Lots of low load knock

Problem: I'm getting ALOT of low load cruising knock. No CELs. I attached some logs, they've been paired down from big ass logs, note the time scale on the left. Basically, at cruise conditions and off idle accel there is a ton of knock.

 

Context: CNT catted DP all else stock @65k miles currenty 105k miles. I have a fairly heavily modified mickeyd stage2 map (opensource) that I have been working on for awhile. Everything has been fine for quite awhile, I log ocassionally and take regular learning views to check the health of the tune and adjust timing accordingly.

 

Recent changes: Timing belt change 500miles ago, group N tranny mount.

 

I changed the timing belt with a gates kit and started driving the car March 3rd, a potential complication is that at one point when trying to wrestle the new belt on, I knocked off my vice grips on the driver side cams, they moved on their own maybe 30-40 degrees. I freaked out jumped onto the forum to see if that'd happened to anyone else had that happen of course found out that yes that has happened and most of the people say no harm done. I moved them back to position using my hands and re-clamped them and finished the job.

 

March 6th I changed to a group n tranny mount.

 

March 10th driving home on the freeway and the car starts stuttering and immediately I realize it's because the ECU is pulling a ton of timing. I did fuel up about 20 miles before this happened. Had the laptop so I hooked it up and threw on a "bad gas" emergency map I made which has 4 degrees of timing pulled across the board, compared to my most recent "no-knock" tune, in the cruise and non-cruise timing maps. Figure it must be bad gas.

 

Take it work on March 11th with the "bad gas" map drove to work to try to understand what conditions were causing it.

 

Drove a bunch more very gently trying to burn out all the gas prior to fill up.

 

Today, March 14th, fill up 15.9 gallons. Drove home 10 miles. Re-loaded the previous map I was running before all this started, jumped on the freeway to go a buddies house and it wasn't good (see march14 attached log) I chopped up the log to show the bad spots. Got to my buddies house was a bit pissed so I reloaded the bad gas map, unfortunately prior to taking a learning view, but I think the log speaks for itself.

 

 

Known issues: Spark plugs have about 40k on them, have a new set was going to change them Monday but after this happened I wanted to get a compression tester and test while they were all out. Also, I have a small vacuum leak at the IM to TGV interface on 3 of the 4 cylinders. Idles real smooth, but when I spray brake clean at the interface the engine bogs a bit.

 

My theories:

 

First one (seems most likely at this point) is that the drivers side valves touched when the vice grips fell off so now I have slightly bent valves and the knock is coming from the hot spots on the valves. This makes a bit of sense seeing as how today the bad knock came as I was going up long hills on I5.

 

Second theory is that the vacuum leak is playing a roll, although my most recent LV showed essentially +8 +5 -4 -4.

 

Third theory is that the plugs are done and they're causing misfire

 

Miscellaneous theories: the group n mount is transmitting extra noise, the knock sensor has failed, other engine noise (noisy cam follower), piston slap is causing false knock.

 

Any ideas from the guru's would be greatly appreciated. I'll probably end up exhausting all the easy fixes before I bother pulling the motor and the heads.

 

Thanks!

romraiderlog_march11_bad_gas_map.csv

romraiderlog_march14_good_map.csv

Whoa. I have never seen that much timing pulled in FBKC before. I also wonder why you have so much FBKC and yet in those areas it doesn't set any value in FLKC. Mmm.

I think normally if your plugs are so bad as to cause timing to pulled via knock correction, you'd see something in the misfire codes or roughness monitor, unless it only happens at high revs where the misfire algorithm doesn't work reliably. OTOH, it's cheap to buy some plugs for diagnosis.

Perhaps you can log boost error to see if the knock events coincide with overboost, which can cause FKBC or FLKC via pre-emptive knock control.

Take an LV to check for misfire or other codes?

Whoa. I have never seen that much timing pulled in FBKC before. I also wonder why you have so much FBKC and yet in those areas it doesn't set any value in FLKC. Mmm.

I think normally if your plugs are so bad as to cause timing to pulled via knock correction, you'd see something in the misfire codes or roughness monitor, unless it only happens at high revs where the misfire algorithm doesn't work reliably. OTOH, it's cheap to buy some plugs for diagnosis.

Perhaps you can log boost error to see if the knock events coincide with overboost, which can cause FKBC or FLKC via pre-emptive knock control.

Take an LV to check for misfire or other codes?

 

I figured this was the case during my per gear tuning, but have never seen a formal discussion about it. If there is one, pointing it out might prevent other people from castrating their timing table trying to fix a problem that's not there, but somewhere else.

 

The "knock" in FBKC and FLKC officially references the Knock Sensor. Yet the corellation between negative Boost Error and pulled timing events is irrefutablely obvious. It takes serious conviction in the face of danger to turn from one foe to another.

 

I wondered if perhaps there is a general acceptance of a missing parameter in the softwares... believing Subaru engineers surely wouldn't muddy such important monitors with unaccountable data. A separate table would make tuning far more direct and accurate.

Thanks for the feedback guys I appreciate the help.

 

I'll take some more logs tonight which include boost error to see if there is any probable cause or at least correlation between my knock events and boost error.

 

I'm going to start the evening with a compression test and then replace all the plugs, I have a set already. I think I'm going to check out the knock sensor too just to make sure it's not loose or broken. From the vacation pics, it looks like it's right underneath the intercooler, hopefully I wont have to remove the intercooler.

 

Thanks!

Lots of knock..i would make sure nothing rattling near the knock sensor and that nothing under the hood is loose. I am curious why fbkc is kicking in...flkc should be handling the bulk of this low level knock. The knock sensor is notorious for picking up noise below 3k rpm. Is there knock on WOT runs?

Which revision of the map did you flash? There was a large timing error in the earlier revisions and a fix was posted at some point.

 

That map was built on my car and I've spent years modifying it to various degrees and for various mods. Your timing advance table appears to be modified from the original based on the difference between total and base timing and considering IAM is at 1.

 

As for the low load knock, I used to get a considerable amount as well but have since been able to tune most of it out. I still have flkc disabled under 1.3 or so loads so it's not pulling timing for what I believe is noise in certain lower load ranges.

Lots of knock..i would make sure nothing rattling near the knock sensor and that nothing under the hood is loose. I am curious why fbkc is kicking in...flkc should be handling the bulk of this low level knock. The knock sensor is notorious for picking up noise below 3k rpm. Is there knock on WOT runs?

 

 

I agree, this is definately something I want to verify, especially since there is no direct probable cause for the knock going crazy.

 

As far as the fbkc vs. flkc I would imagine because there wasn't a huge amount of driving history on either of the flashes for the logs I showed you guys. There was/is minimal knock on WOT runs, but I'm hesitant to do any WOT for fear that there is a serious issue. I'd rather do just the heads rather than the whole motor.

 

 

Which revision of the map did you flash? There was a large timing error in the earlier revisions and a fix was posted at some point.

 

That map was built on my car and I've spent years modifying it to various degrees and for various mods. Your timing advance table appears to be modified from the original based on the difference between total and base timing and considering IAM is at 1.

 

As for the low load knock, I used to get a considerable amount as well but have since been able to tune most of it out. I still have flkc disabled under 1.3 or so loads so it's not pulling timing for what I believe is noise in certain lower load ranges.

 

I went stage 2 around Oct. 2011 and downloaded the map around then. Soon after I made some basic changes to the map after some learning views, then everything appeared to be healthy until my recent issues I ran on the same revision from 3/12 until about february of this year when I started tweaking small things in the name of smoothing everything out. I never added timing just took the single cell spikes out.

 

Seeing as how the majority of my knock is under 1 maybe I should increase the knock threshold, but it seems weird that I would all-of-the-sudden have to do it, especially for the quantity of knock that I'm not getting; something has certainly changed.

 

Thanks guys, I'm still open to ideas. I didn't get home until about 45 minutes ago so I will be doing the compression check, spark plug change, and knock sensor check tomorrow morning.

 

Cheers,

-Trevor

Good luck - I too struggled with some low load knock/noise and I too raised the threshhold (both RPM and load) for flkc. Been running great for 4+ years.

 

On my car it is completely random. Some days, I get very, very few low load knock events, other days I get continual knock sum increments if I hold it at a certain RPM (~2,300-2,500) but my WOT are always clean. I also noticed that wiggling the tactrix connector while ilding would occasionally cause an increment in knock sum. I have long suspected that some of this low level knock/noise that we deal with may be electrical in nature.

 

Good luck.

Updates:

 

Compression was good - cyl.1 = 145psi cyl.2 = 142, cyl.3 = 146, cyl. 4 = 141

 

2 and 4 may have been slightly slower since I had jumpers from the battery to the battery leads.

 

I also borrowed a borescope and looked at all the pistons, and I was even able to look at a small part of the seating face of the intake and exhaust valves. Everything looked good.

 

All the spark plugs looked great too... Replaced them as part of ordinary maintenance.

 

Pulled the intercooler and looked at the knock sensor, the only thing that looked out of the ordinary was the inlet tube from the intercooler to the throttle body was loose at the throttle body.

 

I'm putting it back together tomorrow morning and I'll see if anything changed.

3/18 update:

 

Drove to work this morning on my normal map, and minimal knock, about the same as I got before, which I always attributed to noise.

 

My LV I took this morning from about 100 miles of my bad gas map, after I put everything back togehter, no FLKC anywhere, and the upper two bins of AF learning both zeroed out. Still have some low load leak due to the intake manifold to TGV seal leak.

 

I'll post up my logs from the drive in to work today. I logged some boost error and am a bit suprised with what I was seeing. Time for some more logging and maybe if I can figure it out, boost by gear .

 

Thanks for your help guys, I'll post up the new logs after work.

Would have posted logs last night, but I developed a massive intake leak on the drive home. I visually checked all the vacuum hoses and there were still on, I'm suspecting the intercooler to throttle body hose popped off .

 

Details to come when I figure it out.

Curious...update?