now im a tbi guy!

[buddrow]

You don't have to be hatin' on tbi Steveo lol. It does serve the purpose. Not everything can be as glorious as your beloved LT1. :) Glad you got it lined out though. That VE map looks goofy like maybe the mask is different almost but not quite totally wrong.

Buddrow

[steveo]

yeah i dont know what the hell is going on there without that VE table.

mask ID byte in the bin is apparently at 0x04, and it's 4D. the 4D XDF im using seems fairly well developed.

there's no obvious reason why fuel would be added to a diagonal region across the table like that...? some of the spike is at common driving range too, 2400rpm @ 50kpa... 2800rpm @ 40kpa.. so if it's legit, its some pretty dumb VE..

but i suppose since this is a very wimpy cam with swirl port heads (very odd port velocity profile) and restrictive exhaust/intake; maybe there's a legit reason for extra fuel there?

[steveo]

interesting thing that i'd like to understand

i was chasing an exhaust leak.. you know, the usual TICKTICKTICK under load, but fairly quiet otherwise.

i found it in an interesting place, the o2 sensor ceramic was snapped in half, including the wire being slashed right open.

there was never a check engine light, but wow.

so since this thing was so stuck that my o2 socket broke in half. so i unplugged it while i went hunting for better tools.

it ran exactly the same with the sensor unplugged (never entering CL, obviously?) but oddly enough took about 2 hours of total driving before a check engine light was set.

why..?

when i finally removed the sensor i found that the element was also broken, just laying in the housing. i really doubt it broke from removal of the sensor.

how the hell did it run like this without setting a CEL, obviously it hit closed loop, right? if so, how did it manage to trim without a noticeable difference in how the car runs?

this is important in my understanding of these ol' tbi ecms; every other ecm i've played with runs like crap if you so so much as foul the o2 sensor, and almost instantly throws a code for it..

[RobertISaar]

the O2 sensor circuit is a bit different than the other A/D circuits in that it doesn't have a pull-up or pull-down resistor that will bring the signal to the limits when the sensor is unplugged, because of this, the whole circuit acts like an antenna and voltage will go everywhere randomly. that may have happened long enough to where you don't get a code set and the ECM was in closed loop, based on non-real sensor readings.

[ony]

steveo, what flaver ecm are you running?

[steveo]

8063 4D

no tuning gear or even serial cable yet, being lazy.

[steveo]

even before I get that stuff, I can tell, tbi is like running a friggn trs-80 when everyone else has quad core 3d accelerated supercomputers. I'm getting a tpi rig ASAP, when I find one that isn't four digits in price.

[RobertISaar]

these days, most people would call TPI ancient too....

i'm not one of those people, i feel it certainly still has valid applications.

[EagleMark]

Serial cable? What did you datalog the LT1 with? It should work on this as well.

Don't underestimate the old ECM, it does it's job quite well. The worst part is the old EPROM, do yourself a favor and eithwer get an Ostrich to Emulate and some day burn a final chip... yeah I know, no such thing as a final chip... or the G2 adapter and solder it in so you can run a newer EEPROM 27FS512 flavor works well with the Moates Burn 2.

I also have a GQ-4x programmer so I can burn the old 2732a chips. Gets old erasing them with UV light but it works. You can still buy the 2732a chip in ST brand and I found them to be very forgiving when it comes to multiple erasings and burnings compared to the stock GM chips... AMD IIRC...

There's a few old TBI ECM that would be a lot better off in tuning if someone put the time into the mask/XDF. When I started with $42 there were several of them floating around and even the best one just did not cut it. Luckily there was a great dissasembly and after combining them all and double checking the dissasembly along with adding things it tunes very well! Then of course you can add all the tricks, data tracing to ADX as well as history tables. I enjoy tuning $42 now with all the work done to XDF and ADX.

[buddrow]

We can give you a huge THANK YOU Mark for putting the $42 stuff.

Steveo, i had run into similar situations with totally inop o2 sensors not setting a code in the past but I just figured the early systems were more "forgiving" compared to the newer ecm/pcms.

Buddrow

[EagleMark]

steveo knows the work that goes into them as he's a member of the 100 hour club in $EE. I'm feeling better so if he plans on keeping the $4D I'll give a hand on the XDF and ADX to make them more then useful!

[steveo]

Serial cable? What did you datalog the LT1 with? It should work on this as well.

it was so built into the car with the dash computer, i just sold it with the car. need a new one.

There's a few old TBI ECM that would be a lot better off in tuning if someone put the time into the mask/XDF.

even with $4D the parameters discovered so far put EE to shame, although as i've already discovered, some of that is weird stuff (like that VE table that wasn't quite right)

Then of course you can add all the tricks, data tracing to ADX as well as history tables.

as with EE, im not too interested in tunerpro for acquisition so i wont really be too into working on the ADX.

i tried to make time with the EE one, but if im not going to use it, its hard to motivate myself to work on it.

i still use the software i wrote for all that. it just spits out csv data all the same, but..

if i want to analyze something in particular i can just use gnuplot to graph stuff, or run a c function to build my own tables and do averaging.

with the lt1 i was doing stuff like this for narrowband tuning (pseudocode)

make an array of 15 cells, storing min/max/avg of lterm(left), lterm(right), maf afgs, rpm, map kpa
if(coolant temp > whatever) and (runtime > 120 seconds) and (closed loop = 1) and (blm enable = 1) and (pe enable = 0) and (blm cell < 15) {
  update all averages, min, and max in that cells
}

that gives a cell-by-cell running average, which you can feed hundreds of logs into.

**** BLM CELL vs TRIM, MAF, MAP, RPM RANGE/AVG ****
(Igoring records with records < 30)

* Cell 0 (527 Hits)
        BLM: 120 - 133 (Avg 124.5)
        RPM: 850.0 - 1575.0 RPM (Avg 1242.1)
        MAP: 21.4 - 34.3 KPA, (Avg 28.8)
        MAF: 7.8 - 44.2 AFGS, (Avg 9.5)

* Cell 1 (3181 Hits)
        BLM: 120 - 129 (Avg 124.3)
        RPM: 1400.0 - 2200.0 RPM (Avg 1767.1)
        MAP: 18.5 - 34.3 KPA, (Avg 27.8)
        MAF: 8.1 - 72.3 AFGS, (Avg 14.9)

* Cell 2 (864 Hits)
        BLM: 120 - 128 (Avg 122.5)
        RPM: 2000.0 - 2775.0 RPM (Avg 2260.0)
        MAP: 16.6 - 34.3 KPA, (Avg 26.4)
        MAF: 9.4 - 36.8 AFGS, (Avg 19.9)

i know it's not data tracing, but i think it's better for me to read. i found not only does it give me a realistic view of whether cell boundaries are being used properly, but where it spends most of its time, and easily throws me thousands of records for averaging.

similar for fueling,

**** Wideband AFR (Average) vs RPM during PE ACTIVE ****

RPM    0 -  400       14.0    8 Counts
RPM  400 -  800       13.9    5828 Counts
RPM  800 - 1200       14.4    1293 Counts
RPM 1200 - 1600       14.5    163 Counts
RPM 1600 - 2000       14.8    5 Counts
RPM 2000 - 2400        0.0    0 Counts
RPM 2400 - 2800       13.0    15 Counts
...

and for maf table scaling:

**** Wideband AFR AVERAGE vs MAF AFGS ****

MAF   0 -   5       14.4    2654 Counts
MAF   5 -  10        0.0    0 Counts
MAF  10 -  15       14.5    2381 Counts
MAF  15 -  20       14.6    3477 Counts
MAF  20 -  25       14.6    1954 Counts
MAF  25 -  30       14.5    1237 Counts
MAF  30 -  35       14.4    891 Counts
MAF  35 -  40       14.4    495 Counts
MAF  40 -  45       14.4    436 Counts
MAF  45 -  50       14.5    285 Counts
MAF  50 -  55       14.3    234 Counts
MAF  55 -  60       14.3    185 Counts
MAF  60 -  65       14.4    162 Counts
MAF  65 -  70       14.3    138 Counts

i'll do the same thing with this car, so i probably wont even fire up tunerpro rt.

I'm feeling better so if he plans on keeping the $4D I'll give a hand on the XDF and ADX to make them more then useful!

thats a ton of motivation to get off my butt and get some chip burning gear and a cable.. honestly i dont even know where to start tuning this beast, i think you're the guy i need to talk to. i still haven't gotten around to popping a valve cover and figure out what kinda heads i have.

[EagleMark]

Looking at the titles of the TunerPro RT XDF and ADX I've already done a little work to them. Also found a good looking disassembly to work with and added it to the $4D thread. With that we have what we need to look at the VE table. Unless you have a better looking mask/XDF and dissasembly?

http://www.gearhead-efi.com/Fuel-Inj...Information-4D

History tables work well for things like BLM, then use it and a spreadsheet to dial in the fuel VE tables, and can always export the data log file to CSV or... Excel format. The Data Tracing is more and advanced feature but can show you exactly where you are in bin file while playing back data, great for finding that spot giving you trouble.

Diddin't you find your bin file? Post it up! It'll give me something to start working on...

[EagleMark]

Fueling looks the same as $42...

;~~~~~~~~~~~~~~~~
;-Main %VE table
;~~~~~~~~~~~~~~~~
;
LD278   FCB       0		;row offset
        FCB      16		;Col offset
        FCB       8		;# of cols

;    RPM/1k  .4   .8  1.2  1.6  2.0  2.4  2.8  3.2 ;kPa
;              
        FCB  50,  49,  54,  57,  60,  62,  63,  65 ; 30
        FCB  51,  52,  55,  63,  69,  76,  86,  90 ; 40
        FCB  53,  54,  61,  65,  79,  88,  99, 103 ; 50
        FCB  55,  59,  75,  82,  86,  94, 101, 108 ; 60
        FCB  65,  71,  88,  91,  93,  99, 110, 116 ; 70
        FCB  95, 107, 112, 108, 103, 103, 111, 120 ; 80
        FCB  98, 106, 121, 121, 119, 119, 122, 127 ; 90
        FCB  99, 107, 121, 125, 124, 125, 125, 129 ;100
        FCB  99, 106, 120, 127, 126, 124, 126, 130 ;110
;
;~~~~~~~~~~~~~~~~~~~~
;-%VE adder table
;~~~~~~~~~~~~~~~~~~~~
;
;		%VE		;RPM/25
;
LD2C3   FCB      51		;   0
        FCB      51		; 400
        FCB      74 		; 800
        FCB     102		;1200
        FCB     106		;1600
        FCB     110		;2000
        FCB     112		;2400
        FCB     111		;2800
        FCB     106		;3200
        FCB     102		;3600
        FCB      96		;4000
        FCB     108		;4400
        FCB      99		;4800
        FCB      72		;5200
        FCB      64		;5600
        FCB      57		;6000
        FCB      53		;6400
;

[steveo]

heres what ive been dicking around with so far mark;

4D-steveo is a supposed 4D xdf from the old moates site that has some interesting stuff in it mapped out, but has that bunk VE table and who knows what other mistakes.

i started 'cleanup' on it, but stopped once i found that VE table error. i think i will continue on with it, but everything in there needs verification.

4D works is the one everyone seems to have been using, ve table displays correctly and seems the most...um..correct.

AMYB is a bin i tracked down from the old site that matches the sticker on my chip.

thanks a lot for that disassembly, im going to have a go at it!