Badly need help - random knock after rebuild

[Praetor353]

Hey guys.

 

Car - 05 LGT 5mt running Irving 93 octane

Mods - Up, down, catback, intake, APv3 with stage 2 91 OTS map

 

I've searched until I couldn't take it anymore, and have been sitting, thinking and researching on this problem for over a month with no luck. I'm at a point where I want to drive the car off a cliff, I need help diagnosing this problem. I put a new shortblock in as well as all new gaskets, had the heads redone etc. The problem I'm having is the car goes through random segments of bad perceived knock, with over 12 degrees pulled. The computer eventually drops the DAM to 0 and learns -12+ degree over 2000-3500rpm ish. I've ruled out spark plugs and fuel quality, and the problem is present in the same way when I bump down to the 91 octane OTS map. If i reset the ECU, it doesn't seem to make a difference. It only drops 12 degrees once every 2 or 3 days, its not constant. I drove the car for 3 days this week without any significant knock, and with DAM at 1. The other 25% of the time it will retard timing, drop the DAM and shut off boost control, become rough when cruising as it goes through different degrees of dropped timing.

 

I would appreciate anything here, and don't mind paying for any help (paypal). I can't help but feel whatever is causing this is simple, possibly a sensor. Should I contact Infamous and see what he can do, or are there any other e-tuners who could help here? I'm not sure what my options are here. I took a few logs on the AP - links below. Not sure if I got all the correct parameters. Sorry for the hectic post, I can answer any questions or do more logs. Thanks in advance.

 

https://drive.google.com/open?id=0B6-ieR_Gb48qZUdPR1ZJRlMwZDQ

https://drive.google.com/open?id=0B6-ieR_Gb48qUjNyVXBqVzlNYVE

https://drive.google.com/open?id=0B6-ieR_Gb48qWFZrS2ozZHZ2c1U

[Max Capacity]

Your going to want to contact Mike Kinsman of http://www.tuningalliance.com He's one of the best tuners for these cars.

 

You can tell him I sent you.

[Praetor353]

Max Capacity, thank you - I just sent an email. Will wait and see what they say.

 

If anyone else has any ideas/questions in the meantime or wants to see logs, let me know

[serx7]

Your datalog8 shows knock or FLKC being applied right after you shift, and the last instance of corrections in that same log shows knock being registered about 0.8 sec after you let off from a brief 100% throttle application. Have you gotten into/under the engine bay to see if anything is loose/rattling?

 

I've seen similar low load/low-ish RPM knock behavior in 2 of my Subeys (my prev car, a 2005 Saabaru 9-2X Aero and my current car, a 2005 OBXT). '

 

On the Saabaru it was a marginally loose turbo heat shield. I could actually hear it rattle w/ the windows down during certain shifting and in-gear throttle on/off transitions.

 

The OBXT's causes included:

 

- my IPR TMIC very slightly resting on the pitch stop mount. knock would show up in logs b/n shifts and sometimes when sharply letting off the throttle, often pulling several degrees of timing. re-positioning the TMIC slightly greatly reduced the frequency of knock showing up in datalogs.

 

- Nameless midpipe just barely touching the rear diff. again, knock would occasionally appear in logs during shifts or b/n sharp throttle transitions. the heat xfer also seemed to have degraded the rear diff cover seal which I ended up having to have replaced (the seal, not the diff).

 

- built motor noisy after wintertime cold starts. even after warming up for 2-3 min to let the pistons expand a bit before gently driving off, it would 'randomly' pulls large amts of timing just cruising down a local country backroad. after fixing the above 2 causes, my only solution to this cause was to disable knock control below certain load/rpm in cruise conditions.

 

- shift knock (FBKC, not FLKC). everything i've tried from reading many posts from ppl w/ similar behavior and a lot more tuning experience than i have, has not completely stamped this out. my clutch is somewhat noisy, which I think might be the root cause here. if i have the car idling in the driveway w/ the clutch pedal in and the shifter in neutral and quickly release the clutch pedal, it makes a -very- audible, sharp metallic 'click'. for the time being I'm living w/ it. i've reduced the FBK advance delay in the tune so that if FBKC does get registered, it restores the pulled timing about twice as fast as stock. so far so good, no rebound knock shows up in the logs so it seems the change I made is acceptable/appropriate. the benefit here is that if FBKC registers right after a gear change, it doesn't maintain the -2 to -4 deg timing reduction the whole time the car is running through the next higher gear.

[Praetor353]

Serx, thanks for the ideas. I have checked for anything obvious rattling or causing noise the knock sensor could pick up on. I will thoroughly check again, however I have a feeling it may not be that. It's picking up knock mostly when I'm not shifting, for example today fine knock control retarded the timing around 2500rpm 11 degrees, then I felt a roughness and watched as feedback knock dropped another 12, no shifts just light throttle. Cruising backroads under light throttle at 20 degrees advanced when I'm supposed to be around 40. I will recheck for rattling.

 

I replaced the clutch with the build as well, though its an exedy stage 1 with wrx singlemass flywheel. No chatter whatsoever.

[fahr_side]

Loose heatshields etc. Knock sensor not properly tightened to block.

[serx7]

Take care how you interpret FLKC as seen in datalogs. FLKC is a learned correction, so if you see FLKC in the log, check to see if the Knock Sum value remains unchanged, or increments when the FLKC appears. If Knock Sum remains unchanged when you see FLKC appear, the timing reduction is coming from a learned correction that happened during a previous actual knock event. If you see FLKC in the log -and- see Knock Sum increment at the same time, then the log has captured FLKC being applied as a result of a new knock event. It will then appear in future logs when the engine operates in same load/rpm range.

 

In your datalog8, I saw large FLKC applied but also saw Knock Sum increment after it had already been applied, but that was b/c there was an instance of FBKC (actually it appears to be a triple-count application of FBKC). It showed up right after an upshift, which is what made me suspect shift knock or some other noise occurring as a result of the upshift. See what happens starting at line 68 of datalog8 (FLKC -11.75 with Knock Sum incrementing 1-2 rows earlier in the log), then see what happens b/n lines 71-73 and beyond (FBKC appears, -2, then -4 then -6 with Knock Sum increment each time the FBKC value changes). The FLKC and FBKC both occur as you get back on the throttle after having briefly lifted off the throttle.

[Praetor353]

Very true, I do notice that the computer learns corrections. I'm constantly watching fine knock learned as well as feedback knock when I drive so I know when it's actually knocking versus hitting a learned correction in the table

 

I'll look over the car today for anything that could be knocking tapping or hitting. I'll pull the intercooler and double check the knock sensor itself. Additionally I'll start resetting the ecu and try and capture the event better.

 

How was my datalog? Did I log the right parameters, and the right amount of parameters? It's my understanding the resolution of the log drops the more parameters you have

[serx7]

Just keep in mind that it -is- possible that FLKC you see on the AP gauge -might- be in response to an actual knock event as opposed to the ECU proactively applying a previously learned correction. The main way to distinguish is to see knock sum the instant the FLKC shows up.

 

Normally it's recommended not to reset the ECU b/c if the knock is real and you reset the ECU, the Fine Learning table is cleared, and as soon as the engine is operated in a problematic load/rpm range again, it will actually knock again. In your case, it seems the FLKC is primarily showing up aat low-ish RPM/low-ish load, so you are probably ok. If you do reset the ECU, you -must- log -full time- until you see the FLKC get applied again, and you may want to drive around a fair bit, including shifting/etc for a while, to maximize the chances of catching more scenarios that kick in the FLKC. If you don't, the log may not catch the actual event/s that kick in the FLKC.

 

As for the logging parameters, I think for what you're trying to accomplish currently, your parameter list is OK. You're seeing 15-16 samples/sec which is decent. I log something like 20 parameters and see 11 samples/sec. For day to day use I should probably cut that # back but I haven't really needed higher resolution and if I start removing parameters, I'm likely to forget to add them back in if I decide to log for specific reasons in the middle of a drive. Ex: After installing the GS CAI, and also later after installing the GS dry element filter and replacing the MAF sensor, I was going b/n logging for MAF scaling (closed loop) and logging for WOT AFR (open loop). I can't tell you how many times in the middle of a drive I'd decide to do some MAF scaling logging and would completely forget to add Closed Loop Switch, AF Sensor 1 Ratio, AF Correction and AF Learning 1, so after driving around for like 30+ min steady state across all kinds of airflow ranges, I'd get home and go into a mini-depression once I realized I had forgotten to re-add those parameters.