Ignition timing in conjunction with

[tom cloud]

>
>>There appears to exist a number of sensors to build
>>an excellent self tuning EFI, 

        [ snip ]

>I have thought about this an awful lot, and plan to put
>some of my ideas into practice, once I figure out how
>to actually build the thing...

I 'spect everybody on this list has thought about this.  I
guess maybe I'm too conservative, but 'sadistics' (i.e. statistics)
says that the closer you gets to perfection, the more effort you
expend to get less -- law of diminishing returns.  I figure a lot
of this has already been solved.  I'd like to not have to repeat
probably tens of thousands of hours of research and engineering time
trying to come up with the same answers.

Here's some of my questions:  why do you have to re-calibrate if you
change your throttle body size, or injector, or whatever else?  I mean,
I know why, but why can't the bloomin' confuser figger that out for me?

Here's my proposed solution:  the EGO is purty good at stoich, right?
And, if you'll read the data, it's really not too bad off stoich, it's
just that the curve gets real steep and is probably not perzactly the
same from one to the next.  So, why can't you sense when you're in
'cruise' (i.e. small load -- probably determined from MAP, RPM and
TPS), find stoich by dynamically changing A/F and remember that?
Shouldn't take long to store a table of these points (and each point
would be 'refined' by averaging everytime the same point was experienced).

Then, rich / lean conditions aren't too far off (EGO-wise).  Probably
the best way to determine lean (if you're going for max MPG) would
be to measure EGT -- probably'd only need to measure one cylinder.
On my motorcycle, I put a washer with a thermocouple silver soldered to
it under one spark plug and was able to see what I thought were useful
operating temp changes.

Now, rich is a real problem for me.  It's when you wanna "GO" !  So,
this is where I thought you'd measure acceleration or torque (strain
gauge on motor mount).  But, Todd says no.  I don't know how he would
do a real time feedback system, but he apparently thinks a knock sensor
is good for this.  Mebbe so.  Seems to me this is more a function of
timing than mixture.  [Now, there's a variable I haven't been considering
-- i.e. timing!!]  All I'm thinking of right now is A/F ratios.  The EGO
does give data either side of stoich -- according to the data stuff
I've seen, its repeatability is influenced mostly by temperature,
but you should be able to use it to get you purty close -- certainly
to the point where 'diminishing returns' means you'll spend more time
screwing with it than driving it.  [I know I can get 743.375
horsepower, but if I could make 740 and spend 6 months and $5,000
less, could I forget about the 3-3/8 ponies?  I dunno.  I'd sorta
like to have 'em all!  But, then ....]

The answer to a lotta this seems to me is that we miss one of the
most important feedback elements:  the driver!!!  See, this is where
I come back to the question:  wadz rong wid doz liddl pots under
de dash?  I know you digital jocks hate the thought of an analog
pot -- so, dammit, put a grey code encoded disk out there (and spend
a couple hunnert extra to prove your point!).  Bottom line:  the
confuser can make lotsa decisions, but *YOU* can help it.  Now,
the prob here is that lotsa of this stuff is reel subjective -- like
you can't really tell when you're making max horsepower (you CAN tell
if it's way off, but you can't peak it 'cause it's probably changing
purty fast).  So, I don't have answers.  Just questions.  Seems like
there's gotta be some way the driver can easily provide some limited
feedback to his 'self-teaching' confuser widout setting up a *&$^%&*
laptop and trying to type on that &*&$^()) thing at 140 mph.

(You guyz gotta understand:  the last time I laid a wrench on my steed
was probably in April.  I work lotsa hours.  Not looking for sympathy,
just saying, I guess, that if I had a lotta time and money I'd probably
go for that last .01 ponies.  Certainly like to think about it.  I'm
pretty good at micro-confusers and software.  I've been around long
enough to not be scared of analog -- in fact, I often see things that'd
be better done in an analog way.  What I don't know about is the fine
points of engine control -- esp. the doo-dads used to sense things
and what each of those variables means.  I have done enough engineering
to recognize that, considering IC engines have occupied lotsa time
and money and there sure is lotsa high level stuff written about it,
this must be a pretty deep subject.  So, you guys that DO know,
jump in here and help us out!)