Very high knock count

[Vermont]

Me and Grinder were working on his 07 STI ( Cobb OTS stage two with ECBS), when we decided too do a log. First thing we found was his IAM was at 0.6250 but had zero FLKC an his FBKC was jumping around constantly even under closed loop and just reving in neutral. The knock when it would occur was random, no set load or rpm. At first we thought it might be boost related knock but then we were seeing instances under engine breaking. I monitored cylinder ruffness count but that was stationary at 0. here is a picture of the Learning view I took before we reset the ecu.

As you can see his IAM is down and he was having corresponding random knocks before and right after. But the FLKC table has zero values, which if it was real knock should not be the case. Right now I am suspecting that the sensor might be going out, or something is rattling around/or near the sensor. He has noticed the power being down since about two weeks ago, no recent engine mods have been done too the car in the past few months. Knock has been happening in both CL and OL, although most of the knock was jumping around under CL light-medium throttle, some of the events were reaching -9.30 degrees pulled. This logging was done using a Taktrix cable and laptop. We took one log on the way back but were unable to get any type of knock event.

Any ideas we could get on this would be great. Grinder is planning on doing a track day tomorrow. He has already paid so he is committed and going but some help before tomorrow would be great.

 

[Td_d]

The FLKC table is cleared as soon as IAM mode (rough correction mode) is invoked - so the fact that the table is zero is just a sign that it's switched over. Normally you would see positive values, as the ecu starts adding back timing.

 

[Vermont]

The thing I am asking though is if this sounds more like a knock sensor malfunction issue or rather an issue with the tune that needs to be rectified. I am leaning towards the sensor as the car was fine earlier but now has this issue.

An update on the issue. Grinder called me up from the track. Looks like he threw a code out there on one of the straights. Waste gate too high or something like that. Looks like we are getting more and more problems just popping up left and right.

 

[Vermont]

Also, when you flash the car, what is the IAM starting value? If it is .5, then that IAM may not be a large issue. Just because the ECU doesn't want to advance the multiplier that doesn't necessarily mean there is something wrong. Also, what are IDCs and AFRs looking like now that it's cold out? Remember that there are timing advances built into the ECU based off of temperature. The stock values, if I remember correctly is 3 degrees at the current temps out. So, it may not want all the multiplier because of the intake temperature compensation. Look at total ignition timing in the logs.

We reset the ecu and it went back 0.8 and then stayed there. After the reset we were unable to get the car too consistently knock. On a few of the pulls back to the house we got some light knock but nothing over -1.4 and zero miss fires. When we first started monitoring the car we were getting large knock events up too -9 under CL light throttle driving and yet very few shown knock events under OL, also zero FLKC and zero miss fires (as monitored under roughness count per cylinder). There were no consistent load or rpm ranges at which we could make the knock show up. One pull it would be at 2k rpm with light throttle but next it would be knock free in the same range.

The reason I am doubtful of the tune is that he has been running this same Cobb stage two ots tune for a while now with no problems. I asked if he had noticed any drop in power and he said about two-three weeks ago that he noticed the car being a little slower and more sluggish, which can be explained by the low IAM. I did not check the IDCs or the AFRs that that time due too one time constraints and two I did not have the proper drivers for the cable connecting too his wide band. The next time we log I am going to fix this so I can watch the AFRs.

What would you suggest I look at next too diagnosis this problem? We are going too do a mechanical look over and see if we can find an obvious problems off the bat, but I doubt we will find anything. At this junction I suspect a part/sensor malfunction some where rather than issues with the tune. I also highly doubt that this knock is real, as from what I have seen it seems more like phantom knock that the sensor is picking up. But then again I am by no means an expert thus why I came too you all

 

[Vermont]

The FLKC table is cleared as soon as IAM mode (rough correction mode) is invoked - so the fact that the table is zero is just a sign that it's switched over. Normally you would see positive values, as the ecu starts adding back timing.

I had thought the populations in the table were kept and it is only when the motor is in fail safe mode that the table is ignored. I have seen a few Learning View shots over on rom raider where the table being populated by loads of negative and the IAM being well under 1.0. Could you explain a little more about the IAM mode as you call it? is this mode activated when ever the IAM drops bellow one? I had thought the IAM was only a over all multiplier and not a separate mode. The only time it is supposed too move is right after a reflash when the IAM is set bellow one and then the ecu slowly works its way back up too 1.0 too make sure there are no problems. Once it is out of that mode i thought it only adjusts the IAM if it sees enough knock that it can not be corrected with FLKC and FBKC.

That learning view shot is from before I reset the ecu. After the reset the IAM went back up too .8 but never moved from there during the next 15-20 min due to intermittent knock events.

 

[Td_d]

As soon as the ecu senses the need to drop IAM (due to a widespread sensing of knock, the ecu logic is very complex around this), it clears the FLKC table - basically, the ecu wants to check the true impact of dropping timing across the board - with a clean slate. It can then further drop IAM, or once knock is substantially reduced, slowly start raising timing in various cells of the FLKC table (and if happy, start raising IAM).

I would also check if the sensor is properly torqued, can cause false knock.

 

[Td_d]

Oh, and by IAM mode, I mean rough correction mode - when it the ecu moves from using FBKC and FLKC, and drops timing across the board.

 

[Grinder34]

hmmm, im a total newb to this. But both Fuji and Td say its likely a physical issue? I can remove the intercooler and torque the sensor down after the track this weekend.

If i can do any logs that would help shed light on the issue, please let me know. Its my car, and i'd very much like to NOT blow the engine, so any input, advice, warnings are extremely helpful. Thanks for bearing with my and my intro level knowledge of engine tuning and troubleshooting.

 

[Grinder34]

Ive noticed precious little oil consumptuon. less than half a quart per 3k miles.
that being said there somehow is always oil in the silicone elbow. my next purchace is a leakdown and compression tester.

 

[Grinder34]

Well i have your old ixiz. ill let you know oil consumption after a full track weekend.

 

[Grinder34]

it's plummed so the drain simply empties under the car.

I apparently used no oil at the track this weekend. 6 x 30 min sessions. Still reads full.

 

[Grinder34]

I just did a log. I now have a DAM (same as IAM, just for AP, right) of 1.000. I also have some FLKC.

So maybe i need a tune, but looks like the knock sensor thing figured itself out?

https://docs.google.com/spreadsheet/ccc?key=0Aics16LC4qhLdG14VC1zNnFUQU1tbktsMVVtUzBRLXc

 

[Vermont]

Dam and IAM are pretty much the same thing.

Hold on... If you have an AP, why the fuck is mark pulling a learning view???

Yes, you need some love on the tune.

He asked me too to do a quick log and look it over. I had my lap top ready so I grabbed it and plunked on RR. I used learning view becasue it shows learned values in a very nice and presentable way. I am not a AP user and hate having to figure it out just too log as I never use it.

Yes you could use a tune but honestly it is safe once again if your IAM is back up too 1. Also the amount of FLKC is the question. Most cars will have a little bit here and there. Its if you have a large amount memorized (i.e. -8 in a single cell) that you need to be worried and look over the tune.

This makes me think electrical issue (sensor/wire) as it is intermittent and not constant.

 

[Td_d]

All I see is tip-out FLKC, and only 2 cells of it, nothing significant, and very minor FLKC, that's on it's way out (-0.7).

 

[Vermont]

All I see is tip-out FLKC, and only 2 cells of it, nothing significant, and very minor FLKC, that's on it's way out (-0.7).

basiclly what I was trying too say. I.E. that its ok. When you see a lot of FLKC and deep numbers of them then you need to worry.