infini_IV

iTrader: (4)

Im happy to pull drive recorder files from ANY ecu you want all day long, there is no sum data read via the MUT, ever thought maybe its more then coincedence that as you get your "knock sum" readings timing gets pulled. Did you bother to open the D/R files i posted

donour

iTrader: (6)

What in the world are you talking about? Have you read jcbanks knock routine disassembly thread?

d

GST Motorsports

iTrader: (8)

Yea I remember you posting that awhile back....

infini_IV sounds like he got his hands on a shiny mitsu MUTIII tool and now he thinks he has the power of grayskull.

infini_IV, I really think you should search and do some homework more into the inner workings of the ECU before you keep making posts about this.

- Bryan

fireroasted

iTrader: (3)

OK guys, just to get back to the OP....me....

Is it appropriate to deduce that the boost drop in the logfiles I posted reprents a problem physically, like a boost leak somewhere, and its NOT something the ECU is doing because it is seeing 9 counts of knock?

I just want to confirm this path of troubleshooting with folks. I still can't find anything physically wrong with piping, boost controller, ex mani, etc.

So based on my logs from the OP what are your suggestions to look for?

Some paths/options I have are:

1 - knock is being seen by the sensor from 0 to 9 and staying there because the sensor is bad. (This theory seems doubtful based on all the posts I have read, and that when I clear the code and log, the sensor returns 1-2 counts for a bit until the CEL happens, and then I am flat at 9 again. The logs show this. When you see 1-2 counts its because before that log I cleared the code).

2 - I have a bad wire in the wiring harness. Damaged by heat. This seems way out there, but I had that with the front O2 about a year ago, so its on my list.

3 - I have a boost leak somewhere, or some other kind of physical failure (exhaust) that is making noise more than normal that is tripping up the sensor. Also trouble shot this, by carefully inspecting all of the exterior of the exhaust. Didn't check the muffler packing or the silencer because I figured that is too far downstream to matter. Everything seems rock solid tight from the exhaust manifold down and on.

Are there other possibilities I missed?
Is there something else I should be logging to help me find a clue?
What do you guys think?

<Boosted Tuning> I am not using the Tephra Mod, (sorry Tephra), only because I haven't mastered things as it is yet so I haven't introduced anything else lately. Its on my list though. So no drop boost feature.

Thanks!

mrfred

iTrader: (50)

mbartel - a continuous knock sum of 9 will slowly pulldown the "octane" value (learned knock for infini_IV), and when it drops below a threshold point (octane of 154), it will slowly shut down boost control. The continuous knock sum will also start blending the high and low octane timing and AFR maps.

infini_IV - Guess you can call it knock retard if you want, but its used for other things in the ROM code besides pulling timing when knock is detected.

infini_IV

iTrader: (4)

learn knock is a percentage of the knock curve total. knock CA as i said is not a count of knock but a figure of timing retard "DUE" to knock

Boosted Tuning

iTrader: (12)

Do you disassemble ROM code? The person that you quoted does, and probably knows a lot more about the ECU then you.

Please stop.

03whitegsr

iTrader: (8)

I really think this is just a matter of semantics and he is just being difficult, for god knows why?

On one hand, he is right. Knock sum IS NOT the voltage produced by the knock sensor or anything like that, it is a scalar that is used to calculate the amount of ignition retard (along with other items as mentioned by mrfred) that the ECU feels is needed to prevent further knock sensor activity.

Raw knock output per channel (cylinder) from the aux processor likely is hiding out some where in the ECU. The RAM Addresses may even be in JCSBanks thread, I don't remember? I'm not 100% certain here, but I believe the ECU actually has a digital processor external to the CPU that filters the analog knock sensor signal to a certain bandwidth and then sums (digital integration) the knock voltage over a particular time span relative to the desired crank angle range and converts it to a digital signal. The ECU then uses that filtered data to figure out what we call "knock sum."

I believe the actual signal is based on a summation of the knock sensor voltage, but it happens over a very small and limited time crank position range.

infini_IV

iTrader: (4)

calm down people! no ones suggesting we reinvent the wheel, im merely questioning that what we know as "knock sum" is not knock sum but knock retard. Its not a number in relation to the severity of knock but a timing retard figure. as clearly shown in the drive recorder files. the data list item is expressed as KNOCK CA which is crank angle. The other data item is LEARN KNOCK which is expressed as a percentage of how much the ecu has learnt of the knock curve (volatile memory)

mrfred

iTrader: (50)

infini_IV - At this point, its becoming a matter of personal preference. Those two items have been called knock sum and octane for the last three (four?) years in the Evo community, and I doubt its going to change here.

GST Motorsports

iTrader: (8)

infini_IV,

#33 in your shiny list "Detonation Control Learned Value" is "octane %" in evoscan.

The other stuff, what mrfred said.

I'd still be interested in if you did any search whatsoever for the threads of actual ECU disassembly that has all the cool little knock control bit and bytes and such or if you just keep stroking your shiny MUT tool.

- Bryan

infini_IV

iTrader: (4)

Your right it probably wont change, and it doesnt need to.

The point is what we know as knock sum is not a count of knock severity at all (like most people believe) but a figure of timing retard in response to knock, nothing more nothing less. Why people like GST motorsports got all exicted prematurely and chucked their custard who knows

l2r99gst

iTrader: (2)

Actually, it's both. It is a knock retard and it's based on the knock severity. If you get 2* of knock retard (6 counts) as compared to 10* of knock retard (30 counts), it's a measure of knock severity. More knock will get more knock retard. jcsbanks disassembled the entire routine and has a great thread on it. The knock sum and retard is based on output from the sensor and a moving average that is the knock 'threshold' that once surpassed, triggers knock sum and retard. Octane value is decreased once the knock sum is 6 counts or greater (or retard is 2* or greater), etc, etc, etc.

Knock sum/knock retard is the same thing though. Back in the days with DSMs and DSMLink they called it knock retard, but you could still log the actual knock sum as well. In the Evo community, since the ECU had to be disassembled little by little and loggers were being built, etc, knock sum was what was discovered first and that is what everyone began to log and become familiar with.

But, since the ECU is very similar in coding to the DSM, we always 'knew' about the 3 knock counts to 1* knock retard. They are one in the same, just a different name and scaling so to speak.

infini_IV

iTrader: (4)

Call it want you want, and i suppose you can argue that you can work it out backwards ie. if you know the amount of retard you can work out how severe the vibration is which works. But the figure that we log eg 2 knock sums is actually 2 dergrees of timing retard pulled by the ecu in relation to knock. This helps explain why "knock sum" is not always in proportion to knock voltage and what we have come to call "phantom knock" it is simply timing retard

l2r99gst

iTrader: (2)

Ummm....no. With the scaling that we use for 'knock sum', it is not a 1:1 ratio to knock retard. 2 counts of knock is not 2* of timing retard. It is 90/256 degrees of knock retard per 1 knock count (roughly 1* per 3 counts of knock). I don't know why you are saying they are exactly the same. I think you don't understand what we are referring to as knock sum and what scaling we are using. That's like saying 16 ounces is the same as 16 pounds. It's matters what scaling and resulting units you are using to describe your data value.

If you actually read what I wrote, I am agreeing that they are the same thing. Except the Evo community calls it knock sum, and gives it the correct scaling to be the knock sum. If you call it knock retard, then it needs a new scaling to be logged as knock retard. They are not 1:1.

Phantom knock is a completely separate subject...I think you are getting confused on this whole issue. Phantom knock is simply a case where the ECU is detecting knock, but it isn't real. That will trigger knock counts/knock retard, when in reality there shouldn't be any, but due to noisy valvetrains, etc, the knock sensor picks up the noise, which surpasses the knock threshold, and it triggers the knock counts/knock retard.

Again, if you seriously want to learn everything about the stock EVO ecu and knock control, read jcsbanks' thread on the subject. It's a great thread with every piece of information that you would want. But what we are calling knock count and you are calling knock retard, I agree, it's the same thing...but with different scaling/units, so the exact value won't be the same. But, whatever you call it, it is the amount and severity of knock being detected by the ECU.