'165 ecm running $32B, 2.8 v6, LC-1 wired to nb input with ford t-bird maf.

Can someone take a look at this and tell me why the int is stuck below 128 in any cell other than 0 or 1? I can post a datalogg if needed.

 

So far nothing really stands out to me.

Wait! Here's something strange: Reset Corr/init of Auto Enrich is not set in the 3rd airflow mode word. Address 16, bit 1

Try setting this bit to 1.
LC016: FCB $42 ; 0110 0010 3rd AIR FUEL MODE WORD
;
; Bit 0 = not used
; 1 = Reset Corr/init on A.E.
; 2 = not used
; 3 = not used
;
; 4 = not used
; 5 = Int Reset when B.L.M. cell Chnge
; 6 = Calc Base Inj PW not Tbl (C434)
; 7 = Purge when output energized

I see that you have rather wide BLM airflow boundaries. Your Ld VAR for Scaling LV8 is raised which will increase the normal load by a factor of 1.45, but I don't see this as being a problem.
Your Stoich AFR is 14.69 instead of 14.73. Your closed loop timers are relatively short. I'll keep digging.

 

Are you sure that's not reset corr/init on accel enrich? The way I have the blm cells set up it gives me a an idle park cell (0), idle drive (1), cruise (5), acceleration (6), heavy accel (10) and wot (15). The load scaler is for v6's so I know that's correct. The closed loop timer is set to the warmup time of my wbO2. What I've been trying to figure out is what's happening in the idle cells that allows it to go above 128. Grrr!

 

Some of the hacs say Auto Enrich, some just say A.E. so its not completely clear.

I can check my actual source code printout tonight to clarify the meaning.

From the source, this bit is commented as:
1 = RESET CORRCL/INT ON A.E.

and where it is used is commented as:
TEST AE OPTION

 

Other things that will reset the INT:
DFCO or PE flags?
Cell changed?
Force low pulse width?
TEST AE OPTION:
BRANCH IF AE NOT ACTIVE
BRANCH IF NOT IN PROGRESS
CCP change?

 

Here's the relevant excerpt:

; CK AF MODE WD
; b1 DECELL FUEL CUT OFF ENABLED
; b5 PWR ENR IS ACTIVE
;
LDAA L0043 ; AF MODE WD
BITA #$22 ; b1 & b5
BNE LDBF4 ; BR IF b5 & b1
; ... else
;
; CK AFR WD 3
;
LDAB LC016 ; AFR MD WD 3
BITB #$20 ; b5, Int Reset ON BLM CELL CHG
BEQ LDBB2 ; BR IF NOT b5
; ... else
BITA #$04 ; b2
BNE LDBFA ; BR IF b2
; ... else
LDBB2: LDAA L0033 ; MINOR LOOP MD 1
BITA #2 ; b1, 1 = FORCE LOW PULSE, RESULT OPN LP
BNE LDBFA ; BR IF NOT b1
; .... else
LDAA L003E ;
BMI LDBE8 ;
; ... else
BITB #2 ; b1
BEQ LDBC6 ; BR IF NOT b1
; ... else
BRCLR L0038,$88,LDBC6 ; BR IF b7 & b3
; ... else
BRA LDBFA ;
;
; CK IF CCP Chg FORCE INT TO 128 ENABLED
;
LDBC6: LDAA LC017 ; AFR MD WD 4
BITA #$08 ; b3, ON CCP Chg FORCE INT TO 128
BEQ LDBD4 ; BR IF NOT b3
; ... else
LDAA L003B ;
ADDA L003E ;
RORA ;
BCS LDBFA ;
LDBD4: LDAB L0035 ; MNR LOOP MD WD 2
BITB #$20 ; b5, DIAG SW IN ALDL MODE
BNE LDC05 ; BR IF b5
LDAB L0063 ; Filtered ld value
CMPB LC5F1 ; Dcel lean Param
BHI LDC05 ;
; ... else
LDAB L0057 ; RPM/25
CMPB LC5F2 ; Dcel lean hi lmt param
BLS LDC05 ;
; ... else
LDBE8: LDAA L0044 ; AF MODE WD #2
BITA #$40 ; b6, 1 =RICH, 0 = LEAN
BNE LDC05 ; BR IF b1
; ... else
LDAA L00C9 ; Cls/Lp Int, Add Fuel if > 128
CMPA #1280 ;
BCS LDC05 ;
; ... else
LDBF4: LDAA L0041 ; DIAGNOSTIC MODE WD 3
ORAA #$04 ; SET b2, INT FORCED 128. ERR 44/45
STAA L0041 ; DIAGNOSTIC MODE WD 3
LDBFA: LDAA #128 ;
STAA L00C9 ; RESET Cls Lp Int to 128

 

I also noticed you have bit 7 set in the 3rd Air Flow Mode Word. This bit is not set in any of my bins.

This bit is commented as: 1=PURGE WHEN OUTPUT ENERGIZED. I don't know what this affects.

You may want to try turning it off.

So, you've disabled the AE OPTION, disabled the CELL CHANGE OPTION, disabled the CCP change option. There isn't much left. DFCO, PE or Force low pulse width.

Have you considered going back and enabling these options like the original settings. I admit, I don't see why its not working as is though.

 

Well my guess is that bit 7 would be related to ccp... I'll test that too. I tested the reset corr/init on ae function and that didn't help, just made the int reset like crazy. Explain the force low pulse width function to me.

Changed bit 7 and it didn't help either, just made the target afr a point leaner on startup for some reason.

 

The force low pulse is set at the Min Base PW Check when the calculated pulse width is below the min base pulse width and the default pw is used instead.

This shouldn't be the case, since your min base pw is set to 0.

Here's an example from $32 ($32B is the same with different addresses):

;------------------------
; Min Base PW check
;------------------------
BCLR L0033,$02 ; b1, FORCE LOP PW, RESULT OPN LP
CPX LC3A1 ; base PW
BHI LF9D2 ;
BSET L0033,$02 ; Set LOW PW RESULT OPN LP
LDX LC3A3 ; Default Base PW fm ROM

 

Ah, now I know what you're talking about. Well what's left... I've checked out all the related switches, I've looked over all the constants I can see as related, I've changed many of the tables back to stock without luck, I've even started over with a fresh base bin thinking this might possibly be a copy error. I'm going to keep working on it but I'm considering switching to $6E to see if that makes a difference.

 

ok this may be a total newb question but how are you guys getting the source of the bins ? I am very interested

 

I changed the maf read locations, the O2 thresholds and a few other settings and still no luck. Here's a datalogg I took driving to work. AFR changes look pretty crazy but I don't have any filtering on the wb output and it's only monitoring 1 header. You'll need my aldl file (1227165 $32B (WB).txt) to look at it.

 

I can't properly display the log session with TunerProRT Version 4.13.

The session comments only shows a square box and the date on session 1 shows 13/5/2008. Perhaps the file is corrupt, or I have a compatibility issue with my version.

 

Hold on, that one's screwed up. Session length says 138084352! LOL

 

Alright, try the '2.8 \'165ecm (B)6.adl' file I have in the datalogs and tuning notes section at moates.net.

 

I've been looking into your data file to understand the behavior. Its as if the INT is capped at 128 or that there is no negative error or that the O2 sensor signal is ignored due to the low voltage readings. Very strange.

The only thing that stands out is the rather digital behavior of the narrow band O2 signal, but I still don't see anything that would result in the behavior.

 

And it only happens in cells other than 0 & 1. WTF

 

Maybe try putting the limited error term at L046B into the data stream to see whats going on.

Cells 0 and 1 are at very low airflow < 7 gm/sec. The slow O2 filter coefficient is at its lowest (16) under 8 gm/sec.

Could be related?

Maybe try reducing the rest of the table?

 

I reprogrammed the wb to .1mv at its lowest and set the filter to 16 across the table and that didn't do it either. How do I put the limited error term into the datastream?

 

Replace it something that is already there, that is not very useful, like startup coolant temp for example.

Startup coolant is stored at L005F, The limited O2 error is stored at L046B.

Change this:

LC76F FDB $005D ; 8 COOL DEG
LC771 FDB $005F ; 9 STARTUP COOL DEG C.
LC773 FDB $0081 ; 10 TPS A/D COUNTS
LC775 FDB $0057 ; 11 RPM VAR USED FOR F1 EXTEN LOGIC
LC777 FDB $0095 ; 12 PRESENT MINOR LOOP REF PERIOD FROM ECM
LC779 FDB $0096 ; 13 #12 lsb
LC77B FDB $0065 ; 14 FILTERED MPH VARIABLE (UPPER = MPH FOR DISPLAY)

To this:

LC76F FDB $005D ; 8 COOL DEG
LC771 FDB $046B ; 9 LIMITED O2 ERROR.
LC773 FDB $0081 ; 10 TPS A/D COUNTS
LC775 FDB $0057 ; 11 RPM VAR USED FOR F1 EXTEN LOGIC
LC777 FDB $0095 ; 12 PRESENT MINOR LOOP REF PERIOD FROM ECM
LC779 FDB $0096 ; 13 #12 lsb
LC77B FDB $0065 ; 14 FILTERED MPH VARIABLE (UPPER = MPH FOR DISPLAY)

With the hex editor, change the 00 5F starting at 00000771 to 04 6B.


Scale it like the minumum error to implement INT
Calib = Arg*226

or x/226 or (x * 0.4425) volts dc.

It should be limited to a maximum of 0x88 or 136/226 = 0.601 Volts

 

There's no error term showing up. Just stays at 0.

 

Maybe that address isn't accessible. Another possibility would be to change the stored location from L046B to L005F and then put 5F into the datastream as usual. Maybe this will work.

I see that it is stored twice

DC77: STAA L046B
DC7A: LDAA L00D2
DC7C: LDX #$C599
DC7F: JSR LF344
;
DLC82: LDAB L0082
DLC84: CMPB LC563
DLC87: BHI LDC91
;
DLC89: LDAB L0067
DLC8B: BNE LDC91
;
DLC8D: LDAB LC565
DC90: MUL
DC91: LDC91 PULB
DC92: MUL
DC93: CMPA #$0058
DC95: BLS LDC99
;
DC97: LDAA #$0058
DC99: LDC99 PSHA
DC9A: STAA L046B

So at 1C78 Change 04 6B to 00 5F
and at 1C9B change 04 6B to 00 5F

It should just overrite the already stored start-up coolant value. I don't see any harm in that.

 

That did it. Glad you're helping me because I can't read the algorithim, one of these days I'm going to have to sit down and learn the code. It appears the error term freezes when the int freezes at 128, and the actual magnitude of error doesn't correlate to what the 02's reading. I have another datalog called '2.8 \'165ecm (B)2.adl', same place if you want to take a look.

 

Ok, I've got your data file.

If I multiply the raw signal by 42.5, the positive error seems to be reasonable in units of mv at least when the O2 signal is very high (900-1000 mv). Still seems to be off by a factor of 10 somewhere???

I also see that the INT will increment back up to 128 while the raw error value is frozen at 5, (scaled value is 212.5 mv) and the O2 readings are low (150 mv or so).

It would appear that something is going wrong with the error term when lean as expected from the behavior. But how to fix it? That is the question.

If I have a chance, I'll try and observe the error term in my car for comparison, but the recent snow may make it difficult for me to get out and try it.

From your data, the raw error term only varies between 0 and a maximum of 13. This by itself seems odd. Values of 12-13 cause the INT to decrement, values of 4 and 5 seem to permit the INT to increment back up to 128 if already reduced. It would appear that you have just barely enough error to move the integrator.

Most of the time, its either 5, 12 or 13.

Could this be the effect of the low O2 reduction gain settings?

I also noticed that your rich/lean offset vs coolant temp table has been zero'd out. I don't recall if you've tried adjusting this table back to original settings.

 

I've already messed with the offset, didn't have an effect. It's like the error term isn't scaled correctly- I'm going to look for a related term in the calibration.

 

Here's another thought: Have you tried reducing either of these values for the Rich/Lean window?

LC566 FCB 12 ; Diff Value to Make up Rich/Lean Window for Fast o2.(VDC * 226)
LC567 FCB 23 ; Diff Value to Make up Rich/Lean Window When Air div.(VDC *
226)

 

Alright, back to this thing. I've changed both of those and it just made the int cycle too much, but it's still stuck under 128. I messed with the positive error gain factor as well with no results.

 

Anyone else reading this have any ideas?

 

Looks like its just you and me in on this discussion.

Have you considered just popping a regular narrow band O2 sensor back in for testing? That's about the only other thing I can think of trying.

Sorry I haven't had a chance to check the limited O2 error behavior in my car yet. I still intend to do so.

 

Yea, that's what I figured. I'm going to plug the stock maf back in too and see if that's the problem. Maybe it has something to do with the read location.

 

I've converted back to one of my earlier bins I know was functioning correctly and still encountered the same problem, which leads me to believe more and more that this is a problem with the maf or a/d read location. Still haven't had a chance to plug the stock maf back in, cold weather and work in the way.

 

If its any consolation, I've never had the issue using the same MAF A/D reads (on channel 0).

My experience suggests at least, that this is not the problem, but the source of the problem has certainly been elusive, so I don't blame you for considering this.

I'm also reading the fuel pump voltage on channel 0 and using this to report the MAF voltage in the datastream, but I can't believe that this is responsible for the difference in the behavior.

I hope you can find a solution.

Maybe we should consider the MAF wiring? Maybe something strange is going on with the MAF signal ground/analog ground to the ecm?

 

Well nothing's on pin A11 right now. Maybe there's a routine to limit the int if there's no voltage on the analog ground.

 

I have my MAF sensor signal return (pin 4 on the 05+ Ford style connector) connected to pin A11.

This should also be pin 4 on the 96-04 Ford style connector.

I have pin 5 ground connected to A12. I think this should be pin 3 with the old connector, but the labelling is different. Maybe you want pin 5 on A12 as well.

Here's a link with both Ford maf connector pin-outs:

http://www.vmptuning.com/store/index...=194&parent=32

Edit: I checked the old thread. It seems that you have the even older 4 pin connector. Try pin C on A11 and pin B on A12.

 

I took a look at the v6 wiring schematic and it looks like it grounds the maf outright- no return to the ecm...

See any reason not to try A12 on A11 for gigles?

 

So your V6 harness is missing the wire. I get it.

C and B are likely connected together near the MAF end of the harness.

Maybe jumper A11 and A12 together at the ecm end, or run a test wire from A11 all the way to the MAF pin C.

Moving A12 to A11 is probably the easiest thing to try.

 

A12 to A11 didn't do it. Pin C to A11 is up next. There must be a way to disable it in the algorithim- it runs fine with a digital maf.

 

I ran a wire from maf pin c to ecm pin A11 and that didn't do it either. I took a while to look over my repin carefully and noticed that I forgot to disconnect the high gear signal wire on C8 but changing that didn't have an effect. So the only descrepancies left are pins D3 and D10, both system grounds. After these I'm stumped again.