What I Learned While Diagnosing OBD Fault Codes P0421 and P0431 – Part 2

Part 1 of this article can be found here. The problem with my 2007 Porsche Cayman S was different from the run-of-the-mill cat or O2 sensor failure. I had just spent several months rebuilding the engine (see articles here). BTW my engine designation is M97.21. All the suspect parts were new. I had a new Top Gear exhaust system, which included a new warm-up catalyst, and interestingly, no main cat. I had new upstream wide band air/fuel sensors and new narrow band downstream O2 sensors. These were the OEM sensors used by Porsche. The car, before the rebuild, had a mix of one original warm-up cat, one aftermarket warm-up cat, no main cats, and a no-name aftermarket valved muffler setup. My rebuild was predicated on adding as much reliability as I could, so I went with the new Top Gear exhaust, which gets you everything from the flange that attaches to the cylinder head to the tips that protrude out the back of the car. A very nicely constructed piece of kit, BTW. The cats were “200 cell”, which we will discuss more later.

The ECU had been powered down for many months. I don’t profess to know if it retained anything in its memory but I was not surprised when I got a lot of codes and warning lights during my initial startup and break-in driving. Without going into detail, I was able to make most of them go away with my Durametric Pro. But one nagging issue would not go away, that being that it threw the P0421/0431 codes almost instantly after any driving. I guess they start as Pending Codes, which would show up in typically less than 5 miles of driving. Usually the check engine light came on within 25-50 miles of driving. I live in NC in an urban area where the vehicle inspection process includes a download of stored data in the ECU. My inspection had expired during the long period when the car was out of service and I was anxious to get an inspection completed before I attracted attention from the police.

I started doing some reading. The internet of course but also some Porsche technical documents that had been shared with me. Generally they are called Porsche After Sales Training documents. These have been very useful although they do not come close to explaining the entire story. I would guess they are proprietary but I will refer to them in my story when necessary.

One of the first documents I went to was the Porsche PIWIS version of what OBD codes P0421/0431 mean. There is a lot to digest so I am going to reprint one page below. Note that P0431 has identical wording.

I was able to rule out their primary suspects pretty quickly. No, the sensors weren’t swapped and as noted were brand new. The cat was brand new. Durametric data indicated the valves were fine.

Like many folks have done, the previous owner of my car had ditched the “main cat”. It came to me that way. I drove it over 8000 miles and got it inspected last year in NC with no issues of this sort or otherwise. It remains a mystery to me that OBD requirements don’t look at the function of the main cat at all, just the warm-up cat. Whatever, that is how folks can delete the main cat and not record a fault code. Whatever, I found a fellow in Minnesota on RennList who had put the same Top Gear exhaust on his car. He also got Check Engine lights, that he could clear but which came back. He contacted Top Gear and they said to give it time to “learn” the new cat configuration. His very rough estimate was that after 2000 miles of getting the codes and deleting them, his car finally stopped throwing the codes. But I really didn’t want to wait for 2000 miles with an illegal car, with no confidence that would solve the problem!

So one question is, is a Top Gear 200 cell cat the problem?  Google says 200 cell is short for 200 cells per square inch of cat area. What is the OEM cat, you might ask? I could not find an answer to that question. In the After Sales documents I did find a statement that the warm-up cat was 4” in diameter and 3” long. Noting that I had one OEM warm-up cat on hand, it does indeed have that configuration. The Top Gear unit is of a similar size. So that was kind of dead end.

I’m an engineer. I needed data. But what kind and how to get it? One question in my mind was how does a narrow band O2 sensor know that the cat is working? A lot of reading led me to a working conclusion that if the cat is working, the O2 sensor downstream will typically be sensing on the “rich” side of things, which is another way of saying there is a lack of free oxygen in the gas downstream of the cat. If anyone has a better understanding of this please let me know. Slightly more complicated is the Porsche After Sales document explanation that the waveform of the upstream sensor is very “active” as it chases a stochiometric mixture while the downstream waveform is much calmer. I also read on forums that the efficiency target was in the area of 95%.

My working theory is that the ECU data logs a crapload of things, which are internally parsed and stored as part of the emissions tracking function required by US DOT/EPA. Going back to the image printed above, there are 9 bullet points that represent the Diagnostic Requirements. Taking that one step further, I am guessing that the 50 seconds (cumulative) within a rpm range of 1280 to 2880 rpm and an engine load of 22% to 41% gave me a strong message about how to formulate my testing. The other key parameter was the voltage value from the downstream O2 sensor. Here is a graph of O2 sensor characteristics from Pelican Parts. As you can see, voltages above 800 millivolts are clearly rich and voltages below 200 millivolts are clearly lean. Since my Durametric reports an instantaneous voltage on the O2 sensors, that is what I data logged.

Before I jump into my data logging, I am going to say that I found some discussion on the forums about those that had gone before me with aggressive or even no cats. No cats requires defeat mechanisms. Not my cup of tea. But one trick that came up was bung extensions for the O2 sensors. I bought 2 on Amazon for $18. I’m going to be diplomatic and call them a data smoothing device. Similar to another approach which I found, which was to add capacitors to the O2 sensor wiring. At this point, I am not needing to go down that path but it seems like it would be worth pursuing if the bungs don’t work.

OK, getting back to the data. I have data Before Bungs and data After Bungs. My test setup was to attach the OBD cable for the Durametric to my laptop, drive the car, and collect Actual Valves. I found pretty quickly that on a flat road, if I was in a high gear and shooting for 2500 rpm or so, the Engine Load values generally fell into the require range of 22% to 41%.

If you datalog O2 sensor voltages, they jump all over the place as you shift, accelerate, de-accelerate, etc. But if you keep the Engine Load and RPMs in the Diagnostic Requirements range, they are pretty stable. But not totally. BTW, I read somewhere that this cat efficiency check is only done once per engine start cycle. Anyway, I went to a nearby quiet road and collected data before I put the bungs in. Then I put in the bungs. Initially after installing the bungs, I just drove to see if I was setting Pending Codes P0421/0431. Amazingly, I was not! Then I took some data.

Again, all of this is a little bit of an educated guess. I exported the Durametric logging to Excel. I logged Time, Engine RPM, Engine Load, Upstream Actual Lamda (Banks 1 and 2) and downstream O2 Voltage (Banks 1 and 2). I sorted out and deleted all rpms below 1280 and above 2880. I sorted out and deleted all data with Engine Load below 22% and above 41%. This resulted in 279 lines of data for Bank 1 and 206 lines of data for Bank 2, where each line of data was one timestep. Yes, they should have been the same but whatever! I then sorted the O2 sensor voltages from low to high for both banks.  In terms of lines of data, here is what I got.

	Voltage < 0.5 volts	Voltage < 0.4 volts	Voltage < 0.3 volts
Bank 1 (before bungs)	46	33	26
Bank 2 (before bungs)	47	43	30
Bank 1 (after bungs)	16	13	6
Bank 2 (after bungs)	19	14	9

As you can see, adding the bungs really knocks down the low voltage (lean condition, excess O2) excursions. If the acceptance criteria is 95%, then there could only be 0.05×206 lines of data = 10 allowable excursions. We don’t come close to this limit without the bungs but we meet or exceed that limit with the bungs. Again, I don’t know what voltage Porsche was tracking, it might even be plausibly 0.2 volts or less.

I know that physically, without the bung, the end of the O2 sensor is directly in the gas stream. With the bungs, it is not. When an O2 sensor is tasked with making almost instantaneous adjustments to fuel mixture, being in the gas stream is a good thing. When I am looking for a surrogate to tell me that the cat is failing, I think a slowed down decision making process is in order. For a cat that is failing, the lean oxygen rich gas is going to get up in that bung and eventually throw a code. So again, I am comfortable calling the bung a data smoothing device, which is a well known and accepted practice when collecting and analyzing data.

So that is about it. I will post this article now and update it later to give the long term result. In conclusion the straight bungs worked for me. The car has been inspected and passed. I’m good for at least a year!