Announcement Announcement Module
No announcement yet.
Stock ECU knock control Page Title Module
Move Remove Collapse
Conversation Detail Module
  • Filter
  • Time
  • Show
Clear All
new posts

  • Stock ECU knock control

    I was checking the stock ECU knock control background noise multiplier and adder maps and they look quite different from Evo 9 and Evo X. The overal level of the maps is different because software scales them differently than Evo. Values in both attached pictures are not scaled, just direct values from ROM so ignore the "Load" label in the Colt picture.

    I started looking into this as I get knock sums with E85 fuel starting at around 5000 rpm with very mild ignition advance and 1.2 bar boost. Actually knocksum behaves pretty much the same as with 98 octane gasoline and normally E85 does not knock at all in other cars no matter what ignition advance is used.

    So I found that the knock multiplier table has a big dip in this area. This table is similar, but not exactly the same in JDM, AUS and EUR Colt 4G15T ROMs. I haven't logged the variables yet but I suspect that knockbase is too low in this area because of the big dip in the multiplier.

    Has anyone tried to adjust this map and does anyone have good settings to share that would still allow knock control to detect real knock.
    Last edited by ast; 28-05-2015, 07:01 AM.

  • #2


    • #3
      I had no reason to change, but know for sure that it's "x*100/32" % (in Colt ROM) because
      knock_base = mid16(p_conv32to16(mul16(knock_signal_time_averaged, nop310b4(knock_base_correction_vs_rpm) * 8)) + l_knock_base_noise, 255, 1);
      Last edited by Rcus; 28-05-2015, 02:09 PM.


      • #4
        I recently tuned my car in the dyno knock control turned off (put 255 in each knock multiplier table cell so knock signal it never reaches the knockbase).

        Car was tuned for E85 fuel and I was able to add considerable amount of timing advance each added degree giving more power. I found the point when power was not increased anymore and backed off just a bit from there for peace of mind. This is normal E85 behavior in different cars and the reason why E85 ignition timing tuning is very hard without dyno as knock detection systems are useless indicators for getting timing correct in almost all of the cases.

        Electronic knock headphones were used to monitor the engine and there was no audible knock. These was also nothing in the dyno results indicating knock. More experienced tuner commented that the engine sounded generally very noisy, possibly because of the location of the microphone for the headphones (there are not many free threaded holes in the Colt block that can be used, I used the one next to the dipstick).

        I was happy with the results, car pulls so much better than before. Even engine sound is different during acceleration, much more lively and aggressive and not muted like it was.

        When I put back the stock knock multiplier values to enable knock control after dyno ECU picks big knock sum numbers, regularly between 10 and 20 which pulls ignition timing back and kills the powe. I have now tuned the knock multiplier table to get rid of the knock sums and car feels good again.

        My experiences with Mitsubishi knock control in EVOs is far from great. Colt does not seem to be any better.

        Time will tell if there is real knock or not (blown engine or not ) in the meantime I'm happy with the new level of performance.


        • #5
          If you want to play with the knock control tables at your own risk add this to your XML:

          For AUDM 39670016 ROM:

          <!-- *** Knock related tables *** -->

          <table name="Knock Background Noise Multiplier" category="Knock Detection" address="42c2" type="2D" level="2" scaling="uint8">
          <table name="RPM" address="bfdc" type="Y Axis" elements="27" scaling="RPM"/>

          <table name="Knock Background Noise Adder" category="Knock Detection" address="42e2" type="2D" level="2" scaling="uint8">
          <table name="RPM" address="bfdc" type="Y Axis" elements="27" scaling="RPM"/>


          • #6
            Thanks ast, I'll check them out