RomRaider

This is awsome, very useful

Ditto!

Bill, I'll be very excited when you have this ready to go for the 32bit NAs. The amount of time this would save me is insane.

Thanks for the work, like always!

Ty

Something that is interesting is that I'm not getting any knock events near redline. That is, it has been assumed that the reason why the FBKC and FLKC rpm ranges do not extend all the way to redline in the factory is that there's too much engine noise for the knock sensor to at least somewhat accurately distinguish knock for other noise. But I'm not getting any (remember, my hack isn't dependent on any of the ranges). All the false knock appears to be at low loads/low rpm (especially near idle).

that's pretty cool.
I'll get in the 32 bit line.

I've run feedback correction all the way up to 8400 rpm on my old 2 liter. And I ran it all the way to redline on my 2.5 as well. Its never inaccurately pulled timing on me. Basically in stock form, they turn off the turbo after 6k, so its pretty unlikely to knock anyway. I know its been stated and agreed that the knock sensor cant accurately distinguish knock because of engine noise... but I wonder if its the opposite of what we're thinking. Maybe engine noise covers up the knock? Or high rpm harmonics alter the noise??

Either way, it doesnt just pick up engine noise as knock up top, I am sure of that. I've tried it in a lot of different cars and models.

Hmm, interesting. What we also don't know is how the ECU filters the knock sensor input.

I think that is the case here. A good overview of a common method for knock detection is given in the introductory material at http://focus.ti.com/lit/an/spra039/spra039.pdf.

Several FHI patents involving knock detection provide some algorithms:
US 5,321,973
US 5,088,044
US 5,083,278

It's possible that knock detection is handled by a separate IC, maybe like the one at http://focus.ti.com/docs/prod/folders/print/tpic8101.html.

then perhaps the "sweet" spot (i.e. where the knock sensor is most accurate) is generally the rough correction rpm range - that is, the range in which timing would be globally impacted would ideally be where knock detection would be most accurate (with my 02 WRX 2100-4400 RPM). The factory tune is very conservative at higher RPM to, as indicated by the patents that Jon posted, to prevent the occurence of knocking where knock detection wasn't as good.

i'd love to see this in carberry!

(hint hint )