tip in knock with turbo

[Td-d]

Originally Posted by jamesm

it's likely that map definition is wrong or isn't scaled properly or something, as a value of 1 or 2 degrees will definitely disable coarse correction.

Definitely the correct address, and no scaling required - it's a float value. It's simply set at zero for the BRZs...

The logic behind getting into rough correction mode is quite convoluted - not just Rough_Correction_Minimum_KC_Advance_Map_Value involved.

This is a good reference - http://www.romraider.com/forum/topic1840.html

[jamesm]

Originally Posted by Td-d

Definitely the correct address, and no scaling required - it's a float value. It's simply set at zero for the BRZs...

The logic behind getting into rough correction mode is quite convoluted - not just Rough_Correction_Minimum_KC_Advance_Map_Value involved.

This is a good reference - http://www.romraider.com/forum/topic1840.html

that pretty much mirrors the other docs i've read, actually suggesting that the last applied FLKC value be >= 3.9 degrees... and that the advance value in the map be greater than 4.9 (which i hadn't seen before, i'd only read about the FLKC value). it doesn't seem to mention anything about a value in a table, unless they just skipped over that bit:

To exit from fine correction mode to rough correction mode, ALL of the following requirements must be met:

• Engine speed and load must be within the ranges specified by the 'Rough Correction Range' tables.
• Timing advance (maximum) map value is greater than 4.9 degrees.
• Some FLKC value change (positive or negative) occured last execution.
• The last FLKC applied value (|x|) is greater than 3.9 degrees (that is, the absolute correction -> ex. -4 = 4)
• The last FLKC raw difference (|y| * 2.84) is greater than last timing advance (maximum) map value.
• (IAM > 1) or (IAM <= 1 and last applied FLKC was positive).

Of course this all applies to a car that isn't ours... it could totally be wrong. I have found through experimentation that using less than 3.9 in the advance table keeps it from dropping, that's all i know for sure. then again, i often set it to zero in those areas just coincidentally, so maybe that's why it doesn't seem as though the table matters. i could swear i remember running 0<value<3.9 in the advance and not seeing it drop, though i cant be sure about that without correlating old roms to logs or testing it again.

[Td-d]

Yeah, it is odd that that's the case. Maybe setting that value to 0 changes the way the logic handles the advance tables - I've never really looked at it closely. The logic is pretty much the same as other Subarus, incidentally.

[s2d4]

If (in fine correction) {
           If (pulled more than 4 degrees or the ECU is unhappy) {
                  Goto (IAM correction)
            }
            If (Large/rapid change in engine load) {
                  Goto (FBKC)
             }
} else {
           Check to see if any adjustments need to be made
           Stay in FLKC correction mode
           }
If (in FBKC mode) {
      While (engine load is rapidly changing) {
            Check for knock and pull timing if needed
            When (Done) {
                     Go back to previous mode
             }
        } 
}

According to the above, it would be part of fbkc as tip in would be where the load changed rapidly. This falls in line with my earlier log as knock correction as according to shiv is fbkc.

Looks like I'll need access to flkc @shiv@vishnu to confirm for myself or @Td-d /anyone with previous subaru platforms to clarify if the number in the map that dictates where coarse correction starts is at 4, and since ours is set at 0 so behaves differently.

[steve99]

Originally Posted by s2d4

Looks like I'll need access to flkc @shiv@vishnu to confirm for myself or @Td-d /anyone with previous subaru platforms to clarify if the number in the map that dictates where coarse correction starts is at 4, and since ours is set at 0 so behaves differently.

I did log FLCK as well as FBCK (via tactrix logging) very few occurrences of FLCK highest -0.67 compared to many occurrances of FBCK with values up to -2.3.

As for where course correction starts well got no idea

[jamesm]

Most of the tip in knock I see shows up in fbkc. It'll write into fine learning sometimes though, particularly when tipping in to a long pull.