648whp on Pumpgas
The KS is just a piezo electric 'microphone' tuned for sensitivity in a narrow frequency band, which is determined by the bore diameter. Restricting its sensitivity to the frequency range of knock is its passive method of filtering noise.
As the level of noise in this range increases, so does KS voltage. The ECU monitors this voltage, and has an algorithm that aims to differentiate a knock transient from background noise, which it does by 'listening' for voltage spikes typical of knock. The multipliers affect the sensitivity of the threshold that differentiates knock from noise.
If KS voltage is pegged due to an excessive noise floor, it seems logical that a knock transient would be lost within the background noise, in which case the ECU can't flag it. If that becomes the case, we can't tune using Knock Sum as a guide. The only way to be aware of this situation is to look at knock voltage. In the case of the OP, JB's setup is seeing quite a lot of baseline noise in that region, so he's (wisely) noted the need to look at other indicators to be assured there is no knock.
As the level of noise in this range increases, so does KS voltage. The ECU monitors this voltage, and has an algorithm that aims to differentiate a knock transient from background noise, which it does by 'listening' for voltage spikes typical of knock. The multipliers affect the sensitivity of the threshold that differentiates knock from noise.
If KS voltage is pegged due to an excessive noise floor, it seems logical that a knock transient would be lost within the background noise, in which case the ECU can't flag it. If that becomes the case, we can't tune using Knock Sum as a guide. The only way to be aware of this situation is to look at knock voltage. In the case of the OP, JB's setup is seeing quite a lot of baseline noise in that region, so he's (wisely) noted the need to look at other indicators to be assured there is no knock.
Tscompusa, I'm not getting into a dick swinging competition that will simply draw away from the topic at hand. I will simply respond with "I have tuned enough cars that make enough power to have recognized this important detail LONG ago."
Thank you Ted B for that explanation. Ted B understands how the knock sensor actually works and how it is not "adjustable." If you max the SENSOR out with noise, you'll NEVER detect detonation. It has nothing to do with the multipliers at that point.
As I said, if everything is peachy, yes, the knock sensor can provide a level of protection. The point I am trying to make is that it is most wise to verify that everything is peachy with the knock sensor if that is how you want to tune. Personally, I rely very little on the knock sensor. Yes, you have to keep it happy, otherwise it is going to step in and jackup your tune. But I don't rely on it at all to tune, merely I use it as a safety net should something go wrong.
If you want to see proof, you can simply look at Aarons post in this thread.
One thing I use to really like about DSMLink was that you could datalog individual cylinder knock voltage levels. THAT really helped on understanding how the system works and it helps greatly with determining what is and what is not excess engine noise. I think that channel was logged at like 200Hz so it had excellent speed that could pick up detonation events most of the time. With the individual channel logging, you could easily see the interaction between knock retard and detonation control. You could also pin down when you had one cylinder that was sensitive to detonation for what ever reason. VERY helpful.
Thank you Ted B for that explanation. Ted B understands how the knock sensor actually works and how it is not "adjustable." If you max the SENSOR out with noise, you'll NEVER detect detonation. It has nothing to do with the multipliers at that point.
As I said, if everything is peachy, yes, the knock sensor can provide a level of protection. The point I am trying to make is that it is most wise to verify that everything is peachy with the knock sensor if that is how you want to tune. Personally, I rely very little on the knock sensor. Yes, you have to keep it happy, otherwise it is going to step in and jackup your tune. But I don't rely on it at all to tune, merely I use it as a safety net should something go wrong.
If you want to see proof, you can simply look at Aarons post in this thread.
One thing I use to really like about DSMLink was that you could datalog individual cylinder knock voltage levels. THAT really helped on understanding how the system works and it helps greatly with determining what is and what is not excess engine noise. I think that channel was logged at like 200Hz so it had excellent speed that could pick up detonation events most of the time. With the individual channel logging, you could easily see the interaction between knock retard and detonation control. You could also pin down when you had one cylinder that was sensitive to detonation for what ever reason. VERY helpful.
Last edited by 03whitegsr; Dec 27, 2010 at 07:54 AM.
Account Disabled
iTrader: (38)
Joined: Aug 2005
Posts: 9,319
Likes: 1
From: Virginia Beach, Virginia
JB in this thread we've been focusing on what I assume are 3rd gear pulls right? How's the car looking when you generate more heat by rowing through the gears? I often need to shave off a little timing etc... for 1/4 mile runs after doing a base tune in 3rd gear. FYI, when I post a customers chart, it's always after we've determined that it's safe to run hard in the 1/4 or for extended periods if our customers are going to do longer sessions.
One aspect of the knock sensor not covered is the window of crank angle at which the knock sensor is listening. This can also be tuned, but it may be hardcoded and not programmable.
food for thought.
food for thought.
I don't know if anybody has found anything on that topic in the stock ECU? The stock ECU does have individual cylinder knock levels though, so it has to be looking at it in some form of a limited time window.
I believe the strategy behind the stock sensor is over kill, the AEM knock table is super easy to set up and does a fine job of handling the safety of the car. Same sensor, different algorithm, much less headaches.
The new Haltech has full knock control in it, waiting on the box to be released for the EVO8's, it's out for the EVO9 now. I hear it's fantastic.
The new Haltech has full knock control in it, waiting on the box to be released for the EVO8's, it's out for the EVO9 now. I hear it's fantastic.
Tscompusa, I'm not getting into a dick swinging competition that will simply draw away from the topic at hand. I will simply respond with "I have tuned enough cars that make enough power to have recognized this important detail LONG ago."
Thank you Ted B for that explanation. Ted B understands how the knock sensor actually works and how it is not "adjustable." If you max the SENSOR out with noise, you'll NEVER detect detonation. It has nothing to do with the multipliers at that point.
As I said, if everything is peachy, yes, the knock sensor can provide a level of protection. The point I am trying to make is that it is most wise to verify that everything is peachy with the knock sensor if that is how you want to tune. Personally, I rely very little on the knock sensor. Yes, you have to keep it happy, otherwise it is going to step in and jackup your tune. But I don't rely on it at all to tune, merely I use it as a safety net should something go wrong.
If you want to see proof, you can simply look at Aarons post in this thread.
One thing I use to really like about DSMLink was that you could datalog individual cylinder knock voltage levels. THAT really helped on understanding how the system works and it helps greatly with determining what is and what is not excess engine noise. I think that channel was logged at like 200Hz so it had excellent speed that could pick up detonation events most of the time. With the individual channel logging, you could easily see the interaction between knock retard and detonation control. You could also pin down when you had one cylinder that was sensitive to detonation for what ever reason. VERY helpful.
Thank you Ted B for that explanation. Ted B understands how the knock sensor actually works and how it is not "adjustable." If you max the SENSOR out with noise, you'll NEVER detect detonation. It has nothing to do with the multipliers at that point.
As I said, if everything is peachy, yes, the knock sensor can provide a level of protection. The point I am trying to make is that it is most wise to verify that everything is peachy with the knock sensor if that is how you want to tune. Personally, I rely very little on the knock sensor. Yes, you have to keep it happy, otherwise it is going to step in and jackup your tune. But I don't rely on it at all to tune, merely I use it as a safety net should something go wrong.
If you want to see proof, you can simply look at Aarons post in this thread.
One thing I use to really like about DSMLink was that you could datalog individual cylinder knock voltage levels. THAT really helped on understanding how the system works and it helps greatly with determining what is and what is not excess engine noise. I think that channel was logged at like 200Hz so it had excellent speed that could pick up detonation events most of the time. With the individual channel logging, you could easily see the interaction between knock retard and detonation control. You could also pin down when you had one cylinder that was sensitive to detonation for what ever reason. VERY helpful.
The way it works from what i gather last night if im correct is a voltage spike can indicate detonation and if it doesnt pass the threshold of the knockbase it will never register as knocksum, so in some circumstances it is not a bad thing to register knock voltage.
im going to log voltage on cars as a 2nd reference for now on just to keep tabs on it.
im going to pm jcsbanks to get some more detail about how it works since he pretty much discovered and disassembled the knock control.
I dont think there is a single evo or 4g63 in that matter that has the capability to peg all 5 volts on the knock sensor though. that would require the engine to be falling apart pretty much.
I will log my car.. its a good candidate. a 2.3L, no balance shafts, little noisey head, clutch, etc.. when i set this car up it was pegged with phantom knock.
I believe the strategy behind the stock sensor is over kill, the AEM knock table is super easy to set up and does a fine job of handling the safety of the car. Same sensor, different algorithm, much less headaches.
The new Haltech has full knock control in it, waiting on the box to be released for the EVO8's, it's out for the EVO9 now. I hear it's fantastic.
The new Haltech has full knock control in it, waiting on the box to be released for the EVO8's, it's out for the EVO9 now. I hear it's fantastic.
what year was this? i stopped using dsmlink around 2007..... if i had a dsm id use it to this day tho, its a nice advanced ecu control for sure.
i will login my forum account and search and see what i find about that, im interested.
ya it gets into that alittle bit on that thread, but not so much.. theres a lot more going on then we think with the knock sensor. i dont see how you can beat the stock ecu knock control when its setup properly.. cant make something better when its as good as it already gets.
Newbie
Joined: Jul 2008
Posts: 72
Likes: 0
From: Finland
In my opinion you wont find nothing as advanced as the stock knock control that the evo ecu has to offer.. even jcsbanks said it, and hes been decompiling/disassembling and rewriting ecu's for years. right now hes working on opening the nissan GTR ecu's, so for someone to say that about the knock control means a lot to me. Its not as standard as people think it is.. theres a lot to it. its pretty advanced.
More advanced system uses ionization current from spark plugs to detect knocking.
That allows to "hear" invidual cylinders.
What system is best for tuning purposes is another thing.
From 1982
http://web.inter.nl.net/users/turbo-team-europe/apc.htm
Tscompuse, last time I used DSMLink was probably '05? V2 was when they first introduced individual cylinder knock level datalog channels. I had chip #11 on DSMLink V1 so I had used it since it first came out.
Something to realize, you don't have to max out the 5V analog input to "max out" the knock sensor system. I don't have the raw numbers on hand here, so I'll follow up when I get home so I can get the actual knock multipliers and adder to show what I mean.
To me, keeping tabs on the knock voltage only needs to be done occasionally. Once you establish the voltage is reasonable, it's not going to change much unless something serious happens; sever engine damage, cam change, heavily worn valvetrain, etc.
There are other electronic ways to detect detonation too, as mentioned above using ionization technology. Another method that is more limited to high powered motors and is more of a fail-safe then a tuning device is to monitor the distance between block and head. If the distance increases (very small change on the order of 0.0001"), cylinder pressures have gotten high from either over advanced timing, or detonation.
Something to realize, you don't have to max out the 5V analog input to "max out" the knock sensor system. I don't have the raw numbers on hand here, so I'll follow up when I get home so I can get the actual knock multipliers and adder to show what I mean.
To me, keeping tabs on the knock voltage only needs to be done occasionally. Once you establish the voltage is reasonable, it's not going to change much unless something serious happens; sever engine damage, cam change, heavily worn valvetrain, etc.
There are other electronic ways to detect detonation too, as mentioned above using ionization technology. Another method that is more limited to high powered motors and is more of a fail-safe then a tuning device is to monitor the distance between block and head. If the distance increases (very small change on the order of 0.0001"), cylinder pressures have gotten high from either over advanced timing, or detonation.
Last edited by 03whitegsr; Dec 27, 2010 at 12:24 PM.
