Custom EFI project

[Ric Rainbolt]

Hi.  I've been lurking this group for about a month and now I have some
questions.  I am building a custom EFI computer for a car of mine that
currently has mechanical fuel injection and a simple electronic-advance
system with distributors.  I would like to give you some background
information on my endeavours.

The engine that I am using as my test bed is an aluminum 3.0 litre 90 deg.
V-8, with aluminum two-valve DOHC heads.  I am currently in the process of
rebuilding the motor to strengthen the bottom end.  When done, it will have
a billet forged crankshaft, forged aluminum pistons with a 7.2:1 compression
ratio.  I am converting it to electronic port fuel injection and
distributorless ignition.  After I have the engine operating with normal
aspiration, I will be adding a pair of Garrett turbochargers (hence the low
C/R) plumbed through at least one air-to-air intercooler (air-to-water is
also a possibility).  Also, I am converting the lubrication system from wet
sump to dry, though this has no impact on the EFI! :-)  I will be using
eight Bosch fuel injectors (19lb/hr @ 3 bar).  Fuel pressure will be 3-6
bar, controlled by the EFI.  Oh, yea, pump gasoline 91 octane or better.
Redline will be 8500 RPM.  Boost pressure will be limited to 1 bar.
Expected output is 380-400 HP, up from 205 HP.

The CPU I have selected, due to exposure and access to tools, is a Phillips
SAB80C167 (or C167).  If you are not familiar with the C167, it has:  16-bit
10 MIPS sustained, 16-channel 10-bit A/D, 5 timer/counters, PWM, 2
capture/compare units, 50+ I/O lines, etc.  Should be more than adequate, eh?

Here's how I'm going to proceed (once the engine is rebuilt):

** STAGE I **

First, I'm going to get the engine to crank and idle (800 PRM).  At this
time, I will be using the stock ignition system, which is known good.  So,
basically all I'll need to implement is a group fired (1 bank of 8; i.e. all
together) injection scheme to meter fuel the same way an "electronic
carburetor" works.  The sensors that will be used are: Throttle position
sensor (TPS), system voltage (SYSV), exhaust Lambda (O2), top-dead-center
(TDC), manifold pressure (MAP) and engine speed (RPM).  At this time,
coolant and air temp will be ignored. Any guestimates as to initial pulse
duration (duty cycle) for the injectors to idle??

** STAGE II **

Next, I will create a "basic" fuel map that is linearly proportional to
manifold pressure and RPM (in other words, estimated airflow).  Then I will
run the engine at varying engine speeds from idle to 6500 RPM under no load
and tweak the fuel map to achieve stochiometry.  After this, "Throttle Pump"
and "Decelerative Fuel Cut" will be implemented.

** STAGE III **

Next, I will rewrite the fuel injection control to drive the injectors as 4
banks of 2.  The OEMs do this and I have read that, other than cutting the
driver count in half, it provides cooling for the intake valve by wetting
with the out-of-phase pulse.  Is this true or should I opt for true
sequential injection??

** STAGE IV **

Then, I will add adjustments for the fuel curves for changes in intake air
temp (IAT), coolant temp (CLT), just-after-start enrichment and system
voltage (SYSV).

** STAGE V **

Saturn V? No, stage five.
I will now to remove the stock igniton modules and connect them to a test
harness which will be controlled by a PC.  I will step through all MAP and
RPM values and record the OEM ignition advance map.  Then, I will program
this in as my "base" map into my EFI and test my EFI using the same harness
to verify the implementation.
Now to install my ignition components on the engine and say "goodbye" to the
distributors forever!

** MORE LATER ***

Any questions, comments, flames or information??  Any help welcome.  I'm
willing to share results of anything I learn from this, including choice of
electronics and sensors.

Ric Rainbolt
ricrain at computek.net