Scott T

I've been tuning a '747 on a 327 using GMECM, TunerCat, and Win ALDL. The thing's driveable and getting stronger as I zero-in on the VE table(s) and the spark. However, I've had a consistent problem with an "oscillation" showing up in the O2 measurements from both the 1-wire sensor and the WB. It's about 2 cycles per second, at all speeds, most noticeable at low RPMs: the mixture runs up and down across the full range of the sensors. I can feel it when driving (and of course now I'm obsessed with it).

I've been studying the 7747 hack, and I came upon a table referred to as "Int Delay Vs. Airflow". My research in the archives has led me to believe that it is the "transport time" for the mixture signal from combustion chamber to O2 sensor, and is used for predictive calculations. My O2 sensors are located in the collector section of the headers, so my logic was that a hysteresis was developing due to the difference in lag from the stock (just outside the exhaust port) location of the sensor compared to the location of mine (roughly 10 times the distance, thus 10 times the transport period). So, I figured I'd increase the transport time table by an order of magnitude (x10). I programmed this last night, and today on the way home from work updated the romulator and drove home, logging all the way.

It did have a bit of the expected effect - the oscillations, though still present, have a noticeably smaller magnitude, though the same period. However, when I switched over to the BLM table in WinALDL, what did I see but a whole bunch of 126's - 23 cells had entries in the "Narrow Avg" table, and all but *two* were 126 . Two days ago when I last logged, the numbers were not too bad (ranging from 120 to 135 or so), and I did tweak them last night, but I think I smell a snake.

Am I interpreting the "Int Delay Vs. Airflow" table correctly? I'm thinking not, but what did I do to hit such a perfect VE? I'm thinking the thing was running in open loop (though it was fully warmed and a 20 minute drive), though I didn't have the brains to check the flag table in WinALDL to confirm it. So, am I really good, or am I really dumb?

RBob

iTrader: (1)

Interesting stuff, isn't it. The fueling loop is a PID algorithm set up to oscillate due to high proportional gains. The "Int Delay Vs. Airflow" is used to adjust the delay or lag of exhaust stream. This is from the exhaust port to the O2 sensor.

What you probably need to adjust are the proportional gains. If you can feel it then it is too much gain. As programmed the O2/AFR is supposed to swing high/low.

These two tables are used to change the proportional gains:

PROPORTIONIAL GAIN FLOW FACTOR vs AIR FLOW
PROPORTIONIAL VALUE vs SLO o2 ERROR

The math for the proportional gain correction is:

Correction counts = = (PrpSloO2 * PrpGanOff) / 256

The counts are aded/subtracted from the PW. If set to zero the INTegrator will then be used to push the O2 through the calibration window.

RBob.

Scott T

RBob, where could I find the code that generates that equation? Is it in the hack? I reduced the PropGainFlwFactorVsAirflow and PropValueVsSloO2Error tables by 25%, as well as returned the IntDelayVsAirflow to its initial value, with little change in behavior - except that the BLMs are varying now, which is good. I think I need to study up on the algorithms used, it may be time for me go back and study up on Motorola 68HC11 reference manual.

RBob

iTrader: (1)

The ASDZ hac is the place to go for the information.

RBOb.