PorscheG96

Since buying my 2004 GT3 a year and a half ago, I've been trapped in a sort of California smog hell. Having learned far more about O2 sensors and related CEL than I ever wanted to know in order to completely resolve my issues, I thought this info would be beneficial to the community. Much of what I learned below is already on the web somewhere with enough digging and/or you may already know about it, but I also believe there are some interesting findings here that shed new light on the behavior of our exhaust emissions equipment. At any rate, I've tried to aggregate everything here.

1. Background

When test driving the GT3 prior to my purchase, I noticed a brief surge of 50-100 RPM's at idle once or twice and mentioned this to the owner. He didn't seem to know anything about it and otherwise the car did great at PPI so I bought it. Within a week or two of taking ownership, perhaps within 200 miles, I got my first CEL code. At that time I didn't have a code reader so I drove the GT3 to Tony @ Heyer Performance to read my error code: P0420, catalyst efficiency low bank 1. Tony's advice was that O2 sensors might cause this code instead of the cats which would be a less costly fix, but first we cleared the code to see if it re-appeared and, of course, it did within a couple hundred more miles.

As a newly afflicted GT3 owner, I couldn't take my mind off of exhaust upgrades to improve the sound and drop some weight off the tail. After many nights of reading and learning I decided that the best value / reward would be to install 997.2 RS headers+cats and Ti center muffler on my 996 GT3. This setup combined with side muffler bypass and carbon cup bumper drops approx 70 lbs from the rear of 996 GT3, closes the gap somewhat between the 996 GT3's 381PS and 415PS of 997 GT3 due to the much improved equal length header design, and could address my catalyst efficiency problem as a bonus. Thanks to Steve W for the RS exhaust setup and another Rennlister for the cup bumper, I was in business...



2. OBD Reader

A few days later I had everything installed -- car ran great and sounded just AWESOME! I drove a few hundred miles and all was right in my automotive world when suddenly there was another CEL! By now I'd decided it was a good idea to have an OBD reader so recommendation #1 is for everyone to buy one of these handy gadgets:


This OBDII dongle lets you connect via bluetooth from your Android phone using applications like OBDLink or Torque to view output from various engine sensors, smog readiness monitors, and CEL codes. Alternatively for iPhone owners, you can try BlueDriver or Veepeak software although I haven't tried either of these.

So back to the story, the new error code was P0139: O2 sensor slow response, bank 1 sensor 2. This code would continue to plague me for over a year as I had other priorities in life at the time and would try occasional fixes that all resulted in failure until recently.

3. Ideal Behavior of the Catalyst & O2 sensors

Using the Torque app, I was able to configure a window showing real-time output voltages for all 4 of my GT3 O2 sensors. With this being my first time looking at these values I had no idea what they meant or what problems to look for. I spent some time reading about how O2 sensors work and the main takeaways were the following:

-O2 sensors compare oxygen INSIDE the exhaust system to oxygen OUTSIDE the exhaust system - there's typically a path to detect oxygen on the outside of the sensor bung at the wiring base.
-The INSIDE/OUTSIDE O2 differential is what generates the O2 signal voltage.
-If the quantity of exhaust gas O2 is high then it's roughly equal to the outside O2 level, hence the differential is low and the sensor generates a very low voltage. This condition will occur when fuel is the limiting combustion reactant, hence there's not enough fuel to burn all the O2 so low voltage is a lean indicator.
-OTOH, when the exhaust gas contains little O2 then the differential between inside/outside O2 is high which generates a high voltage since O2 was the limiting reactant, leaving extra fuel behind meaning a rich condition.

Now that I knew what voltages between 0v and 1v meant, I began to interpret what was happening at the front and rear O2 sensors. In a properly functioning system at partial/constant throttle, the front O2 sensor rapidly oscillates between rich and lean as pictured here:



The way to tell that the catalysts are doing their job is by looking at the rear O2 sensor readings which indicate perfect stoichiometric around 0.5-0.6 volts. Things get more complicated / tricky during the drive cycle and readiness test sequences that I'll explain later.

4. P0139 & P0159 O2 Sensor Errors

As for the O2 sensor problems, P0139 and P0159 are the same error for slow response of the post-catalyst oxygen sensors but on different banks. P0139 is bank 1 [driver's side of engine for LHD cars] and P0159 is the same error on bank 2. I found that O2 sensor slow response is likely to be triggered under certain conditions:

-A sudden lift from constant to closed throttle where the ECU will begin a count-down from full lean on the forward O2 sensor to full lean on the rear O2 sensor. A normal response value based on my empirical data is about 1.5 seconds from front to rear O2 and the banks are compared independently from each other. In other words, lifting off the throttle won't necessarily result in banks 1 and 2 going lean at the same time. There might be a 1 second difference and so the rear O2 sensors need to detect this difference individually or an error code will be triggered.
-During the smog readiness drive cycle, the ECU will make some lean / rich adjustments on its own and expects to see the correct reading from the rear O2 sensors within a couple of seconds or the error will trigger:





In general, the rear O2 sensors will trigger errors if they don't respond to the front O2 sensor mixture change quickly enough or with enough amplitude to be analyzed properly by the ECU.

5. P0420 & P0430 Catalyst Errors

In these cases, the voltage readings of the rear O2 sensors will appear the same as, very similar to, or at less than or equal period/interval to the front O2 sensor voltages. I don't have a screen capture of the bad catalyst since I was unarmed with the OBD reader at that time, but was actually able to trigger this error again using an O2 sensor simulator. In playing with the adjustments of the simulator, I began with a 5 second period which promptly triggered P0430 on bank 2. The ECU was seeing voltage readings from the rear O2 sensor [simulator] that were more frequent than the readings from the front sensor, so it figured that the catalyst wasn't working and emitted this CEL [the simulator generated wave form can be partially seen on bank 2, sensor 2]:



In general, if the rear O2 readings follow a similar period or amplitude as the front O2 sensors then it's probably time to replace the catalyst.

6. Smog Readiness Monitors & Drive Cycle

For OBDII cars, a series of tests are executed before the smog monitors can be deemed ready. I'm sure that many of you have experienced the frustration of going to your smog test and being turned away because the system isn't ready yet. All of the monitors get reset whenever the current fault codes are cleared or the battery is disconnected.

In the first readiness test, the O2 sensors and O2 heater elements are tested for resistance. When toying around with an O2 simulator I realized that if only the O2 loop was hooked up to the car then the O2 heater test would fail. We decided to wire in the O2 sensor heater but intercept the O2 voltage loop with the simulator which looked entirely Frankenstein-ish:



When the engine is first started it's running in open loop mode, ie. the O2 sensors are skipped in favor of a rich ECU map. Once the engine reaches a certain temperature then the ECU enters closed loop operation and the O2 sensor + O2 heater diagnostics are performed. When these tests pass then the next couple of tests are for the Secondary Air Injection and Evap Systems. It turns out that the secondary air injection test is performed at idle and this explained the unexpected idle surge that I noticed on my initial test drive. My warning to others out there who are shopping for GT3 or any OBDII Porsche is to bring an OBD scanner with you and check to see a) whether there are any stored fault codes and b) whether the smog monitors are properly set. I wish I'd made certain the the readiness monitors were all good before pulling the trigger on my purchase but I didn't know that the quick idle surge was a diagnostic test to ready the monitors. Apparently a mechanic had worked some voodoo to make it pass smog but it was shortly thereafter when the catalyst monitor failed and caused this entire headache.

Anyway, the final test is the catalyst which is the big one. I've seen various publications of drive cycle recommendations but I'll share what I did and maybe others can chime in with drive cycles that are faster and easier:

-At least 3 cold engine starts, so the cat readiness took me 3 days
-Drive several minutes between 25-29 mph under 2800 RPM
-Drive several minutes between 55-59 mph under 2800 RPM
-Drive several minutes between 25-29 mph under 2800 RPM
-Idle for 5 minutes
-Shut down and restart engine
-Repeat at step 2

I wasn't sure what the reasons were behind such mundane driving techniques until watching my OBD readouts for the O2 sensors. It turns out that the ECU performs several tests at constant speed & throttle operation so that the driver cannot manipulate the results. You can see in the following graph that the ECU goes to full lean on the fuel trim independently causing front O2 sensor voltages dip to reflect the lean condition. This is followed by the ECU going full rich as evidenced by the front O2 sensor plateaus which results in the rear O2 sensors slowly stepping from lean to rich again.



These tests are performed many, many times before the catalyst readiness is passed -- ultimately, it took 186 miles of mixed driving for my system to be ready.



7. Brief word on O2 sensor spacers and simulators

An O2 sensor simulator is a piece of hardware configured to generate 0v-1v just like an O2 sensor would, but at random intervals. I was fortunate to have access to an EE who built a custom board for me with adjustable period and voltages, and although this resulted in more error codes, the simulator proved to be a valuable new tool in diagnosing my problem, which I found ironic. The resulting wave form can be seen here on bank 2, sensor 2:



Based on my understanding of the smog readiness tests on 996 GT3, I firmly believe that O2 sensor simulators won't work for our cars because the smog readiness diagnostics are too advanced for the simulator to anticipate. Perhaps they work for cars built very shortly after the intro of OBDII [ie. 1996-1999] but the manufacturers have made the diagnostic software more advanced since then. When the ECU is tuning the fuel mixture at random intervals and expects to see a certain result, you're possibly going to fool the period test but not the lean/rich response test.

If you're going to try an O2 simulator anyway, I cannot emphasize enough that a volt meter should be used to verify the functionality of your simulator on the bench, especially the output voltages as they relate to your O2 voltages, to avoid frying your ECU.

The O2 sensor spacers can probably work for systems with the catalyst removed but I don't have any experience with them, although logically I can see how they'd reduce the amplitude of the O2 detected by rear O2 sensors but I could imagine triggering O2 slow response codes as a consequence.

8. And finally, the solution to my problem

The 996 and 997 GT3 engine wiring harnesses differ in that, among other things, oxygen sensors are located on the same side as their respective engine banks for 997 but the rear harnesses are located on opposite sides for 996 because of the wrap-around exhaust routing design. I remember asking a couple of other 996 GT3 owners what they did about the O2 sensors when using 997 GT3 exhaust and they said the original sensors were fine and although different lengths, they simply re-routed the sensors with no problems, so initially I did the same. There may be no apparent problems immediately and this may be OK for track cars but turns out to be a problem for road cars that needs to pass emissions.

The correct solution involves extending the secondary O2 sensors by about 30" and crossing over the rear sensors to opposite sides. My ECU was triggering P0139 because the rear sensor of bank 2 was being logged as rear sensor of bank 1, so the subtle timing differences between banks 1 and 2 were getting flagged as errors and throwing the light.

I believe another contributor to my O2 sensor errors was the design of 997 O2 sensor vs 996 O2 sensor. I noticed at one phase of my investigation that not only are the harness connectors different but the 997 sensors have openings that measure about 4x the surface area to allow more O2 into the sensor element which I think is a result of the improved catalyst design. Additionally, the internal O2 element changed between 2007/2008 GT3 and 2010/2011 GT3 as I think the catalyst element changed to the HD versions, but I'm not sure about this. In the end, I used 996 harness connectors on extended 997.2 rear O2 sensors.





Since I was buying O2 sensors like crazy and experimenting with different things, I collected a list of generic Bosch sensors which are ~2/3 price of those from Porsche and completely plug-and-play, no modification necessary. Here's the list:

9. Generic Bosch O2 sensors for different generation GT3

996 GT3
Front:
Rear:
997.1 GT3
Front:
Rear:
997.2 GT3
Front:
Rear:
So after all that, my GT3 is running around with 997.2 RS exhaust and no CEL's.

Turbo Jonny

Great post , thanks for sharing .
My boxster has a similar annoying cel problem though the
Car runs fine .

Macster

Yes, great post.

Couple of things: I have an Actron OBD2 code reader/data viewer/data logger. Besides these functions it offers the abilty to initiate some O2 sensor tests. These tests are very helpful when trying to confirm a sensor is bad vs. just replacing a sensor. And I'm not a fan of this. My philosophy is to replace sensors in pairs if not all 4 at the same time.

The official factory reference for P0420 and P0430 error codes is to if there are any aging O2 sensor codes to replace the indicated sensors then road test the vehicle. If the converter errors come back then replace the indicated converters.

The aging O2 sensor error codes may be also active but converter code is a higher priority error code. If the OBD2 tool is not sophisticated it might not read beyond the 1st error code. Or the aging sensor codes may be pending. However, it reads like the tool you used is pretty sophisticated. It is under these circumstances then having a tool that can run some of the O2 sensor tests comes in handy. While the sensors may not fail outright their reponse to the various tests can suggest they are wearing out.

Welty

Thank you for the excellent post. To share my similar California pain. (I have two OEM replacement cats that are less than (18) months old.) Multiple entities (myself, my mechanic (s), Porsche dealership) have driven my car the many prescribed various drive cycles to reset the moniters. Many miles, many opinions, much time & $s. I even put the car on a dyna for a part of a day. With the hopes of reseting the monitors. All many $s. All, without success. Finally last week to end my frustration. The local Porsche dealership clipped me $3,400 to replace the computer in my car. Fortunately, my car has been blessed with many Mooty hand me downs (none related to this issue). Or I was about ready to get rid of this beast. The wife would have been very happy and would have put Mooty back on her Christmas card list.

gt3'er

Yes, interesting post. Not sure I understand it... But what I gather from this: "The correct solution involves extending the secondary O2 sensors by about 30" and crossing over the rear sensors to opposite sides. My ECU was triggering P0139 because the rear sensor of bank 2 was being logged as rear sensor of bank 1"... is that the sensors were simply installed in the wrong place.

My CEL just came on - so, its time to get an OBDii reader. Fortunately, I just passed smog after swapping back to my OEM cats from the Fabspeed 200's, which I enjoyed (except at Laguna Seca.)

Mr. Welty, it seem unthinkable that your mean blue machine would need a new computer. Did they bench test it and pinpoint a problem? Could it not be "rebooted" or something? My word, that's horrendous. BTW, I'd suggest getting Mooty back on the Xmas list - no doubt, he's got loads more fun stuff.

rbahr

Good article - subscribe...

yelcab

Hats off to you, man. Great knowledge generated here.

rebrewer

Nice article, especially interesting since I got catalyst efficiency low bank 1 & 2 codes coming back from a weekend at Sears Point Sunday.

powdrhound

On the 996GT3, are the pre-cat sensors Bosch LSU4 wideband and the post-cat sensors switching type LSF4 narrowband? I know that went to wideband primaries on the turbocharged TT/GT2 engine but not sure about the GT3. The rest of the normally aspirated 996 cars had narrowband primaries if I'm not mistaken.

Did you also find out if the rear post-cat sensors had any engine control functions or simply a catalyst efficiency function. In other words, do the post-cat sensors have any control of fueling or do they simply play a monitoring role?

PorscheG96

Pretty sure the pre-cat sensors are narrowband just like the post-cat sensors on GT3. I accidentally purchased 996TT pre-cat sensors when trying to decode the generic Bosch PN's and they had an extra wire + different harness compared to GT3 which I think is consistent with wideband. Considering that the catalysts are involved in numerous chemical reactions, it's a safe bet that the post-cat sensors aren't reliable or necessary for engine fueling logic...

Macster

The DME constantly monitors the #2 sensor readings to ensure the converters are operating at the proper level of efficiency. (This is determined by the sensor voltage levels each converter can store oxygen during the lean cycles of the incoming mixture.)

If the #2 sensor readings start to deviate too far from what is expected the DME will adjust fueling in an attempt to get the readings from the #2 sensors it expects. In some cases -- in say the event of a mechanical failure related to the VarioCam Plus system -- the DME can take the fueling too far afield and misfires can occur.

With a factory stock engine both the #1 and #2 sensors are critical to a properly fueled engine.

powdrhound

Can you please post a source of this information? Is this conjecture or something concrete actually backed up by materials posted by Porsche in any any of their literature / maintenance manuals? If so, PLEASE LIST YOUR REFERENCE.

I have documentation relative to the 6GT2 where Porsche goes into 5 full pages of detail on the inner workings of the primary wideband pre cat sensors and their interaction with the DME. This is followed up by a single sentence description of the post cat sensors and their operation as seen here:

"Modified LSF4 (offset) oxygen sensors are used
downstream of the primary catalytic converters
(previous vehicles had LSH25 offset sensors)"

It seems to me that if the post cat sensors played a part in engine fueling / tuning, Porsche would devote more than 19 words to it when they go into pages worth of info in the primary sensors.

aryork

Great info!

G96-->Just to make sure 1x2 means Bank 1, sensor 2 - is that right?

I have a GT4 (okay, in wrong forum ), but have a fabspeed sport headers (200 cells) and just got CEL. Codes are:
P2097 - Post catalyst fuel trim system too rich bank 1
P2099 - Post catalyst fuel trim system too rich bank 2
P2272 - O2 sensor signal biased/stuck lean bank 2 sensor 2

I had P2097 once before after I used a bad fabspeed header gasket and half my manifold screws fell out. Put screws back in and code went away. Now it's back with a vengeance. After seeing your plots I discovered my DashCommand app could also collect that data, so I went out to the car. When I cranked it cold, both post-cat sensors showed about 0.5V, just like your plot. But after warming up, my sensors did not act the same. The right one (bank 2) seems off. Here's a plot after I created a log file (when revving slightly) and brought it into Excel. Note, my OBDII dongle only x-mits once/second, I believe, so data is a bit choppy - looked better on the app itself, but message was the same.


I have an inspection this month and believe I should just put the OEM headers back on, get the codes cleared, and get the car inspected. It sounds like this can turn into a lengthy science project!
Update: digging more into this and realize what data I should be looking at. Ordered a higher data rate OBDII dongle which should be here this week...

aft86

Great thread! Don't mean to resurrect an old thread but I am in the same situation as you were but in a 2014 991 TTS (wrong thread as well but I haven't found much on the 991 turbo side) Hoping to find a solution here since I have to register and inspect my car soon, and as is it won't pass inspection.
Long story short, I have a catless exhaust with a GIAC tune. I have reached out to GIAC and they don't make a file to keep the CEL from coming on. I've tried two types of o2 spacers, a straight and angled ones (at one point I had both on at one time) and light always comes on. I just picked up a pair of o2 simulators from Magnum Tuning, the #795 to be exact, and installed them. These connect to the voltage cables but use the heater cables from the sensor. The CEL still came back but as soon as I restart the car, it turns off. The codes I am getting now are P013F and P014B - which are O2 Sensor Delayed Response - Lean to Rich (Bank 1 Sensor 2) and (Bank 2 Sensor 2). I am wondering what I am doing wrong or if I need to adjust the simulator's voltage. Below is a screenshot of the charts with the CEL on.

aryork

There is another thread somewhere (maybe GT4 forum) where I posted a lot more info/data. I collected some data at constant driving conditions on the highway. There's some rationale for doing that...can't recall all the details. Also, I installed Fabspeed mini-cats (can't recall if I had mentioned that in this thread or not). It did not fix the CEL problem, but definitely changed the readings, and possibly gave me an additional "too lean" code. Anyway, long story short - I ended up re-installing OEM headers to get the code to clear. While the sport headers were off I washed them out with Dawn dishwashing liquid, which is *not* supposed to help, but I figured I would try anyway and got them ceramic coated to keep some heat in. After re-installing them (after passing inspection) without the mini-cats, I have gone a long time, nearly a year, without a CEL. There are so many variables, not sure of cause and effect. I thought that the gas at VIR could be sketchy...meaning maybe they put leaded gas in the unleaded side or something like that. But, all this seems related to my CATS which you do not have. There are several folks (like AlanC) on the GT4 forum who run Dundon catless headers and have never had a CEL. Seems to be a black art to this stuff...