You can always compare voltage vs knock sum as well.

Voltage is amplitude so while its not 100% the real answer it will at least help you make some educated guesses about whats real and whats not in a built motor.

Aaron

 

Aaron...I've been doing that for a long time and people ALWAYS tell me I am wrong. The key to is to see the large spikes over the threshold as real knock and only little noise that breaks the threshold as false knock.

 

Where are these voltage readings? I haven't logged them yet, can it be setup with evoscan?

 

The voltage reading will tell you nothing as far as real or phantom knock. The ECU sees knock, whether phantom or real, as anything that breaches the knock threshold. There is no way to definitively tell whether knock is real of phantom simply by the voltage signal.

What you can get from the voltage signal is a baseline of how noisy your motor is, in general. For example, a built motor may be noisier, on average, than a stock motor. And you can create your own knock threshold based on this average noise level.

A very, very good thread that explains all of this well is the thread that jcsbanks posted on how the ECU calculates knock and the knock table definitions. He gives all of the paramaters to log, etc.

Yes...look for the thread by jcsbanks. It is a very good read. But, read what I wrote above regarding the voltage.

Let me see if I can find John's thread and link to it here.

Here is John's thread. Read the whole thing...it's a great thread that will thoroughly explain how the ECU determines knock and how to tune the tables, if necessary.

Here is a graph from John showing the knock signal against the threshold, against the resulting knocksum from page 5 of that thread:


Another, zoomed in a bit:


One final one...showing everything that can be logged...John gives an explanation of all of this in his thread:


John's explanation for above graph:
Further explanation:

 

One side note (that you will find out if you read the thread), is that many of these parameters can only be logged accurately with DMA. Something like EvoScan won't be able to accurately log some of this. But, nonetheless, it can still be used to get an idea of what's going on.

Once you read that thread and understand what's going on, you will have a perfect understand of knock control in the stock ECU and how to tune it, if necessary.

 

DMA? I hate to sound like a noob... but what is that? Where can I purchase that software?

 

DMA/LiveMap is a patch that jcsbanks made for some of the ECUs.

 

+1

 

l2r99gst...I use the method you described. I guess I don't explain it in the great detail you do. Yes, you first have to determine how "noisy" your motor is. Then you can make the adjustments as necessary.

 

Something to consider is that Evo 9s tend to have phantom knock at ~2500 rpm at low loads. On my own car, I typically see 2-5 knock counts when crossing ~2500 rpm at around 80 load. Its happening at higher load in your car, but its something to consider.

 

Well, have ironed out all the really bad spots through timing and fuel changes. Still two areas with three counts and I have come to the assumption that these are likely phantom. I have pulled as much as 5* more in those areas and they won't change. So here are my multiplier tables:


So what do you think, should they be edited for that RPM band?

 

Didn't re-read the whole thread. Where are you getting the knock ???Rpm and load

 

Anywhere from 2000-3000 rpm and 100-130 load

 

Ok since you are 100% convinced that the knock is phantom. Raise the "Knock Sensitivity Load Threshold" in the problem area to like 80. Try again. If you are still getting the phantom knock then you should raise the Noise multiplier by a point or 2 in the problem area.

 

Well, I thought I was all good and certian the rest was phantom and only 2-3 counts then on my way home this bad boy popped up.... What. The. F.



Any ideas? Or should I just go with upping the multiplier tables. I have tried everything at this point...