Preemptive feedback knock correction

Leafy · **Posted:** Sun Nov 04, 2012 3:46 am

throttlehappy wrote:

I have narrowed it down to a few 'triggers'

* Engine g/rev or g/sec increasing rapidly in under 250ms(believe I found the timer value near the PEKC retard value)
* Total Ignition timing values changing by more than 3 degrees with some sort of g/rev and rpm reference(I suspect a 3D table controls this?)
* Boost AND vacuum error

Look for delta throttle position as well.

GM uses a knock scheme very close to what you guys are seeing and its during transients. The computer is predicting that knock is likely to happen in certain transient situations and automatically pulls timing even when there is no knock present. I believe they use delta tps, but I havent looked into it on the newer torque based ecus. I like how subaru adds in the timing changing rapidly element, since thats another thing that can cause knock, I dont like the maf engine load usage though, it seems like it would be too slow for this correction to work.

redwagon · **Posted:** Sun Nov 04, 2012 4:32 am

Leafy wrote:

Look for delta throttle position as well.

I've seen timing pulled under sudden before WOT also, with no obvious explanation for it, not even overboost or heading toward overboost.

dschultz · **Posted:** Sun Nov 04, 2012 5:07 am

Easty wrote:

Quote:

Where (address) in your ROM are to looking now at the switch?

FFFF76CE Pre_Emptive_FBKC_Switch

Have a look in sub_33954 at the end.
0x00033CCA
0x00033CEA

Easty · **Posted:** Sun Nov 04, 2012 5:18 am

Quote:

Have a look in sub_33954

Bingo! You found it!! thanks!!!!

how on earth did you find it?

dschultz · **Posted:** Sun Nov 04, 2012 5:28 am

I searched on addresses near it and then looked for data move instructions with an offset of 1 or 2 or 3, etc.
So the search on FFFF76CC was the key.

Easty · **Posted:** Sun Nov 04, 2012 5:40 am

Nice Find!

Easty · **Posted:** Sun Nov 04, 2012 6:48 am

After i quick peak at my rom i found some activation thresholds

Attachment:

Untitled2.png

My naming scheme may not be appropriate given the way they are used, for example;

If Unknown Value(Could be delta load or MAFV) Value is greater than 0.25
or
If RPM is greater than 1700
or
If RPM is less than 4400
or
if load is greater than 0.799
or
if load is less than 2.8
or
if primary ol enrichment is less than 1
or
if coolant temp is greater than 78 degrees c
then
PEKC is active

There is also a 2D table with an IAT axis at the beginning of the routine

Attachment:

Untitled21.png

its result is compared with an unknown value when i get a chance i'll do some logging to figure out what it is.

nsfw · **Joined:** Thu Nov 23, 2006 2:23 am **Posts:** 2565

EDIT: I thought I had something insightful to add here, but I was wrong.

td-d · **Posted:** Sun Nov 04, 2012 7:48 am

Nice work!

dschultz · **Posted:** Sun Nov 04, 2012 1:19 pm

Easty wrote:

if primary ol enrichment is less than 1(7.35AFR)

Remember that AFR is specified in lambda in the ROM.

throttlehappy · **Posted:** Sun Nov 04, 2012 2:14 pm

Leafy wrote:

throttlehappy wrote:

I have narrowed it down to a few 'triggers'

* Engine g/rev or g/sec increasing rapidly in under 250ms(believe I found the timer value near the PEKC retard value)
* Total Ignition timing values changing by more than 3 degrees with some sort of g/rev and rpm reference(I suspect a 3D table controls this?)
* Boost AND vacuum error

Look for delta throttle position as well.

GM uses a knock scheme very close to what you guys are seeing and its during transients. The computer is predicting that knock is likely to happen in certain transient situations and automatically pulls timing even when there is no knock present. I believe they use delta tps, but I havent looked into it on the newer torque based ecus. I like how subaru adds in the timing changing rapidly element, since thats another thing that can cause knock, I dont like the maf engine load usage though, it seems like it would be too slow for this correction to work.

Yeah, that was my initial find of PFKC. My DBW is not stock so it throttle tables showed it much more obviously to have -1.4 fbkc but my FBKC value was -2.00! lol

If you have ever seen a GR STi timing ntable, it explains why Subaru needed the timing change function... timing table looks like it belongs in Nepal with snow on it

td-d · **Posted:** Sun Nov 04, 2012 3:33 pm

I believe there's a throttle plate opening % delta threshold in there - would be interesting to log to note (it's checked against a 2d table)

0xFFFF6524

This is the 2D table - IAT is the x-axis:

Code:

ROM:00086CD8 word_86CD8:     .data.w h'10            ; DATA XREF: sub_33972:loc_339DEo
ROM:00086CD8                                         ; sub_33A92:off_33B94o
ROM:00086CDA                 .data.w h'800
ROM:00086CDC                 .data.l flt_C7EF4
ROM:00086CE0                 .data.l word_CA14A      ; = 1.0e1
ROM:00086CE4                 .float 0.0019073777
ROM:00086CE8                 .float 0.0

td-d · **Posted:** Sun Nov 04, 2012 5:56 pm

And whilst we're checking out FBKC:

Code:

FBKC_Fine_Rough_learning_mode_enable_load_delta_max   00000000000CE088 
FBKC_Fine_Rough_learning_mode_enable_load_delta_min   00000000000CE084 
FBKC_Fine_Rough_learning_mode_enable_min_coolant_temp 00000000000CE07C 
FBKC_min_AC_start_delay_                 00000000000CDF84 

RAM:FFFF7C68 Ram_filtered_load_delta
RAM:FFFF7CB0 Ram_AC_start_timer

Edit: renamed titles for clarity - these are part of the trigger points that allow the ecu to switch from FBKC to FLKC or rough correction mode.

Easty · **Posted:** Sun Nov 04, 2012 7:53 pm

dschultz wrote:

Easty wrote:

if primary ol enrichment is less than 1(7.35AFR)

Remember that AFR is specified in lambda in the ROM.

So what does an OL enrichment of 1 equal?

td-d · **Posted:** Sun Nov 04, 2012 7:56 pm

It means stoich - 14.7afr

RomRaider

Preemptive feedback knock correction

Who is online