Thots cont'd

[Mike Palmer]

Somebody wrote :-)...

>> 02 sensors react relatively slow - maybe 10 readings a sec? 
>> TPS - sensors - at least 10 readings a second if not more
>> Temp sensors - 1 reading a second
>> RPM sensor - at 10,000rpm (why not?) - that 600,000 degrees per second.  4 
>> bumps on the crank shaft (ie - crank triggered ignition) would mean less 
>> than 1000 per second...
>> MAP or MAF sensor - 1 per second - maybe  more... 10?
>     
>For temp, intake air temp would also be good.
	 
For rates-of-computation, I think the P4 in my Z24 does portions of
it's fuel calcs once every 12mS (80+ times per second).

BTW, it's 60,000 degrees per second, not 600,000 :-)

For speed/density systems MAT is not "good" it's mandatory. Air
temp is an important player in the S/D equation. 

>> The system would have to work at:
>> Startup
>> Idle
>> part throotle cruise
>> WOT
>> anything else?
>     
>How about deceleration fuel cut.
	 
It may be necessary to also consider wall-wetting constants to
prevent lean-out and flat-spots on post-decel tip-ins.

>One thing I have noticed that has been missed so far is battery voltage.  
>You need to measure this to correct the pulse width sent to the injector.  
>Perhaps there are some new injector driver circuits out there that do this?
>
>I have been using the MC 3334 form Motorola.  

Yep. It will be necessary also to scale battery volts into the
fuel delivery. One way to establish a crude correction table would be 
to have an injector fire into a measuring crucible for a specified
number of pulses at a constant fuel pressure and pulse-width. For
various battery volts from 6.0 to 17.0, measure the quantity
delivered and then apply a scale factor to the applied pulse width
based on the battery voltage to achieve constant deliveries.

Don't forget too that injector operation takes time (i.e. for the 
solenoid to actually pull the injector open). At higher engine speeds
this may become a factor as far as timing goes. This is why
I like PFI systems conceptually over SFI systems - they are
less timing sensitive. (Some of the injectors on my 2.8 fire
onto the backs of closed intake valves...).

>My experiments so far are probably a little crude to what I have read.  I 
>have been using a PC parallel port connected to a A/D converter and then 
>used a lookup table to calculate the pulse width.  To date the system has 
>idled the car, but fails to accelerate properly ( engine stumbles and dies 
>).

Crude? Not at all! Actually sounds cool! 

I know nothing of your system, but can I suggest a few ideas?

1) Filter your TPS and MAP readings using a first order low pass
algorithm. It's a little esoteric but it will make the system less
susceptible to transients.

2) You'll need to have a process that monitors the delta-MAP and 
delta-TPS values (i.e. delta = filt(value_now) - filt(value_last_loop)). 
You can use these delta's to adjust the value of an asynchronous 
fuel accumulator (AFA). The AFA would be used to adjust injector on-time
to mimic an accelerator pump. Your system probably already
increases fuel-pressure automatically as a function of manifold
vacuum so a look-up table may be necessary to scale the delta
contributions accordingly. The AFA would normally add 0mS to the
injector on-time when del-MAP and del-TPS are low or zero but
when the AFA is adjusted by higher deltas, it would add *or subtract*
on-time accordingly. Subtract? Yes, to account for lift-throttle
conditions. Pretty complicated no doubt.
	 


--

- Mike