Knock Sum

[Hondata]

I trying to understand this knock sum thing.

 

If i see 33 count in my log, means the car is knocking badly at that point right.

 

In order to have a perfect running car the knock sum should be 0 in the log right?

romraiderlog_20120707_004344.csv

[KCwagon]

Knock sum includes all percieved knock events, including false knock. You can get a false knock event from simply engaging the clutch at a stoplight. It is not uncommon for a healthy car to have a big "knock sum" numbers.

 

I would look at WHEN the knock is occuring. I.E.-partial throttle, under boost etc.

 

As an example - "Fine knock learned" will tell you where the ecu has learned to adjust timing as a result of previous knock occurances.

[Hondata]

If you see my log, the knock sum is from the start to the end even at boost.

 

No matter how i retard the timing it still knocks in the data log.

 

nothing to do with cam angles when knock occurs right?

[TheOneDoubleN]

I've always kind of wondered about this....

 

I always just long feedback knock and fine knock learning. I never understood the reason behind logging knock sum if it's going to count false knock events or extremely minor cases. Shouldn't we only be interested in knock events where timing is actually pulled? Or is there some rationale behind looking at Knock Sum?

 

TIA!

[ssbtech]

When the car is using feedback knock correction, how does it determine the difference between real knock and false knock?

[iNVAR]

If you see my log, the knock sum is from the start to the end even at boost.

 

No matter how i retard the timing it still knocks in the data log.

 

nothing to do with cam angles when knock occurs right?

 

Stop messing with your tune. If you don't understand what "knock sum" is, you're ill equipped to be tweaking stuff at this point.

 

As mentioned before, knock sum is the count of knocks your car has detected (both false AND real knocks) since you started the car. When you shut the car off, it goes back to 0. It maxes out at ... 35 I think, and then it goes back to 0 again.

 

It is not uncommon to see that number go up very high even on a perfectly healthy car due to false knock readings. If the ECU isn't pulling timing with those knocks, it is likely a false knock reading, or basically harmless.

 

I started my car up this morning and within 1/4 mile, my knock sum was at around 7 due to traffic lights and stuff.

[iNVAR]

When the car is using feedback knock correction, how does it determine the difference between real knock and false knock?

 

It doesn't. If it had the ability to determine what was real and false, it wouldn't read it as a knock at all then would it?

 

The general consensus seems to be that if your KS # goes up, but no FBKC or FLKC registers with it, it should be harmless.

[TheOneDoubleN]

It doesn't. If it had the ability to determine what was real and false, it wouldn't read it as a knock at all then would it?

 

The general consensus seems to be that if your KS # goes up, but no FBKC or FLKC registers with it, it should be harmless.

 

So would it be correct then to say that the knock sum shows how often the knock sensor picks up a signal, regardless of severity, and feedback knock would represent only knock severe enough to warrant pulling timing?

[bmx045]

Hondata, you're on your base map from Shamar, leave it alone and let him finish the tune, you clearly don't know what you're doing. It may take a year for him to finish the tune but stop f*cking with it.

[iNVAR]

So would it be correct then to say that the knock sum shows how often the knock sensor picks up a signal, regardless of severity, and feedback knock would represent only knock severe enough to warrant pulling timing?

i believe that's essentially right but if someone wants to chime in here...

[Hondata]

Hondata, you're on your base map from Shamar, leave it alone and let him finish the tune, you clearly don't know what you're doing. It may take a year for him to finish the tune but stop f*cking with it.

 

I am not messing with the tune, just worried about the knocks.

 

I do not want the engine to blow even the tune can be completed.

 

Turbo cars with knocks at high end is no joke.

[CasopoliS]

I am not messing with the tune, just worried about the knocks.

 

I do not want the engine to blow even the tune can be completed.

 

Turbo cars with knocks at high end is no joke.

 

see my response in your other thread.

[fishbone]

I am not messing with the tune, just worried about the knocks.

 

I do not want the engine to blow even the tune can be completed.

 

Turbo cars with knocks at high end is no joke.

I agree. You should probably take it easy until you get a new tune revision.

Just to clarify, knock sum incrementing does not mean real knock.

In other words, knock sum can increment and the ECU can very well not take action (no timing pulled).

[spec.B Dream]

One of the big reasons knocksum was important back in the day that that people believed that a knock event could occur and the ECU would adjust the timing and then go back to normal so quick that it wouldn't be caught in the log. This drove me crazy because I tried to tune out every instance of knocksum increment, which was impossible.

 

Now, with RomRaider fastpolling, there is no need for knocksum. Every feedback or learned knock correction I see now exists across multiple (even tens of) fast-polled lines in the log.

[G5892]

The knock sum number is going to increase as you drive the car. Take note of the knock sum number. Then do a log from 2k to redline then see if the knock sum number has increased. If it hasn't, then your fine. That's the only useful way to read it.