NefMoto

When measuring knock voltage via blocks 26 and 27 what is an acceptable limit?

I too would appreciate some wisdom on this.

That's a good question for seasoned stage 3+ tuners/users. You can find this information indirectly on EcuX plots where knock voltage is usually graphed. If my memory serves me well, it is not uncommon to get 40-50v with significant timing corrections:

http://forums.quattroworld.com/s4/msgs/13648.phtml
http://forums.quattroworld.com/s4/msgs/11637.phtml
http://forums.quattroworld.com/s4/msgs/10458.phtml
http://forums.quattroworld.com/s4/msgs/98949.phtml

and so on...

There's a gulf between these guys and the casual stage 1/2 hobbyist.

The question was about acceptable levels... so here you go.

Keep in mind that knock sensor voltage gets taken consideration and correction factor CF is used to indicate timing retard ECU is instituting to lower the knock... Too much knock, ECU retards timing until the knock is acceptable. You can control aggressiveness of this behavior by one of the maps... But keep in mind that the correction goes down by maximum 12deg so if you put in some stupid timing values you will still kill your engine if timing causes massive knock.

You're right, thanks for the links.

Very helpful guys. Thanks!

I would forget about the Knock voltage as an indicator for how close you are to the knock limit.
It's way to complicated to explain here when you have no background on that, but if you are interested in how knock detection works you can read and try to understand what is written in KRKE in the ME7 Funktionsrahmen.

By far more important is to keep the CF under control. Always make small steps when you ajust timing and log your CFs. Although there is also a load limitation when continous knock occurs that can't get under control with ignition retard you shouldn't push the limits here.

As I bored over the weekend anyway I can do a short writeup on how knock control works on ME7 if there is demand for it.

Please :)

OK first of all I strongly advise you to leave all the maps I will mention alone because you can cause serious damage to your engine if you change something and you don t know what you are doing !!!!
What I have read so far on Audi forums implies that hardly anyone knows how knock control works in a modern ECU so I will start with some basics first and then describe it based on a calibration workflow.  
For a better understanding I will do some simplifications but if anyone has some deeper questions feel free to ask.
What is knock: When talking about knocking combustion you refer to the phenomenon that parts of the fuel/air mixture, that has not been reached by the flame front that was initialized by the spark plug, self-ignites due to the rising pressure/temperature in the combustion chamber.  This causes pressure vibrations in the combustion chamber which leads to the characteristic pinging noise. The shockwaves (as well as normal engine noise) induce structure-borne acoustic oscillations that can be measured with a knock sensor.
How does knock detection work:
Main problem is to distinguish normal engine noise from knocking noise.  
To do that the signal from the knock sensor is amplified, filtered and then integrated over a certain range after tdc (knock window) for each cylinder on each combustion cycle.  
For calibration the engine is equipped with cylinder pressure sensors and an indication system to monitor knock while operated on an engine dyno.

(http://img703.imageshack.us/img703/9938/indicom.jpg) (http://img703.imageshack.us/i/indicom.jpg/)

As a first step you usually determine the knock windows which are KFMAKR and KEMLN in ME7 by letting the engine knock and log the cylinder pressures and Knock sensor signal.

(http://img541.imageshack.us/img541/8093/knockwindow.jpg) (http://img541.imageshack.us/i/knockwindow.jpg/)

The next step is to find a proper frequency for filtering the knock sensor signal that is least impacted by general engine noise. You identify this by a frequency analysis of knocking combustion cycles with 3D graphs that have the knock intensity and frequency as x and y and the knock sensor voltage as z (sorry can t upload an example for that). You can calibrate 4 frequencies (FMFKRBx) depending on rpm (NKRFMx) on ME7. On S4/RS4 files only one is used (12kHz).
When you have your knock windows and frequencies you can calibrate the noise levels that define the threshold between normal noise and knock. The values are calibrated for each cylinder separately in KFKEx depending on RPM and LOAD. They represent the quotient of the current integrator value to the reference level (ikr(x)/rkrmx).

(http://img8.imageshack.us/img8/5820/kfke1.jpg) (http://img8.imageshack.us/i/kfke1.jpg/)

The reference level for each cylinder is limited to the values in RKRMX1N and RKRMX2N. This two maps group the cylinders in quiet and noisy cylinders by a bitmask. The limit is used to ensure a good recognition under noisy engine conditions.

(http://img402.imageshack.us/img402/9017/rkrmx1n.jpg) (http://img402.imageshack.us/i/rkrmx1n.jpg/)

What is also taken into consideration is rate of change in load (KFELDY) and rpms (KFENDY) which is multiplied with the values out of KFKEx.
When Knock is detected the ECU retards the ignition angle with every knock even by the values in KRFKN/KRFKLN on the affected cylinder.  When knock isn t present any more the ignition angle is phased back to normal with KRVFN/KRVFSN.
As you can see there is no easy way to link the knock voltage to how close you are to the knock limit as first of all it is dependent on the general noise level of every individual engine and second highly relative to the engine operation point and transient condition.

This is excellent. Perhaps you should add this to the Wiki?

Really good post!

I defined these maps in my xdf some time ago.  I'm looking at tuning them soon, as I think the ECU is too sensitive to knock from what i have found listening through det cans.

Rick

I have never used det cans, but how do you seperate normal engine noise on 6000+ RPM from knocking noise without cylinder pressure sensors? Sure the OEM calibration is always a compromise to cover the deviations in series production. It's more on the safe side but you have knock endurance tests with statistics to validate the calibration which don't tolerate a great number of false knock recognitions. I can hardly see how you can improve this with the method you will use without a high risk to blow up your engine.

This is a huge topic, always causes of debate.

Firstly, knock isn't as dangerous as often people make out.  It can kill and engine pretty quickly especially when it leads to pre ignition, but light knock has virtually no effect on engine durability. 

You need to train your ears for knock with det cans, but you can clearly hear the onset of det through the head phones even at 7k.  There is a slight change in engine tone too.  I have never damged an engine yet using this method - so fingers crossed!  I'm actually tempted to disable knock control completely, but not going to - not yet anyway. 

Last night, I increased the knock threshold by 10% on all cylinders.  Not done any logs yet, but will be interested to see what comes of it.  It would be nice if the timing was more stable - the CF can jump about a bit, which I don't think is a true representation ok what is actually happening in the engine.

Rick

The tune I'm using has KFKEx raised slightly on all cylinders from stock, I assumed this was normal practice but it makes me worry a little that my CF's are no longer representative of reality.

To that end I'm interested in your results.

@Silentbob, are the knock voltages post-filter & intergation or representative of the raw signal?

Carlossus - how much have they been changed by?

The voltage is a raw signal. 

Rick

+12.5% across the board. My CF's are usually <6-7 at full load.

I think that's probably close to optimum.  Just been out with mine at +10% and i've  probably gained a couple of degrees.

Rick

It's actually very easy... Det cans + the human brain are very capable of determining knock sounds even @ 30psi/8000+RPM.

I always use det cans to tune cars with standalone, even on the dyno.

what are the memory addresses for KFKEx on the mbox?
i would like to add this to my xdf as well.
thanks.

should be relatively easy to find. I'll look for them this weekend time permitting if nobody else posts em up.

i'm looking for them as well... this is new to me so i probably won't find them.
would you start with the potential maps in winols or would you look elsewhere?

I've attached an XDF with them in.

Thanks Rick!

It's not that easy. The knock voltage you see in VCDS, ECUx s.o. is the variable rkrn which is used for diagnoses purposes. It's the reference integral value muliplied with 64/(2^amplification factor).

On the hole knock topic. Maybe I'm a bit of a sissy but if you have seen what consequences even minor hardware changes can have on the previously calibrated knock recognition you act a bit cautiously here.
When I have some time I will give the det can method a try and compare it to the pressure sensors. 

Something I want to add:
I have noticed that almost everyone is running ignition angles on high load which result in permanent ignition retard per knock control.
Is everyone aware of the fact, that the knock control has also an adaption function which learns values if you do this?

Can you elaborate a little? Ignition is retarded by adaptation at those load points where  correction is always required?

^x2
is this an intergral function of kr? In which map would you find these values?

Yes it is, but only in certain conditions, and it can and is reset during certain conditions.  Again, there a maps to tune this!

Rick

Buy why do you want to always have correction active? Maybe a degree or two but seems to me you don't want to be relying on 6-7 degrees of correction factor all the time. For one the adapts can take place. Second you leave yourself less room for retard if you really need it (got a bad tank of fuel, meth ran out, etc). If I was going to play with the knock control I'd get my spark as close to ideal as possible with knock control off and det cans and then with knock control on I'd want 1-2 degrees of correction in normal situations.

I'm not saying this is a desirable situation, but it's worth knowing that if you were running too much advance at max load and after a few weeks you checked your CF's they would look OK, but in reality adaptation has knocked off a few degrees of advance already; but only at that specific load condition.

I'll do a short writeup on that topic till the weekend.

Depends in which environment you are tuning. Personally, I like to tune timing in the worst case enviroment (high IATs, etc) and have enough correction so the car is faster when conditions are good.

In any case, I don't think you'll EVER max out correction if you are generally running 6-7 degrees.

YMMV.

Yes, please!

I will add that to the s4 wiki as well since i know such a thing exists, but i was never clear on the details.

This thread may cross the line into epic.

Definitely. FWIW here is all i have that refers to long term timing advance (retard?) adaptation

http://s4wiki.com/wiki/Timing

See the "questions" section.

First I want to mention, that unfortunately you can t directly see the activities of the adaption with the variables you can log with ECUx, VCDS as the corresponding ones are not included. Easiest would be to have the proper A2l but maybe the software disassembling guys will dig them up some time ;)
I ll again simplify some stuff for better understanding.

What you see in the logs is the variable dwkrz(x) which is the delta ignition angle that is passed on to the ignition module in the software. Depending on operating conditions it can represent different variables.
Most important one is when knock control is active (not meaning that it pulls ignition angle but that it s not inactive due to cold engine(TMKR), low load(LKRN) for example), it s not in idle control and there is no safety retard due to failures for example in the knock sensor ciruit: dwkrz(x) = wkr (x)
wkr(x) is the RAM area where the ignition retards are summed up for each cylinder.
Without adaption knock control retards the ignition angle individualy for every cylinder by KRFKN(KRFKLN in stationary mode) with every recognized knocking cycle and sums it up in wkr(x).
wkr(x) is limited to KRMXN (which is the max value incl. adaption) on retard and 0 on advance of ignition.
For advancing the values in wkr (x) back to 0 the counter zkrvf(x) is used. It is initialized on every knocking cycle with the values out of KRVFN which controls how fast the ignition is phased back to normal. For every non knocking cycle were the ignition angle is limited due to knock control activity, this counter is decremented by 1. When the counter is 0 the ignition angle is advanced by 1 degree and reinitialized with KRVFN.
Now the adaption comes into play. The map for the adaption values in RAM is wkra (x). The axes are load (KRALx) and RPMs (KRANx).

How does the adaption learn the values:
When adaption is active ( TCO>TMKRA + knock control is active) and writing into the adaption map is not forbidden due to rl < LKRAN, TCO<TMKRAS, N<NKRAMIN, N>NKRAMAX, ASR, VMAX,NMAX, then the values of wkr(x)+KRDWKLA (which is 0 on S4/RS4) is written into the adaption map if the value is more retarded than the saved value in wkra(x).
If the current ignition retard in wkr(x) is more than KRDWA earlier than the stored value in wkra(x) and zkrvf(x)=0 the value in the adaption map is advanced by KRDWSA.

How are the adaption values used for ignition angle calculation:
As already mentioned the variable that influences ignition calculation is dwkrz(x) = wkr (x).
The values of wkr (x) are overwritten with the values out of wkra (x) when the adaption is active (conditions see above).

Summed up:
What you see in your logs are applied ignition corrections of the knock control including adaption values. The maximum retard due to knock control is limited to KRMXN! If you learn high adaption values it will take the adaption quite a while to get ign <back to normal>.
The mean value of the last 6 ignition corrections dwkrz(x) due to knock control also influence the max. allowed load in LDRXN. It has a long term adaption which becomes active above RLKRLDA.
The correction factor for LDRXN can be found in KFFLLDE. How fast the adaption works can be influenced by TLKRLDAB and TLKRLDAU.
If there are any specific questions just feel free to ask.

I find it useful to make KRDWA small.  What this means is that if your knock tendancy reduces, the adaptation map changes quickly, so you should always be optimum.

Rick

Couldn't you clear dtc's and repeat logging to catch the system in an unmodified state (as far as the overall correction factor of overall ign timing)? Taking note of cf factor and overall ign timing? I apologize if this sounds stupid!

KRDWA is not defined in XDF floating around forums. Where can I access data that would show address locations for different maps/constants/etc in M-box?

Thanks.

I have

4Z7907551R 0x11252
F 0x10fde
G 0x110a9

I will look for it in M box time permitting

Also, is there a way to read the current long term KR adaptation?

If not, but if it does exist somewhere in ecu memory, can you convince nefmoto to add it to his ecux alternative :)

KRDWA
Knock control adaptation to differential current ZW map
0x1109D
Factor : 0.750000

The adaption values are (wkra(x)).

thanks! It will appear in my next map pack release

Hi.
I increase KFKEF maps and actually I see more timing retardation.
Then I take stock-10% and my cf's dops a lot.

So do I have to reduse it when I want to make it less sensitivy on knock?

You raise it to raise the detection threshold. In your case it could be that you are well into spark knock before it pulls timing so cannot retard fast enough, meaning many knock events are detected and force severe retardation. With a lower threshold, the first slight knock is detected and a minimal retard stops any further events.

I would suggest caution. If the lower threshold has similar power, I'd stick with it. :)

I would also avoid touching KFEF

I have a question about knock voltages. I've done som very mild tuning, and logged the knock voltages just to see what's going on. It seems that cylinder 3 always is a bit higher than the rest. For example, if the others show 15 V, cylinder 3 says 19 V. Any ideas about this?

I don't know much about knock detection, but could it be just a sensor deviation? The result is the same when I use the stock file.

On a V6, the highest voltages will be 2 and 5, because the two sensors are mounted nearest to the center of each bank. Another reason to NOT trust voltages; just monitor your CFs.

Are you looking at an inline 4? If not and you have a v6, then yes, something might be amiss.

Yes, this is an inline 4. Sorry, what is CF?  ???

when did you last do your spark plugs and / or coil packs?

I havn't owned the car for very long, but it has been serviced according to the manual. Coming up on 100.000 km now, and I think the plugs where replaced at maybe 60.000. I don't think anything has been done to the coil packs.

Sorry, old (pre FR) term.

KR ignition retard

CF - correction factor which is how many degrees of ignition is engine pulling out of requested value.

So I'm going to bump this old thread as I've been playing with the knock control quite a lot.
I'm currently running with the KFKEF tables all raised by 20% from stock.
I have disabled the knock adaption.
I have KRDWA set to 1.5
I'm also limiting the knock that can be pulled using KRMXM - it starts at -8.25 and tapers down as rpms rise.

Car is running very smooth and is making great power.

I find that by default the ecu pulls way too much timing and for way too long.

Anyone else have any other knock tuning methods that haven't been talked about here:?

I wouldn't numb the knock recognition w/o detcans...

I tweaked hysteresis a bit on my tunes but not much else.

They all do, just before they melt the engine.
Every engine will make the most power with some knock, it just won't live very long.

And you have verified this using headphones on a load bearing dyno?

Haha that is very true.
No to the use of a load bearing dyno... I don't have easy access to one, that sure would be nice though!
The ecu is still pulling timing, and I have also reduced the requested timing in the areas that would start the knock retard anyway, I forgot to mention that.
The biggest change between the stock knock regulation and how it is behaving now is that the retard gets limited on the top end more which helps with top end power.

I'm actually planning on making a DIY DetCan this weekend and checking things out.

That sounds very bad. Request less timing, don't numb KR.

So the answer is no then. You are basically just randomly +20%-ing maps without really understanding what you are doing.
Turning off knock adaptation I also do not see the point in.
If you read about how the knock control works in these, you will see that it has a reference level. In fact two reference levels, one for the louder cylinders and one for the quieter cylinders. And knock is detected based on deviation.
The knock control is calibrated from factory using in-cylinder pressure sensors. In fact the FR lists the pressure spikes in bar that need to be recognized by the knock control.

Calibration might be needed when changing pistons and so on, but with stock pistons, especially at the upper level of the range, the factory knock control can usually hear the knock before, or at the very edge of where you can due to the engine being very noisy.
Of course this can no longer be correct if the engine has forged internals for example, but just +ing up the maps like you are just means that your engine is most likely running with some knock in the upper rev range. And yes, of course that will make more power...

Wow relax guys... more than anything I have been experimenting with the knock control... the numbers I posted above are my latest test.
I HAVE been reducing the timing requested and tweaking AFR in the areas that would cause the knock retard in the first place.
But what I have found from many many logs is that even though the knock voltages would then drop way down after that at higher rpms the ecu would still keep the retard that it got at first through the wot pull. The result was numb timing values at high rpms.

I've been experimenting with the different settings to try and reduce this effect.
I plan to also experiment with DetCans as well.

I was trying to stir up some debate on this matter because there is very little info out there.

Oh and the adaption was turned off after much discussion on the matter with phila_dot, he was actually the one who suggested it.

Knock voltages are not relevant at all.
They only show you how noisy your engine is, they have little to do with the ECU detecting knock.

I don't think you quite understand the theory behind how an engine control unit detects knock...
Looking at absolute values is a complete waste of time, you are looking for a spike within the measuring window, a deviation from the normal levels.
There is a reason why these data acquisition devices with pressure sensors cost a fortune.

I've been tuning turbo cars for over 10 years and was involved in megasquirt code development so I know a thing about two about knock...

Just because I'm new here doesn't mean I'm an idiot.

I am not saying you are an idiot, I am pointing out that your reasoning shows lack of understanding.
Reducing or advancing timing and so on is not too relevant to calibrating knock control. Knock control calibration is going to first and foremost depend on the engine hardware, engine revolutions and cylinder charge.

As for ECU maintaining timing retard - there are variables for how soon it steps timing out and back in.
You are blindly adjusting thresholds instead.

Fact is, you are now running up top with some knock. You can reduce the time after which the ECU will phase timing back in, but if you phase timing back in too soon, then the cylinder temps will still be high from previous knock events and you are going to go right back into knock.

I really love nefmoto.

That is all.

Rick and others here have previously said what I am saying. That the factory setup is too sensitive.
In my experience this is always the case with factory tunes. They tend to be on the safe side of knock, pulling timing at the very first time.
Just my experience and my observations. You don't have to agree with them...

But let me ask this. Who here has put set cans or sensors on the engine and compared the factory knock tune to audible knock that they could detect? Its one thing to say leave it stock and dong mess with it... but its entirely different to say that based on testing and data.

Thoughts?

I think PRJ is probably the best possible person to have answer that BT.

I'm hoping prj will answer that without yelling at me haha :-)

You can adjust the amount of timing it pulls, but you are not adjusting that.
What you are messing with is the ECU's definition of knock.
The maps you are changing are not timing pull or amount to pull. The maps you are changing tell the ECU whether knock has occurred in a cylinder or not.
Either the flame front propagates through the cylinder uniformly or it doesn't.
If you want to modify how the ECU reacts to detected knock, then you are tuning the wrong maps! This is my point from the start.
I don't see how it's possible to dial in these maps cautiously or not cautiously - the OEM has pressure transducers hooked up to each cylinder, and they can exactly see when a spike occurs, and this is what they calibrate them to. If a reasonably sized spike occurs you are probably not going to make more power, because the peak is likely going to be before TDC.

I've done this plenty of times on many different engines. And there are different factory calibrations as well.
I have experienced sometimes that the ECU detects knock when none is present, but that is usually on tunes where the engine block is not even the same...

I think what you want to tune in your case is the way it reacts to knock, not how it detects knock.
KFKEF are coefficients which basically tell the maximum permissible deviation from the base levels. The only right way to tune those is with in-cylinder pressure transducers.
You can do headphones and a multiplicative addition in a pinch, but I would not touch those until you verify that the ECU is seeing knock where there is none.

Look at the KRRA module, specifically KRFKN/KRFKLN, which tells how much to retard on a single knock event, as well as KRVFN/KRLVFN/KRVFSN, which is the speed at which the ignition angle is phased back in.
You can try to reduce the amount if pulls right off the bat, but if the amount you pull is not enough the first time, and you end up pulling more the 2nd time, you will lose power. As the cylinder heats up more and more due to knocking combustion, the fuel will be more and more prone to detonation.

I have been doing extensive testing over the last week or two adjusting all of the standard knock variables including KRFKLN etc.
I've probably made about 15 tunes in the last week trying various things.
I'm always open to more information and others opinions but I do have quite a lot of experience tuning high boost engines... i've been flown across the country to help people tune... I'm not totally new at this.

I'm not sure I have any of KRVFN/KRLVFN/KRVFSN defined for the M Box though... Anyone have those locations handy?

But having said that I plan to make a detcan and use that for testing alone with a knocksense unit that I have laying around from a Megasquirt install I tore down.
I'll let you know what I find...

OK I've found all but KRLVFN now I think.

I limited my changes to KRALH, RKANH and KRFKLN

ymmv, i haven't looked at the ones prj is suggesting.

How is knock control affected by a larger bore?

Say we bore the cylinders 2mm, reduces wall thickness by 1mm. Would the knock control become more or less sensitive? Would it be affected at all?

Bore is the main variable concerning knock resonance frequency. So it will be affected, for how much, i cannot say.
I guess it depends whether the sensors are broadband, or tuned to a certain frequency range. Maybe somebody could do a vibration analysis :)

If I put it this way, has anyone tuned an engine which is bored out and ran into problems with knock control?

http://www.feec.vutbr.cz/EEICT/2004/sbornik/03-Doktorske_projekty/03-Kybernetika_a_automatizace/06-klusacek.pdf

Not sure if that helps, but might be of interest.

It looks like the range the knock sensor can detect is between 5 and 15khz. Given your increase in bore size, It looks like the knock frequency will have been reduced slightly, but still within range of the knock sensor.

Knock frequency can be calculated (theoretically) by using:

900/(Pi * r)

Given that, your new knock frequency would be roughly 6900hz, while the old frequency would be 7100. Not an appreciable difference (roughly 2.8 percent) considering the calibration of the factory sensor. I'd be willing to bet solid motormounts affect the sensor more than that.

Any thoughts on looking at a 2.8 file and comparing the values of the maps outlined in the beginning of this thread with your own?

Lastly, I have to add that this is just my take on the information I've assimilated. It may or may not be entirely correct.

Thanks for the reply DD, very interesting info :)

Im currently working on a 1.8t which is bored and stroked to 2008cc, but I will compare 2.7 > 2.8 binarys. Is the only difference between 2.8 and 2.7tt the bore?

I will chime back in if I find anything valueable.

Did anyone ever compare the 2.8 to 2.7 values?

So CF = correction factor, which is ignition retard.  I want to log this and have a few of these type of variables defined in my IDB file.

wkrm is it?

OK also found dwkrz_0-7 in my IDB, referenced via a pointer in KRRA.  Guess I'm set for knock logging.

I know it is old post and old topic but I do not think it would be wise to start a new one.

I would like to kindly ask if it is in fact 1 degree by which ignition angle is advanced when counter initialized with KRVFN rolls? I think it should be 0.75 degree or KRFK(L)N value.

Is wkr(x) value signed or not? I am asking because 06A906032FC (HW/SW 0261207204/360294 at least) has KRDWKLA set to -1.5* and I would like to be sure if adaptation is using current wkr(x) increased by 1.5* or decreased by 1.5*.
Also I do not understand last sentence - does it mean that if wkr(x) is greater than KRDWA (2.25*, which should mean wkr(x) is unsigned) on each zkrvf(x) counter roll adaptation value is advanced by KRDWSA (1.5*)?

I would also like to kindly ask for KFFLLDE ignition advance correction axis location in mentioned above 1.8T ME7.5 file and also for confirmation of TLKRLFAB/U addresses (0x01EB36, 0x01EB46) in same file.

You guys are lost between theory and practical application. Forget knock voltage, it is completely irrelevant (I said this on VWkotex and everyone about lost their minds  ;D ). PRJ is right on the mark as far as what is happening.

Now how do we apply this in a practical manner? I have spent a lot of time around Dynos and that is the answer none of you want to hear, but it is the right answer to the question (how do we find the knock threshold?).

Step number one.... dial out all knock that the ECU sees on the dyno. Step number two.... shut off knock sensors and do a pull. Step number three.... add two degrees to the timing and repeat until adding timing no longer gives more power....Step four.... bring your knock sensors online and increase from the lowest setting until it detects the knock that is causing you to not make any more power.

This is how all the pro tuners I have worked around deal with tuning knock sensors.

This is bad advice.

All engines run best with some knock. Just because it's present doesn't mean it'll stop making power at this magical threshold you seem to think exists. What you're talking about is ideal for tuning E85, nothing else.

Following that advice while using a fuel that cannot reach MBT without experiencing knock (read: pretty much all fuel other than E85 and race gas) will almost guarantee a grenaded engine.

I do not even want to know those "pro tuners" you are talking about. The process you described is not calibrating knock sensitivity it just killing an engine by setting ECU so it will not react to knock as you insist you tune it to a point it does not show any ignition retard due to ongoing knock, disable knock sensors and advance timing back (so use timing which you know is too advanced). IMHO without proper equipment noone should mess with knock sensor calibration. Only thing which could be adjusted is KRKF(L)N.

Old discussion, but i don't think it's worth creating a new one.

If we are talking about non-stock engine (with stronger -so heavier and different design- rods), factory settings off KFKEF should become wrong, and may detect knock when it's not knocking? So raising KFKEF 10...12perc. maybe not so bad idea, if i have more timing retardation comparing with stock internals(after engine rebuild)?

honestly IMO if you have a road car that you are tuning yourself without detcans and dyno , you should probably not touch factory knock settings ever, pulling false knock few degrees is always a better thing than not not pulling IMO. With different fuels and stuff maybe you can mess with the recovery but yeah..

Raising something by percentage when you have no clue how it is working is always a really bad idea!
No, it is not going to detect false knock even with OEM settings, and if it does it's the sensor fault.

But hey all the tuning these days is done by geniuses who think "let's raise everything in a square by 10-12%". Better nip this shit in the bud.

This why i found this old topic - i don't change blindly all maps, first trying to understand how it works. It's a little bit strange for me, that only H rods replacement couses more knocking (engine was build responsibly - every clearances in OEM spec, no mechanical issues after 15000km). Maybe its absolutely normal?

The algorithm is pretty clever as it looks at the instantaneous voltage vs a weighted average of the last few cycles. It only refers to the generic baseline maps in some extreme circumstances. So if it knocks under WOT in a high gear it most probably is a knock, as it detected a large change from that of the last few cycles, not a generic stronger noise from the rods. In low gears false knocks can occur if the car is accelerating much faster than it used to , then changing KRFTPx can help. Read the FR as it not quite as simple as I just described it.
Forget looking at knock voltages from the logs for anything but baseline values. The real knocks from the sensors are too fast to be logged by the logger.  Use an after market knock detector to be safe.

If the engine was rebuilt properly and the head came off then also the head was skimmed and possibly the block decked, so the result is possibly a slightly higher compression ratio.

Yes, head and and block was skimmed(it was almost year ago, not sure, but it was about 0,1-0,2mm from both surfaces). Stock head gasket was instal, so sligty higher compression ratio i got. Logically, I didn’t think about it .
Thanks everyone, i will stay at stock knock maps.

Just to chime in on the knock v issue.

there was some mention of high knock v on cyl 3.
and I think a lot of people see this.
this is on 1.8t.

I have that as well.
my cyl 3 is always way higher.

I honestly believe this is because that is where the turbo is!
Think about it. Your k0xx turbo is tucked right up against cylinder three.
so that red hot exhaust turbine is cooking right next to where that chamber is trying to make power.
Those temps are going to interfere with the combustion for sure.
Probably makes it much more knock prone than the other cylinders.

I'm going to try and stick some heat shielding inbetween the block and the turbo and seei f that effects things.
Maybe there is power to be gained there.

Has it worked?

Disregard what this guy says, has no clue about anything just types a lot.

The issue with cyl 3 is that they have the highest EGT. This is internal, not external. EGT is affected by a bunch of factors, but distribution of air is the biggest one, and also location in the engine block affects it a little.
If you look at what the stock intake manifold looks like, it is clear that cylinder 2 and 3 are going to get the most air, and cylinder 3 is going to run the hottest.
Exhaust backpressure affects it as well. So you have a combination of intake manifold and exhaust manifold shapes. The better the path of the air to the intake manifold and the better the flow on the exhaust for a cylinder, the more actual cylinder charge it gets and the higher the dynamic compression is - hence more knock.

Suggesting that some kind of infrared emissions from turbo affect EGT is plain and simply retarded.

You can adjust the FZN and FZWN maps at higher RPM's to give a little more fuel to a cylinder and cool it down more, this will reduce knock.
IIRC they were firing order, too lazy to look up.

learned this today. Thanks!

Hi, i'm back to topic with a few questions. Recently stroked my engine 1,8t>95,5x82,5 and sometimes see (fake?) knock at low loads/rpm in one cylinder. Decided not to mess up with knock maps without proof what it's really not knock.
1. I'm going to make detcans. What is the best option for 1,8t block - two hoses from 2 different points on block, or 1hose/point is enough?
2.If i leave knock sensors in original place, so where best location for screwing detcan "sensor"?
3.If i confirm, what its not knock but just engine noise, how to set kfkef(x) properly? Just raise maps still ecu didin't see knock? But maps for all cylinders different - how to determine which map I should increase more than other maps?

How much is it retarded by?  Just manually retard the area of concern yourself by more than the KR did and the knock, ,if real ,will go. If its fake, it will still be there. Log zwbas, to make sure the timing changes asked for did actually take place.
...

This is exactly what i done. Retarded timing(tried adjust from 1 degree to 10 degree) doesnt help in some areas, still pulling timing like before. Also tried add timing(increase values in kfzw), and got less timing pull. Complete opposite of what it should be.

Is it just one knock event or many in succession? If just one forget it. If many in succession, make sure B_krldy or/and B_krndy are actually triggered at wot, (B_krndy only in low gears) to make the next trigger less sensitive for while. If they aren't triggered adjust KFDYES and NGKRWN wisely. Also set CWKRREF =1 .The effect of   B_krldy and/or B_krndy is to increase kek. Also  B_krndy with CWKRREF=1 together, increase the rkr (baseline) for the triggered cylinder.
Both the increased kek and higher rkr, make it harder for KR to trigger quickly after the first knock on that cylinder.

Thanks for post, now i should read more about that you wrote. Just want to confirm: rkr -> RKRMX(1)N, kek -> KFKEF(x) ?

Update. And another question after reading FR. If in understand right, when CWKRREF=1, B_krldy and B_krndy aren't included into knock detection at all. Also didin't find any disadvantages of CWKRREF=1, it is also recommended in FR (for automatic transmisions).

Updating my progress.
Made detcans, and can confirm what stroked engine is much noisier than stock (a have both cars, so no problem with comparing). Extra sound looks like from pistons(i think due to higher piston to wall gap, ant shorter piston skirt).
So first, i was right about fake knock - i can't hear any knock at low rpms/load, but ecu pull timing about 10 degrees. So what i done and where i ended(?) :

1. Set CWKRREF to 1
2. After reading FR, decided not to touch KFKEF maps at all, left it stock. Maybe it doesnt fit for stroked anymore, but without cylinder pressure sensors its impossible to set this values.
3. Made a lot of pulls 3/4 gear with different boost level and timing to find where real knock appears.
4. According rkrn values from from log without real knock, increased RKRMX1(2)N. This values adjusted after few puls, just to get point, there is no real knock.
5. Slightly increaseded boost level(with new RKRMX1(2)N values), and knock apears, but ecu also detects knock and pulled timing. This knock not continuous - appeared and gone (after pulled timing and extra fuel via KLAMKLRL). Log attached.

How do you think, did i do everything right?

Best to do your own baseline for each cylinder and then use that to see whats going on.
Log ikr(i), vkr(i),rkr(i)

 Calculate 64*ikr(i)/(2^vkr(i)) . This is the instantaneous signal from the knock sensor for cylinder i.
Do the same calc for rkr(i).

Compare the normalised rkr(i) above to RKRMX1/2N (check which cylinder is mapped to which of these two maps). If the calc value or rkr(i) is< map then the map is never used, which is what you want really. You want the true rkr to be used, not some fixed map value which is there for safety.

Unfortunately, logging is usually not quick enough to catch the moment the knock occurs, but if you do these baselines when the load is partial, you can get the true knock free baseline of the motor and can start looking at the kek values that are now too close to the baseline due to the mechanical noise. You can also use the readings to make your own RKRMX1/2N maps as described in the FR.

Do make sure that B_krldy =1 under load in 3/4 gear first. It does a few things, 1) decrease knock sensitivity and 2 makes rkr(i) follow ikr(i) quicker. It does other things too.

Thanks for explanation. B_krldy =1 should be all the way during pull in 3/4gears ? From logs i can see what it =1 only at "noisier" load/rpm range.

It must be 1 before any knock is detected.

Thanks for help. I think i finished my knock detection. krndy trigerring exactly before real knock.

I bought supertech pistons without any coating, next time i think worth to pay more for coating; it should reduce extra noise and and maybe no need to bother with knock detection adjusting