[Diy_efi] Re: Minimalist micro based EFI ?

[Bevan Weiss]

----- Original Message -----
From: "Mike" <erazmus at iinet.net.au>
To: "List for general do-it-yourself EFI talk" <diy_efi at diy-efi.org>
Sent: Tuesday, March 11, 2003 9:03 PM
Subject: Re: [Diy_efi] Re: Minimalist micro based EFI ?


> At 08:55 PM 3/11/03 +1300, you wrote:
> >It is most defineately possible, it's just the limitations that it puts
on
> >the features available.  ie the ignition/fuel table would have to be
small.
> >And it wouldn't be able to handle that many different operations
> >psuedo-simultaneously.
>
> mmmm, true, though if the critical timing aspects were more or less
> synchronous with eficient interrupt handling then the simultaneaity
> wont be that much of an issue... But the map resolution would be
> a difficult one to get around or alternative interpolation method.

Perhaps if you move away from the mapping method, and move more towards a
mathematical model type approach.  You've all seen on map table graphs how
everything tends to be rather smooth in it's transitions.  Perhaps if you
use coefficients and various mathematical models for the actual changes.
ie normally you've got the table Fuel (rpm, tps, map), then modified by some
other values for air temp, coolent temp, etc.
Why not just group it into a very large function Fuel(rpm, tps, map, air
temp, coolent temp, etc) where everything relates in a mathematical sense.
ie
Fuel = 0.06*rpm + 0.3*tps + 0.2*map etc etc etc.
Some more complicated models may be needed, ie squares and cubes, perhaps
even other polynomials.  This would certainly reduce the amount of memory
required.

> >Without all the extra hardware, ie hardware PWM, TPU etc then trying to
get
> >the performance that you desire out the processor will be that much
harder
> >to achieve.
>
> >From what I've done in embedded since late 70's - the processor
> most often spends a lot of time in idle loops or recalcing b/ground
> variables - I wonder what wew really mean by the term 'performance'
> and what sort of metric applies to a synchronous EFI data-model... ?

If you go to a mathematical model as before, then you could use up all the
'extra' processing cycles refining the coeffiecents.  Assuming some kind of
feedback system...


_______________________________________________
Diy_efi mailing list
Diy_efi at diy-efi.org
http://www.diy-efi.org/mailman/listinfo/diy_efi