[mdlug-discuss] OBD fuel monitor fine tuning

[Drew]

At 03:51 PM 10/15/07, Robert Meier wrote:
>Were both of the above measurements at the same fuel station?

     No. However, I have devised a funnel which lets me get the fuel level in
the main tank up to a consistent mark, without the possibility of the hose
sucking back any of my fuel.Basically, I fill the tank normally most of the
way, then I pull the hose and put in the funnel, and dispense fuel into
the funnel carefully while watching for the level and keeping the end of the
hose well away from the bottom of the funnel. As soon as I see the level
settle at the top of the filling tube, I stop and consider it a full fuel load.
The state of fill this way is to within plus or minus about a cup (1/16 gallon)
of fuel.

>Have you considered trying the same and different fuel stations to
>discover/deny a correlation?

     Only thing I do there is try to avoid buying ethanol-blended fuels. And
this because I don't get enough of a price break to make up for the reduced
BTU content per gallon which would result.


>A gallon of avgas masses 6 pounds under specified conditions.
>
>I understand that typical gasoline density varies from about 5.5 to 6
>pounds per gallon depending on temperature (expands with rising
>temperature), air pressure (expands with decreasing pressure),
>humidity (expands with decreasing relative humidity).

     I got a density measurement: In a metal can that I had previously
calibrated with a mark for one pint, a pint of fuel from the emergency
container weighed in at 355 grams, plus or minus 5 grams. Water
to the same level weighed 475 grams, same precision. Results which
I trust to two significant figures are: Fuel = 747 grams/L = 6.237lb/gal.
The water checked out at 1003.85 g/L.

     The fuel is a bit denser than either of us suspected - off by about
4 percent, and using the updated value would change the results *in
the wrong direction* - dividing the fuel mass flow by the larger result
would produce a smaller volume flow result, and the measured usage
is already larger than the tabulated results.


>Still looking, I've been unable to find a better description than
>copied by http://en.wikipedia.org/wiki/OBD-II_PIDs .


     That much I already have in the Python code to OBDGauge. I am,
however, confused about the directions of Rich and Lean in regards
to the sign of the values. Both the OBDGauge User Guide and the
Wiki page state that negative values for the fuel trim numbers (ie, raw
values less than 128) indicate a Rich mixture. However, When I first
start the engine, the values that I get for the short term fuel trims start
out very strongly *positive* (raw values much more than 128), and
gradually work their way to stoich; and I was under the impression
that engines tended to run rich when first started. I just started my
engine to double check this, and both short term fuel trims read over
40/128 to positive. (I display these numbers in 128ths instead of
calculating "percent".)


>Please let me know if you find a better description.
>
>My currently uncertain understanding is that the fuel trims (my
>Saturn SL2 has only 1 bank) are PI states in the fuel metering
>/oxygen sensor servo loop.

< math snipped >.

     But this still doesn't tell me, given values for fuel trim, how to
calculate the actual deviation from stoichiometric mix that the
engine is running at the time. Especially as there are *two sets*
of short term and long term Fuel Trim values. Do I add them all
together, or average them, or some other combination?

----

- Drew.