[dirk37]

I've been fighting with a tuning issue with my Turboed 89 Taurus sho since I finished the build. When below about 3k if I get into boost the car will lean out as the boost increases until I let off. I did some datalogging and research in the GUFB manual and I've come to the conclusion the X2J has an FN037 table which is capping the aircharge below ~3k rpm. As you can see in the datalog below, the aircharge follows the maf curve until the turbo spools, then it caps out and begins following the rpm curve.

Now the strategy file for the X2J doesn't have the information for the FN037 table. I've done some tuning by modding bin files with a hex editor before and I'm wondering if anyone could share with me what the FN037 table may look like or where to look for it. I've done some poking around in bin files before but I'm new to actually reverse engineering them to discover parameters, functions, and tables.

Thanks for the help, dirk37

[decipha]

http://info.efidynotuning.com/dis.htm

[sailorbob]

IIRC the strategy the X2J ecu uses doesn't have the FN037 function but something similar to the FN1036 table used in later strategies.

[mpaton]

It's a similar concept to FN037, but FN037 is a simple function of N.
X2J uses an Alpha-N table and then corrects that with ACT and normalizes with BP, at

Code: Select all

[code]

687C EF68F5 L687C call FLUW_UU
687F 0801 word FN070(N) (#$C108)
6881 AA00 vect N
6883 A03830 ldw REG_30,REG_38
6886 EF5EF5 call FLUW_UU
6889 D083 word FN044(TP_REL_X) (#$CA48)
688B 5C02 vect TP_REL_X
688D A03832 ldw REG_32,REG_38
6890 AD0A34 ldzbw REG_34,#$0A
6893 457300F838 ad3w REG_38,REG_F8,#$0073 ; T_5 FNALPHA_N(N,TP_REL_X)
6898 EF19F6 call TLUB_Ub
689B 5F72AF3B3C ml3b REG_3C,REG_3B,BP ; BP
68A0 EFC3F4 call FLUB_SS
68A3 A4C0 word FNAPLHA_N_CORR(ACT) (#$CCFE)
68A5 AE00 vect ACT
68A7 090138 shlw REG_38,#$01
68AA 6C3C38 ml2w REG_38,REG_3C
68AD 8DEF0038 divw REG_38,#$00EF ; 29.875
68B1 D503 jnv L68B6
68B3 BDFF38 ldbse REG_38,#$FF
68B6 C3763638 L68B6 stw REG_38,ALPHA_N_MAF ; ALPHA_N_MAF[/code]

What I've called ALPHA_N_MAF is the limiting value, the equivalent of AIR37 in A9L.

The limiting code is at

Code: Select all

569C  A376C250              ldw   REG_50,AMINT			; AMINT
56A0  3BE371                jb    MFMFLG,L5714
56A3  6776B050              ad2w  REG_50,ARCHLI			; ARCHLI
56A7  DB06                  jc    L56AF
56A9  8B763650              cmpw  REG_50,ALPHA_N_MAF		; ALPHA_N_MAF
56AD  D104                  jleu  L56B3
56AF  A3763650      L56AF   ldw   REG_50,ALPHA_N_MAF		; ALPHA_N_MAF
56B3  C376B250      L56B3   stw   REG_50,ARCHI			; ARCHI

while ARCHI is written to at 0x2D70 in A9L.

[dirk37]

Thank you very much for disassembling and finding that for me! I've been messing with the disassembler all morning without much success.

So can I just nop 56AF then and have it disable the check then?

Edit:

I got it working!! I overwrote the code with FF (NOP) and now the aircharge follows the maf curve under all conditions It's no exaggeration
when I say the car has run better than it ever has. It boosts way faster and there's not a bunch of unexplainable changes in the air fuel ratio. Thank
you all so much for the help!

For others reading this, the data to be replaced is A3763650 and can be found at offset 36AF. Replace it with FFFFFFFF. This can be done with a hex editor,
but I'm going to contact the creator of the X2J strategy and see if he'll add it in.

[sailorbob]

Why don't you just populate the table with appropriate values?

[dirk37]

I didn't want to reverse engineer the functions. I'll probably do that anyways for fun, but in the mean time patching was much easier. Plus with patching I don't have to worry about table limits.

[decipha]

by disabling it, if your maf ever fails youll be walking home

the pct_load table and its variants are also used for failure management when things go too far stray

Id recommend dialing it in, its easy to populate the table with approximate values

[sailorbob]

The table is not used for MAF failure management; there's still FN040 and what seems to be a variant of FN1358 in this strategy for that.

The FN1036 variant is used for air charge clip and backflow calculations.

[decipha]

Thats good to know, the taurus strat seems to be pretty advanced given its age.

[LeadHead]

Thats good to know, the taurus strat seems to be pretty advanced given its age.

Wonder if Yamaha had any input into the programming?

[dirk37]

I made a patch for the Quarterhorse using Tunerpro RT and uploaded it here. Remove the .txt extension and then you can open it with Tunerpro RT. Once you've got it open in Tunerpro, download the bin from Emulator into Tunerpro, apply the patch, then upload it back to the emulator and you'll be good to go! Unfortunately every time you make a change with Binary Editor this will be overwritten so you'll have to repeat the process each time.