I dont believe you can max out the stock knock sensor easily on a evo if you know how to adjust the multipliers properly .. until i hear it from jcsbanks i will not agree with that.. not with what ive seen when tuning lots of cars and adjusting the settings. talking 40-50 cars adjusting it. and all built motors / noisey as hell, etc. stock motors require it also if the person runs like a quartermaster, or noisey cams, etc.. or other misc vibrations.

I will however start logging voltage based on what ive read though (since its part of the sequence that determines actual knocksum).. and im very interested to see what my personal evo shows voltage wise around redline. i will probably get it out by next week, waiting for the temp to go above 35f here.

When I spun a rod bearing (due to running it 2quarts low on oil) and was driving my car with a spun bearing it was tipping off the knock sensor, but still didnt max it out.. the knock sensor constantly adjusts itself also.

im gonna continue to research this alittle more.. i cant pm jcsbanks his pm's are full.. maybe i can hunt him down on GTR forums like a stalker lol.

 

Tscompusa, I feel the very first post in this thread has an exact display of what I'm talking about.

I don't think the ECU could detect knock in this case simply because the average knock voltage level is so high. Aaron, any chance you could grab a log with both the knock voltage and the knock threshold value? I have a feeling the knock threshold will be pegged above 7000RPM.

I'm not saying there is detonation, I'm only saying that if it did detonate, I don't think the ECU could detect it.

 

I tune in 3rd at this level, past 700 I have to tune in 4th on the dyno.

Anytime I tune in 3rd I will do a pull in 4th on the street and if practical (i.e. car making enough power vs length of pull) I will make a pull in 4th. The other thing is I will do 2-4 on the dyno or 3-4 on the dyno to get a rough idea of whats going on.

 

Interesting, I can look for the 4G64 knock sensor I didnt realise they had them. I have 2 complete builders out back to go stare at.

03White-

I agree with what you are saying and the stock setting as far as the knock are there above 5500 but thats like 20-21 on the scale so its pretty deaf (well factory anyway). This is the first time it had THAT much noise, but its the first time I pushed it that hard on pump too. I looked at all my prior logs and havent seen more than 3.8v spike (same run had one 3.6v). There were some 3.0 and 3.2 on ethanol and even one 4v spike that ended up not being real. Took the timing away and it dropped 20whp (there was no recorded knock at that time).

I have noticed that a stock turbo car with too much timing will knock on pump past 7k, so the settings should be good. The problem as you stated is the voltage total vs what it can hear. Voltage is still just amplitude it doesnt reflect the frequency it can hear. Its still looking for 63-6700hz and then double that at the second harmonic as far as I know for the actual event. I am presuming if it "heard" knock it would immediately go to 36 counts based on the amplitude.

 

lower your multiplier 1 point at high rpm and lets find out. im interested to see. i honestly dont think its going to just go to 36 like you say, but im curious.

 

I didnt bring the car with me this week so it will have to wait until friday when I had back home. I will take it out on the street and make a pull or 3 and see what it does.

 

ok thanks Aaron. im sure it will be appreciated by a lot of people.. it will give us a better understanding how the knock sensor actually works. (me atleast)

i just dont think its gonna spike to 36 from 1 multiplier lower, but i very well could be wrong due to the voltage levels up there.

if it does tho, thats gonna make me sad, because then i wont trust the sensor as much as i do now.

I just remember on my car when i was removing phantom knock it was slowly going away,, like i didnt hit a certain multiplier where it went from 20 to 0.. it slowly removed it.. so i would imagine it will slowly show the knock not just instantly show 36 counts but i could be wrong since the higher rpm is different then down low.

definitely interested to see the results though for sure.

 

I don't think you would necessarily need to do so. So long as the noise floor is high enough to mask transients, that might be all we need to make knock 'invisible' to the system.

 

Ok, not 100% sure these are stock values, but they are close.

RPM Knock multiplier
0-3800 15
3800-4800 18
4800+ 20

Knock Adder
Single Gain 2
Triple Gain 7

So let's say we are interested in above 4800 RPM

Knockbase=(mult*knockvar2/8) + Add2
Knockbase = (20 * knockvariable2/8) + 7

Rearranged to solve for knockvariable2
knockvariable2 = (Knockbase - 7)*8/20

So if Knockbase equals 255, there is no way detonation can be detected at all so that is the value we are most interested in.

(255-7)*8/20=99 -> If we take this to be voltage, 99/255*5=1.94V

Thus, if the AVERAGE knock sensor voltage reaches 1.94V, there is no way for the ECU to detect detonation at all.

Now, "knock voltage" is not the average knock voltage value. It is only value of the detected signal at that instant. Thus, you would need to log the knock threshold value to determine if you have maxed it out. Also, the sample rate is very poor using MUT logging and if you are truly interested in this, you'll need to use DMA logging. For example, the log posted here has a sample about every 150ms, or about 7Hz, which IMO is worthless for this type of data. That 3.9V might be maximum, or i might be near average, you have no idea since there isn't enough data to properly judge.

If you have a good data rate, 200Hz for example will grab almost every ignition event at 7000 RPM. That data would be MUCH more reliable. FWIW, DSMlink logged at ~160Hz on the knock channels. VERY good data can be grabbed from it.

 

Original thin crust or pan?

 

Guido angelo, and vinni hooked me up! Forgetta bout it!!!

 

yep i remember reading that. you cant capture it easily when it does its process of determining actual knocksum where the voltage has to be spiked over by the knockbase if i remember correctly to produce a knocksum.

typically most 2.3 are more noisey then anything else.. usually end up with values of 18/20/20 or 19/20/20 etc.

the fact the way the ecu works the voltage is pretty important and should be logged .. so i am gonna start logging it when need be now just so i can see more data whats goin on even if knocksum isnt produced.

lots of great info in this thread and looking forward to Aarons test with higher RPM and lowering multiplier a little.

also stock settings are 13/18/20 low/mid/high

the 9 show more detail table but its pretty much the same multiplier

 

David,
Are your pump gas numbers on the 3582 or 3586?