Stg 2 OTS = Knock?

[AF_3C2]

So- just had a bunch of work done to my car and am experiencing a lot of knock events- not certain if they are phantom or real, but wanted to run it by you guys while waiting for shop time.

 

Car:

- 05 MT, 96k miles

- Original owner babied, kept it dealer maintained

- 87k miles on stock tune, went Stg 1 at 90k, Stg 2 at 96k

- Was experiencing Cyl. 2 misfires on idle, suspected upper intake manifold gaskets as the misfire would get worse in summer, not so bad in cooler temps (below 50 degrees F), but was always constant at idle and would throw an occasional CEL- monitored with AP (cylinder roughness).

- Car eats mainly Shell 91 octane, every once in awhile Costco 91 (which original owner always ran on)

 

Work done at 96k-

- STI UP

- Grimmspeed catted DP

- Valve lash adjust

- Valve cover gaskets

- Upper intake manifold gaskets

- Swapped #2 injector with #3 (to test if misfire related to injector)

- One step colder plugs (NGK Iridium)

- CEL for removed EGT sensor removed from Stg 2 OTS map

- Gates Timing belt kit

 

So, after I picked the car back up I live monitored all cylinder roughness counts (looking for missfire issue), DAM, and boost. The DAM was at .312, but worked itself back up to 1.0 after a few driving cycles. No trace of the misfire could be felt- car idled great, ran great, and pulled like a freight train, but I could hear what sounded like an exhaust leak coming from underneath the car. Shop already closed for the holidays, so I could not take it back to have looked at. I suspected it was the UP, since it needed the heat shield to be re-welded. I verified later that it had be welded.

 

60 miles later, while cruising up a hill, the DAM suddenly falls to .1xx. No miss fires or noticeable changes to how the car was performing. I start changing my monitors to look for fine knock learning and knock feedback.

 

When cruising, below 3k rpms the fine knock learning fluctuates between -3 and -14. Any time I accelerate hard to above 3k rpm or hit boost, the fine knock learning goes to zero- sometimes just above. Knock feedback has occasionally hit -14 to -17 (never see this in real time). On a long highway trip, the DAM did eventually recover to first .6xx and then 1.0.

 

Last night I bring it to the shop for a quick inspection. Exhaust leak sounding noise is minimal, so I let the tech do some aggressive driving to see if it comes back- and of course it was not very noticeable on the drive. However, the DAM plummets back to 0.000, and fine knock learning is constantly in the negative when not hard accelerating. Also, slight missfires can be felt through the drive train, but are not registering as cylinder roughness yet.

 

Now I am a bit freaked- all the horror stories of cracked ringlands and burnt valves has me worried about this car. Owe too much to get rid of, wanted more performance over stock, but now worried there is some serious issues.

 

Shop claims the car needs custom tune- just some don't cope well with OTS maps. Initial recommendation is to run a 91 ACN map to see if that helps with the knock detection, and bring it back in the for the mechanical rattle noise. Based on what they might find, compression test (and leak down if compression bad) is in order to see what I might be in for.....

 

I already dropped $2500 on this work so far. Custom single map tune starts at $500- money a bit tight, but I understand the whole pay-to-play mentality, was just hoping I wouldn't come across an issue like this.

 

Any advice on what to log for to help determine what might be wrong here? No current CELs, car still runs strong- just the crappy exhaust leak-type noise, the obvious issues the ECU is seeing by pulling timing.

 

I do have a third party warranty on the drive train- when I had the shop look over the paperwork (to see if any work I had done would be covered) it appears the warranty only covers severe failure of lubricated parts- ie. the car pretty much has to throw a rod to be covered. Cracked rings and burnt valves do not apply......

[serx7]

I don't think running the ACN 91 map will make much difference at least not for cruising (it runs the same timing in cruise, but does run slightly richer w/ a little less timing and less boost under load).

 

Since you're seeing big FLKC corrections during cruise, my 1st suspicion would be something rattling underhood/underchassis that's tripping the knock sensor (assuming the knock sensor is working properly).

 

I'd probably have the compression and leak down test done anyway for peace of mind, or to reveal a problem.

 

Seeing the FLKC go to zero during acceleration is probably b/c you're operating the engine in different load/rpm cells and no FLKC has been learned/stored there. Seeing FLKC go positive means DAM < 1.0 and the ECU is experimenting w/ adding small amts of timing back (0.35 deg steps) to see if the previously-observed noise returns or stays quiet.

 

Keep in mind that the FLKC table gets cleared out if/when DAM changes, so if you just ran the car through a certain RPM/load range, saw FLKC and later don't see any FLKC, it could be that the FLKC learned itself out (incremented back to 0 correction) or the FLKC table got cleared b/c of a DAM change.

 

If you aren't doing so already, do some datalogging to see if anything suspect appears in the logs.

 

For general stuff I usually monitor the parameters below. If your AP is running the latest firmware your sampling rate should still be in the vicinity of 10-12 samples/sec, which IMO is still pretty good.

 

A/F Sensor 1 Ratio (AFR)

AF Learning #1A (also B, C, D for a brief idle log to see what these are)

Boost

Calculated Load

DAM

Feedback Knock (FBKC)

Fine Knock Learning (FLKC)

Gear Position

Knock Sum

Ignition Timing

Inj Duty Cycle

MAF Voltage

RPM

TD Boost Error

Throttle Position

Wastegate Duty Cycle

[AF_3C2]

Thanks for the heads up.

 

The shop really wants me to part out $500 for a custom dyno tune- something I am not really prepared to do right now (early Christmas, still paying for a complicated birth, and some home repairs). Plus, I have been planning on letting Cryo do my tune.

 

When I get my logs, just post them on here for review, or do I need to email them? I wouldn't have much clue as to what I am looking at other than some basics, and I don't have much personal time at home (school, kids) to really get into it and learn more about tuning.

 

Talked to the shop today- they did not find anything obvious when they took the DP back out for inspection- but they did mention a bracket that might have possibly been the source of the mechanical noises I was hearing. They moved it and re-secured it out of the way, and will still proceed with loading the ACN map and test it tomorrow.

[covertrussian]

Learning view and a log of a pull would be very useful.

[nm]

Since you're seeing big FLKC corrections during cruise, my 1st suspicion would be something rattling underhood/underchassis that's tripping the knock sensor (assuming the knock sensor is working properly).

And as this implies you could have a dead knock sensor. Mine was going nuts and a cheap part fixed everything.

 

That said, long term a dyno tune will be healthier for the car.

[serx7]

OK sounds like a plan. I've read/heard that the OTS tunes sometimes knock a -little- in cruise but I'm pretty sure that the OTS would not cause the amt of knock you're seeing.

 

You can probably just attach the logs to this thread. The OTS starts w/ DAM = 0.7 IIRC, so after the reflash and w/ a warmed up motor, a moderately heavy throttle spurt from 2500-4000 rpm in 3rd or 4th will probably bring DAM up to 1 (initially). If the cruise log is clean/significantly better than before, try a 3rd or 4th gear 2000-whatever WOT log.

[AF_3C2]

Don't have the car back yet- it's getting a compression check done, but they did find the following issue that is interesting......

 

With the 91 ACN, boost is still hitting around 17 psi, which the shop feels is a little high for stock components like the intercooler. I know others have run over 20 psi on stock intercooler for long periods of time with minimal issues, but the shop is concerned this is too high and blames the catless up pipe since the tune itself calls for no more than 15 psi as target boost.

 

The real worry part is the ECU is still pulling around 5 degrees of timing after seeing knock events on deceleration. No knock is seen during acceleration to the best of their knowledge.

 

From my reading here, and I may be mistaken in my assumption, is that going to a catless should not cause so many of these issues. Does my combination of doing UP, DP, and the tune combined with my altitude providing a perfect storm for cause these issues?

 

My rough idle has come back- and once more the ECU is starting to see roughness on Cylinder 2- it had been silky-smooth after all the work had been done, but had started getting rough again by 80 miles after picking the car up. When I started noticing it, it wasn't enough for the cylinder roughness monitor to pick it up- the shop says it can see it now....

 

My fear is cracked ring land or burnt valve on #2.......compression results will be posted when I get them.

[serx7]

Peak boost/target on the Stg 2 91 ACN map is 15.66 psi b/n 2400-4000, tapering from there to about 13 psi by redline.

 

It's overboosting and I'd prefer observed peak to not exceed ~16.5 psi in the higher gears, but 17 psi isn't that that bad, but could certainly be getting you too close to the knock threshold. How long does it see 17 psi?

 

Datalogging shows -5 deg being pulled right at throttle lift? or during actual deceleration (which I wouldn't expect since there's like no engine load off throttle)?

[AF_3C2]

I am not sure- the tech just said there was timing being pulled due to knock detection on deceleration.

 

That was what made some head scratching, and that plus the rough idle has got me thinking the compression test is going to come back with bad news for #2.....

 

Will be calling them back shortly since they have not called me here at work yet.

[AF_3C2]

Compression results are in:

 

#1 = 112

#2 = 105

#3 = 112

#4 = 107

 

Shop says they think most of the roughness and knock can be tuned out.

 

I'm a little worried about #2- maybe the beginnings of a burnt valve?? All values are lower than what might be seen at sea level- I'm sitting at almost 6k feet.

[serx7]

I found a compression/altitude chart online which indicates at ~6000 ft elevation, the correction factor would be 0.8395 compared to sea level.

 

I presume that means your #'s @ sea level #'s would be something like this:

 

#1: 133

#2: 125

#3: 133

#4: 127

 

These #'s are marginally better than my built motor's sea level compression #'s (120 in cyls 1, 2, 4; 115 or smt like that in #3.)

 

The only issue my car exhibits is moderate oil consumption in the range of ~1 qt per 1000 mi with mainly highway driving and a daily handful of boosted intervals b/n 3000-5500 rpm typical.

 

I'm still curious that your shop only saw knock upon deceleration, not ACCeleration. Did they show you the logs? Seeing is believing...

 

You can probably pseudo test whether your knock can be mitigated by reducing global timing by -3 to -5 deg via your AP (it's in the Tuning / Adjustments menu). I would hope that on the Stg 2 91 ACN map w/ 'good' 91 octane gas, you'd see an improvement in the knock signature with that much timing pulled across the board.

 

You won't want to run w/ globally reduced timing long-term, but I found it useful when testing around suspected false knock on my car. My IPR TMIC was slightly resting on the pitch stop mount and was causing false knock. Initially I thought it was knock near peak boost (since that's where it always showed up), but I was able to zero in on the false knock assessment when reducing global timing by -3 to -5 deg via the AP had -no- effect on reducing logged knock.

 

[Edit: can you get a Live View? If not, do a quick log (warmed up, idle) while logging AF Learning #1A, B, C, D. That'll show you long term fuel trims and can help you see if you might have a vac/boost leak. You'll probably want to drive normally for a day or 3 before doing this, since the fuel trims, esp the C & D ranges, can take a little while to settle in depending on driving conditions.]

[AF_3C2]

Will be picking up the car tonight. Hopefully starting tomorrow I can start grabbing some logs on the morning commute- 20 miles one way, half in town, half highway.

[nm]

Compression results are in:

 

#1 = 112

#2 = 105

#3 = 112

#4 = 107

 

Shop says they think most of the roughness and knock can be tuned out.

 

I'm a little worried about #2- maybe the beginnings of a burnt valve?? All values are lower than what might be seen at sea level- I'm sitting at almost 6k feet.

Why didn't you do a leakdown? Shit can hide in a compression test.

My last compression test on this:

http://notajetta.com/lgt/engine/IMG_4444.JPG

(the missing metal was there when they came out)

was just fine. Only the leakdown told the tail.

[serx7]

Oy... out of curiosity, what were your leakdown #'s (what were the compression #s too)?

[nm]

Oy... out of curiosity, what were your leakdown #'s (what were the compression #s too)?

I don't recall them exactly, but the #2 and #4 had bad leakdown.

 

Compression was all in the low130-high120s IIRC.

 

It wasn't as bad as it looks, the ringland bits were still there until the piston was removed. Thankfully, they didn't work loose.

I was going through about 10qts of oil per 4000mi. That was the only tip off until the leakdown. No CELs. Drove fine and even pulled hard (hard enough to blow my transmission)

[AF_3C2]

The shop didn't believe in doing a leak down on top of the compression- wanted an extra couple hundred to do it as well. I did ask.

[AF_3C2]

So- I tried to log this morning on the commute, but the AP would not let me log as many parameters as serx listed in post #2. From the live monitor gauges of fine knock learning, knock sum, feedback knock, roughness #2 and roughness #3 and DAM, I saw something else this morning....

 

I waited until my DAM came up to 1.000 (started out at .500 from the ECU reset), I noticed that whenever my engine was under load and at least half throttle applied while the engine RPMs were between 2500 and 3000, the knock sum counter would go up, feedback knock would vary from 0 to -12. After a few tries of harder acceleration in gears 2-4 in this RPM range, the fine feedback knock eventually entered a -4.25 for this RPM range. Did not see any issues for deceleration like was reported yesterday- nor did I see any live monitoring of cylinder roughness, but could feel it at idle when sitting at stop lights.

 

MY DAM was still sitting at 1.000 at the end of my commute- but now the -4.25 is stored in the fine knock learning table for the RPM range of 2500-3000. Does not appear to knock at any other time.

 

False knock/external noise? Too sensitive of a sensor?

OR

thinking there might be something else going on internally- especially with this idle roughness that I have done everything else except coil, injectors and fuel pump to fix? Any chance a weak fuel pump could cause some of this?

[nm]

The shop didn't believe in doing a leak down on top of the compression- wanted an extra couple hundred to do it as well. I did ask.

 

They were wrong. Find a new shop, anyone familiar with the EJ would know that a compression test isn't good enough.

[serx7]

^agree. I think I paid 150-200 for compression + leakdown on my car before it went on the dyno for a protune last year.

 

Do you still see the big timing corrections during cruise like you mentioned in your OP?

 

Based on your recent post it looks like you have an AP3. If you tried logging the parameters I listed -plus- the AF Learning #1A, B, C, D along w/ several gauges displayed in the live view, you'll probably exceed the max # of parameters. Try removing those and see if it'll let you log. W/ my AP3, I log 15 parameters and have 5 gauges displayed in live view, and can successfully log.

 

If you don't have access to a laptop and Learning View, you might still want to do a short idle log that only includes AF Learning #1A, B, C, D to see if your long term fuel trims are OK. Alternatively just see what they are via live monitoring. The B, C, D trims won't change at all during idle, while the A might fluctuate a little bit (A covers idle airflow conditions).

 

Does your car eat any oil? Did you try removing the oil filler cap w/ the motor warm idling, to see if you see any vapor puffs emanating from the filler tube?

 

Assuming your knock sensor is OK and you don't have motor probs, I'd still suspect something mechanical is triggering the knock sensor.

 

Get the leakdown test done by a reputable shop.

[AF_3C2]

Alright- finally got Accessport Manager installed without hanging on my computer. Shop left some of the old pull logs on it, so I downloaded.

 

Datalog10.csv shows the knock events with pulled timing on deceleration.

 

They had swapped my injector from #2 (was previously seeing roughness on this cylinder) with #4 and there is some roughness now on #4- they told me it was coming back on #2..... maybe a starting to fail injector??

 

Would be better than a cracked ringland or burnt valve.....

datalog10.csv

[AF_3C2]

Datalog15 here appears to show a cold start after an ECU reset- the DAM is .5 and the RPMs hover around 1500- notice roughness on #4- I do not see this often on the live monitor.

 

None of the logs show the large timing corrections I was seeing on the live monitoring. Now that I got the AP squared away, I hope to be able to log a few better runs in the morning.

datalog15.csv

[serx7]

Alright- finally got Accessport Manager installed without hanging on my computer. Shop left some of the old pull logs on it, so I downloaded.

 

Datalog10.csv shows the knock events with pulled timing on deceleration.

 

They had swapped my injector from #2 (was previously seeing roughness on this cylinder) with #4 and there is some roughness now on #4- they told me it was coming back on #2..... maybe a starting to fail injector??

 

Would be better than a cracked ringland or burnt valve.....

 

that doesn't look like deceleration knock to me. would be nice if they had included throttle position.

 

if you look several lines above where the knock gets logged, you can see boost and RPM rise quickly (to nearly 6000 rpm) along w/ high WGDC, then RPM drops briefly w/ associated vacuum (negative boost) and 0 WGDC, followed immediately by a return to yet higher boost and (briefly) increasing RPM again.

 

looks to me like that was an upshift, followed by knock. might be shift knock or knock due to the loaded condition in the next gear (appears to be a 1-2 upshift since peak boost before the shift was only around 14 psi while boost after the upshift hit 17.6 psi).

 

immediately after the large knock #'s appear, the RPMs and boost values drop off rapidly. they probably aborted the pull upon seeing the knock.

 

the logging sample rate is very slow (3 samples per sec). if you apply the latest firmware update, you should see logging sample rate increase to 10-12 samples per sec w/ the # of parameters your AP is currently configured to log.

[AF_3C2]

AP is updated to the latest firmware I believe- will check it again tonight when I pull the logs from this morning's commute.....

 

I bought a used injector set from a member here, and will probably send the whole set off to DW for servicing so I can replace them all- certainly think I have #4 problem now.....

[serx7]

I just remembered you were on the Stage 2 91 ACN OTS map right? That has boost targets 15.66 psi (2800-4000 rpm, 15.37 psi @ 4800 rpm) and your datalog10 showed a peak of 17.64 @ 4165 rpm.

 

Logging sample rate is too slow to know if that was the actual peak boost right after the upshift (it could have been higher but didn't get picked up by the log).

 

IMO 2 psi is too much overboost, and could easily cause knock on a consistent basis if not addressed. If that was a 1-2 upshift as I suspect, you're likely to overboost by as much if not more in higher gears (or in higher gear upshifts).

[AF_3C2]

This is something the shop was concerned with as well- they claim the catless up pipe is to blame and needs to be addressed with a tune.

 

I am considering a 3 port EBCS as well....

 

And yes, this is currently on a 91 ACN Stg 2 due to the overboost concerns seen with the normal Stg 2.