Wierd knock back-off issue
Thread Starter
Joined: Aug 2003
Posts: 14,094
Likes: 1,092
From: Mid-Hudson, NY
Wierd knock back-off issue
Hey all, I had a weird thing come up on the dyno this weekend.
I currently have my knock compensation set that at 2.1v knock, add +6 fuel and take out 4 degrees of timing. Again, I have this set at 2.1v.
So on the dyno today, the 1st pull had a normal wideband AFR curve on the dyno chart, but the 2nd and 3rd pull had this weird dip in the AFR curve.
So naturally I thought that was weird unless I knocked alot there to get 2.1v.
So I checked the logs after. I never hit 2.1v of knock.
Check out my log here:
http://www.norcalmotorsports.org/use...dynotest03.ecd
at 5827rpms (x12.32), I have 1.8v of knock, then shortly after at 6027rpms (x12.60) the log goes from -22 fuel to -16 fuel (added +6 fuel) and it "adjust timing" shows -4 degrees of adjustment. So basically it did what I asked it to do at 2.1v, but at only 1.8v of knock.
Also at 6187rpms (x12.76), it shows only 1.66v of knock but then at x13.36 it does the same adjustment.
Any ideas or explanations? Did I find a bug or a snafu?
I currently have my knock compensation set that at 2.1v knock, add +6 fuel and take out 4 degrees of timing. Again, I have this set at 2.1v.
So on the dyno today, the 1st pull had a normal wideband AFR curve on the dyno chart, but the 2nd and 3rd pull had this weird dip in the AFR curve.
So naturally I thought that was weird unless I knocked alot there to get 2.1v.
So I checked the logs after. I never hit 2.1v of knock.
Check out my log here:
http://www.norcalmotorsports.org/use...dynotest03.ecd
at 5827rpms (x12.32), I have 1.8v of knock, then shortly after at 6027rpms (x12.60) the log goes from -22 fuel to -16 fuel (added +6 fuel) and it "adjust timing" shows -4 degrees of adjustment. So basically it did what I asked it to do at 2.1v, but at only 1.8v of knock.
Also at 6187rpms (x12.76), it shows only 1.66v of knock but then at x13.36 it does the same adjustment.
Any ideas or explanations? Did I find a bug or a snafu?
Thread Starter
Joined: Aug 2003
Posts: 14,094
Likes: 1,092
From: Mid-Hudson, NY
Originally Posted by tlcoll1
Is your knock display set for peak-hold with decay, or instantaneous? I'm guessing the latter.
Tom
Tom
So does that mean that peak-hold actually shows accurate peak numbers and instantaneous does not?
Which means my 1.6v in instantaneous was actually a 2.1v?
That's a little scary.
Originally Posted by razorlab
Instantaneous.
So does that mean that peak-hold actually shows accurate peak numbers and instantaneous does not?
Which means my 1.6v in instantaneous was actually a 2.1v?
That's a little scary.
So does that mean that peak-hold actually shows accurate peak numbers and instantaneous does not?
Which means my 1.6v in instantaneous was actually a 2.1v?
That's a little scary.
It's a long explanation, so bear with me. The true knock value is grabbed after every firing cycle in the engine. The ECU+ plugs that knock number into the instantaneous value, and also updates the peak-hold w/decay value. The peak-hold w/decay number is used to decide whether to do the knock-related fuel and timing back-off. I do it this way because it's safer- that is, the peak-hold w/decay number inside will "hang around" longer than the instantaneous value, and if you're really getting knock, it's important to implement the back-off until the problem has solidly gone away rather than for just one engine cycle.
When you're datalogging, you're seeing a sampled version of the knock (instantaneous or peak-hold w/decay) value. At high RPMs, the datalogging samples are much slower than the rate that the values are being updated inside the ECU+. In particular, you may not see (on a datalog) a spike over your threshold even though it actually occurred inside the silver box. That's what probably happened - you got a spike over the threshold, but it went away before the next datalogging point. However, since the ECU+ noticed the spike, it did its job and applied the back-off values. And since the ECU+ is using the peak-hold w/decay value for back-off, you saw that the back-off hung around for a few datalogging samples.
So the lesson here is that the peak-hold w/decay number is more accurate if you're trying to tune your back-off voltage, simply because the rate of datalogging won't really let you see the true per-firing knock value.
Make sense?
Tom
When you're datalogging, you're seeing a sampled version of the knock (instantaneous or peak-hold w/decay) value. At high RPMs, the datalogging samples are much slower than the rate that the values are being updated inside the ECU+. In particular, you may not see (on a datalog) a spike over your threshold even though it actually occurred inside the silver box. That's what probably happened - you got a spike over the threshold, but it went away before the next datalogging point. However, since the ECU+ noticed the spike, it did its job and applied the back-off values. And since the ECU+ is using the peak-hold w/decay value for back-off, you saw that the back-off hung around for a few datalogging samples.
So the lesson here is that the peak-hold w/decay number is more accurate if you're trying to tune your back-off voltage, simply because the rate of datalogging won't really let you see the true per-firing knock value.
Make sense?
Tom
Originally Posted by tlcoll1
It's a long explanation, so bear with me. The true knock value is grabbed after every firing cycle in the engine. The ECU+ plugs that knock number into the instantaneous value, and also updates the peak-hold w/decay value. The peak-hold w/decay number is used to decide whether to do the knock-related fuel and timing back-off. I do it this way because it's safer- that is, the peak-hold w/decay number inside will "hang around" longer than the instantaneous value, and if you're really getting knock, it's important to implement the back-off until the problem has solidly gone away rather than for just one engine cycle.
When you're datalogging, you're seeing a sampled version of the knock (instantaneous or peak-hold w/decay) value. At high RPMs, the datalogging samples are much slower than the rate that the values are being updated inside the ECU+. In particular, you may not see (on a datalog) a spike over your threshold even though it actually occurred inside the silver box. That's what probably happened - you got a spike over the threshold, but it went away before the next datalogging point. However, since the ECU+ noticed the spike, it did its job and applied the back-off values. And since the ECU+ is using the peak-hold w/decay value for back-off, you saw that the back-off hung around for a few datalogging samples.
So the lesson here is that the peak-hold w/decay number is more accurate if you're trying to tune your back-off voltage, simply because the rate of datalogging won't really let you see the true per-firing knock value.
Make sense?
Tom
When you're datalogging, you're seeing a sampled version of the knock (instantaneous or peak-hold w/decay) value. At high RPMs, the datalogging samples are much slower than the rate that the values are being updated inside the ECU+. In particular, you may not see (on a datalog) a spike over your threshold even though it actually occurred inside the silver box. That's what probably happened - you got a spike over the threshold, but it went away before the next datalogging point. However, since the ECU+ noticed the spike, it did its job and applied the back-off values. And since the ECU+ is using the peak-hold w/decay value for back-off, you saw that the back-off hung around for a few datalogging samples.
So the lesson here is that the peak-hold w/decay number is more accurate if you're trying to tune your back-off voltage, simply because the rate of datalogging won't really let you see the true per-firing knock value.
Make sense?
Tom
Trending Topics
It's a long explanation, so bear with me. The true knock value is grabbed after every firing cycle in the engine. The ECU+ plugs that knock number into the instantaneous value, and also updates the peak-hold w/decay value. The peak-hold w/decay number is used to decide whether to do the knock-related fuel and timing back-off. I do it this way because it's safer- that is, the peak-hold w/decay number inside will "hang around" longer than the instantaneous value, and if you're really getting knock, it's important to implement the back-off until the problem has solidly gone away rather than for just one engine cycle.
When you're datalogging, you're seeing a sampled version of the knock (instantaneous or peak-hold w/decay) value. At high RPMs, the datalogging samples are much slower than the rate that the values are being updated inside the ECU+. In particular, you may not see (on a datalog) a spike over your threshold even though it actually occurred inside the silver box. That's what probably happened - you got a spike over the threshold, but it went away before the next datalogging point. However, since the ECU+ noticed the spike, it did its job and applied the back-off values. And since the ECU+ is using the peak-hold w/decay value for back-off, you saw that the back-off hung around for a few datalogging samples.
So the lesson here is that the peak-hold w/decay number is more accurate if you're trying to tune your back-off voltage, simply because the rate of datalogging won't really let you see the true per-firing knock value.
When you're datalogging, you're seeing a sampled version of the knock (instantaneous or peak-hold w/decay) value. At high RPMs, the datalogging samples are much slower than the rate that the values are being updated inside the ECU+. In particular, you may not see (on a datalog) a spike over your threshold even though it actually occurred inside the silver box. That's what probably happened - you got a spike over the threshold, but it went away before the next datalogging point. However, since the ECU+ noticed the spike, it did its job and applied the back-off values. And since the ECU+ is using the peak-hold w/decay value for back-off, you saw that the back-off hung around for a few datalogging samples.
So the lesson here is that the peak-hold w/decay number is more accurate if you're trying to tune your back-off voltage, simply because the rate of datalogging won't really let you see the true per-firing knock value.
Tom
Never forgets a feature
Thread
Thread Starter
Forum
Replies
Last Post
blk-majik
Evo X Engine Management / Tuning Forums
23
Aug 16, 2012 04:20 PM
