Thank you, I was waiting for someone to say this. Don't the rally racers not use a knock sensor at all?

 

The worrying thing I had with a very noisy forged engine is that I couldn't hear minor knock over the general clatter of the thing. Knock sensor was useless.

For the slushy daily drivers amongst us that want a smooth tune with reliability for different gas quality the stock system rocks though.

Different horses, different courses.

 

If they are using anti-lag it is not used. They also use race gas of varying octane in a restrictor motor, so its a whole different category of tuning anyway.

 

Well yes, which is the point here anyway. Street cars , factory knock control makes sense. Once your street car starts moving towards more of a race car with crazy rattly noisey multiplate clutches, rocks for motor mounts, stiff suspension, etc something else needs to be thought of.

 

I think we have to touch the knock sensor filter map for these annoying false knock counts. but there are 1 to 12 maps. Has anyone tried it? where do you begin?

 

A new thread should be started so the important topic can be found easily.

 

That is not a problem in DSM's for some reason.

I've got Prothane motor mounts, prothane sub frame mounts, prothane suspension bushings, open cutout exhaust, and stiff as **** shocks. I can hit my block with a metal hammer and the knock sensor signal stays put.

 

Its not a problem in DSMs because DSMs are as old as 18yrs old. Most of them have had the knock sensors insulative material melt and ooze out, rendering it less than useful. You may want to replace yours.

 

I believe the knock sensor is also filtered for certain frequencies that would be produced by a detonation event...it doesn't simply trigger timing to be pulled from any noise, although certain noises have historically been able to trip the knock sensor, like tifter lick, etc.


Eric

 

As far as I can "Decipher" each of those maps represents a frequency range (I have no idea what ranges they represent though)

I would like to log the engine's knock sensor with a 3d audio spectrum analyzer, it is probably the only way to see what the filters would be looking for..

The best I can tell is the values are discretion values, if it goes above that point, it sees it as knock, and the combination of frequencies represents a signature.. But thats about all I can extrapolate..

 

I wonder what TAR(unit) means.... it seems to be related to the load value?

 

I did replace it and when I did the new sensor acted exactly like my old gooey with respect to sensitivity.

 

I'm reposting some results I have that show evoscan knocksums going to zero when going from 91 to 100 oct.

-----
So I finally got around to doing some data logging of my custom reflash. I've never
been all that happy with either the dyno results (305 whp, 285 wtq on 91 on dynojet) or how it performed on the street.

'03 evo: TBE, hiflow cat, cams, MBC, short ram intake, 21 psi boost.

Here are some 3rd gear "pulls" on 91 oct.

RPM TPS Knock Timing Load AFR

5000 100 12 07 220.794 11.4
5156 100 12 05 219.300 11.2
5343 100 11 06 218.625 11.0
5531 100 11 08 219.934 11.0
5750 100 11 09 214.363 10.9
5875 100 15 10 210.714 10.9
6031 100 15 11 210.587 10.9
6156 100 15 11 209.793 10.8
6343 100 17 12 207.816 11.0

and

RPM TPS Knock TIM LOAD AFR

4937 100 00 09 215.991 11.8
5187 100 04 08 220.045 10.9
5531 100 10 09 216.059 11.1
5781 100 10 11 214.131 11.0
6062 100 09 12 208.617 10.9
6250 100 14 13 208.361 10.9

I didn't particularly care for those knocksums so my next step was to fuel up with 100 oct and
repeat.

Here are some 100 oct results.

RPM TPS Knock TIME LOAD AFR

5187 100 0 10 215.913 11.7
5312 100 0 10 219.911 10.9
5468 100 0 12 213.628 10.8
5625 100 0 14 211.505 11.1
5750 100 0 14 214.363 11.0
5937 100 0 16 209.399 11.0


RPM TPS Knock TIME Load AFR

3750 100 0 11 181.494 missing
4343 100 0 11 231.944 "
5437 100 0 14 223.726 "
6125 100 0 19 216.113 "

I should point out that the knocksums drop to zero using 100 oct.

The car definitely pulls much harder with 100 oct. No surprise considering the much more
aggressive timing.

All data taken at 3rd gear and 20-21psi boost. AFR measured using zeitronix wideband. Knock and timing measured using evoscan software and tactrix cable. Load calculated using load = AirFlow*852/RPM. Reflash has both the high and low octane maps equal to each other for both fuel and ignition timing. Ignition timing maps is precisely what the above data shows for 100 octane at the given rpm and load.

Looks like its retune time for 91 oct. Those knocksums just floored me.

Sadly, neither the AFRs or the timing seem all that agressive to me. In order to reduce knock I'll make even less power, and I'm sad about my torque as it is. Any advice is appreciated.

 

Please log the octane value, this is exactly the situation we need to see it! See the thread in this forum where a data.xml file was posted which will do it - place this in the datasettings Evoscan directory.

I would suggest you take your low octane ignition map and subtract 1 degree from the entire 100 load column, 2 degrees from the 120 load column, 3 degrees from the 140 load column, 4 degrees from the 160 load column, 5 degrees from 180 column and 6 degrees from 200-260 columns.

Then repeat the datalogs and please post the results On 91 with the octane number logged.

 

Thanks for the feedback jcs. I haven't really been following the octane number logging debate. I'll take a look at the thread you mention and see how I can start logging the octane address.

Could you elaborate on exactly what you're trying to see? If the octane address changes depending on the fuel I use? Something slightly more sophisticated?

Are the timing value changes you suggest important for the experiment you want or are you just advising me on how to change my tune to eliminate knock with 91 octane?

In any case it'll take me a few days to repeat these results at 91 and 100 oct.