From Cranking to Idle

[John Dammeyer]

Hi Phil,

Great response. Thanks.  I'ved add some more questions.

>Date: Wed, 04 Aug 1999 03:28:28 +1000
>From: Phil Lamovie <injec at ains.net.au>
>Subject: Re: From Cranking to Idle.DIY_EFI Digest V4 #443
>
>Hi All,
>
>The problem of Injection timing at cranking is one of engine
requirements.
>
>The problem is not as it seems. The engine doesn't run on petrol it
runs on
>vapor.
>When the engine is cold very little vaporization takes place in the
cold
>air stream.

I thought one of the injector's jobs was to atomize the fuel though.

>
>In fact the only physics available to aid  us is low pressure. Thus the
>time to inject is during the exhaust stroke. This allows maximum time
for
>latent heat absorption and thus vaporization and allows for a vacuum
build
>up during intake as long a you can convince the customer to keep their
foot
>of the acc.  pedal.

Not a problem.  Most cars instructions are to leave the throttle alone.
On the Honda,  there is a cold start system that increases the air flow
so that it acts as if the throttle has been cracked a bit.  However,
we've removed that to save weight and a couple of extra coolent hoses
that could crack or break at altitude.  (This is an remanufactured
engine turning a propellor on a hovercraft or aircraft).  So we have to
crack the throttle a little bit at startup.  This has been the biggest
obsticle to cold starting.  If we open the throttle too far the MAP is
wrong compared to RPM and mixture is off.

>
>You will need approx. 300% excess fuel to start the engine. This is due
to
>the partial vaporization of the fuel. A cranking engine often has HC
>components in the exhaust of 50-60% raw fuel. Don't assume that the
>compression of the air will supply sufficient heat as the pressure
builds
>up very late in the piece. And the dwell time at that temp is only a
few
>ms.  Worse still if you get the A/F wrong you have just chilled your
>combustion chamber with a dose of raw fuel. Makes it important to
really
>over fuel.
>

We've found that we do need that much extra fuel.  We inject just after
the intake valve starts to open and pulse width at cranking is about
6ms.  Once the engine starts and while the engine is cold the warmup
enhancement keeps the pulse width at about 3.5 to 4ms.  Once warmed up
it drops to 2ms at idle and the loop is closed because the O2 sensor is
used for closed loop control.

>
>You can either multiple fire the injectors by dropping the input
divisor or
>simply have a table with very big numbers. You may still run out of
pulse
>width if your injectors are too small. (See Toyota for several hundred
>million examples) they still fit cold start injectors in place with
very
>low winter temps.


Yes.  I'm somewhat worried about really cold temperatures.  My problem
is that on Vancouver Island we rarely drop below 0 C.  However,  the
injector pulse width at starting is tied to engine coolent temperature
and the pulse width is also tied to engine intake manifold air
temperature so hopefully it should work when cold.  We'll find out this
winter.


>
>The aim of the game is to use the smallest  injector possible that
fulfills
>the requirements. Often 85% duty cycle is the target. that is of course
85%
>of 720 Deg.
>or 612 Deg.


Based on the last dynamometer run,  we found we were at about 12ms at
6000RPM and pulling 111 HP and our goal is 125HP;  We've been able to
get over 250ft/lbs Torque. This, at a little less than 75% duty cycle.
As I understand the reason for 'the name of the game' is to provide a
smooth unloaded idle.  Since we turn a prop idling at about 900RPM we
have a little bit of a load so a slightly larger injector might keep me
under 75%.  (I have a hardware limitation on pulse widths longer than
75% if I plan my injection at low rpm for the intake stroke).


>
>Due to the acc of the engine 0 to 600 in 1 to 1.5 seconds the pulses
must
>be linked to the crank. This will ensure that there are no missed
pulses.
>This also means that the crank trigger event must result in an injectio
n
>event even if the output buffer has not been updated with a sensible
rpm
>measurement and thus  in fact the ecu may not know how fast the engine
is
>tuning.

And in fact I do that.  I use a double Hall Effect Sensor buried under
the CAM Shaft timing pulley so not only do I detect TDC for each
cylinder but the pattern is such I know which cylinder since only one
Hall Effect sensor changes at each TDC.


>
>Most manufacturers use a cranking table that cuts out at say 350 rpm.
Below
>that
>each trigger pulse is greeted with the same ms injector output
regardless
>of vacuum..

Now that's a good idea.  Just say the heck with MAP,Water Temp, Air Temp
and Throttle position and inject at fixed value under about 350 to 500
RPM.


>
>If you hard link the events then it doesn't matter how fast the acc of
the
>engine really is. If the fueling is  wrong it will simply acc more
slowly.
>
>The clue is to use the micro to set up a timer/off event from the
>instigation  of injection.


Yes.  I do that too.

>Your electric motor should not be able to unsync a magnetic trigger
firing
>a gate on a transistor. Though it would confound the heck out of an
input
>capture rpm reading
>as it is using clock events to determine rpm. During cranking it would
be
>very late in it's determination. By the time it has worked out the rpm
it
>could be 200% wrong in
>its phasing as the engine speed could more than double on a 1/4
revolution.
>

It doesn't unsync the injectors, since I tie injector on with the TDC of
the intake stroke.  What perhaps isn't working as well as I'd like is
the mixture at cranking due to the high acceleration from cranking to
idle.

But you are right about figuring out RPM during that initial high
acceleration moment from cranking to idle;  doesn't work.

>Hope this cures some headaches and creates many others.


Yes,  thanks again.  I still have an issue of when to actually trip the
injector on.  I've seen carburated cars start after only a half a
revolution although as they get older it usually takes a turn or two.
This implies that the fuel leaving the carburator gets into the cylinder
easily enough although a cold engine leaves lots on the intake manifold
walls.  Explains why intake manifolds have a hot water path in them and
gobs of fuel are needed when first starting.  8-)

But,  injection is supposed to get rid of the need for a lot of the
excess fuel because it's sprayed almost directly onto the intake valve.
Spraying onto a closed intake valve when the engine is cold or just
warming up doesn't seem logical for a low pollution fuel efficient
engine.

My SAAB 99 used to be multi-port injection with the K-Jetronic Bosch
system that varied fuel flow via fuel pressure.  Engine ran fine but had
lousy altitude compensation and it had a cold start injector.  My 900 is
much more sophisticated with multi-port sequential and has the luxury of
having a MAF but no cold start injector; runs fine at any altitude but I
think the turbine helps a little there. 8-)

My JEEP uses MAP but has to be cranked a while before starting and I
don't think it has a cold start injector.  If I knew where to tap into I
could sync the scope on the TDC sensor and look at the fuel injection
pulse to see when they do it.

But,  isn't the idle quite a bit rougher if injection were to be started
on the POP part of the stroke.  Obviously at high RPM and 85% duty cycle
it no longer matters but in-between....

Thanks,

John