WPSLPGExhaust Gas Oxygen (EGO) sensor Information

[Steve Schwartz]

Sorry, Greg, but I must disagree.

The 'bang-bang' control you refer to would be fuel_on - fuel_off.  The fact
that the EGO can only accurately report RICH or LEAN, has no bearing on the
control algorithm.  The ECU maintains an injector DUTY CYCLE, which is
constantly adjusted, in small increments, up and down, to keep the EGO
crossing the stoich point.  PROPORTIONAL control, possibly with an INTEGRAL
term, under some conditions.  That is the ONLY way it can know that the
AVERAGE mixture is NEAR stoich.  The control algorithm has nested loops, if
you will.  I suppose you COULD look at TPS response as a DERIVATIVE term,
but we digress...

New and interesting info:

> three way CAT PERFORMS BEST if the gas coming through it
> alternates between
> a bit rich and a bit lean.

I didn't know this!  But it has a ring of truth.

Some of this is splitting hairs, and watching dozens of angels dancing on
the head of a pin...

Best,
Steve

>
> The reasons why the oem ecu's oscillate across stoich:
>
> It allows use of a very cheap (in terms of ecu capability) bang/bang
> control strategy.
>
> It also allows a very cheap way of checking sensor performance (I believe
> this is an EPA requirement), by counting the frequency of crossings.
>
> By using such a crude control strategy, it also plays into the fact that a
> three way CAT PERFORMS BEST if the gas coming through it
> alternates between
> a bit rich and a bit lean.
>
> Nothing at all to do with sensor performance.
>
> Greg
>
> >At 11:59 PM 8/2/00 -0500, Tom Meagher wrote:
> >
> >>Oscillation duty cycle, averaged with a low pass filter to produce a DC
> >>voltage is evidently the operative parameter.  Perhaps you
> could sample the
> >>signal with the PIC and do FFT's on it to produce additional
> information to
> >>better characterize AFR.
> >
> >If all you wanted was an average AFR over a few seconds, go ahead.
> >It's a waste of time imho. It still won't be any more *accurate*
> >than an instantaneous reading due to the sensor's output being so
> >nonlinear, and varying with EGT, backpressure, etc, none of which
> >are compensated for by any amount of averaging over time. You'd be
> >averaging over many many cycles, slowing response time greatly,
> >and gaining nothing in the bargain.
> >
> >
> >>Does anyone know the physics behind the oscillating output?
> >
> >There's no physics of it, that's the ECU adjusting AFR to try and
> >maintain stoich, aka closed loop operation. Look at the timebase
> >on the scope traces. The sensors can show some interesting things
> >at much higher bandwidths, but those traces are not showing any
> >of them.
> >
> >Really this has all been beat to death in the archives. People
> >are almost lucky Gar is away for the moment. I can easily see
> >him dispensing a brace of new bodily orifices when he returns.
> >(And if you think I'm exaggerating, you haven't been reading
> >the archives enough.)
> >
> >   Chris C.
> >
> >-----------------------------------------------------------------
> -----------
> >To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without
> the quotes)
> >in the body of a message (not the subject) to majordomo at lists.diy-efi.org
>
>
> ------------------------------------------------------------------
> ----------
> To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without
> the quotes)
> in the body of a message (not the subject) to majordomo at lists.diy-efi.org
>
>

----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org