Evo Knock Sensor Filter Maps and their purpose
 

I am using the 9653 Tephra ROM on my 03 lancer, and I'm concerned that the knock sensor is not functioning entirely as intended.

There are various posts about "Knock Sensor Filter Maps 1-12", but none of the XML files I see have them. Some posts point to the "inaccuracy" of the titles of those tables, and say that they are "not to be used". My base lancer rom does have "Knock Sensor Filter Maps 1-5" and I'm, along with a lot of other base lancer guys, trying to determine how to get our exact same settings into the Evo ECU.

So far we have the Multiplier is 18 for low/middle/high rpm and that our single gain is 3, and triple gain is 5. The decay timer is the same between the roms I think, but the fact remains that we see a ton of knock in the lower rpm's like 2250 or so, at very low load. So, the next thing would be compare those filter maps, but it looks like they've all been removed from any Evo Rom's I can find (perhaps they were initially mislabeled and that's why...that's the only thing I've heard so far). Most of us also use the knock load from our roms instead of the evos, which is 40 load up to 2500, and 50 load beyond....because the evo's is basically unreachable on our engines with 8psi boost or so, anything under 160 load is not reported with the stock knock settings.

If you've got any info on those maps, or where they've all gone to, let me know. I saw a post of someone who graphed all 12 in a spreadsheet, and they are indeed fairly different than our lancer base ones.

The search tool is a beautiful thing... there are a couple threads that go into detail exactly what you are asking...

Here is one that took me 2 secs to find via the search tool:

https://www.evolutionm.net/forums/ec...ble-found.html

Just FYI, I'm talking about the Knock Sensor Filter maps, not the Knock load. The tables I'm looking for are 2D tables, not the 1D table that is just a "floor" where anything below is not reported. Please read the whole question portion above.

Yes, read the actual thread I linked... all the info in there

The tables you are looking for are like this (these are evo 10 but evo 8/9 very similiar)

http://norcalmotorsports.org/users/b...les_072709.jpg

Since I'm feeling generous,

Here are the tables for your rom, 9653:

<table name="KnockSum decay timer" category="Knock Control" address="1784" type="1D" scaling="uint16"/>

<table name="Knock Load (knock ignored below these values)" category="Knock Control" address="421a" type="2D" level="2" scaling="Load8">
<table name="X" address="6576" type="Y Axis" elements="10" scaling="RPM"/>
</table>

<table name="Knock Multiplier - Low RPM" category="Knock Control" address="177a" type="1D" level="2" scaling="uint16"/>

<table name="Knock Multiplier - Mid RPM" category="Knock Control" address="24aa" type="1D" level="2" scaling="uint16"/>

<table name="Knock Multiplier - High RPM" category="Knock Control" address="21de" type="1D" level="2" scaling="uint16"/>

<table name="Knock Adder - Single Gain" category="Knock Control" address="1780" type="1D" level="2" scaling="uint16"/>

<table name="Knock Adder - Triple Gain" category="Knock Control" address="177c" type="1D" level="2" scaling="uint16"/>

Thanks for responding. Let me add a picture to help explain what I am asking. I have the settings up top from my base lancer rom, and on the bottom, the stuff I've had in my Tephra ROM....but I still see knock in places I wouldn't with my base lancer rom. That table on the far right is the one that I am trying to find in the Evo roms, if it even exists. In a few threads I've heard "don't use that, it's not actually a knock table" or "just set all that to 0" or many other random things...but never the info to define that table, or the 11 others that ought to exist in the evo rom if my info I have thus far collected is correct.

*edit* I have 5 of these 2D tables like the one in the upper right of this pic, the evo has 12 apparently, although I don't know how I would copy that data over....no idea how they are referenced.

The "knock sensor filter" tables are incorrect and aren't knock tables at all. Don't use them. ;)

The defs I provided above are the correct knock tables

Ok, that's what I was wondering. They are still all over our base lancer xml files, so if they aren't correct I bet they should be removed. So, with the same settings as I find in my base lancer rom, inserted into the tables you mention, I see different knock behavior.

Let me elaborate on "different". With basically the same timing/AFR's at WOT, I see relatively similar knock results (typically being none). But, down low, at low load, low rpm spots....I see significant knock. If I bump it up to a higher gear, that knock won't show, and typically I can be cruising along at low load/rpm and the knock sensor will suddenly start going off at a slow blink (3 counts) and then if I hold my speed fast blink (6 counts) ....and if I let off for a second, and then go back to the same load/rpm the knock light is calm and quiet.

The reason I suspected those knock filter maps is because they seemed to get less aggressive the lower the load....but if they aren't knock related at all, what else could cause such a drastic knock sensor difference? BTW, thank you for helping me out on this, it's good to know those tables called "Knock Sensor Filter 1-12" or 1-5 in my case probably don't do anything knock related, that explains why they were removed from evo roms.

Our initial thought was get an evo knock sensor, since the threads are the same...but I don't know if that's a good idea. The engine is a different engine, so it makes sense if you swap in a 4g64, but for my 4g94...I don't think it's the right solution.

I've heard things like, "knock sensors are calibrated to filter certain frequencies that are "known engine noise". Is it possible there is something like that, that isn't typically defined and will probably be my "dead end" on this?

I want to know what their real purpose is for though.

They are for detecting errors in engine rotational speed, thus they are mis-fire detection 3D maps.

and, as has been pointed out

dont play with them.

The tables Bryan posted are all you need, actually, more than you need.
The three multipliers are the go-to tables.

I guess we can turn knock detection totally off with the multiplier maps, correct? I'm sure my knock detection is totally useless and out of tune anyway.

Hey, I'm happy to know what they are...one less thing I need to try and troubleshoot.

I'm thinking, as everyone said, those multipliers are "the thing". So, any thoughts on how or when the different multipliers kick in? I've heard about 3 different guestimates, all very similar, but not exactly the same. Also...is there an exact formula for how they work? something like...

knock voltage * knock multiplier * gain = X and if X > "something" knock gets registered? I mean I realize it's probably more complex, but any idea?

I'm also going to make a "det can" set, and use that to hopefully audibly hear the knock, that way I can tell if the sensor is too sensitive or not sensitive enough.

jcsbanks has a thread that details exactly how the knock routine runs in the ECU. A simple search should find it. Actually, Bryan linked you to it in post #2 in this thread. All of your answers are in that thread if you just read it.

There is a table that has the knock sum ceiling, which is 36. Put that to 0 and you basically turn it off.

I'd have a new motor ready to go shortly after....

Yup, I read that one about a month back, and it was too much information that I didn't understand. After spending a couple weeks trying to tune the car, it makes much more sense, thanks to the jscbanks fellow and people like you as well, I think the knock stuff makes enough sense I can hurt myself with it.

So, the solution for our 4g94 cars is going to be determining how the knockadc and knockbase all correlate to each other with our settings vs evo settings. I'm thinking using our single and triple of 3 and 5 ought to be the best thing, but the 18 multiplier accross the board probably needs some adjustment. Det cans as well, no other way to figure out if it's real or fake as I make multiplier adjustments.

Thanks guys, sounds like we've got a chance.