basic fuel metering equations
ulan at ee.ualberta.ca
Fri Jul 1 20:37:30 GMT 1994
> In the prototype controller, AE should simulate the accelerator pump of a
> traditional carburetor. That would make it an additive term to the equation
> of (1.2.2) that would add a pre-set quantity of fuel as a function of
> throttle change. Additionally, it would have an adjustable "decay parameter"
> that would be similar to the hole diameter of the "shooter". In the future,
> it may be beneficial to include MAP and coolant temperature into this
> term. For now, I will just represent it as a yet to be defined additive
> function of AE. (need more info here --- need to define the controlling
It's actually pretty important to have the 'shooter size' and 'decay rate'
controlled by engine temperature.
I've seen systems that have two accel pumps: one for delta MAP and the other
for delta throttle.
Ideally, you would model the fuel thickness in the intake manifold between
the injector and valve, which would require air temp, manifold temp, manifold
vacuum/pressure, and probably an estimation of the willingness of the air
to accept the gasoline as a vapour. This is something for someone who has
done research on this area... if you can model it, you don't have to spend
as much time tuning...
> 1.3.2 --- Deceleration Enleanment (DE)
> I don't believe this term is necessary in the first prototype of our
> controller. Ultimately this term will lean the engine during deceleration
> much as the acceleration enrichment term adds fuel during acceleration. It
> should have the same inputs as the acceleration term.
Exactly. Just acceleration enrichment turned around. It only adds about
fifty or so lines in my code (assembler). Makes a big difference when
you're coming out of a corner and you don't have to let the engine cough
out the raw gas it sucked in on the overrun.
> 1.3.3 --- Close-Loop Feedback (CLF)
> This should be a multiplicative term representing the integrated error from
> the oxygen sensor. The conditions for when this term should be included are
> yet to be defined. Conditions to consider are cold O2 sensor, cold engine,
> acceleration, deceleration, power enrichment. (others?) (need more info here,
> need to define the controlling algorithm)
Actually, I've seen it done as a combined additive and multiplicative term.
The short-term feedback (oscillator) is additive, the long term (block-learn)
is multiplicative. I guess this simplifies control calculations...
Most FI software I've seen has a lot of compensation for the temperature
effects on the O2 sensor. A lot of engine control software attempts to
predict the temperature of the exhaust and the exhaust system components
to compensate for this. I thought it was wierd, too...
More information about the Diy_efi