Instructions to change fuel maps on 14CUX Griffith, Chimaera

Dave,
Well done, That's excellent work and explains a lot about my very early test results. I always assumed it was in the code as R3383 seemed to hold the idle higher but now thinking about it I never copied my R2422 full data structure in to R3383 like I do now into R3652. I'm already looking forward to having some more data to test when the weather improves.

This sounds like what Mark Adams refers to as engine breaking that he improves in his Tornado chip and therefore must be configurable data after all.

Stevesprint I thought you were on holiday as you’ve not posted on here for a while! And there’s me thinking ‘Steve’s away so perhaps I’ll be able to get a word in edgewise now’. Good to see you back.

I think you’re going to be busy testing the last eight positions (x200A - X2011) in the fuel map data sections as they show some interesting trends:

Tune R2422
FM2 $91, $1B, $04, $82, $2C, $2D, $68, $3C
FM5 $B9, $1B, $05, $6C, $6E, $2C, $5C, $64

Tune R2967_chim450_CAT
FM2 $B2, $1B, $03, $6C, $24, $23, $68, $3C
FM5 $91, $0F, $04, $BC, $6A, $27, $5A, $64

Tune R3361
FM2 $B2, $1B, $03, $6C, $24, $23, $68, $3C
FM5 $B2, $1B, $05, $6C, $7A, $2C, $23, $64

Tune R3652
FM2 $B2, $1B, $03, $6C, $24, $23, $68, $3C
FM5 $B2, $1B, $05, $6C, $7A, $2C, $51, $64

Looking at R3361/R3362 and Joel’s issue with high RPM it appears X2010 values as low as $23 in FM5 were in favour for those tune versions. It’s noticeable that the FM2 value of $68 is consistent throughout. Some other wide variations occur in X200E (a fuel map value) and X2011 (a value for calculating throttle delta)!

Dave,
I'm pleased to hear you missed me, made excellent progress and had an exciting time. I’ll have to go off-air more often.

Joel,
Thanks for sharing your important discoveries with us all. I’m sure you realise your first point was a huge help as it pointed Dan and Dave in the right direction, plus I’m now looking forward to testing the data variables but I'm sure you'll beat me to it.

Your second point, I've also noticed exactly the same AFR fluctuations in closed loop at idle and agree it seems wrong that the open loop idle AFR is more stable than closed loop. At first I thought something was wrong until Mark Adams educated me that the AFR fluctuates by design to help the CAT’s chemical reactions. There's also a really good explanation on Wikipedia under the heading ‘Three-Way’. http://en.wikipedia.org/wiki/Catalytic_converter.

Joel,
I'm very impressed you've been modifying the code, do you use Dan's rebuild project? It's my favorite toy. Do you mean you've modified the data at C222 & C223 or the code below that uses the data at C222 & C223? I guess this could vary depending on which revision you are working on.

;------------------------------------------------------------------------------
;                      MAF Fault Substitution Code
;
; This code executes when the MAF Fault Bit is set. Since the load based row
; index cannot be calculated without a valid MAF reading, an alternate method
; is needed. This method estimates the fuel requirement based only on throttle
; position and engine temperature. The data values that are used to tailor
; the results are stored in the data portion of the PROM. The code is set up
; for different open and closed map values, however, both sets of values are
; identical. The end result is a 16-bit value in the AB register set. The
; normal fuel map code block is then bypassed.
;
;------------------------------------------------------------------------------
                ldd         throttlePot         ; load 10-bit TPS value
                subd        throttlePotMinimum  ; subtract TPmin
                bcc         .LE87A              ; branch if result is positive

                ldd         #$0000              ; else, limit value to zero

.LE87A          lsrd                            ; shift right twice so that
                lsrd                            ; top bits are in B register
                ldaa        fuelMapNumber       ; load fuel map number
                beq         .LE885              ; branch if map 0

                cmpa        #$04                ; compare with 4
                bcs         .LE89E              ; branch if less than 4
;-------------------------
; Fuel Maps 0, 4 and 5
;-------------------------
.LE885          ldaa        coolantTempCount    ; load ECT sensor count
                cmpa        $C221               ; value is $47 (56 C or 138 F)
                bcs         .LE895              ; branch if ECT is hotter

                ldaa        $C222               ; value is $1E
                mul                             ; mpy top 8 bits of TPS by $1E
                addd        $C223               ; value is $1000, add this 16-bit value
                bra         .LE8B5              ; branch ahead

.LE895          ldaa        $C225               ; value is $24
                mul                             ; mpy top 8 bits of TPS by $24
                addd        $C226               ; value is $0E00, add this 16-bit value
                bra         .LE8B5              ; branch ahead
;-------------------------
; Fuel Maps 1, 2 and 3
;-------------------------
.LE89E          ldaa        coolantTempCount    ; load ECT sensor count
                cmpa        $C228               ; value is $47 (56 C or 138 F)
                bcs         .LE8AE              ; branch if ECT is hotter

                ldaa        $C229               ; value is $1E
                mul                             ; mpy top 8 bits of TPS by $1E
                addd        $C22A               ; value is $1000, add this 16-bit value
                bra         .LE8B5              ; branch ahead

.LE8AE          ldaa        $C22C               ; value is $24
                mul                             ; mpy top 8 bits of TPS by $24
                addd        $C22D               ; value is $E000, add this 16-bit value

Joel,
Thanks again for all your valuable help.

Hum, prepare to be unimpressed. What I did was change the code in Dan's rebuild project from this:

;---------------------------------------------------------------------------------------------------
;              Determine if Rich or Lean
;
; X201B (right) and X201C (left) are offset or bias values that are added to the threshold value
; The data value XC098 (usually zero) is also added/subtracted as an addition bias.
;
; From Wikipedia:   lean mixture causes low voltage (excess oxygen)
;                   rich mixture causes high voltage (depleted O2)
;
;---------------------------------------------------------------------------------------------------
.LE21D          pulb                            ; pull the HO2 reading reading
                ldaa        $00BD               ; this is working value from X00BE (right) or X00BF (rt)
                bne         .LE23A              ; branch if value is not zero

                ldaa        o2ReferenceSense    ; load the HO2 threshold value
                tst         $0088               ; test bank indicator bit
                bmi         .LE22E              ; branch ahead if bit is 1 (left bank)

                adda        $201B               ; Right:  add X201B to HO2 reference value
                bra         .LE231

.LE22E          adda        $201C               ; Left: add X201C to HO2 reference value

.LE231          cba                             ; A (ref plus X201B or X201C) minus B (HO2 reading)
                bcs         .LE243              ; branch if HO2 reading is higher than reference value

                suba        $C098               ; Lean: data value is zero, subtract from reference
                cba                             ; result still positive, since value is zero
                bcc         .LE25A              ; branch to jmp to lean_condition code

.LE23A          ldaa        $0089               ; code jumps here if X00BD value is non-zero
                bita        #$40                ; test X0089.6
                beq         .LE2B0              ; if zero, branch to jump to LE459 (into lean code)

.LE240          jmp         .LE2EF              ; jump down (into rich_condition code)
;--------------------------------------------------


To this:

;---------------------------------------------------------------------------------------------------
;              Determine if Rich or Lean
;
; X201B (right) and X201C (left) are offset or bias values that are added to the threshold value
; The data value XC098 (usually zero) is also added/subtracted as an addition bias.
;
; From Wikipedia:   lean mixture causes low voltage (excess oxygen)
;                   rich mixture causes high voltage (depleted O2)
;
;---------------------------------------------------------------------------------------------------
.LE21D          pulb                            ; pull the HO2 reading reading
                ldaa        $00BD               ; this is working value from X00BE (right) or X00BF (rt)
                bne         .LE23A              ; branch if value is not zero

                ldaa        #23                  ; load the HO2 threshold value
                tst         $0088               ; test bank indicator bit
                bmi         .LE22E              ; branch ahead if bit is 1 (left bank)

                adda        $201B               ; Right:  add X201B to HO2 reference value
                bra         .LE231

.LE22E          adda        $201C               ; Left: add X201C to HO2 reference value

.LE231          cba                             ; A (ref plus X201B or X201C) minus B (HO2 reading)
                bcs         .LE243              ; branch if HO2 reading is higher than reference value

                suba        $C098               ; Lean: data value is zero, subtract from reference
                cba                             ; result still positive, since value is zero
                bcc         .LE25A              ; branch to jmp to lean_condition code

.LE23A          ldaa        $0089               ; code jumps here if X00BD value is non-zero
                bita        #$40                ; test X0089.6
                beq         .LE2B0              ; if zero, branch to jump to LE459 (into lean code)

.LE240          jmp         .LE2EF              ; jump down (into rich_condition code)
;--------------------------------------------------

I then did a compare of the .bin files and saw that the code was altered at bin locations X6222 and X6223. When I said I changed the data at C222 and C223 I was trying to translate my .bin locations into what I thought were the ECU addressing. Maybe I should have said E222 and E223, or maybe I am just demonstrating more ignorance? Anyway, I altered the machine code at the same locations in my R3362 bin to force an immediate address and load the value of #23. Don't ask why I chose 23. Once I had the car idling on this code I altered the value and watched the AFR for changes. I was able to force the AFR to around 14.0 by using a value of around 59 decimal which I believe is close to 1.15volts. These results don't really make any sense to me since that seems quite out of range of the O2 sensor voltage. I was going to go back in to the code and remove the bias values at x201B and x201C and retry this test. I think these bias values might be responsible for the switching of the O2 sensors, i.e., the effect you mentioned in your previous post, which might be making it difficult to alter the AFR by changing the o2referenceSense.

Steve,I only managed to fix that board using a microscope and special soldering equipment,if I had not been so hamfisted in removing that original prom I would have been ok.
I just hope others learn from my mistakes and do EXACTLY what your instructions say when removing the original prom.
On the plus side is the fact that the board is tough enough to undergo 2 removal cycles and still work ok if you are careful!

Well the weather today was good enough for me to check over and fire up the old v8.Might as well tell you now it is in a TR7!
After it had warmed up a little I fitted the repaired ecu with an EEPROM burnt with Steve Sprints R3652 TVR 400 hybrid tune.
I fired it up and set airflow meter for smoothest idle,this corresponded with 1.4 volts on ROVERGAUGE as he recommends,what a difference its made.
Gone is the little bit of hesitation on pressing the accelerator pedal and the engine sounds so much smoother.Nice work Steve.
What surprised me the most was the fact my engine is only a 3500,admitedly with mildly cleaned up heads and a Kent h200 cam.
Will not be able to test it on the road till Sunday(1 FEB)but if it goes well I will be a happy man.(I altered the idle speed down to 700 rpm in the software as the idle seemed too high but its still smooth).Will report when I have more info.
I still cannot get over how excellent ROVERGAUGE is,especially as a diagnostic tool.If it cost 50 quid it would still be a bargain,don`t know how I managed without it to be honest.Tunerpro is bloody good as well,makes altering software quicker.
A BIG BIG THANK YOU TO DAN,COLIN,STEVE and everyone else who has contributed to this thread.

I am working on mapping a tuned Bowler Tomcat with a 3.9L V8. Thank you for the efforts guys, as this saves me the hassle from swapping ECU's with an open system. I am just starting but are able to connect to the ECU, change mappings and rebuild the project of Dan. While reading the thread I've been updating the TunerPro XDF for this interested. I've added all tables for all tunes and corrected some offsets (should now be all above 0x4000). I also added the TUNE ID, MIL enable and overrun fueling. If someone could provide me with an e-mail address I'll be glad to send the XDF for sharing.

Meanwhile I hope to receive my Ostrich 2.0 and 20AM MAF sensor somewhere next week so I can start adjusting for real. Hope that the injectors can keep up (ERR772A) otherwise I might need to swap the injectors. Can anyone provide the number of the TVR430 ones, I assume these are bigger given the higher power output?

PS
I also have dumps of a R2666 ROM and a R3157 ROM, where remarkably the R3157 ROM uses the "old" offsets and the R2666 ROM uses the "new" offsets.

Erik
I'm delighted to hear you've got off to a flying start with TunerPro and added more tables and tuning options.

If you email me your XDF file I’ll make the main table column headings dynamic, so they display the correct rpm points for extended main fuel tables, and I'll upload it to http://www.stevesprint.com/remap-14cux/gadgets.htm... . I’ll also add the throttle, start-up and temperature tables that Paul requested.

R3157 is a huge surprise and sounds interesting, does the fuel table display correctly in RoverGauge? Please can you email me a copy and I’ll also upload it to my website for Dan & Dave to see. I’ve PM’ed you my email address.

Thank you in anticipation, Cheers
Steve Sprint
P.S. Bowler Tomcats look amazing fun!!!

I don't know if the maps are displayed correctly in rovergauge as I haven't had the car running with this ECU (was a spare ECU that came with the Tomcat). I have no access to the car for 1 week so until then I cannot test it. The EPROM (Intel 27C128) had a handwritten label on it with "3157" and a green dot. The label had a vague red watermark on it with "GP" in italic letters.

The Tomcat has a rover V8 with a TVR exhaust system, John Eales stage 3 heads, a 40D engine base, a flowed inlet manifold shorter turrets and....
an ECU running the stock R2666 map 1 . Therefore I'll be putting some effort in adjusting the fuel map as I suspect it is running lean and hitting the speed limiter (gearbox is from a defender TD5, gears replaced with straight-cut).