More Questions on cranking

[John Dammeyer]

>
>Date: Thu, 05 Aug 1999 03:31:24 +1000
>From: Phil Lamovie <injec at ains.net.au>
>Subject: More Questions on cranking
>
>Hi All,
>
>John wrote :-
>
>>I thought one of the injector's jobs was to atomize the fuel though.
>
>atomize =  very small drops   vapor = molecules

>
>Petrol drops burn like a candle. They have essentially two layers of
>combustion the  ratio of which  is controlled by chaos. As the drop
>surface vaporizes
>there are instants of both rich and lean combustion in very pretty
>patterns. The resultant gases are all bad news. Keep in mind the
>surface area/volume ratio.


Got it!


>Don't even look at the Map sensor during cranking the signal from the
>sensor
>is useless as the engine is "cogging"  and slowly to boot. The signal
>has both positive and negative excursions. For a high class test fit a
>vacuum gauge, and  turn the engine over on the starter  without fuel
>or spark. Don't borrow a gauge you can't afford to replace.
>
>Pick an appropriate amount say 7 ms and just squirt. If the engine
>fires it may go
>from 0 vacuum to 70 kPa  and 600 rpm in 5 or 6 revolution.


Good point.  My starting algo. has a pulse width from 10ms with MAP ==
Ambient to 4.5ms at 8".  So The cranking gives it a couple of shots of
rich mixture and once the engine accelerates the mixture leans out to my
cold running value.  Hot starts are even more efficient and start with
only a tiny puff of black smoke.

>
>If this aircraft is flying above 5000 ft you would be well advised to
>reconnect the throttle body water supply. This will prevent freezing
>as the aluminum will out shrink the brass and lock it solid. This will
>make landings difficult and/or messy.


I'm not sure the throttle body water supply actually warms the throttle
body.  What happens as far as I can see is this:  Bi-metal strip inside
assembly when cold allows a certain amount of air to bypass the throttle
plate.  As the engine warms up the coolent warms the strip which slowly
closes  (or reduces the opening anyway) and in essence reduces the idle
speed.  Connected to this interesting assembly is the idle air bypass
motor which doesn't allow air past the throttle plate but allows warm
water past the bi-metal strip which opens and allows air past the
throttle plate.  Convoluted?  Yes.

>
>Check your default for O2 sensor failure in closed loop and make sure
>it's not lean.
>Aircraft  engines have a habit of operating under much higher average
>loads than cars.
>Thus a 70% load is not unusual. In most cars the average engine load
>is 10-15%
>This is not an appropriate load for 14.7:1 A/F in closed loop. If you
>have a broad band sensor then you will be able to target 12.5 - 13.0:1
>A/F ratio.
>
>Under no circumstance should the engine boost at A/F of more than 13:1
>it will
>simply melt after a few minutes. Imagine under what circumstance you
>would keep the throttle of a turbo car at 3/4 for 3 or 4 hours in
>closed loop.



Boy you've got that right.  I mades sure that the engine ran slightly
rich (1100 degree EGT readings) open loop.  Then when I enabled the O2
closed loop,  the mixture leans out and I reached 1350 degrees.  I'm
told they like to run the engines at 70% to 80% at about 1450 degrees so
I have some room for tweeking.

The ambient air pressure sensor should handle changes in altitude to
ensure that the mixture doesn't go lean or rich with changes in altitude
or weather.

>
>I know that auto makers are "dumb" but they have read the Bosch Blue
>Book.


Bosch Blue Book????

>
>And the cost of the hoses was less important than a throttle stuck
>after a long run on the highway in freezing conditions. Remember that
>partial vacuum = cooling just as increased pressure = heating


Good point.  I know there is warm water in the intake manifold,  I'll
check into how far that heat goes.

>
>By the way how did you determine the injection amount vs map points
>for rpm vs load at 50, 100, 150 etc. rpm ?
>


I worked backwards.  Not knowing the size of the stock Honda Injectors I
made assumptions on Honda's  HP and Torque ratings along with
displacement and RPM.  Then calculated, based on what I figured the
injectors should be, and installed a series of percent values into a VE
table indexed by RPM.  I have an initial mixture based on RPM;
cranking(9%), idle(14.5%), running(14.7%) and accelerating(?).

MAP/Ambient modfies the mixture.  The VE value also modifies this
mixture.  There is also a cold enrichment based on engine temperature
that decreases as the engine warms up.  Once the engine is warm the O2
loop closes.  If acceleration occurs (TPS change),  the O2 loop is
broken until either n strokes or m milliseconds.  Haven't got the Accel
quite right yet; engine stumbles a bit under really quick throttle
increase.  Also my O2 loop is only using the P of a PID algo.  I'd
actually like to change that to a fuzzy logic algo.

I also used a number of different articles from Bruce Bowling,  Al
Grippo,  Ed Lansinger as the starting point.

>
>If you were firing the injector for 12 ms at 6000 rpm you were
>probably
>in another dimension.    Lets see 6000 rpm = 100 rps or 1 revolution
>every
>10 ms. I think you were missing injection cycles because you can't
>start the
>new event until you finish the old.



OH! I get it.  This is why you suggested injecting during the exhaust
cycle;  you inject once every revolution.  It brings up an interesting
question too.  When an injector is rated at 80% duty cycle to avoid
overheating  I've often wondered: 80% of what.  80 minutes on for every
100 minutes.... silly.

I'm running a 720 degree engine cycle.  At 6000 rpm this is 3000 CAM RPM
which is 20ms per revolution and 12/20 is 60% duty cycle.  My run log
for this afternoon shows 6200RPM and 12.49ms duty cycle which is about
65%.  O2 sensor was at exactly 0.5V and this stayed steady for about 2
minutes.  No EGT sensors available for this run.  We were turning a
propellor of undetermined origins but the injector pulse width suggests
that we were not at the maximum horsepower compared to our Dyno runs.  I
would have richened up the mixture at 6200 RPM had the muffler not
fallen off and melted through the fuel pump wiring.  sigh......  second
time today.  For the next test we've added bailing wire.  ;-)

Tomorrow we will finish installing the ducted fan used in the
hovercrafts.  With Allison Throttle Bodies or SU carburators we've
achieved 125HP but the SUs aren't really good for airplanes and the
Allisons are expensive and require the pilot to adjust mixture at
altitude.  My barometric pressure sensor should dispense with that
problem.

>Why not use a single hall and a chopper with a differential width for
>the TDC
>indicator. Start a timer with the rising edge and stop it with the
>falling edge.
>The short one is TDC.


We ran into a physical size problem for mounting the sensors etc. My
first proto. used a slotted wheel exactly as you described for initial
board and algo testing on the electric motor drive.   Our reduction
drive is at the Flywheel end of the engine and that's where the
distributor used to be.  We now have an end plate over the hole.  By
switching to small magnets and latching hall sensors we were able to use
our older model ignition with mechanical fuel injection (allison
throttle body) (or SU carb) by only changing the position of the
magnets.  I then don't have to try and figure out by checking timers for
short verses long on edges etc. as to where TDC is.  My sensors are
configured as a 2 bit Grey Code absolute position encoder.  Interrupting
on the edges therefore tells me TDC for each cyl.  i.e. the two sensor
values translate directly into a 2 bit integer representing the
cyclinder #.


>
>If you leave out the units you won't pick the core issue


Understood.

>
>The smoothest idle has the smallest cycle to cycle variation in
>combustion and that is hard to control if your target A/F is stoic. If
>you vary even a little say 15% leaner you will get a miss fire. This
>will decrease the vacuum level and cause the fragile chain to break.
>The rich tolerance for a "cold" engine is probably in the 100% range.


When we are cold the engine runs a little rough.  I probably haven't got
the mixture perfect for warmup.  But, once warm,  soooo smooth.  Lack of
any intake/exhaust valve overlap up to 4000RPM probably has a lot to do
with that and then,  at 4000RPM when the CAM profile changes... wow!
the engine just wants to go and go it does.

I'll keep you informed.

Regards,

John