Long story short, I have an issue showing up in data logging that has been narrowed down to likely/possibly bad oem cats, through a process of elimination. I'm wondering whether it's possible to disconnect from the headers and just drop that down enough to peek into them with a flashlight to see how they look, without taking off the whole bumper and exhaust. (Which the local shop tells me is a *7 hour* job... Not sure if I buy that, but first want to see if it's even necessary...)

Oh, also, if it is possible to just disconnect the cats, could the car be driven like that? (Obviously would be loud..) Looking for the easiest way to get a data log without the cats to see if they're the culprit.

 

I believe you should be able to remove all 4 o2-sensors (pre-cat and after cat) and insert an inspection camera into the o2-sensor bungs. I know removing the after-cat sensors should be easy without removing anything. Not sure about the pre-cat ones.

btw removing the rear bumper is very easy if needed for the pre-cat ones.

 

Disconnecting the headers wont help, those are from the block to the turbo, you need to disconnect the exhaust at the turbo outlet.

This is a stock exhaust, yes? It's all 1 piece, so you will need to remove the 4 nuts on both turbo flanges and drop it down enough to poke an inspection camera in there.. I got one at Harbor freight that works pretty good.

Disconnecting the exhaust at the turbo is possible with the bumper in place but you need a modified wrench to get at the top nuts.. if it were me I'd just pull off the bumper. When I work on my car I like to not be frustrated and removing the bumper & intercoolers makes access much easier.. and I can do it in 20 min on my lift now that I've done it a few times.

I suppose taking out the O2 sensor just past the turbo and looking thru there is possible, but you would still need to remove the bumper & intercooler to get to the O2 sensor threads.

As far as driving w/o cats, since this a 1 piece exhaust, not really an option.

What are you seeing in the logs that makes you think the cats are suspect?

 

Of course, yeah, need to disconnect from the turbos, not the headers.

Basically I'm getting a fair amount of knock across all cylinders under boost. Sometimes one or two more than the rest, but it varies from around 6 to 12 across all. I _think_ it's generally a bit worse on 1-3 than 4-6, but not 100% sure.

Have already changed plugs and coils, and ruled out gas by trying it with 95 octane race gas available at the track here. (Knock was basically the same as on pump gas.) Talked to Mitch @ Cobb and he suggested the OEM cats might be the problem, as they've got about 50k miles on them.

Only local shop experienced with these cars is the dealership unfortunately, and they had no clue what could be wrong. They've got a request in to some central resource to help, and have sent them my logs, but haven't heard anything back yet. I suggested the inspection of the cats to them, but they said it was a 7 hour job, which sounds insane. Even if they were completely removing and reinstalling the exhaust, which shouldn't be necessary, it shouldn't take more than 3-4 hours, right? And just to disconnect and peek into the cats, even if they're removing the bumper, should be less than that.

 

Never ever heard of a bad 997.1 TT OEM cat, let alone both cats.

Can you share the CEL codes?

 

No codes. It's pulling the timing a bit under boost, but not enough to trigger the cel. Knock sums are 'only' in the range of 6 to 12 or so. Highest I've logged are in the 13s.

I agree that bad cats seems unlikely, but I can't think of anything else to check, and neither can Sam @ bydesign or Mitch @ Cobb, but both agree that the level of knock is not normal. (And I've compared to logs from other people's Turbos, which back that up.) Fuel trims look fine, so injectors should be ok. Valves looked good when the plugs were changed recently. O2 readings look consistent across both sensors, although that would be another straw to grasp at. Gas has been ruled out, as I mentioned. I'm really not sure what else could cause elevated knock across all cylinders.

 

Check the actuator rods of your turbos, the ball joints might be stuck and need to be lubricated.

Depending on the position of a seized rod the boost pressure will be either too high or too low.

If the boost pressure is too high you’ll get knock.

The rod should go back to its original position again automatically when you let it go.

On the right side turbocharger the rod moves upwards:







On the left side the rod moves downwards:

 

Thanks Stef.

Before I dig in there, the data logger does log target and actual turbo duty cycles, as well as measured turbo pressure before throttle. If one of the actuator rods was stuck, it should show up there, right? What's the expected pressure before throttle measurement?

In my case, the duty cycles start at 40% cruising at just under 3000RPM, briefly rise up to about 65%, then quickly drop back down to ~30% around where knock starts to get bad (with actual matching target very closely). The pressure starts at 15PSI before the pull, increases up to about 28 while the turbo duty cycles are elevated, and then holds at that level for the remainder of the pull. Here's a log: https://drive.google.com/file/d/0B0H...ew?usp=sharing

 

I would actually suspect it is the O2 sensors that are giving you that trouble. Have you checked into those? They are only about $ 150-180 a piece, and they do go bad - how long have you had those? If they are the original ones, I would try that first...

 

They are the originals I believe. I figured if they were going bad, the readings between the two wouldn't be so consistent (as seen in the logs). I suppose if they tend to degrade in sync though, they could both be reading similarly wrong. Would be a cheaper thing to try, anyway.

 

Mine had gone bad...and I would get those type of flashing cel's - constant missing. The reading on the scanner sometimes will not be obvious and it will only trigger the CEL after many backfirings...

 

Did you data log them? Both Sam and Mitch said it's pretty unlikely for them to go bad and still have the two sensors be so similar in their ratings (as in my log there). It's possible, but seems unlikely. Of course, it's unlikely for both cats to be bad too, but less so.

Anyway, I'll see what the Porsche factory suggests first I guess, then go from there.

 

Yup, and that is how we caught it - through datalog.