean

I never said that. What I said was IF it was truly running 15% lean, FI engines would be going kaboom. I have yet to see FAOSC (on) running anywhere near 15% unless the tooner f'd up.

This whole thread is moot anyways. Getting rid of the CATS has zero, I repeat, zero detrimental affects, other than the CEL coming on, which the logic ignores anyways. Trims are great, timing is great, MPG is the same or better, extra power always a bonus.

 

[K4fxd]

I would think there is something in the code that shuts off FAOSC if the signal is not what it should be with a healthy cat.

 

[junits15]

I would think there is something in the code that shuts off FAOSC if the signal is not what it should be with a healthy cat.

I’ve thought the same thing but I have no way to prove it.

welding in a second wideband would probably do the trick

 

[engineermike]

So the belief is that if this individual had kept running the car the trims would have gone back to 0 as the bias was applied to the 02 sensor reading and the error was obfuscated into the lambda sensor bias?

I think where I'm getting hung up on is if the bias is applied to the lambda sensor data directly I would think (all speculation) that we shouldn't be able to detect that with fuel trims. If the lambda sensor is being biased such that 1 lambda looks like .8 lambda to the ECM and we are commanding 1 lambda, in real life the exhaust stream will actually be 1.2 lambda. I would think (again I'm not totally well versed in this) that in that scenario the fuel trims will be near 0, as the ECM is not aware of the fuel mismatch.

If the bias is applied to the fueling, it would make sense that we will see the issue in the trims. Assuming the same .2 lambda bias as in the previous paragraph, if its biasing fuel that would push the base fueling lean by .2, which would be detected by the lambda sensors. The feedback loop would correct and generate that +20% fuel trim.

I'm not sure I completely understand what you're saying but I don't believe it learns bias, though bias is the issue here. A different bias table is used for all 4 O2 sensors and if you remove cats, the bias for the rear 2 are suddenly incorrect. There is no secondary check for the rears (since they ARE the secondary check) so the PCM has no way to know they are incorrect.

I wonder if this is why people claim to get better fuel economy with headers, when they're actually just running slightly lean

This definitely crossed my mind. We know the pumping losses at part throttle are near-zero, so deleting cats won't reduce backpressure any appreciable or measurable amount in normal circumstances that would affect fuel economy. However, if lambda were incorrect, then it could.

 

[engineermike]

No Im logging it correctly.

I can't say that I've ever seen an active, working LTFT cause logged lambda to not follow commanded.

 

[engineermike]

I never said that. What I said was IF it was truly running 15% lean, FI engines would be going kaboom.

You said if it were lean it would hurt motors and I explained was that's not necessarily the case. If the timing was offset to counteract knock or it was set up rich (like target .72), then it wouldn't hurt the engine. And as I also explained, even the GT500 can run 1.0 lambda at full boost and it's perfectly safe all day long so it just depends on the how the tune is set up. If they targeted .77 lambda and it's 15% lean then it's running .88 which isn't a death sentence.

I have yet to see FAOSC (on) running anywhere near 15% unless the tooner f'd up.

So let's just say FAOSC is on and cats are missing, then fuel trims go high. Tuner sees this and changes the MAF curve to get the trims back down. This is not fixing the problem but most enthusiasts and some tuners would assume so. All it did was hide the problem.

This whole thread is moot anyways. Getting rid of the CATS has zero, I repeat, zero detrimental affects, other than the CEL coming on, which the logic ignores anyways. Trims are great, timing is great, MPG is the same or better, extra power always a bonus.

The detrimental effect is that your lambda reading is no longer accurate as long as FAOSC is on and cats are off.

If you manage to turn the CEL and FAOSC off, then you'll always get "not-ready" OBD status and can't pass inspection.

If you leave the CEL and FAOSC on and use an anti-fouler, then your wideband readings will be incorrect and you aren't running the lambda you think.

 

[junits15]

I'm not sure I completely understand what you're saying but I don't believe it learns bias, though bias is the issue here. A different bias table is used for all 4 O2 sensors and if you remove cats, the bias for the rear 2 are suddenly incorrect. There is no secondary check for the rears (since they ARE the secondary check) so the PCM has no way to know they are incorrect.

I understand now I was way off base you are completely correct you can see this in the fuel trims. I was thinking that it wouldn't be possible to see this error in fuel trims because I forgot that the ECM knows how much fuel it should be delivering for a set airflow.

The amount of fuel to hit a certain lambda is not negotiable. Assuming the lambda sensor is calibrated we can say

If:

• fuel is e10 gasoline
• Airflow = 14.1 pounds/min
• desired lambda = 1
• measured lambda = 1
• reported lambda = measured lambda
• applied fuel flow will be = 1 pound/min
• actual required fuel flow will be = 1 pound/min

However! if the lambda sensor is biased wrong, lets say that it reports 1 lambda when its actually measuring .8 lambda. If we run the same example again...

• fuel is e10 gasoline
• airflow is 14.1 pounds/min
• desired lambda = 1
• measured lambda = .8
• reported lambda = (measured lambda * 1.25) <-- 25% error from bad bias
• applied fuel flow will be = 1 pound/min
• actual required fuel flow will be = 1.25 pounds/min (25% more fuel)

This is because the sensor will not report 1 lambda until it actually sees .8 lambda, so the car will have to inject an additional 25% more fuel in order to make the sensor report 1. This will be 25% more than expected to get 1 lambda at the fixed airflow and will definitely show up in the fuel trims.

In this example the car is running rich, its running at .8 when it thinks its at 1, but this can easily go the other way, the car can report 1 lambda when it's actually measuring 1.2 lambda. FAOSC needs to be off for non-stock cats/catless.

 

[junits15]

So the conclusion is: if your fuel trims are good and you haven't mucked with the MAF curve then you can sleep easy.

The holy grail would be to find out how to alter the downstream O2 sensor biasing system to apply 0 offset in all scenarios. Then we can leave FAOSC on and be "ready".

 

[rolfe.oliver]

Yeah I have that disabled in my tune. Along with all the 0420/0430 codes. Also have COT disabled as well.

Right here, officer. This is the one you are looking for.

Jesus Christ, who is posting this shit nowadays...

Sponsored