Some ideas...

[Dale Ulan]

> type, but I have seen some messages where people have stated that they wish
> to use a speed density system. Why?

S-D is more reliable. You also need to calibrate both of them, no matter
what, since MAF has to calculate reversion.

> mass of air curve? If I understand what I've read correctly, the computer
> determines the current necessary to maintain a constant temperature accross
> the wire in the sensor. Is this a part of the controller, or is it done
> automatically inside the sensor? What sort of support circuitry would be
> required to use a MAF sensor if it doesn't have the built in "smarts"?

A GM MAF sensor gives a frequency output, which can be measured using an
input capture function on the 68HC11.
>  
> Now for O2 sensors. I've heard (incorrectly?) that Honda uses an O2 sensor
> that gives you a better idea of the mixture, rather than just rich or lean.
> I'm sure that it would cost quite a bit (if this puppy even exists) but I
> have a feeling that it would be worth the extra effort and cost to implement.

It costs about $400 Canadian, and uses a wierd driver circuit. I don't
know exactly how it works....

> Lastly, how many timing marks should the wheel (or whatever) have to provide
> accurate info for the computer. The article on a distributorless ignition
> system from PEM uses only 8 marks. At say 8000 rpm, how accurately can the
> computer determine the angle of the engine, and how accurate does it need to
> be? One degree? Two? Would it be better to have many marks, say each
> representing 2 crank degrees, so that the computer doesn't have to
> constantly calculate the speed of the engine? Instead, all that you would
> have to do is tell it to count so many marks from a known positition,
> activate injector number 1 for x number of counts (depending on MAF and RPM
> etc), count Z many more marks, fire coil, repeat for next cylinder, etc...
> Which would require less processing power? I have a feeling that up to a
> certain point (RPM wise), one would be better than the other.

If you know the speed of the engine, degrees can be translated into
timer ticks, which is how most engine computers do it. For example,
a timer tick is 250 usec on my ECM, so if it takes 12000 timer ticks per
cylinder on a 4-cylinder.... well, you can easily figure out speed with
a division, and the spark angle is translated to, perhaps, 10000 ticks
after the first reference.

If you want to use a 68332 processor, you *could* use a degree wheel, becuase
in a 68HC11, the processing overhead of too many ticks can get heavy. I
found that more than 8 ticks per revolution tends to slow things down
too much. 
>  
> Does any of this make any sense, or am I way out in left field playing my
> own little ball game?

You sound like you're kinda on the right track. The biggest thing is
different processors have different timers, so the HC11, which is
useful for almost all ECM needs, lacks in here...
the 68332 series looks NICE. I was going to try one...

-Dale