*** Knock, do you have any?

[fishbone]

iNVAR, my memory may be failing me, indeed, but I have one thread about my Stg2 tune. It hasn't been particularly active. No matter. I thought the purpose of this thread was not to figure out whether or not knock was there, but to investigate or get direction as to solving it, or in general talk about when it happens, how, why, etc etc. Yes, you are coming off as an anal retentive jerk but now I see why.

Granted back when this started we didn't have nearly as many logging parameters on the AP from Cobb and we had to kinda figure it out the hard way.

I know what needs to be done, the reason I am still engaging is to figure out why this damn thing is a repeating scenario come Spring. That, and to learn a thing or two by talking to everyone else and seeing what other's experience has been. Wouldn't adding me to the ignore list stop sending you notices?

 

bmx045:

 

LE:

We can (and likely will) get more complicated, but I want to keep this thread simple for the noobs so they know what to look for.

I get it. I get how the knock control strategy works etc so this thread is no longer for me then.

Edited May 12, 2011 by fishbone

[iNVAR]

http://legacygt.com/forums/showthread.php?t=112604

http://legacygt.com/forums/showthread.php?t=90258

So 2, not a bagillion I guess. You started 2 for your car specifically but it seems like you're talking about the IAM on your car dropping below 1 in a bunch of other threads too. Blah, sorry I snapped at you man, but my point still stands.

[fishbone]

I landed in this thread on page ... definitely not 1 so that's why I may have missed the original intent.

Those 2 threads of mine were actually 1, but I asked the mods to split it since the first section was hardware, and second was the tune. Once I got the hardware figured out I was hoping I'd get more exposure by splitting one and moving it to the tuning section.

 

Anyways, no such thing as bad gas, just bad tunes for the given gas, right?

[LittleBlueGT]

Could it be the FBKC burning off? For instance..assuming knock sum is at 0 and there is a knock event. knock sum goes to 1 and FBKC says -2. On the next row of the log, knock sum remains at 1 but FBKC says -1.05 as it burns off off the pulled timing?

 

Read the RR stickies, but yes, knock sum increments once, FBKC will go to -2 (unless you changed the value) then it slowly decays down to nothing. If you had another knock event and FBKC had already decayed down to -1ish then it would go up to -3.

[bmx045]

I reset the ECU the other day and proceeded to do a 3rd gear pull 2k-6k, I noticed my knock sum fluctuate under normal driving, sometimes up to 23, however there was no fbkc or flkc --- all 0's. When I did the pull, the knock sum was steady at 7 and still, all 0's for fbkc and flkc. I did a brief pull on april 23rd, without resetting ecu, and knock sum was 1 and fbkc and flkc was 0. Is this purely because I reset the ECU prior to my pull?

[spec.B Dream]

Yep. Resetting the ECU (and also every time you start your car) knocksum is reset to 0.

[spec.B Dream]

Also, knocksum randomly increments during cruising. It drove me crazy at first, but I have now learned to accept it. Your wot pulls are free from knock, so that is good. As long as LearningView is clear you have a healthy running car (from a knock point of view).

[fishbone]

Knock sum counter goes back to 0 once it hits 34 or so, doesn't it? At least that is what it does on the AP.

 

Question for clarification regarding the propensity of the knock sensor to pick up false knock.

False knock is likely to be reported at low load and low rpm, correct?

In other words, if there is flkc at higher rpms but still low load, (below 1.6) that is true knock, correct?

What I need to better understand is ... the alarm over seeing learned knock at random places in the Learning View table. I am not dismissing it as alarmist, I am just seeking to understand. Knock at over 2800rpms for example, even though it is under 1.6 load can still be destructive? Because I have spotted it here before, the train of thought "don't worry about it, even if it is true knock, at that load it is not destructive and nothing to worry about" in the context of debating whether or not a certain event was true knock or not.

Edited May 19, 2011 by fishbone

[iNVAR]

Knock sum is exactly what it sounds like: it's the total number of knocks since you started your car. The knock sum counts goes back to 0 when you shut off the car, or like fishbone said, when it reaches something like 34... I forget the exact number but it's around there.

 

If I'm not mistaken, FBKC (FeedBack Knock Control) is what will cause an increment in your Knock Sum count. FLKC (Fine Learning Knock Control) does NOT increment knock sum count. A FBKC event occurs when your engine experiences a knock event... it's detected by the knock sensor. When enough knocking at a specific load/RPM range occurs, your ECU learns to retard the timing in the future. You can see this table by pulling a Learning View. When your engine next hits that load/RPM range that has been learned, your engine retards the timing and you now have a FLKC and because no actual knock occurs, Knock Sum does NOT increment.

 

I'm trying to go off of memory here so bear with me and correct me if I'm wrong.

Edited May 21, 2011 by iNVAR

[Legend]

Knock sum is exactly what it sounds like: it's the total number of knocks since you started your car. The knock sum counts goes back to 0 when you shut off the car, or like fishbone said, when it reaches something like 34... I forget the exact number but it's around there.

 

If I'm not mistaken, FBKC (FeedBack Knock Control) is what will cause an increment in your Knock Sum count. FLKC (Fine Learning Knock Control) does NOT increment knock sum count. A FBKC event occurs when your engine experiences a knock event... it's detected by the knock sensor. When enough knocking at a specific load/RPM range occurs, your ECU learns to retard the timing in the future. You can see this table by pulling a Learning View. When your engine next hits that load/RPM range that has been learned, your engine retards the timing and you now have a FLKC and because no actual knock occurs, Knock Sum does NOT increment.

 

I'm trying to go off of memory here so bear with me and correct me if I'm wrong.

 

Mostly correct. FBKC does not increment knock sum. Knock sum is incremented by the knock sensor detecting sounds that either are knock or similar to knock. One consequence of incrementing knock sum is FBKC, which will immediately retard timing (by 2 degrees in stock tune). Then that retard amount will decay to zero and/or get decremented again in the event of another knock signal.

[fishbone]

Right. Also, knock increment can be seen with no FBKC or FLKC. At least that is what I have seen in some logs.

[LittleBlueGT]

Knock sum results in one of 4 things:

 

-nothing

-FBKC (usually seen when load is quickly changing)

-FLKC (usually seen when load is stable)

-IAM change (usually seen after FLKC has already retarded things a bit)

[bmx045]

What will bad gas cause? An usually high FBKC? if, during a bad tank of gas, there is a lot of FBKC and the ECU decided to pull timing and register FLKC, will the ECU always stay like that? or will it periodically relearn? I.E. will a learning view change periodically if you don't reset the ECU?

[fishbone]

The ECU always learns. It will pull FBKC or FLKC whichever is warranted, but not both at the same time. it's either one or the other. FBKC is not learned. FLKC is learned.

If it is FLKC it will learn it but the ECU's single goal in life is to hit the targets you give it. So, over time, it will experiment with advanced timing hoping that it will eventually get itself back up.

IAM will also drop accordingly to reduce advance across the board.

Normally you would never see positive advance (in blue) unless timing has previously been pulled so the value is less than it should be and it's trying to get back up.

If anyone knows how long exactly it takes for the ECU to go through it's experimentations I'd like to know, but I do know that if you know it was because of bad gas and now you got good gas, it's easier to just reset the ECU.

Bad gas can cause both FBKC and FLKC, depending on how you are driving and what the ECU is seeing.

Edited May 27, 2011 by fishbone

[sumfoo1]

Still on my ecutek tune and logging the snot out of it before i re-tune the car,

 

Today it was 96 degrees, my iam dropped to .875 , the ecu is pulling 6 degrees at peak torque is this pretty normal for a hot day and a crappy inter-cooler or should i pull some timing off my tune?

[iNVAR]

I don't think that's normal. Yeah, it's hot...but I think there's a temperature compensation table that you need to tweak or something.

 

It's probably from heat soak...

 

Someone else jump in here.

[LittleBlueGT]

Still on my ecutek tune and logging the snot out of it before i re-tune the car,

 

Today it was 96 degrees, my iam dropped to .875 , the ecu is pulling 6 degrees at peak torque is this pretty normal for a hot day and a crappy inter-cooler or should i pull some timing off my tune?

 

That is not normal. Properly tuned that will not happen. (unless you had a bad tank of gas or something)

[Deadleave]

Low load knock, no IAM drop, no FLCK, but still felt like car was going to die.

 

I have been noticing very high knock sum at start up/idle/or at low rpm/low load when car when first starting from cold. In the span of pulling out in reserve, and swing the car around and moving 100 feet, and down a ramp, 60+ knocks. It felt rough and felt like it was going to stall. Driving to a stop light 1/2 mile away at ~1500 rpm, I looked over the saw that count was 200+ counting all the resets

 

Rest of the drive once on high and car up to temp, no knocks. I even went partial WOT to 4.5k and no knocks.

 

Odd, and this is been happening for a few days now.

Edited June 4, 2011 by Deadleave

[SeeeeeYa]

Low load knock, no IAM drop, no FLCK, but still felt like car was going to die.

 

I have been noticing very high knock sum at start up/idle/or at low rpm/low load when car when first starting from cold. In the span of pulling out in reserve, and swing the car around and moving 100 feet, and down a ramp, 60+ knocks. It felt rough and felt like it was going to stall. Driving to a stop light 1/2 mile away at ~1500 rpm, I looked over the saw that count was 200+ counting all the resets

 

Rest of the drive once on high and car up to temp, no knocks. I even went partial WOT to 4.5k and no knocks.

 

Odd, and this is been happening for a few days now.

 

This is only a suggestion, an idea to explore, from some recent observations.

 

AVCS can cause this... that is, AVCS tuning absolutely affects knock sum just like you describe. Maybe, if this is a totally new and different running condition with your car, something in the AVCS is different now... like an improperly working device or a change in software settings that occurred without your knowing. Pulling your tune and inspecting it is an easy step. Logging your AVCS for proper behavior is another.

 

As reference, minor adjustments to my own AVCS reduced knock sum behavior, exactly like you observe above, to very low numbers that incremented logically, and significantly improved low rpm operation. YMMV, of course.

Edited June 5, 2011 by SeeeeeYa

[sumfoo1]

I had a bad tank of gas... last fill up was much better, and the current tank i'm on hasn't knocked but once or twice and i think it was false knock as there wasn't much load or timing on the engine at either point.

[ChemicalWaste]

Won't try to fight it: noob question because, even though I've been reading I'm not absorbing quite yet until I get into it. I don't think it's knock but I also don't think I want to be wrong about it. I haven't played with the ATR software yet but after reading this thread I think it would help me out to learn what I'm looking at. So: is this knock or is this not?

 

datalog9.csv

 

Thanks ahead of time.

[rtbrjason]

Won't try to fight it: noob question because, even though I've been reading I'm not absorbing quite yet until I get into it. I don't think it's knock but I also don't think I want to be wrong about it. I haven't played with the ATR software yet but after reading this thread I think it would help me out to learn what I'm looking at. So: is this knock or is this not?

 

[ATTACH]102458[/ATTACH]

 

Thanks ahead of time.

 

Anytime knock sum increments you might have knock, but if the load and rpm are low it doesn't matter. In this log it went from 23 to 24 at so low an rpm that it's not dangerous and could even just be noise. Of course, it knocked at least 23 times before this log, though they could have all been at low load and rpm. Still this log is fine.

[ChemicalWaste]

Good enough for me. Grazie.

[SeeeeeYa]

On my 11 WRX I found AVCS tuning lowered knock sum significantly. It also removed the glaring disparity between the knock sum of cyl 4 vs cyl 2... where 4 would always be double or more of 2.

 

So when I began tuning my 11 STI I naturally went to the AVCS and saw similar results. However, with the STI's dual AVCS things get tricky and with only ~500 mi on it there has been little time for map revisions. Monitoring knock sum showed the normal increments at low load, and nothing of concern given the very low numbers.

 

Yesterday I made a very slight change that made a relatively huge change in the knock sum, lowering it to zero in cyl2 and only 1 in cyl 4 after a ride in ambient temps that approached 100F (120 to 140 IATs) during which I hit max boost several times. It was the lowest numbers I have seen in either car.

 

Between a longer trip in the morning when I'd been monitoring performance and the afternoon ride I had altered the tune due to LTFT, not knock sum. When I got home and checked the knock sum I was absolutely amazed at the incredibly low number and reviewed what I'd changed. One was the MAF CAL for the AEM, but that wasn't it, of course, as the changes there were subtle. The AVCS was the same as the morning map.

 

The only other change between maps was the change from 8 in the CL/OL transition delay (down from the stock 750 and an experimental value in the previous map) to 0. That was it. No more sitting still watching knock sum increment with tiny blips of the throttle. No more watching it increment merely backing up, or any other low-load maneuver. One knock count after a ride I purposefully explored boost in heat that could melt lead.

 

YMMV, as we say, but it totally makes sense now. Not only does entirely removing the CL/OL delay remove the soft rubber bushing from the DBW throttle, it stops the ridiculous knocking during the waiting period as well.

[MasAyinde]

I'm currently attempting to fix knock that I see as soon as I get on the throttle hard. This is the only knock that causes FBKC I see anywhere in my logs. The rest is low load knock that appears quite often, but I'd like to see about this first. I've seen some other people having knock as soon as they make a rapid throttle change, whether it be lifting off or giving it gas quickly. Attached are two logs from this morning where it seems to be happening.

romraiderlog_20110728_070840.csv

romraiderlog_20110728_070932.csv