Subject: EFI block I/O diagram
From: Rotary Engine
Date: 2/18/2007, 7:20 PM
To: AARotary Engine



     >
     >     Attached is the Input-Output block diagram for the simple EFI
    system.
     >     This diagram is intended to discover any missing sensors or
    engine
     >     outputs (i.e., other than for pilot information).
     >
     >     Please comment, especially omissions. I would rather provide
    for a
     >     sensor up front rather than have to patch or edit the code
    later on. I
     >     realize that not everyone will want all the sensors.
     >
     >     Fuel flow can be calculated from the fuel pressure given a
    calibrated
     >     hole in the slide throttle. The two pumps on each side
    (tank?) can be
     >     operated in parallel, i.e., both on all the time; in the
    event one
     >     fails, the uC can notify the pilot and increase the duty
    cycle of the
     >     remaining good pump in order to maintain proper fuel
    pressure. Or, a
     >     switch could be provided for the pilot to switch to the other
    pump.
     >     Builders choice.
     >
     >     The uC system will have total redundancy a la Tracy, i.e., an
    A uC and
     > a
     >     B uC. If one fails, switch to the other and land.
     >
     >     Doc Custer
     >     Retired Software Engineer and sometime Audiologist
     >
     >     -----------------------------------------------------
     >     Looks OK to me Doc. I am glad you did that diagram.
     >
     >     I am thinking out loud now.
     >     You can get fuel flow from the pulse width you are feeding
    the pumps.
     >     It must be calibrated because all pumps do not in general
     >     pump an amount of fuel directly proportional (in a linear
    manner)
     >     as a function of the the pulse width. Even tho the average
    voltage
     >     applied to the pump is directly proportional to the pulse width.
     >
     >     In other words applying a 50% wide pulse to a 12 volt
     >     pump is the same thing as applying 6 volts DC. In any event
     >     you may need a voltage regulator on the 12V supply. If the
    battery
     >     is 14 volts the pump gets 7 volts.
     >
     >     Another way of doing it is feeding the battery voltage to a 1/3
     > divider
     >     and feed the output of the divider to one of the A to Ds in the
     >     chip so the computer knows the battery voltage at all times.
     >     It then uses a look up table to adjust the fuel pump pulse width
     >     according to the available battery voltage.
     >
     >       This is the cheap
     >     way to do it as you don't need a high power voltage regulator
     >     on the 12 volt supply. If the battery voltage is low you widen
     >     the pulse to compensate. Pumps can take up to 7 or 10 amps
    and as far
     >     as I know there are no  single chips regulators that will handle
     >     that. Well at least the last time I looked. In any event you can
     >     display the battery voltage on the instrument panel as a trouble
     >     shooting aid.
     >
     >     The other possibility is to ignore the voltage level as most
     >     alternators have a regulator that is pretty much set to 13.5
    or 14
     >     volts or so. Assume that will always put you in the mixture ball
     >     park and the pilot in the loop will fine tune with the
    leaning knob.
     >
     >     The EGT is a bit hard to use as a feed back sensor as the
    response
     >     time is very slow and
     >     it changes direction as you pass through peak EGT. You are
    better off
     >     with the O2 sensor IMHO if you are looking for a closed loop
    system.
     >     I would forget the closed loop and keep  the pilot in the
    leaning
     >     loop. The engine will run over a wide mixture range. 10:1 to
    20:1.
     >     Air/fuel.
     >
     >     EGT rises as you lean the mixture and then drops as the mixture
     >     gets leaner and leaner past peak. A tough nut to crack as far as
     >     programming is concerned. You can lean the mixture and see if it
     >     rises. If it rises you are on the rich side of peak EGT. If
    it lowers
     >     you are on the lean side of peak EGT. This takes a lot of time
     >     in addition to the slow response of the TC. .
     >     the only way to speed up the response of a thermal couple is
     >     use a smaller one. That has life limit implications.
     >     --
     >     Paul Lamar ...No rotor no motor.
     >
     >
     > I thought of a way to handle the EGT issue a while back..   If
    running
     > auto fuel or 100LL+TCP, include a cheap narrow band O2 sensor to know
     > which side of stoich you are on and then use the EGT to set the
    mixture
     > from there.  The programming gets much simpler then.  The only
    remaining
     > problem I can think of is that as you change manifold pressure
    peak EGT
     > changes - at lower mp, there's less total energy in the exhaust
    stream,
     > at max boost there's more.
     >
     > I think the computer would have to either use a lookup table to guess
     > what peak EGT is at the current m.p.-rpm  operating point (probably
     > close enough) or periodically cycle the mixture to stoich so that it
     > knows what the peak temp actually is.  I'd prefer the lookup
    table method.
     >
     > I'd still have a mixture control..  But instead of it directly
     > controlling fuel flow it would control how far from peak EGT you
    are,
     > lean or rich.  Set the mixture knob straight up gets you max power
     > (50deg ROP), left is even more ROP, slightly to the right of
    straigth up
     > is peak, further to the right is LOP.  As you climb and descend the
     > computer keeps the fuel flow where it needs to be to keep that
     > relationship.
     >
     > Using the computer to accomplish this is nice because on the LOP side
     > the engine response curve is a much different shape (temp/power
    vs fuel
     > flow) than on the ROP side.  The control is very sensitive on the LOP
     > side - temp/power drop quickly for a small change in fuel
    flow.  On the
     > ROP side the curve is much flatter.  A mechanical mixture control
    while
     > LOP is always so twitchy that managing it precisely is
    annoying.  A good
     > job for a computer.
     >
     > The mixture feature is less important on a turbocharged engine
    which can
     > maintain constant MP over most altitudes...
     >
     >
     > Matt-
     >

    Thanks Paul & Matt for the valuable suggestions!

    Attached is an updated diagram with the changes you suggested.
    Notice the
    voltage sensor and the mixture control.

    Anybody have a recommendation for "a cheap narrow band O2 sensor?"
    Experience from the field is much more valuable than a catalog.

    Parts recommendations for all the sensors and actuators is greatly
    appreciated! Saves me a lot of time and results in superior choices.

    Doc Custer
    Retired Software Engineer and sometime Audiologist




I have not played with an O2 sensor but have read a fair amount about
them.  The narrow band flavor have a sharp change in voltage when
transitioning through stoichiometric mixture.  I think something like
this one would work properly:

http://store.summitracing.com/partdetail.asp?autofilter=1&part=CTO%2DO2%2D13&N=700+115&autoview=sku
<http://store.summitracing.com/partdetail.asp?autofilter=1&part=CTO%2DO2%2D13&N=700+115&autoview=sku>

That's summit racing http://www.summitracing.com part number CTO-O2-13 $25.

The input circuit will need to be fairly high impedance, to an A/D
probably would be best.  Or a simple comparator would work pretty well too.


Matt-


Thanks, Matt! I'll check it out. Could you please cite some of your
reading material? I am in need of an education. :)

Doc Custer
Retired Software Engineer and sometime Audiologist


The Rotary Engine NewsLetter. Powered by Linux.
ACRE NL web site. http://www.rotaryeng.net
Copyright 1998-2006 All world wide rights reserved.