New list member with a project (very long)

[Andrew Brownsword]

Hello,

  I just joined the list a couple of days ago.  I am in the middle of a
project that I thought you folks might be able to offer advice on.  I tried
posting a description of this to the list already, but I think I got the
email address wrong last time.  If you get two copies, I apologize in
advance.  Before starting I'll just let you know that I'm a relative novice
in all things automotive, but am a software engineer in the computer games
industry so I know something about programming and microprocessors.  Anyway,
on with the adventure:

  My project car is a 1993 Ford Probe GT.  It has a Mazda engine, the KL-03.
This is a 2.5L DOHC V6, OBD-I, and of course EFI.  The airflow sensor is a
vane airflow meter (VAF), which is flapper door style.  The ECU is a version
of the MC68HC11, with an assortment of support chips.

  About 6 months ago I had a local shop install a turbocharger (Garrett T04E
52).  The original plan for fuel delivery was to use a CarTech adjustable
FMU.  That is what the car has been running on since then, but the stock
injectors turned out to be fairly anemic (220cc/min) and even doubling the
fuel pressure doesn't allow me to run at my target boost level (6-7 psi)
without going extremely lean.  In addition it raises the fuel pressure at
idle slightly, causing the car to idle more poorly than stock.  I've been
running at 2-3 psi while I work out a solution.

  Part 1 of the solution was to obtain higher flow injectors from the only
source available -- VENOM.  It took longer than necessary, but thats a
different story.  Suffice it to say that I now have 6 350cc/min injectors
sitting on my shelf.  The first attempt to install them ran into,
predictably enough, problems.  The stock ECU obviously doesn't understand
high flow injectors and thus the car can't possibly run without some
modification to the injector control logic.

  Part 2 of the solution, therefore, was to reprogram the ECU.  The initial
naive attempt was to modify the one table that the shop's ROM mapping
software knows about in these cars -- it turns out to be a Load vs. RPM fuel
enrichment table.  Unfortunately this table can't be tuned "downward", and
it certainly doesn't help with idle.  Clearly a more extensive reprogramming
was to be required, especially since completely replacing the ECU is not
something that I really want to entertain.

  My next step (part 3, if you're keeping track) was to obtain a bunch of
hardware and ROM information from a very helpful and knowledgeable fellow
PGT owner (who happens to be on the same PGT mailing list that I am on).
>From him I obtained a lot of hardware information, an initial ROM image, and
a few starting points.  I also got the ROM image from my own car and wrote a
MC68HC11 disassembler (rather cool, if I do say so myself), and another
program which makes scripted changes to a ROM image (it scales values
according to the injector ratio, recalcs the checksum, etc).

  I've spent the last month working backward from the functions which
actually control the injectors.  At this point I have figured out the 7
places from which the injectors are fired, and located a ROM value or table
which is used to scale the value before it is passed to the injector
functions.  Next week I'm going back to the shop and we're going to try
installing the program to see if it actually leans out the mixture like I
expect it to.  If it does we'll drop in the high flow injectors and
hopefully we'll be in business.

  The only (known) additional likely complication is that the stock airflow
meter (the VAF) may not be able to read values > 1 atmosphere.  If this is
the case I'm going to need a supplement or replacement.  I have several
options at this point:  piggyback computer, alternative VAF, lean the
injectors out a bit more and keep the FMU.

  There are two piggyback computers that might do the trick.  Neither of
them support this engine directly, but there is another PGT owner out there
that has tried them both.  They are the SuperAFC and the Link AFM.  The Link
AFM includes a MAP sensor, the SuperAFC purportedly has inputs for a MAP
sensor.  Both of them modify the VAF's input voltage to the ECU, tricking it
into a longer duty cycle.

  There is one possible replacement VAF:  2nd gen 87-88 RX-7.  It looks like
exactly the same design, but is used in a car which runs about 6 psi.  This
seems like a very appealing solution if the VAF on the PGT can't handle
boost.  Hopefully we could just recalibrate the VAF's interpolation table in
the ECU and thing would work (yeah, it sounds optimistic to me too).

  Keeping the FMU is my least favourite solution, but it is the easiest one
and will likely work in the short term.  By telling the ECU to lean it out a
bit more we can compensate for the slightly higher idle fuel pressure that
the FMU causes.  Basically its a hack until a better solution is available.

  Whew, this turned out WAY longer than I intended.  Thanks for reading this
far, and if you have any suggestions, tips, experience, advice, comments,
insights or trivial... please let me know.

  Andrew Brownsword

PS:  The end result is definitely worth it.  The KL-03s biggest problem is
that it doesn't breathe well and delivers only about 135 hp to the wheels in
stock form.  We dyno'd mine @ 5psi up to 5500 RPM where it really leaned out
and measured 210hp at the wheels.  Even at only 2-3 psi it hauls.