Gads Spark attack-decay

[garfield at pilgrimhouse.com]

On Wed, 20 May 1998 00:08:44 -0400, "Bruce Plecan" <nacelp at bright.net>
wrote:

>On some ecms they have a spark attack, and decay rate.  Meaning
>the longer there is spark knock the more timing retard is applied,
>right?.  Also after some period of time that retard decays, and full
>timing is restored.  
>  So now if I had a perfect way of detecting knock, then I could instantly
>apply full, retard, and very very quickly let it erode right?

Hmmm, are you "waxing theoretical" on us now, Dr. Pelican?

You might be right somewhat, in that this "ratcheting" adjustment
downward of the timing, and then seeing if that helps, and if not,
further reducing the advance, MIGHT have something to do with the
imperfection of the knock sensor, but MAYBE an even stronger issue is
that the *presence* of knock doesn't tell you how "bad" things are, so
you don't really know how little/much advance decrement is needed to
"calm the raging storm".

I'm not sure that either acoustic or ion-based knock detection can tell
you well enough just "how bad" the knock is, so that you could predict
the *amount* of advance to remove to nip it in the bud in one fell
swoop. Hence, probably this "adaptive" routine of decrementing in small
jumps until the knock goes away is NECESSARY by the nature of the beast,
regardless of what kind of knock detection you have. I dunno, but that'd
be my scientific wild-ass guess.

Take as a mental exercise, the way ION works to detect knock. At it's
most basic level, you see that a plasma is produced in the cylinder when
it SHOULDN'T be there, and that's "knock". OK, then in addition, I've
seen in the patents that by possibly looking at the LEVEL of the
ionization current, you can tell how MUCH of a full detonation you're
seeing, like in the case of right after the spark, one patent alleges
that by looking at the level of the ionization current, you can tell if
the plugs are sooty or fouled somewhat, cuz the combustion won't be as
complete, hence the plasma won't be as dense nor long-lived. Sorta like
making it a trinary variable, so it's fire/sorta-misfire/really-misfire.

But I'd bet you're right on the bleeding edge to hope for some "measure"
of how BAD/MUCH detonation you're getting, and so determine from that
how much to back off on the timing. In lieu of that MEASURE, you just
have to take away some and see if that does the trick, and if not, take
away somemore. Same goes for returning to full advance. Consequently, I
think it's quite possible that even with much better knock DETECTION,
you still may not have any better knock MEASUREMENT, and so are still
stuck with the fairly stodgy and slow retard time constants. Besides, I
thot you weren't up for completely brain-transplanting the code in these
ECMs!? Seems to me, in order to change the rates at which the spark is
retarded/brought-back, you'd have to mangle the code, anywho?

>  Also on some engines there are some cylinders that are much more
>prone to detonation then others, so since no system is perfect, and
>simplicity is the key to design, would it make sense just to monitor
>those cylinders.   For lab use, and engineering stuff ya monitor all
>of them as close as possible, but for real world use what ya all think?

Well, me, I think the trick is KNOWING which cylinder is the bad guy. A
pretty impossible trick, methinks, overall. Piezo knock pickups are
placed on the case not so much nearest the known bad boy, as they are
attempting to find a spot that's sorta "equidistant" acoustically, from
all the cylinders, so no one cylinder has a louder say in who's
knocking. It's true in aviation engines (less so in water-cooled
automotive engines, but still partly true, no doubt) that given the
particular cooling/air-flow arrangements, etc. etc. there will be one
cylinder that's warmer than the rest, and that's the one that you watch,
but that all by itself is an AWFULLY simple model. And even here, the
hotter cylinder can change, depending on speed! Ooops, could ruin yer
whole paradigm-day, couldn't it? Also, lots of preignition phenom have
to do with carbon buildup in one cylinder, or a particular sharp edge in
one of the cylinder's that creates a slightly hotter spot than the
others, etc. NONE of these phenom can be predicted by "who's the
systematic bad boy of the bunch". You just don't know apriori. Soooo,
why skimp, when the electronics and the sensors are, especially with ion
detection, neither hard nor expense to have on a per-cylinder basis.
With acoustic pickups you're probably stuck with a SINGLE central
detector, or on the engines I'm familiar with, a pair of them, one per
bank of the boxer. But this is hopefully a "cost tradeoff" that ION will
allow us to avoid.

Bottom line, since yer polling the populace, is for me, I'd think it's
penny-wise/pound-foolish to NOT use sensors-per-cyl. And I'd bet if you
watched even the quintessentially IONIC Saab's when they start to ping,
I'll bet you'd see them ratcheting down the timing advance, just like
everybody else. Only in their case, they know alot more certainly WHEN
knock has started, AND when it's been quenched. A distinct advantage, to
be sure, but not necessarily getting away from the spark retard/advance
time-constants needed for "adaptive spark timing".

Just me dos centavos amigos. Operatin on a wing ana prayer in these
territories, I freely admit, but since yer solicitin opinions, there's
mine. You got what you paid fer!  B)

Gar