Dyno Tune knock Gadget and ECU datalog knock
First, That e-mail that you shared with me from Dustin disappointed the heck out of me. It was really horrible to hear that Vishnu has no problem with 15 counts of knock.
Second, let us assume that Sean is correct and we do not know how much knock causes harm to the engine. We do know for CERTAIN that when our logger register 1-3 counts of knock we see timing get pulled. And pulling timing is NOT the optimal way to make power. I often ask myself this question when I see tuners letting the engine ride the knock sensor: What is the point of advancing the timing so much if the knock system in the Evo will pull the timing down to where it should be?
Third, as Dan suggested, the Evo has a very stout engine and can take a beating and keep on running. The damage from knock happens over an extended period of time as your rod bearings get pounded until they fail and you get rod knock.
Second, let us assume that Sean is correct and we do not know how much knock causes harm to the engine. We do know for CERTAIN that when our logger register 1-3 counts of knock we see timing get pulled. And pulling timing is NOT the optimal way to make power. I often ask myself this question when I see tuners letting the engine ride the knock sensor: What is the point of advancing the timing so much if the knock system in the Evo will pull the timing down to where it should be?
Third, as Dan suggested, the Evo has a very stout engine and can take a beating and keep on running. The damage from knock happens over an extended period of time as your rod bearings get pounded until they fail and you get rod knock.
First, That e-mail that you shared with me from Dustin disappointed the heck out of me. It was really horrible to hear that Vishnu has no problem with 15 counts of knock.
Second, let us assume that Sean is correct and we do not know how much knock causes harm to the engine. We do know for CERTAIN that when our logger register 1-3 counts of knock we see timing get pulled. And pulling timing is NOT the optimal way to make power. I often ask myself this question when I see tuners letting the engine ride the knock sensor: What is the point of advancing the timing so much if the knock system in the Evo will pull the timing down to where it should be?
Third, as Dan suggested, the Evo has a very stout engine and can take a beating and keep on running. The damage from knock happens over an extended period of time as your rod bearings get pounded until they fail and you get rod knock.
Second, let us assume that Sean is correct and we do not know how much knock causes harm to the engine. We do know for CERTAIN that when our logger register 1-3 counts of knock we see timing get pulled. And pulling timing is NOT the optimal way to make power. I often ask myself this question when I see tuners letting the engine ride the knock sensor: What is the point of advancing the timing so much if the knock system in the Evo will pull the timing down to where it should be?
Third, as Dan suggested, the Evo has a very stout engine and can take a beating and keep on running. The damage from knock happens over an extended period of time as your rod bearings get pounded until they fail and you get rod knock.
i just hope that my engine will last long and that the prior bad tune it had wont cause any future problems with it.
the the above poster said no use tuning for high timing when the ecu will pull it down to curb the knock it gets from the Sensor.
we have seen that at 3 KC it drops the timing and at 6KC the car goes in octane map as the octane level drops from 100.
so we should be aiming for 2-3 knock sums when tuning.
we have seen that at 3 KC it drops the timing and at 6KC the car goes in octane map as the octane level drops from 100.
so we should be aiming for 2-3 knock sums when tuning.
Evolved Member
Joined: May 2006
Posts: 2,399
Likes: 6
From: UK
jrsimon27, sorry if I've been contradictory or confusing. I'll try to confirm my present understanding which has also developed over time...
The only knock issue I've had on the Evo has been lift off knock. I had a knocklink in and used det cans before Evoscan was first released. I was getting knock episodes that I could sometimes hear in the cabin, but it was only when I had proper datalogs when Evoscan appeared that I could see three things:
1. All the knocks I heard in the det cans and could see on the knocklink had a recorded knock sum which gave me confidence that I didn't need to continue using the det cans or knocklink.
2. All the stuff I could hear turned out to be starting within one log row of starting to lift from 100% throttle - ie lift off knock.
3. Lift off knocks of IIRC about 10 could be heard in the cabin.
I can't tell you absolutely what level of knock sum will damage your engine and over what time, and I can't promise anyone that lift off knock doesn't do any harm either (although I suspect considerably less than full throttle knock).
What I'm sure on is that by over doing the timing map you end up with less net ignition timing than you'd have by cooperating with the knock control system. The way the standard cars are tuned I honestly don't believe knock sums that are less than 6 are likely to do any harm at all. It is very possible that knock sums of 15 or so don't either. With knock sums this high of course the ECU will migrate towards the low octane maps so the ECU is always trying to get the knock sums in the 4-5 range. If some parts of the map result in the octane value creeping back up and then you have knock sums of 0, 0, 0, 15, 15, 14, 14 etc then it wouldn't surprise me if these repeated sudden 15s might do some damage because there has obviously been quite a sudden loud noise. It may well be that you can hear these in det cans (I would expect so), but it would need such a poorly designed ignition map to set it off that I haven't triggered it.
I don't tune to knock sums of 5 because I can usually end up with another degree and a smoother drive if I just remove 1 degree from the ignition map, and then end up with say 0 or 1 knock sums. Sometimes just 1 degree can dramatically reduce the knock sums. It may not just be the ECU over-reacting (and even if it was that would be sensible), but once the conditions for knock have been setup then I feel anecdotally that it needs more ignition retard to get rid of it than if you'd prevented it in the first place. It seems to be an avalanche phenomenon IYSWIM.
The ECU probably throws fuel in in response to knock sums as well, but that is a whole different discussion.
The only knock issue I've had on the Evo has been lift off knock. I had a knocklink in and used det cans before Evoscan was first released. I was getting knock episodes that I could sometimes hear in the cabin, but it was only when I had proper datalogs when Evoscan appeared that I could see three things:
1. All the knocks I heard in the det cans and could see on the knocklink had a recorded knock sum which gave me confidence that I didn't need to continue using the det cans or knocklink.
2. All the stuff I could hear turned out to be starting within one log row of starting to lift from 100% throttle - ie lift off knock.
3. Lift off knocks of IIRC about 10 could be heard in the cabin.
I can't tell you absolutely what level of knock sum will damage your engine and over what time, and I can't promise anyone that lift off knock doesn't do any harm either (although I suspect considerably less than full throttle knock).
What I'm sure on is that by over doing the timing map you end up with less net ignition timing than you'd have by cooperating with the knock control system. The way the standard cars are tuned I honestly don't believe knock sums that are less than 6 are likely to do any harm at all. It is very possible that knock sums of 15 or so don't either. With knock sums this high of course the ECU will migrate towards the low octane maps so the ECU is always trying to get the knock sums in the 4-5 range. If some parts of the map result in the octane value creeping back up and then you have knock sums of 0, 0, 0, 15, 15, 14, 14 etc then it wouldn't surprise me if these repeated sudden 15s might do some damage because there has obviously been quite a sudden loud noise. It may well be that you can hear these in det cans (I would expect so), but it would need such a poorly designed ignition map to set it off that I haven't triggered it.
I don't tune to knock sums of 5 because I can usually end up with another degree and a smoother drive if I just remove 1 degree from the ignition map, and then end up with say 0 or 1 knock sums. Sometimes just 1 degree can dramatically reduce the knock sums. It may not just be the ECU over-reacting (and even if it was that would be sensible), but once the conditions for knock have been setup then I feel anecdotally that it needs more ignition retard to get rid of it than if you'd prevented it in the first place. It seems to be an avalanche phenomenon IYSWIM.
The ECU probably throws fuel in in response to knock sums as well, but that is a whole different discussion.
Last edited by jcsbanks; Sep 4, 2007 at 02:18 AM.
Evolved Member
Joined: May 2006
Posts: 2,399
Likes: 6
From: UK
Going back to the original post, it could well be that the ECU is keeping the external knock devices quiet by effectively doing a good closed loop job of managing the noise that the knock sensor will pick up - ie doing what it is supposed to.
The alternative to cooperating with the knock control system is to turn it off and just map the car with open loop ignition timing. Trouble is you have to leave a degree or two off so that the engine never knocks in different conditions. Unless you have too much piston or other noise you can't get rid of or calibrate around I think this is a step backwards.
The alternative to cooperating with the knock control system is to turn it off and just map the car with open loop ignition timing. Trouble is you have to leave a degree or two off so that the engine never knocks in different conditions. Unless you have too much piston or other noise you can't get rid of or calibrate around I think this is a step backwards.
thanks jcsbanks maybe i misunderstood you,you where talking about lift off knock and i got it as if it was full throttle knock my mistake sorry
jrsimon27, sorry if I've been contradictory or confusing. I'll try to confirm my present understanding which has also developed over time...
The only knock issue I've had on the Evo has been lift off knock. I had a knocklink in and used det cans before Evoscan was first released. I was getting knock episodes that I could sometimes hear in the cabin, but it was only when I had proper datalogs when Evoscan appeared that I could see three things:
1. All the knocks I heard in the det cans and could see on the knocklink had a recorded knock sum which gave me confidence that I didn't need to continue using the det cans or knocklink.
2. All the stuff I could hear turned out to be starting within one log row of starting to lift from 100% throttle - ie lift off knock.
3. Lift off knocks of IIRC about 10 could be heard in the cabin.
I can't tell you absolutely what level of knock sum will damage your engine and over what time, and I can't promise anyone that lift off knock doesn't do any harm either (although I suspect considerably less than full throttle knock).
What I'm sure on is that by over doing the timing map you end up with less net ignition timing than you'd have by cooperating with the knock control system. The way the standard cars are tuned I honestly don't believe knock sums that are less than 6 are likely to do any harm at all. It is very possible that knock sums of 15 or so don't either. With knock sums this high of course the ECU will migrate towards the low octane maps so the ECU is always trying to get the knock sums in the 4-5 range. If some parts of the map result in the octane value creeping back up and then you have knock sums of 0, 0, 0, 15, 15, 14, 14 etc then it wouldn't surprise me if these repeated sudden 15s might do some damage because there has obviously been quite a sudden loud noise. It may well be that you can hear these in det cans (I would expect so), but it would need such a poorly designed ignition map to set it off that I haven't triggered it.
I don't tune to knock sums of 5 because I can usually end up with another degree and a smoother drive if I just remove 1 degree from the ignition map, and then end up with say 0 or 1 knock sums. Sometimes just 1 degree can dramatically reduce the knock sums. It may not just be the ECU over-reacting (and even if it was that would be sensible), but once the conditions for knock have been setup then I feel anecdotally that it needs more ignition retard to get rid of it than if you'd prevented it in the first place. It seems to be an avalanche phenomenon IYSWIM.
The ECU probably throws fuel in in response to knock sums as well, but that is a whole different discussion.
The only knock issue I've had on the Evo has been lift off knock. I had a knocklink in and used det cans before Evoscan was first released. I was getting knock episodes that I could sometimes hear in the cabin, but it was only when I had proper datalogs when Evoscan appeared that I could see three things:
1. All the knocks I heard in the det cans and could see on the knocklink had a recorded knock sum which gave me confidence that I didn't need to continue using the det cans or knocklink.
2. All the stuff I could hear turned out to be starting within one log row of starting to lift from 100% throttle - ie lift off knock.
3. Lift off knocks of IIRC about 10 could be heard in the cabin.
I can't tell you absolutely what level of knock sum will damage your engine and over what time, and I can't promise anyone that lift off knock doesn't do any harm either (although I suspect considerably less than full throttle knock).
What I'm sure on is that by over doing the timing map you end up with less net ignition timing than you'd have by cooperating with the knock control system. The way the standard cars are tuned I honestly don't believe knock sums that are less than 6 are likely to do any harm at all. It is very possible that knock sums of 15 or so don't either. With knock sums this high of course the ECU will migrate towards the low octane maps so the ECU is always trying to get the knock sums in the 4-5 range. If some parts of the map result in the octane value creeping back up and then you have knock sums of 0, 0, 0, 15, 15, 14, 14 etc then it wouldn't surprise me if these repeated sudden 15s might do some damage because there has obviously been quite a sudden loud noise. It may well be that you can hear these in det cans (I would expect so), but it would need such a poorly designed ignition map to set it off that I haven't triggered it.
I don't tune to knock sums of 5 because I can usually end up with another degree and a smoother drive if I just remove 1 degree from the ignition map, and then end up with say 0 or 1 knock sums. Sometimes just 1 degree can dramatically reduce the knock sums. It may not just be the ECU over-reacting (and even if it was that would be sensible), but once the conditions for knock have been setup then I feel anecdotally that it needs more ignition retard to get rid of it than if you'd prevented it in the first place. It seems to be an avalanche phenomenon IYSWIM.
The ECU probably throws fuel in in response to knock sums as well, but that is a whole different discussion.
The alternative to cooperating with the knock control system is to turn it off and just map the car with open loop ignition timing. Trouble is you have to leave a degree or two off so that the engine never knocks in different conditions. Unless you have too much piston or other noise you can't get rid of or calibrate around I think this is a step backwards.
detcans is good for supercharged V8s and cars that handle knock that would otherwise melt a 4g63.
Right now our cars ar still slaves to the knock sensor, unless you use AEMEMS in which case you better be able to make it work as good as the mitsu knock control system
Right now our cars ar still slaves to the knock sensor, unless you use AEMEMS in which case you better be able to make it work as good as the mitsu knock control system
What about knock counts at low throttle and 1st gear when engine is quiet. Does this occur because background noise is less and the knock sensor is more sensitive?
I always get a part throttle knocksum in first gear. Knock sum usually 2-8 and load is between 60 and 120%
I always get a part throttle knocksum in first gear. Knock sum usually 2-8 and load is between 60 and 120%
Evolved Member
Joined: May 2006
Posts: 2,399
Likes: 6
From: UK
I do believe you, I did a ring land on an EJ257 with sudden unexpected knock at 150mph on an ECU that didn't have knock control, although I was at 100 BHP over what people thought was the sensible limit at that time on that lump. I also fiddled with a J&S Safeguard on this engine, but it couldn't be set so that it would pick up det satisfactorily at high RPM whilst not picking up false knock at low RPM. It needed to be fully mappable even if the gated, filtered sampling idea was good. I fed my findings back to J&S. This was on the GC8 Impreza - obviously the GDB ECU can handle full closed loop knock control very well, the GC8 OEM ECU and aftermarket options did not do anywhere near as well.
I would have thought Bez Bashni would be in a position to run knock control with his noisy forged pistons, but he gave up on it and mapped it open loop. If you look at his knock control algorithm stuff on aktivematrix it seems pretty difficult to follow exactly what the OEM ECU is doing. It seems to sample at specific crank angles and use a differential between the present and previous values.
Others just fit their forged bits and don't have trouble.
If I get a forged motor I will only go with parts and engine builder that have turned out identical examples that don't false knock. I had this nightmare before with a forged engine again on a Subaru, it pegged the knock correction to max retard as soon as it hit the knock control zone and there was no way to recalibrate it. The engine was so noisy I returned it, but I believe the noise floor was so high that you could easily have knock that would be drowned out by the noise.
There are successful high power European Evos running on MoTeC and Autronic without knock control though, but the mapping must have an extra safety margin in there over a knock control system that can be leaned upon where necessary even if you don't intend to all the time?
I would have thought Bez Bashni would be in a position to run knock control with his noisy forged pistons, but he gave up on it and mapped it open loop. If you look at his knock control algorithm stuff on aktivematrix it seems pretty difficult to follow exactly what the OEM ECU is doing. It seems to sample at specific crank angles and use a differential between the present and previous values.
Others just fit their forged bits and don't have trouble.
If I get a forged motor I will only go with parts and engine builder that have turned out identical examples that don't false knock. I had this nightmare before with a forged engine again on a Subaru, it pegged the knock correction to max retard as soon as it hit the knock control zone and there was no way to recalibrate it. The engine was so noisy I returned it, but I believe the noise floor was so high that you could easily have knock that would be drowned out by the noise.
There are successful high power European Evos running on MoTeC and Autronic without knock control though, but the mapping must have an extra safety margin in there over a knock control system that can be leaned upon where necessary even if you don't intend to all the time?
Don't even suggest this. Motors are lost quick on the street without knock control (PM me if you want pictures). I've ran 9 thou piston/wall clearance forged pistons and played nice with the knock sensor. You can always play nice with the OEM knock sensor up to at least 600whp because thats the highest HP car I've messed around with. I theorize that it works well even up until 800whp.
Last edited by jcsbanks; Sep 4, 2007 at 08:01 AM.
Has Bez or anyone for that matter developed a patch to disable off boost knock control?
I am about to develop a boost switch activated knock sensor bypass relay, but damn a code fix would be a much better deal.
I am about to develop a boost switch activated knock sensor bypass relay, but damn a code fix would be a much better deal.
I never understood why someone would tune using something like an HKS knock amp anyhow because I have many data logs showing the knock sensor spiked in voltage, but never registered a knock count. If you plot out the knock sensor voltage versus knock count, you will sometimes see a very low voltage but a high knock count, and vise versa.
With that kind of knowledge from those data logs, why would anyone use any kind of voltage regulated knock thingy-ma-jig to tune with when the knock sensor doesn't appear to determine knock solely based on output voltage?
And that's exactly what I mean. So, if a local "tooner" was doing this with an HKS knock amp, when what was he really accomplishing? Anything? Bad practices?