FW: Re: fuel map discussion

[Jonathan R. Lusky]

Ciciora Steve writes:
> I agree.  I'm not shure where to draw the line between fast and less readable
> and slower and easer for someone else to understand.  
 
IMHO, draw the line all the way over to the fast side :).  You can smooth
things out in the calibration software (looking at raw hex gets OLD
real quick :).

> > OT = BPC * 1/AF * VE * ATC, where OT = injector on time, AF = 
> 
> I believe the above approach above will work, but it is not how I think I
> understand it. I'm not sure why VE would be in it's own 3D look up table.  The
> VE of an engine is not something I have a good feel for, know how to calculate
> or know how to measure.  Please correct me if I'm wrong, but besides some
> sensors (thermistors) VE and the fuel injectors are the only non-linear things
> in the equation.  I envision a 3-D 'fuel map' (for steady state conditions)
> that wouldhave RPM on one side, a combination of MAP and air temp on the other
> side, and would output the base pulse width.

Take air temp out as a seperate fudge factor and you have basically what
I put above.  The VE table isn't really VE, its just close :).  VE is
really the ratio of the actual mass of the air drawn in to the ideal
mass of the air that would have been drawn in if the cylinder were to be
filled with air of the same density as the air as the inlet (of the
throttle body, I think).  In other words, the % full of air that
cylinders are.  Since you are firing the injectors once per intake
stroke, you don't need to know actually mass flowrate of the air, you
just need to know the mass of air in the cylinder after a given intake
event.  Anyway, the pulse width table you describe is basically a VE
table if you multiply it by the appropriate constant.

> Other factors (cold start, power,
> etc) would modify this base pulse width.

I agree.  I just think it makes more sense to fudge the base pulse with
air temp instead of fudging the MAP value with air temp.

> If it would help clear things up,
> mabe the fuelmap would output the amount of fuel (in grams or something) which
> would go to anothertable that accounted for the non-linear fuel injectors (and
> output the pulse duration in ms).

As long as everything is being multiplied, don't worry about carrying
through units.  Scaling the factors to fit in 8? bits with the best
resolution is whats important.  You correct the units with a constant
tagged on near the end (BPC in my equation).

> I doubt that this would help since the
> injector response time changes with age.  Na, forget what I just said about a
> second table to linearize the injectors.  Right now I don't think its a good
> idea anymore (tomorrow I'll change my mind... probably).

Actually it'd be nice to have some sort of injector correction table.
The way the EFI Technologies ECU does injector correction is just a
multiplier for each injector (1 = no correction...  thats a one on the
user end, not sure what the ECU gets fed), and minimum and maximum RPM
for the correction table to be used.  This will let you correct for
distribution problems (very useful in some applications with single
plane manifolds that have adjacent cylinders firing back to back).

> It it were not for VE, the fuel map would be a flat slab with no nice looking
> shape.  I think what I described is what you described with the VE map
> incorporated.  

If VE was constant, your pusle width would be constant except for
air temp correction and injector linearization.

> I think for 'tuning' an engine, having one 3D fuel map would be easer for
> someone to play with and see the results.  "Seems a bit lean between 4k and
> 4.5k rpm, I'll just push up these data points a little..."
 
Yeah, same thing as I proposed.  Except you also need to specify a MAP
range in addition to RPM or you'll end up making a mess (been there,
done that).

> Correct me if I'm wrong, but wouldn't you want 14.7 A/F most of the time (all
> rpms and pressures) except under cold start, power, etc?  Would my base 'fuel
> map' be to get you at 14.7 A/F and other modifiers pull you away from 14.7?
 
Yeah, most of the time you'd run at stoich.  Cold start, warm up, and
power enrichment will all be richer, and you'll probably want them all
in tables as F(water temp).  Your fuel map has all of the values
multiplied by 14.7 where my method multiplies by 14.7 (actually 1/14.7)
after pulling a value from the table.
 
> These are just my opinions, and I make no claims as to their accuracy.  I
> just made them up.

Ditto :)  Actually, I'm pulling a lot of this out of GM docs and the
calibration software for EFI Technologies performance-series ECU,
and I'm combining ideas from both (BTW, the EFI-Tech ECU has a table of
pulse width as a function of RPM, MAP instead of a table actually called
VE).

Now I finally remember what irk'd me about the pulse width table..  it
gives you a false sense of knowing whats going on--once the pulse width
has gone through all of the corrections, its ALOT different from whats
in the table.


-- 
Jonathan R. Lusky  --  lusky at knuth.mtsu.edu
 "Turbos are nice but I'd rather be blown!"
    68 Camaro Convertible - 350 / TH350
       80 Toyota Celica - 20R / 5spd