fed up with knock control
Account Disabled
iTrader: (38)
Joined: Aug 2005
Posts: 9,319
Likes: 1
From: Virginia Beach, Virginia
I have a question. I will occasionally get a huge KS spike (sometimes up to 20 counts)in the dreaded 3K-3.5K range when the car transitions from closed loop to open loop and passes the 0 psi mark. I've run my car thru this range a million times with no knock and I'm confident in the timing and fueling. The issue I see is the taper. I have had it peak at 20 counts at around 3200rpm and slowly taper like a set of stairs all the way down til 5K rpm. The taper is my issue. Once I go past my 'ignore' cells it will immediately drop timing when if falls on the knock taper and still ruin the run.
Someone was talking about knock decay time a while back and wondered if the ECU intentionally has the KS taper down slowly rather than instantly disappear and cause the timing to advance by a huge amount in an instant. I see the reasoning behind this but wonder if my thoery is correct at all. If so, can this decay time be adjusted for a mosre promt recovery? It would be cool if the knock sensor was literally "deaf" up until a certain RPM. If it didn't even start listening until a certain range, that would rock.
Someone was talking about knock decay time a while back and wondered if the ECU intentionally has the KS taper down slowly rather than instantly disappear and cause the timing to advance by a huge amount in an instant. I see the reasoning behind this but wonder if my thoery is correct at all. If so, can this decay time be adjusted for a mosre promt recovery? It would be cool if the knock sensor was literally "deaf" up until a certain RPM. If it didn't even start listening until a certain range, that would rock.
I have a question. I will occasionally get a huge KS spike (sometimes up to 20 counts)in the dreaded 3K-3.5K range when the car transitions from closed loop to open loop and passes the 0 psi mark. I've run my car thru this range a million times with no knock and I'm confident in the timing and fueling. The issue I see is the taper. I have had it peak at 20 counts at around 3200rpm and slowly taper like a set of stairs all the way down til 5K rpm. The taper is my issue. Once I go past my 'ignore' cells it will immediately drop timing when if falls on the knock taper and still ruin the run.
Someone was talking about knock decay time a while back and wondered if the ECU intentionally has the KS taper down slowly rather than instantly disappear and cause the timing to advance by a huge amount in an instant. I see the reasoning behind this but wonder if my thoery is correct at all. If so, can this decay time be adjusted for a mosre promt recovery? It would be cool if the knock sensor was literally "deaf" up until a certain RPM. If it didn't even start listening until a certain range, that would rock.
Someone was talking about knock decay time a while back and wondered if the ECU intentionally has the KS taper down slowly rather than instantly disappear and cause the timing to advance by a huge amount in an instant. I see the reasoning behind this but wonder if my thoery is correct at all. If so, can this decay time be adjusted for a mosre promt recovery? It would be cool if the knock sensor was literally "deaf" up until a certain RPM. If it didn't even start listening until a certain range, that would rock.
Yes knock decay rate. It changes with different loads as well usually on most if not all ECU's. The knock sum should be picked up before it enters the routine for the decay rate. It is probably going to do a compare to load and branch one way if higher and another if lower for the different decay rates.
Awesome, I'll have to track it down. So this does what I think? Adjusts how long the ECU allows the KS to linger after the "actual" knock occurance? Ironically, I would imagine if the decay time was too quick it could actually CAUSE knock from the drastic changes in timing.
Yes knock decay rate. It changes with different loads as well usually on most if not all ECU's. The knock sum should be picked up before it enters the routine for the decay rate. It is probably going to do a compare to load and branch one way if higher and another if lower for the different decay rates.
It seems like bypassing the decay routine will be good enough for your concerns Mellon.
Last edited by Jack_of_Trades; Nov 10, 2008 at 06:53 PM.
mm that will be a lot harder todo.
unfortunately the knock routine is uber complex, so the best way (and safest) is to intercept the knocksum affect on ignadvance.
anyone got any logs of these phantom knocks?
unfortunately the knock routine is uber complex, so the best way (and safest) is to intercept the knocksum affect on ignadvance.
anyone got any logs of these phantom knocks?
You do not want to bypass the decay routine. You would want to take the knock sum and do the ignore calculations. If it is above this knock as usual with decay as usual. You need the decay.........
BTW, thanks for helping me out.
Perhaps there needs to be a control to base and view what is really happening with these nuisance knock sums we're all getting (wether phantom or exasterbated mild knock).
Something such as datalogs of VIII and IX's bone stock on, say 93 oct on stock boost and controlled boost. Just to see if aftermarket cams play a role. If a particular pattern is noted compared to like aftermarket cams (with load % set similar), we might be able to guess what some of the knock filters mean.
Something such as datalogs of VIII and IX's bone stock on, say 93 oct on stock boost and controlled boost. Just to see if aftermarket cams play a role. If a particular pattern is noted compared to like aftermarket cams (with load % set similar), we might be able to guess what some of the knock filters mean.
Jamie - knock decay info
https://www.evolutionm.net/forums/sh...1&postcount=23
Not much info really, but should give a starting point for other disasmers OR testers
https://www.evolutionm.net/forums/sh...1&postcount=23
Not much info really, but should give a starting point for other disasmers OR testers
Evolved Member
Joined: Apr 2008
Posts: 604
Likes: 0
From: Sydney, Australia
I can give you an endless supply of 36 knock sum logs if you want them. :-) All with no real knock at all.
See if you can follow me on this. Is everyone sure the knock sum function is what its name says? Are we sure it is not actually the scaled calculation of ignition timing to be pulled.... back to front to thinking certain knock sum is what causes the timing to be retarded. I feel we think backwards on it, if you understand my way of thinking. Because, how can knock sum decay if there is no knock. It should either be knocking enough to cause a certain knock sum, or nothing at all if that is its function. The amount of retard on the other hand will decay.
See if you can follow me on this. Is everyone sure the knock sum function is what its name says? Are we sure it is not actually the scaled calculation of ignition timing to be pulled.... back to front to thinking certain knock sum is what causes the timing to be retarded. I feel we think backwards on it, if you understand my way of thinking. Because, how can knock sum decay if there is no knock. It should either be knocking enough to cause a certain knock sum, or nothing at all if that is its function. The amount of retard on the other hand will decay.
I'm not sure I follow, but I will try to answer what I think your asking.
I by no means have been able to dissect the evo code yet. However I have spent 8 years playing with the knock system on DSM's and about 5 of those working with the machine level code. In the 3 years I've been logging evos I've noticed similarities. All I'm doing is applying my working knowledge from the DSM and correlating it to the Evo. It turns out that the evo behaves the same as the DSM with a few vaiables changed here and their and a few extra functions from time to time (like lean spool).
At any rate, the way the 4g63 knock system is like this.
1. The knock sensor doesn't work at idle type situations. IE: below a certain load and rpm. The reason is so that the car won't stall out if it gets "knock" at idle. Find this variable and you can effectively disable the knock sensor.
2. At low load situations but not high load the car will sense knock and apply a timing retard. I've noticed that IX's have different retard for different RPM and or load ranges. At any rate the knock decay is much faster, usually around 10-100 times faster during low load cruising. Without said knock and full octane the car will apply full timing.
3. At full throttle high load (usually over 100% load) if you get a really quick ping at the right moment (say a harmonic vibration that is in sync with the knock sensor) the knock sum will spike, in some cases it will max out. If and only if knock fully disappears will the decay routine run. If knock continues the sums will rise. If the knock slowly tapers away the knock will stay the same. I've noticed that IX's will pull 1 deg per count some times and 1 deg per 3 counts sometimes. Whereas 8's tend to pull what looks like Z degrees per count of knock. On DSM's it was approximately 1/3 deg timing pull per count of knock.
I am 99.9999% sure that knock is its own entity and not a scaled version of timing pull.
I by no means have been able to dissect the evo code yet. However I have spent 8 years playing with the knock system on DSM's and about 5 of those working with the machine level code. In the 3 years I've been logging evos I've noticed similarities. All I'm doing is applying my working knowledge from the DSM and correlating it to the Evo. It turns out that the evo behaves the same as the DSM with a few vaiables changed here and their and a few extra functions from time to time (like lean spool).
At any rate, the way the 4g63 knock system is like this.
1. The knock sensor doesn't work at idle type situations. IE: below a certain load and rpm. The reason is so that the car won't stall out if it gets "knock" at idle. Find this variable and you can effectively disable the knock sensor.
2. At low load situations but not high load the car will sense knock and apply a timing retard. I've noticed that IX's have different retard for different RPM and or load ranges. At any rate the knock decay is much faster, usually around 10-100 times faster during low load cruising. Without said knock and full octane the car will apply full timing.
3. At full throttle high load (usually over 100% load) if you get a really quick ping at the right moment (say a harmonic vibration that is in sync with the knock sensor) the knock sum will spike, in some cases it will max out. If and only if knock fully disappears will the decay routine run. If knock continues the sums will rise. If the knock slowly tapers away the knock will stay the same. I've noticed that IX's will pull 1 deg per count some times and 1 deg per 3 counts sometimes. Whereas 8's tend to pull what looks like Z degrees per count of knock. On DSM's it was approximately 1/3 deg timing pull per count of knock.
I am 99.9999% sure that knock is its own entity and not a scaled version of timing pull.
I'm not sure I follow, but I will try to answer what I think your asking.
I by no means have been able to dissect the evo code yet. However I have spent 8 years playing with the knock system on DSM's and about 5 of those working with the machine level code. In the 3 years I've been logging evos I've noticed similarities. All I'm doing is applying my working knowledge from the DSM and correlating it to the Evo. It turns out that the evo behaves the same as the DSM with a few vaiables changed here and their and a few extra functions from time to time (like lean spool).
At any rate, the way the 4g63 knock system is like this.
1. The knock sensor doesn't work at idle type situations. IE: below a certain load and rpm. The reason is so that the car won't stall out if it gets "knock" at idle. Find this variable and you can effectively disable the knock sensor.
2. At low load situations but not high load the car will sense knock and apply a timing retard. I've noticed that IX's have different retard for different RPM and or load ranges. At any rate the knock decay is much faster, usually around 10-100 times faster during low load cruising. Without said knock and full octane the car will apply full timing.
3. At full throttle high load (usually over 100% load) if you get a really quick ping at the right moment (say a harmonic vibration that is in sync with the knock sensor) the knock sum will spike, in some cases it will max out. If and only if knock fully disappears will the decay routine run. If knock continues the sums will rise. If the knock slowly tapers away the knock will stay the same. I've noticed that IX's will pull 1 deg per count some times and 1 deg per 3 counts sometimes. Whereas 8's tend to pull what looks like Z degrees per count of knock. On DSM's it was approximately 1/3 deg timing pull per count of knock.
I am 99.9999% sure that knock is its own entity and not a scaled version of timing pull.
I by no means have been able to dissect the evo code yet. However I have spent 8 years playing with the knock system on DSM's and about 5 of those working with the machine level code. In the 3 years I've been logging evos I've noticed similarities. All I'm doing is applying my working knowledge from the DSM and correlating it to the Evo. It turns out that the evo behaves the same as the DSM with a few vaiables changed here and their and a few extra functions from time to time (like lean spool).
At any rate, the way the 4g63 knock system is like this.
1. The knock sensor doesn't work at idle type situations. IE: below a certain load and rpm. The reason is so that the car won't stall out if it gets "knock" at idle. Find this variable and you can effectively disable the knock sensor.
2. At low load situations but not high load the car will sense knock and apply a timing retard. I've noticed that IX's have different retard for different RPM and or load ranges. At any rate the knock decay is much faster, usually around 10-100 times faster during low load cruising. Without said knock and full octane the car will apply full timing.
3. At full throttle high load (usually over 100% load) if you get a really quick ping at the right moment (say a harmonic vibration that is in sync with the knock sensor) the knock sum will spike, in some cases it will max out. If and only if knock fully disappears will the decay routine run. If knock continues the sums will rise. If the knock slowly tapers away the knock will stay the same. I've noticed that IX's will pull 1 deg per count some times and 1 deg per 3 counts sometimes. Whereas 8's tend to pull what looks like Z degrees per count of knock. On DSM's it was approximately 1/3 deg timing pull per count of knock.
I am 99.9999% sure that knock is its own entity and not a scaled version of timing pull.
Point #1 is untrue. I don't know if you've ever gridded up behind a loud domestic car before, but if you do then you tell me if the knock sensor isn't working. It is. At idle if you sit behind a loud V8 that revs your car will stumble and stall sometimes due to the knock it sees. I've stalled on grid before because of this.
Point #1 is untrue. I don't know if you've ever gridded up behind a loud domestic car before, but if you do then you tell me if the knock sensor isn't working. It is. At idle if you sit behind a loud V8 that revs your car will stumble and stall sometimes due to the knock it sees. I've stalled on grid before because of this.
No sure if I am right but I don't think what you are saying makes any sense at all, my car has never done this since I switched to a MAF translator. However when I had the factory MAF there it would do as you say sometimes.
~Zach
Evolving Member
Joined: Dec 2005
Posts: 261
Likes: 2
From: Silverdale,WA
+1 on the maf vibrations, a car with a good set of 12's or a harley with straight pipes will f--k my idle up. It's not as bad with my intake cause the maf kinda floats on a little bendy-alumnium bracket. not connect as ridigly to the frame anymore, so most vibrations dont get to the maf now.
