simple analog fuel injection comments from CSH, HQ

[Raymond C Drouillard]

I have been watching the analog EFI thread with interest.  I haven't
commented 'til now.

A few good ideas have spun off, like using itty-bitty controllers do to
the job.  Also, there has been some verbiage expended towards explaining
why it will be hard to do, or why it shouldn't be done.  I can appreciate
an experienced person offering the fruits of his own errors, but I don't
like to see a willing experimenter discouraged from his experimenting.

Anyhow, I have a few suggestions that may be useful.  It sounds like an
interesting project.  I would try it myself, if I needed another hobby :)

I will keep this fairly general.  No guarantees - these are just ideas
that you can try.  It's fairly long, but worth the reading if you want to
try this.  You might need to look up some of the circuits that I have
mentioned.  Let me know how it goes, OK?


First of all, visit Radio Shack and pick up those little "engineers'
notebooks".  There is one on op-amps that is excellent.

It should be noted that op-amps were first designed as voltage adders for
analog computing applications.

I like the idea of pulsing the injecter with each spark.

It shouldn't be difficult to use a high-impedance amplifier to pull the
signal off of the relucter or points.  If you are using a relucter,
you'll probablly want to use a differential amp.  Feed the signal through
a capaciter to detect the leading or trailing edge (whichever you
prefer).  Feed resultant signal into one input of a compariter, and use a
pot to set a referance voltage that is above the noise floor.  The result
will be a digital signal - one pulse for each sparkplug firing.

You can use an op-amp as a compariter.  I have found that a real
compariter is faster, but that probablly isn't an issue in this
application.

Use this signal to trigger a ramp generater (sawtooth wave generater). 
When the pulse triggers the ramp generater, the output will start at zero
volts and rise linearly until it hits the supply voltage or is
retriggered.  Use a pot to adjust the slope of the ramp.  This will be
your overall mixture adjustment.

If you feed the ramp into the "-" input of the compariter, and a variable
reference voltage into the "+" terminal, you will get a pulse for each
pulse that triggers the ramp generater, and the width will be
proportional to the variable voltage reference.  A higher reference
voltage will yield a wider pulse.

The output of the compariter can be amplified with something like an
emotter-follower amp using a 2N2222 and a 2N3055 in darlington
configuration.

Now, let's discuss the reference voltage (master reference voltage).

You'll want to mix several signals together to get the master reference
voltage.  You can use a bunch of op-amps to make a mixer.  Each of the
signals will be weighted.  The varying of some will cause the reference
voltage to change more that the others.

For the simplest controller, use a MAP sensor to generate the reference
voltage.  As the Manifold Absolute Pressure goes up, so will the
reference voltage.  This signal will have the greatest effect.  Your
system will probablly work somewhat satisfactorily with just this signal.

You'll want to add other signals to make the vehicle more drivable.

Use the MAP signal as a reference for mixing the other signals.  All the
other signals will either be a little above or a little below he MAP
signal.

You'll want to simulate an accelerater pump.  To do this, you'll need a
TPS (throttle position sensor).  This is simply a pot hooked up to the
throttle.  Hook it up so that increasing the throttle opening will
increase the voltage.  Use a capaciter and a resister to create a circuit
similar to the VOX circuit on a transmitter.  Rapidly pushing the gas
petal will cause a brief voltage spike that is mixed into the master
reference voltage.  If you're worried about emissions, or just don't want
a black puff of smoke when you suddenly get off of the throttle, let the
circuit work both ways so that lifting your foot off the throttle will
cause a negative spike.  You'll want to play with the time constants in
both directions.  You'll need a couple of diodes in this circuit.

If you want to get really fancy, you can turn the ignition pulses into a
square wave, and run it through some filters with a frequency response
that resembles the torque curve of the engine.  Rectify and smooth this
signal (detect it like an AM radio signal) and mix it into the master
reference voltage.  I wouldn't bother with this one, especially if the
engine has a fairly flat torque curve, or you want to add the following:

Probablly the best thing to do is to build a feedback circuit using an O2
sensor.  Amplify the signal from the O2 sensor and use it to try to
charge or discharge a capaciter.  Mix the capaciter's voltage with the
master reference voltage.  When the sensor detects a rich mixture, the
capaciter will slowly be discharged which will lean out the mixture. 
When the mixture goes lean, the sensor and amp will start to charge the
capaciter.  Again, you'll need to play with the time constants.

You'll want to build this thing wit a lot of well-labelled pots that you
can play with.  Also, whether or not you want to go with feedback, an O2
sensor will be useful for tuning.

Ray Drouillard, BSEE

PS:
If you want some diagrams of the drivel I hage handed out, send me email.
 It probablly won't be fast, but I should be able to scribble something
out and either scan it or fax it to you.


_____________________________________________________________________
You don't need to buy Internet access to use free Internet e-mail.
Get completely free e-mail from Juno at http://www.juno.com
Or call Juno at (800) 654-JUNO [654-5866]