[s-cars] Fuzz logic vs adaptive learning

Kirby Smith kirby.a.smith at verizon.net
Mon Jan 26 09:47:39 EST 2004


Thanks, Scott, for your insight.  

I think it premature to argue that Motronic is more sophisticated than
the A'pexi, for example, without our knowing just what functions are
embedded in the A'pexi.  A large-valued n-dimensional table is not
necessarily more sophisticated than an n-dimensional high-order
function.  Unfortunately, I don't know what is in the guts of the A'pexi
programming.  If you have some specific knowledge of the A'pexi inner
workings, feel free to enlighten us.  For now, I will accept that it
cannot know, or at least react to, engine degradation, but as a WG
controller for a functional engine (running Hoppen Stage 1), IME it does
very nicely.   

Do I agree that optimized Motronic code with your favorite waste gate
valve (whatever that is) can do as well or better with less engine risk
-- Yes.  But to argue for engine protection over tranny protection is
short sighted, IMHO.  It would cost me a lot more for a replacement
tranny than a valve job.  And burned valves would require a lot of
detonation, which I would hear early and my right foot would then become
the overriding WG controller.  Now when running with an open exhaust in
competition, I agree the driver is not a good WG controller.

kirby


QSHIPQ at aol.com wrote:
> 
> Please
> Don't do anything for my peace of mind Kirby.  My modus on these lists is to
> provoke thought and alternative views to skin cats based on my experiences.
> What I've seen is a lot of copying:  Software, hardware, controllers, turbos,
> etc.  That all assumes what others have done is the right or only way.  Hap
> should be the first to admit that what he has isn't sorted, which means "right"
> is in a context.  In the case of the HP wars, I suspect more HP will come at
> the expense of less comfort in doing so.  It's the nature of 2.3liter motors.
> 
> In the context of controlling boost, the matrix in the motronic is bigger
> than the best fuzzy logic controller on the market today.  Which means to me,
> that I take great peace of mind in knowing that those without controllers are
> doing more for the life of the engine, than one possibly focused on the first
> gear boost levels of an S car (lots of torque, lots of mass, lots of boost, lots
> of traction = something gives).
> 
> Also knowing that most of the software tweeksters we are familiar with are
> lurking on these lists, means you have the makings for better programming and/of
> better hardware.  Let's also not forget that previous to the motronic, the
> MAC 11/14 computer controlled boost was 'top feeder', so we know it does work on
> the I5.
> 
> To date, I see a lot of money spent on hardware that could be software, or in
> software that could be hardware.  Just because an external boost controller
> makes more boost sooner, doesn't make it a better tool.  It's quite possible
> the tool one is using is more sophisticated at the task (by matrix definition -
> motronic is), it just may be hampered by a software OR hardware problem.
> Either/both of which could be pretty easily addressed IMO.  That said Kirby, I'm
> not sure that a top feed control boost is necessarily better than a bottom
> feeder (knowing the hi/lo nature of frequency valves in general), but certainly
> Minhea confirms that it's not that big of a deal to test it.  Me, I'd rather see
> someone program in a stepper motor to a bottom feeder WG, but that's just for
> my peace of mind.
> 
> IMO, no boost controller on the market today works harder than Motronic,
> getting motronic to work smarter is the trick.
> 
> My .02 arbitraged thru the peso
> 
> Scott Justusson
> 
> In a message dated 1/25/2004 7:08:26 PM Central Standard Time,
> kirby.a.smith at verizon.net writes:
> Mihnea:
> 
> In my spare time?  :)
> 
> Thanks for the offer.  I think such an approach would have to start with
> determining what WGFV would be best to use.  There have been comments by
> Scott that the OEM one is rather sluggish and has limitations in its
> duty cycle bounds.  Perhaps the A'pexi valve (already installed) would
> be faster/better ranging.  I don't know whether it could be adapted but
> I could measure the voltage it gets when it is considerably warmer
> outside.  Does the OEM WGFV see an approximately 12 Vdc signal or is it
> driven from a current limiting circuit?
> 
> I left the OEM valve connected to the ECU with caps on the ends so it
> would remain clean inside.  If it is disconnected electrically, rumor
> has it a code is thrown.  If the OEM WGFV is good enough, it would
> certainly be more widely applicable.  I guess I wouldn't terribly mind
> having the A'pexi reduced to an ultimate meter to gain Scott's peace of
> mind.  ^_^
> 
> But, the Motronic approach does beg the question of first gear boost
> limiting.  The Motronic would have to compare vehicle speed with rpm to
> switch maps.  Or, I could just wimp out in first and never, ever get
> carried away.
> 
> kirby
> 
> Mihnea Cotet wrote:
> >
> > Kirby,
> >
> > The Motronic uses a 10x16 points boost map, as well as a 10x16 points WGFV
> > duty cycle map. The boost values are "target values" and the FV duty cycles
> > are set to reach the desired boost levels with a given WG spring tension.
> >
> > Programming the Motronic for top feed control sounds easy enough to me as
> > long as a WGFV (or sort of) is still there, it's a matter of about one hour
> > of real-time tuning with my on-line emulator plugged into the ECU. Let me
> > know if you feel like experimenting this spring :-))))
> >
> > HTH,
> >
> > Mihnea
> >


More information about the S-CAR-List mailing list