AF#3 (Rear O2 sensor) correction limits

td-d · **Posted:** Wed Feb 22, 2012 9:02 am

Ok, confirmed

I have indeed found the AF#3 Rear O2 correction limits. Stock is set to -50% / 50% (!), so it's always on basically. I set it to 1% to test, and stuck nicely to those limits. I believe that if you set these limits to zero, you basically remove the effect of the rear O2 sensor. I am in the process of uncovering the AF#3 learning limits, rear O2 sensor limits and the so-called 'adders' for AF#3, and will post as soon as I have digged them up.

For the '08 EDM Sti (AZ1G201 /2G), the Ecuflash definition is as follows:

Code:

<table name="AF 3 Correction Limits" address="2fae0" type="2D" level="1" scaling="A/FLearning#1Limits(%)">
      <table name="X" type="Static X Axis" elements="2">
         <data>High</data>
         <data>Low</data>
      </table>
   </table>

EDIT - How to find
Here's a general guideline on how to find the values in your rom.

In IDA, do a search for 3F 00 00 00 BF 00 00 00 (all results). You will get 4 or 5 results depending on your Rom. These are the -0.5 and 0.5 sequences. Usually it is the third one that's the AF3 correction limits. In some roms (like A8DK100P) it can be the first one. EDIT - looks like if you search for 3F000000 BF000000 4F22C79A F808D59A you get the exact location (as it is always followed by a routine that leads to the primary open looping fueling routine).

You will know it the right address by looking for the following - there are two subroutines above the sequence (that check the outer ranges) and usually 4 offsets for maps that are 2d, 9 cells and with a multiplier of 0.000015259022 and an adder of -0.5. E.g.

Code:

ROM:00085998 word_85998:     .data.w 9               ; DATA XREF: sub_2F4F0+708o
ROM:00085998                                         ; ROM:off_2FD78o
ROM:0008599A                 .data.w h'800
ROM:0008599C                 .data.l h'C9680
ROM:000859A0                 .data.l dword_C96A4
ROM:000859A4                 .float 0.000015259022
ROM:000859A8                 .float -0.5

You will also see a float above the 0.5 and -0.5 of 0.000019073486.

List of addresses for various roms:

AZ1J500G :31398
AZ1E400U: 2FD90
A8DK100P: 30E10
AE5F301C: 31F7C
AZ1G900P: 2FE4C
AZ1G101N: 30040
AZ1G201: 2FAE0
A8DH200O:2D464
A8DH200V: 2D464
AE5IA10V:3447c
AZ1G400W: 30040

MFB · **Posted:** Wed Feb 22, 2012 9:20 am

Thanks td-d! Ive updated the sti map Im working on tonight. What parameter do I log to see if the correction holds? I turn it both 0.

td-d · **Posted:** Wed Feb 22, 2012 9:33 am

This is the one I use on my logcfg.txt

paramname = A/F_Correction_3
paramid = 0x0000D0
databits = 8
scalingrpn = x,.078125,*,5,-

td-d · **Posted:** Wed Feb 22, 2012 11:29 am

Alrighty, it seems it's relatively easy to track the AF#3 correction limits down, as there are a very limited number of float sequences of 0.5 and -0.5 in most roms.

Mark is testing some definitions for the 09 Forester, once I'm comfortable with 1 or 2 more 'it's working' responses, I'll start posting up some addresses, and how to find it in your rom.

Strangely, my logged AF#3learning parameter (before I zeroed it, of course!) never read anything other than zero. Not sure why - I'm using P66:

paramname = A/F_Learning_3
paramid = 0x0000D1
databits = 8
scalingrpn = x,128,-,100,*,128,/

I might need some assistance with those who definitely log it to narrow down the learning range. In any case, the correction limits at zero seem to completely switch the oscillation caused by the rear O2, so it achieves the same aim.

td-d · **Posted:** Wed Feb 22, 2012 4:03 pm

I can also now confirm that setting the limits to zero basically switches off AF#3 corrections:

Code:

sample   time      rpm       maf    afr       engine_load   throttle   fi1pw   atm_psi   map       A/F_Learning_3   A/F_Correction_3   cl_ol   ol_fueling_target   Final_Fueling_Base   Lc1_Afr
3067     1363.48   3110.25   2.16   14.8916   0.67894       17.2549    2.56    12.1882   7.05182   0                0                  8       14.7                14.4898              14.8617
3068     1363.52   3117.5    2.18   14.7216   0.67309       17.2549    2.56    12.1882   7.04069   0                0                  8       14.7                14.489               14.8617
3069     1363.56   3117.25   2.18   14.5802   0.667941      17.2549    2.56    12.1882   7.01842   0                0                  8       14.7                14.4891              14.6559
3070     1363.6    3096      2.14   14.474    0.65603       16.8627    2.56    12.1882   6.91821   0                0                  8       14.7                14.4914              14.6559
3071     1363.64   3107      2.16   14.4269   0.642959      16.8627    2.56    12.1882   6.89594   0                0                  8       14.7                14.4902              14.3472
3072     1363.68   3122      2.14   14.4117   0.640869      16.8627    2.56    12.1882   6.90707   0                0                  8       14.7                14.4885              14.3472
3073     1363.72   3136.75   2.14   14.4254   0.647546      16.8627    2.56    12.1882   6.87367   0                0                  8       14.7                14.4869              14.3325
3074     1363.76   3144      2.16   14.4875   0.642432      16.8627    2.56    12.1882   6.87367   0                0                  8       14.7                14.4861              14.3325
3075     1363.8    3146.75   2.18   14.5178   0.633985      16.8627    2.56    12.1882   6.87367   0                0                  8       14.7                14.4858              14.4795
3076     1363.84   3184.25   2.18   14.518    0.643247      16.8627    2.56    12.1882   6.90707   0                0                  8       14.7                14.4817              14.4795
3077     1363.88   3192      2.18   14.533    0.647215      16.8627    2.56    12.1882   6.91821   0                0                  8       14.7                14.4808              14.4795
3078     1363.92   3196.5    2.12   14.5785   0.636607      16.8627    2.56    12.1882   6.89594   0                0                  8       14.7                14.4803              14.5089
3079     1363.96   3202.25   2.1    14.5332   0.612287      16.8627    2.56    12.1882   6.72892   0                0                  8       14.7                14.48                14.5089
3080     1364      3181.75   2.12   14.4573   0.592046      16.8627    2.56    12.1882   6.60644   0                0                  8       14.7                14.482               14.553 
3081     1364.04   3186.75   2.12   14.3829   0.587391      16.8627    2.56    12.1882   6.5619    0                0                  8       14.7                14.4815              14.553 
3082     1364.08   3189.5    2.14   14.351    0.582272      16.0784    2.304   12.1882   6.5285    0                0                  8       14.7                14.4812              14.3766
3083     1364.12   3193.5    2.12   14.3814   0.582119      16.0784    2.56    12.1882   6.53963   0                0                  8       14.7                14.4808              14.3766
3084     1364.16   3225      2.12   14.4117   0.578952      16.0784    2.304   12.1882   6.53963   0                0                  8       14.7                14.48                14.3325
3085     1364.2    3230.25   2.12   14.5028   0.576225      16.0784    2.304   12.1882   6.5619    0                0                  8       14.7                14.48                14.3325
3086     1364.24   3242.75   2.16   14.6242   0.584171      16.0784    2.56    12.1882   6.53963   0                0                  8       14.7                14.48                14.4648
3087     1364.28   3229.5    2.14   14.7424   0.592871      16.0784    2.56    12.1882   6.50623   0                0                  8       14.7                14.48                14.4648
3088     1364.32   3238.25   2.08   14.8064   0.58477       16.0784    2.56    12.1882   6.46169   0                0                  8       14.7                14.48                14.7882
3089     1364.36   3245.75   2.12   14.8276   0.572841      16.0784    2.304   12.1882   6.37262   0                0                  8       14.7                14.48                14.7882
3090     1364.4    3277.75   2.06   14.6864   0.559203      16.0784    2.304   12.1882   6.26127   0                0                  8       14.7                14.48                14.9058
3091     1364.44   3277      2.02   14.5787   0.536847      16.0784    2.304   12.1882   5.9495    0                0                  8       14.7                14.48                14.9058
3092     1364.48   3273.5    2      14.4118   0.516352      16.0784    2.304   12.1882   5.73795   0                0                  8       14.7                14.48                14.6559

MFB · **Posted:** Wed Feb 22, 2012 9:16 pm

What af 3 corrections were logged before zeroing out? Interesting to see what the change in afr was. Better to use afr sensor 1 instead since its close loop.

td-d · **Posted:** Wed Feb 22, 2012 11:42 pm

I'm on the road now, otherwise would post a log - pre-zeroing, I was getting swings of up to 3% either way.

AFR sensor 1 is in there (AFR in the log).

dschultz · **Posted:** Thu Feb 23, 2012 12:16 am

But the log is too short to see a trend... :wink:

MFB · **Posted:** Thu Feb 23, 2012 4:58 am

So I got to log today, take a look. Just remove the other parameters you dont need to see

td-d · **Posted:** Thu Feb 23, 2012 2:44 pm

Hmm.. I'm not sure which way to analyse the data to be honest - if you plot the AF#1 corrections versus AF#3, it looks like the rear O2 basically compensates for any under or overshooting by the front O2. Plotted against AFR there's certainly a direct relationship as well.

What would be interesting is to disconnect the rear O2 sensor, see what happens to AF#3 and AFR with the stock values, and then change the values to 0 and see what happens - to see whether it shuts down the problem with rich idle with the rear O2 delete.

eminehart · **Posted:** Thu Feb 23, 2012 2:49 pm

Problem is on the newer cars disconnecting the rear o2 goes into a sorta limp mode where the the CL targets are about 13.3-13:1.

td-d · **Posted:** Thu Feb 23, 2012 3:03 pm

That's exactly what I'm interested in - whether switching off these limits has any impact on that effect or not.

I've still got to track down the other limts (rear O2 sensor AFR limits, AF 3 learning and the adders) - I'm sure the idle richness emanates somewhere in these routines.

MFB · **Posted:** Thu Feb 23, 2012 8:16 pm

td-d wrote:

That's exactly what I'm interested in - whether switching off these limits has any impact on that effect or not.

I've still got to track down the other limts (rear O2 sensor AFR limits, AF 3 learning and the adders) - I'm sure the idle richness emanates somewhere in these routines.

Idle richness also comes from the close loop load comp tables too. Looking forward to the adders being found and disabled.

MFB · **Posted:** Fri Feb 24, 2012 7:14 pm

Yesterday, we did an afternoon run to log including af#3 correction. On the 2nd wot run, we got a -2.734 correction all through out. Question is if it affected open loop fueling. Ive turned off the table and its also confirmed that correction is now always zero. Just odd though that af#3 Learning has always been zero, for the forester and sti.

td-d · **Posted:** Fri Feb 24, 2012 9:10 pm

Yeah - only corrections were in action on my Sti as well - learning was always zero.

RomRaider

AF#3 (Rear O2 sensor) correction limits

Who is online