Intro.... questions & thoughts

Howard Wilkinson owly at mcn.net
Fri Apr 23 16:59:10 GMT 1999


Hi:
    I briefly participated on this list a couple of years ago..... I'm
not really in the same league as you electrical wizzards so a lot of
the circuitry and stuff is way over my head.  I've recently become
interested in simply building FI systems based on factory parts which
is without a doubt the simplest program for such people as myself.
    Some inexpensive ECUs are already out there and some documentation
is available on some of them such as the unit used in the GM Geo
series cars.  These units are being used with different injectors and
fine tuning of fuel pressures to operate a wide range of engines.

    None of this is I'm sure news to anybody on this list.

    I am specifically interested in information on the systems used on
snowmobile engines with a view toward adapting them to the Rotax
aviation engines using some of the same principles.
    The fellow who has done most of the work on the Geo ECU
conversions (recently deceased) claimed that this ECU would not work
dependably on small engines due to the poor quality of the power
produced.... voltages that were inconsistent, etc.... nothing that
couldn't be overcome by the use of a battery.  However it makes more
sense in my opinion to work with an ECU and the various parts designed
for use in this environment.
    If anybody could steer me in a good direction to get started on
this sort of project it would be greatly appreciated.

<><><><><><><><><><><>

Thoughts on programming DIY EFI systems:

    Over the past year or two I've given a lot of thought to your DIY
EFI projects and the programming of the ECUs.  I've reached some
conclusions in my own mind which I will relate here.

1:  Fuel flow in general is proportional to airflow, thus a MAF based
system should be the prefered form.  This is born out at least to my
mind by the Bosh system used in the VW Rabbit which is a totally
mechanical system based on a mechanical MAF sensor and modulated only
by engine temp which counteracts the motion of the mechanical MAF
sensor.  This system works very well and dependably and produces good
power and good economy.
2:  Most EFI systems are batch fired.... thus injector firing rate is
irrelevant unless it goes below engine RPM
3:    Due to the above it should be possible to work with a fixed
injector firing rate, or one with only a couple of steps.  This
reduces the number of variables in the program.
4:  Thus one should be able to operate an EFI system based on MAF
modulated only by ECT  and atmospheric pressure MAP which could modify
the MAF signal appropriately.  This would allow the use of a single
simple lookup table which would only contain pulse width figures
related to MAF data.  It is not really unrealistic that one could
modulate fuel pressure directly based on ECT and MAP so that a very
simple program could be used.
    Perhaps this is a bit simple minded, but experience has shown me
that if something is simple it is more likely to get done.

5:  A properly designed system should allow a self programming mode in
which feedback from the OS would modify the lookup table as required
to fine tune itself, also a simple user lean/rich shift button could
contribute significantly to initial programming.  Ideally flash memory
or something similar would carry the actual program information so
that it could easily be modified, though a EEPROM should be available
to reload the original program.
    With the above strategy the developer could begin with a simple
program and use the self programming mode to add layers of complexity
as needed.  In fact if one used the right programming strategy it
would only be necessary to get the engine running at idle after which
the envelope could be expanded by feedback which could fill in the
lookup table.  Your lookup table begins with a single pulse width.
When the airflow increases the feedback loop detects a lean condition
and changes the value incrementally until correct mixture is achieved.
At some point the firing rate may need to increase.  You set that
point and self program up to that point.  At that point you halve the
value in the table and continue on......
    Atmospheric pressure should be a linear effect so that a simple
mathematical formula applied to the MAF value should be able to
correct for elevation, or it could operate on the lookup table value
prior to being used.
    ECT effect may not be so simple, but is probably pretty close.

<><><><><><><><><><><><><>

The above are just some thoughts.... I suspect most people are working
on timed injection these days....something I consider unnecessary.  I
welcome any comments on the above thoughts.  I believe I could handle
the programming of an ECU though the building of a DIY system is
beyond the scope of my abilities.... except perhaps to hang pieces
together.  The matter of crafting a program with all the ins and outs
really doesn't look like an insurmountable obstacle to me.  It does
however boggle my mind to think that anybody could take ROM code and
reverse engineer and modify the program without source code..... Wow!
..... I know it's been done.
                                                        H.W.






More information about the Diy_efi mailing list