NefMoto

Hello,

I would like to know, how exactly the knock control works on my Audi 200 20V.

What exactly the ECU does, if it recognises the knock event?

Does it everytime set IGA about -3 degrees and than slowly back?

And what the boost does? Is it also everytime somehow controled and reduced?

Thanx!
Karel5000

Can i set the ignition angle on the knock border? Is the knock controll so intelligent, that it will keep the engine safe?

Piston ring land will break when you run at the knock limit.
Soft knocking the knock detection will not recognize.

Tahnx for your answer. So, to go 2 degrees back, wenn the knock control recognises nocking (shiffting the ignition angle about 3 degrees back) is a good idea...

Maybe 2° aren't enought - because knock control only detects strong knocking.
I know a lot of tuned turbo engines where the ring land is broken after 10-20k km.
From my experience with a knock box you have a "wide" range of "slight" knocking before it really knocks very loud.
I tested this with 95 Octane fuel, high CR in summer.
I will repeat it this summer with 102 Octane - I guess there is different behavior in the range of "slight" knocking.
Slight means, you can hear it with extra equipment but "standard" knock detections from the ECM won't detects it.

No it won't.
It will, perfectly unless you mess with it.
In fact I can see and hear every single knock event on these engines.
If it keeps knocking you probably exceeded 7.5 degrees retard on that cylinder - meaning learn to tune ;)
There is no "shifting angle back about 3 degrees"... In fact you can't see any knock control related activity from vag com, because all vag com shows is requested timing, not actual timing.

Again wrong, it detects even the smallest event.
Never had this problem ever on the 200 or so 2.2T based projects I have tuned.
"Knock box" is useless, it detects knock when there is none.
Don't compare knock control on a 3.6 PT or 4.2 ABH with the knock control of 2.2T. Knock control of 2.2T is a completely different world and very sophisticated.
The knock control on the V8 engines is worthless.

Here's a third gear pull from a 400+ hp car with my logger.
You can see knock, you can see ECU reacting, and you can see that requested timing does not change...

You won't have broken ringland because of the forged pistons.
But I have made the experience at other engines (cast pistons).
You also see it when you dismantle the engine, the pistons will be very clean.

The Knock Block is only a small amplifier with a built in bandpass-filter.
The detection is my "brain" - ear.
When I have doubts I record it on the laptop and make a FFT to make the knock visual.
It's also possible to remove other frequencies.
I have recorded it, I will upload it for you.

I have also spoke with different companys which sells and tune with those tools and they confirmed me that also a slight knocking is not the best case.

About knocking, there are different stages. Slight, "normal" and "extrem" knocking.
Extrem knocking is a statistical event, 100.000 fire events without knocking and 2 extrem knockings and engine is damaged. Normal behavior. You can find much more information in books about this.

How good the I5 or V8 Motronic knock control is I don't know - I guess you know it much better than me.
I have see on old Jetronic/Mikroplex ECMs that there is knocking possible which the ECM don't detect (knock threshold voltage to high?! What ever the reason is).
For the M 2.7 (Punto GT) I know that you can have broken ringlands because of slight knocking.

Impact of the fuel in the sound of the combustion/knocking I have to evaluate.

I use stock pistons in most cases. The stock pistons on AAN/ABY/ADU/3B are very strong and capable of 700hp. The only issues is the compression ratio.
We are talking about M2.3 here and the 2.2T engine. None of what you said holds any water in case on these engines or with this engine management.

It's perfectly safe to run M2.3 with some knock events occurring. Nothing will break, because the ECU sees knock perfectly unless you do exotic things ;)

Standard I5 pistons are forged Mahle pistons :) C20LET and Z20LEH use the same material. Very strong piston, I guess the best OEM piston (material) what is available.

I only can speak about the experiences I made.
Most I made with a fiat uno turbo. Since I used the knock block I never had any head gasket damages :)
Ignition is very safe, I drove it at 40°C in the summer with a heat soaked engine and only 95 octane fuel.
The result was a damage free running since 3 years.
Maybe I will be braver this year and use 102 octance (Aral Ultimate) and see how much more ignition is possible.
From real knocking till complete knock free I had an area of around 6-8°.
Maybe it was too safe because you feel the lack of power.
RPM range was 3500-4500rpm.

V8 knock control is worser? Should I install knock block and work with my ears or can I trust the original knock control?

Are there any knock control related maps? Thresholds or what ever?
How they normal look like?

On V8 it's not adjustable, and it's crap. It doesn't have per-cylinder control...

On 2.2T everything is adjustable, as there is a different MCU and software for knock control.
And it's much better.

You speak for 3.6 V8 I guess?
Because 4.2 V8 has newer Motronic and the service manuals says it can.
There is a SSP of the I5 motronic and the new V8 motronic :)
BUT I can't say if this is true what they write there. :(

Yeah, from the V8 I only have experience with 3.6 :)
I don't know about 4.2 but on 3.6 it was really crap at detecting knock.

Yes. 4.2 is better ;)

Here take a look ;) http://www.volkspage.net/technik/ssp/ssp/SSP_143.PDF

Interresting. Thanx for your tips. I am very glad to hear from you.

I suspected the VAG COM, that it shows the ignition setpoint without the additional ignition nocking value only, because i set the ignition about 4° sooner and nothing. Without the Motronic 2.3 specifications it is not easy to do something...

Thats the reason, why i also build such an audio knock detection system - my older knock sensor as a microphone, my laptop and FFT. To hear nocking without a measurement, it is not so reliabe. At least for me. I have not tested the system yet.

I am very surprised, that such an old knock management system is so sophysticated. Can you tune the knock pre-windows, knock windows, frequencies (for each cylinder separetely), gain...?

I can not believe, that it is so reliable, anyway. There is still development of the knock management strategies in the car industry but there are still some problems with it. The nocking is practically detected throw statistics while sometims the knocking appeares but the knock control does not recognises it because of noise and on the contrary...

Yes, as the knocking depends very strong on the combustion area, specially on the materials, sizes or timing of components (knock frequency), and very much on the load (knock windows), it is very important to adjust these maps during tuning of engins. This is probably the reason, why engines blow after tuning not recognising knocking...

I would like to know, if the known tuning companies also adjust the knock maps during tuning...

Yes.
Some "pro-tuners" turn it off.
Haven't seen anyone properly adjust anything.

As for reliability, I have per-cylinder KR in real time and I have car on dyno, and every event I hear I can see in real time in the log.
I don't bother with headphones for these anymore. I only verify at start that everything is OK (incase of heavily modified engine or incorrectly torqued knock sensors) and after that I just look at logs.
Knock recognition works perfectly in this ECU.

It is very nice to hear, that the knock control works so nice. So i will test, if it works also good by me (i have a chip from PK-Motorsport, Germany, i do not know, how well they do it) and than i can eventually set the ignition to the knock border (the border, where a small knock appears from time to time) without problem:-)

I have my engine and the turbo boost a bit modified, also i have a bit more MAF during a full load. I gess about 50ps more. It could already have an influence on the knock window, what do you think?

Are the seven maps arround and behind the addresses ca. 0x1CE4, 0x1D64 in the boost chip these knock maps?

You need to look into how knock control works. Knock window is not something to do with how much torque you are making at all. You seem very confused ...
Mostly no adjustment is required. I mapped a 650hp engine recently and it worked perfectly with the stock settings.

Also, you will be never be able to set your timing correctly on this ECU without good logging ...

Generally, the time, whenn knocking accures, is strongly dependent on MAF or the indicated pressure. This i can guarantee. It is very logical. That is, why the knock window is to be adjusted after some bigger modifications...

No, you are wrong.
Knock window is the time when knocking is the most likely to occur. The window is after the spark event.
The knock sensor is sampled during the window and the highest sample is compared to the average, and knock detection is done from this coefficient.

You do not need to do anything with knock window at all. It depends more on engine RPM.
Read a little how digital knock detection works, because right now it seems you have no idea...

I use a Knock Box for this.
Most tuners which use this, solder the Knock Block at the pins of the ECM and disconnect the pins direct to the ECM.
That they do to know how good is the signal which goes into the ECM and to have the knocking still present during the dyno pull.
At long dyno pulls you have the engine damaged till you see the knock in the datalog (when timing will be retarded) or you don't know if the detection works or not...

Here some information which could help:
http://www.rhinopower.org/knock/docs/AN3772.pdf
http://www.ti.com/lit/ds/symlink/tpic8101.pdf
http://www.rhinopower.org/knock/docs/TPIC8101_pres.pdf
http://www.rhinopower.org/knock/docs/HIP9011_evb.pdf
http://www.rhinopower.org/knock/docs/Prosak_app.pdf

The Bosch CC195 is used in newer Motronics I believe. Somewhere I had information about older CC from Bosch.
Don't find it know.

How ever, Knockwindow is a must have when you don't run fixed ignition. The problem with time constant and crank angle constant events...

Helpful for me would be to know how the "knocking maps" are look like at a M2.3-2.5 motronic.
They should be present in the normal ECM software chip because there I won't find big differences between I5 and V8.

In V8 3.6 PT the knock control is done by a completely different software compared to 2.2T.
It has only internal ROM and is not possible to modify. Does ABH have two chips? Never looked inside those.

ABH has only one chip - 64 KB :)
When I speak about V8, then I speak about my 4.2 ABH :)

I know that sometimes they also used analog circuits for the knock detection.
I have to check the 4.2 ECM circuit. I only installed the emulator and was happy to be able to start working on it.

A friend gave me a 3.6 V8 ECM - I will compare both 1:1.
But I don't have a I5 to be able to compare this too.
In the first step only to identify the ICs etc. - which could be important for knock detection etc.

Possible that knock "software" is on extra small eprom or hard in an ROM.

Yes then it's the same like 3.6... the software is in the ROM of the chip on the bottom board.
Nothing you can do with it ...

Some engines i know have all the knock maps MAF dependent and the windows have definitely nothing to do directly with ignition angle...

You must learn to accept also the ideas of other people. Maybe you are right but maybe others are also right in spite of the fact they say something different as you. It is much better for exchane of experience because there are more engine management systems on the market...

I have found some maps at the addresses at the end of the boost chip, which i posted yesterday in this thread. Because in the lower load range (i gess somewhere till 0.2bar relative by turbo engines) no knock occures, it can theoreticaly be these maps, because some have zeros inside. Maybe it corresponds whith this behaviour. But i only gess, i have not seen it so yet...

Ï think, the knock setting is absolutely different in comparison to 5 or 8 cylinder engines. The blocks are absolutely different, another number of cylinders, another pistons, another compression... If the knock managements are good, they should also have a separate maps for every cylinder...

If you would like to compare also the software from 5T, i can send it to you :)

Dude, knock window by definition IS exactly dependent on ignition angle. The earlier the mixture is ignited, the earlier you need to start to listen for knock and vice versa and this is exactly how it works in the ECU!
I think you don't understand what knock window means.

There is nothing for me to accept or learn from you about this, because you are confusing even the simplest terms.
The time duration of the knock window is mostly dependent on RPM, the start of it is again influenced by RPM and ignition timing.
Don't care how it's done in some other ECU, because that's not the point here, is it?

Also, you don't need to adjust this at all. I already told you it works fine on 650hp engine with std. settings.

Maybe he also sees it load related and miss the information about the ignition?

@karel: Can you upload the complete bin files? top and bottom? At RS2/S4/S6 datasets which I have I have nothing at you adresses.

I guess there is some additional map to get an boost related impact.

There isn't. Everything karel5000 said about knock control in M2.3 so far has been wrong.

Ok :D

But neithertheless I would like to see those maps at other applications to get a feeling for this.
Also how the knock control works, in the SSP it's said that it also makes the fuel mixture richer - is this true?

It is, but that only happens when there is continuous knock.

Ok. I ask because I have noticed a higher fuel consumption in some areas where I raised the ignition.
But I'm not sure if the knock control makes the fuel richer (I didn't noticed at my wideband, but this most not mean anything) or I have raised the ignition too much... argh I must go on a dyno to check the engine output at some points :)

On naturally aspirated car only way to tune ignition is on dyno especially on this older lower compression ratio motor where the cylinder bore clearances are most likely less than optimal.

You are the first one to ask, how i thought it. I was talking about knocking generally. I was talking about the indicaded pressure (pressure in the cylinder). This is a function of many thinks. One of the most important ones are ignition and MAF. Hence, ignition is a function (again) of MAF...

Now it only depends, how the management of ECU is made up. This is a point here. There are different ways...

I like to discuss not quarrel...


The addresses are not exactly beginings of maps, but may be middles... I have not got time to specify exactly the beginings. But emu works there...

I will upload the orig. software. I must first find, how to do it :)

To make a fuel richer because of cooling is a standard function of, probably, all nocking managements.

do you mean this:

(http://content27-foto.inbox.lv/albums/e/eddix/krami/knock-motronic.png)

I have seen this in one tuner work. Was thinking why it has been done like this. Cars mostly runs great. :D

How it is happening if boost chip and fuel chip is two seperate things. In general like two ECUs.

Think about prjs posts ;)

I will leave you to find it out  ;D


Nothing to do with knock.

So you are wrong like I said.
And also cylinder filling is irrelevant, because it is taken out of the game by the fact that knock detection is not absolute value, but rather a coefficient between the base level and the peak. The coefficient depends mostly on RPM and amplification level of knock signal, and the amplification level depends on how loud the engine is ...
Higher cylinder pressure - higher base level, higher peak, same coefficient.

And clearly this works great on the I5 ...

Please, no need to insult me with theories, I am telling you how it is in this ECU. Unless you have a better source you do well to listen - but I kinda have most branches of the knock and adaptive knock routines labelled in IDA by now ;)

I think he is speaking about his understanding of knocking and not about the function in the software.
The software only looks in a window before the next ignition event.
To don't get in trouble with other engine noise.
And also the level - louder combustion and higher revs.
But of course a knock event also depends on the load (VE) of the cylinder.
Lean mixtures need more ignition for example because flame front is slower.
Also the knock event is different "sound level" than at x bar of boost because much more energy knocks (detonation is the wrong word for it).
I guess he means this.

Could or would you give a small hint to find the functions in IDA?
I'm not able to find a real starting point (I know I have to begin at the reset function).

Your ECU does not contain those functions in the chip that you are looking in.

I would also look into I5 for learnings.
But according to the SSP the 4.2 V8 has cylinder selected knock control.
How ever - I only can learn. ;)

Yes, but not in the chip you are looking into!
I already told you where it is. And you can not change anything, because it's in ROM.

Different CPU? V8 has 8051 - that has NO ROM.

Sigh...
The 2.2T has two 80C535's, yours only has one and one chip.
For the third time! There is no knock control detection functionality in the chip you are looking at.

Look around on the board, on the same board next to the 80C535 is a motorola IO extender and a 8051 injector driver. All MCU's.
And none of those have anything to do with knock control either. The knock control is on the bottom board, which is completely different between 2.2T and V8.
And the V8 does not have a writeable chip on the bottom board, it's all embedded in the ROM of the MCU, which is a different MCU on the V8 and the knock control is completely different.

Nothing you can adjust, because you don't have the second MCU like 2.2T!!!

I don't know it. Most ICs Google don't knows - the stuff is too old.
But however, there must be some connection between knock control (in which way ever it works) and fuel injection and DTCs.
Because there were DTCs from out of knock control etc - I guess the counter for the DTC is internal and it would be nice to know this "count" function to maybe see when it knocks...

I think the Philips CC151 V1 should be investigated... :) But didn't find information about this :(

DTC's are a different story, but on 2.2T at least there is no logic to diagnose anything related to knock control in the chip that you communicate with when you read DTC's...
Same like the main chip knows nothing about a MAP sensor. But there is logic for the two chips to talk... So when a fault code is detected the other chip sends a message at very high speed to the main chip and main chip stores it in fault memory.

I don't know how it is done in V8, but I imagine in a similar fashion, where the knock control chip does all the diagnosis.

I don't know.
But I tested the 551B software at my ABH ECM. Engine starts but don't runs very long.
How ever - diagnostic of throttle position works at VAG COM.
So I maybe only have to perform some copy and paste.
When it would be so easy. :(

Yes. Exactly.

The knock windows, frequency and the level depends much on RPM. But it needs to be tuned very seldom...

If the knock maps are directly dependent of IGA, i find it nice and very flexible.

I would like to know, if there are even some corrections necessary but may be none is needed...

Yes, there are some adjustments needed when you change MAP sensor. But not for the reasons you think...
Will work fine on WOT without.

Well, can you tell me please why? The level, the dB bondary because of more noice?
Or because of the correct pressure steps during the knock control or the adaptation?

No, it's not what you think. I will not say more about this, sorry.
If you want to know reverse engineer it in IDA pro.

o.k. Can you give a hint what are these maps all changed to 128?

Found out where are located knock maps.
But don't know how they really works... :(

(http://content28-foto.inbox.lv/albums/e/eddix/krami/aan-knock.png)

everybody quiet. no replies, thoughts?