Received-SPF: none
 receiver=logan.com; client-ip=129.72.10.31; envelope-from=SBoese@uwyo.edu
From: "Steven W. Boese" <SBoese@uwyo.edu>
To: Rotary motors in aircraft <flyrotary@lancaironline.net>
Date: Sun, 6 Mar 2011 17:49:36 -0700
Subject: staging and tuning
Thread-Topic: staging and tuning
Thread-Index: AcvbiKxvvw9fODt8TrCylHdg9WgxIgAM+5sw
Message-ID: <E1AA3B1AF41D8049B1E3FBD5E2256260088739698E@ponyexpress-mb5.uwyo.edu>
References: <list-4892203@logan.com>
In-Reply-To: <list-4892203@logan.com>
Accept-Language: en-US
Content-Language: en-US
acceptlanguage: en-US
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

Bill and anyone else interested:

The issues involved with mixture changes when the EC2 controller stages can=
 be understood by considering two simple concepts:

1. The first is the concept of injector lag, also referred to as injector l=
atency and injector dead time, as well as other names.  The fuel injectors =
are not perfect.  They do not instantaneously respond to the controller pul=
se.  That is, they take a measurable time to open after the application of =
the pulse and they take a measurable time to close after the pulse ends.  T=
hese two times are not equal and the result is that the injector flows fuel=
 for a smaller period of time than the injector pulse width.  In addition, =
the injector flow rate is changing during the opening and closing time peri=
ods.  Measurements of the amount of fuel delivered as a function of pulse w=
idth show that the deficit in the amount of fuel delivered for any pulse is=
 the same whether the pulse width is long or short.  For stock 40 lb Mazda =
injectors with a fuel pressure of 40 psi and a system voltage of 13 volts, =
this loss in the amount of fuel delivered corresponds to an injector lag or=
 dead time of about 1.2 ms.

This 1.2 ms is not trivial.  Consider a rotary engine running at 6000 rpm. =
 The time it takes for one revolution is 10 ms.  Limiting the injector puls=
e to an 80% duty cycle limits the injector pulse width to 8 ms.  The effect=
 of the injector lag is to decrease the amount of fuel delivered by 1.2/8 x=
 100 or 15% at 8 ms pulse width,  1.2/4 x 100 or 30% at 4 ms pulse width, a=
nd 1.2/1.2 x 100 or 100% at 1.2 ms pulse width.  It is obvious that the sho=
rter the pulse width, the larger the fuel deficit.

2. The second concept to realize is that the EC2 controller does not take i=
njector lag into account.  This has been demonstrated with my EC2's at defa=
ult settings by measuring the injector pulse width on either side of the st=
aging transition.  It has also been demonstrated by pulse width measurement=
s taken when engaging the injector back up function where the change from u=
sing two to using four injectors is forced to occur without a change in man=
ifold pressure or change in MAP table address.  These measurements show tha=
t, before tuning, the EC2 staging transition is a transition to using all t=
he injectors at half the pre-transition pulse width.  The essential point h=
ere is that even if the primary and secondary injectors are the same size, =
staging, by using twice as many injectors at half the pulse width does not =
result in the same fuel flow, but rather a substantial fuel flow decrease. =
 Using the above data, this is easily seen by considering what would happen=
 if staging took place at an initial pulse width of 1.2 ms.  After staging,=
 the fuel flow would then be zero.


Before tuning and with primary and secondary injectors of the same size, th=
e effect of injector lag is always for the controller to deliver less fuel =
than it calculates as being needed.  This problem becomes progressively wor=
se as the injector pulse width decreases.  When the secondary injector flow=
 rating is larger than that of the primary injectors, the mixture change du=
e to staging, while predictable, is not intuitive.  This is because the ten=
dency to go lean due to the injector lag effect accompanying the pulse widt=
h being cut in half is now combined with the tendency to go rich due to the=
 two larger injectors now in use.  Whether the mixture goes leaner, richer,=
 or is unchanged upon staging depends not only on the difference in the pri=
mary and secondary injector flow ratings but also upon the conditions (init=
ial pulse width) under which staging takes place.

The discussion which follows is for the case of equal sized primary and sec=
ondary injectors and initial tuning starting with default parameters:

The percent decrease in fuel flow upon staging is most severe for low initi=
al fuel flow rates.  This explains the difficulty in achieving a smooth sta=
ging process if the staging threshold is set too low.  Even for reasonable =
initial fuel flow rates in the range of 6 to 10 gal/hr, the fuel flow will =
decrease significantly.  If the mixture is rich enough when staging takes p=
lace, the decrease in fuel flow may not result in a large change in engine =
power output.  However, if the mixture is already somewhat lean when stagin=
g occurs, the mixture may become very lean and result in a marked decrease =
in power.

I've used three methods to compensate for the injector lag.

1.  The first method is to follow the instructions in the manual.  Tuning i=
s started by adjusting modes 3 and 2.  After modes 3 and 2 have been adjust=
ed, then mode 1 or 9 is used to adjust the mixture correction table.  Tunin=
g the mixture correction table can compensate for the fuel flow decrease on=
 staging caused by injector lag.  This results in a discontinuity in the co=
rrection table at staging.  Since defining the step in the mixture table ca=
n be difficult, programming the mixture somewhat rich across this region is=
 helpful.  Cruise operation at MAP values close to the staging threshold th=
en simply requires some adjustment of the manual mixture control.  Because =
of the step in the mixture table at the staging point, changing the staging=
 threshold also requires retuning of the mixture table in this region. Tuni=
ng the rest of the table compensates for other factors such as changes in v=
olumetric efficiency in addition to injector lag.  Injector lag affects tun=
ing of the entire table although it is just not as obvious as its effect on=
 staging.

2.  A second way to compensate for the effect of injector lag on staging is=
 to use the mode 6 secondary injector differential adjustment even with pri=
mary and secondary injectors of matched flow rates.  Since the mode 6 setti=
ng only has effect above the staging threshold, and since the effect of inj=
ector lag is to make the injectors act as if they became smaller upon stagi=
ng, mode 6 can be used to eliminate the step in the mixture correction tabl=
e.  This is possible because mode 6 has the ability to compensate for small=
er as well as larger secondary injectors compared to the primaries.  Compen=
sation for the effect of injector lag and other factors is still necessary =
throughout the table, but the changes are gradual and much easier to implem=
ent than a discontinuity.  This method removes the injector lag effect upon=
 staging and works quite well.

The tuning procedure then becomes the same whether the primary and secondar=
y injector flow rates are different or identical:

a.  Set the staging threshold at a MAP corresponding to the high end of the=
 primary injector flow limit using mode 7.  (For 40 lb injectors, about 20"=
 MAP works well.)

b.  Adjust Mode 3 to get a mid scale O2 sensor reading at a MAP just below =
the staging threshold MAP.

c.  Adjust mode 6 to get a mid scale O2 sensor reading at a MAP just above =
the staging threshold MAP.

d.  Adjust mode 2 for best operation at minimum idle MAP.

e.  Adjust the mixture table throughout the useable MAP range using mode 1 =
or 9 to keep the O2 sensor reading mid scale.  (To be honest, I skip step "=
e" and simply use the manual mixture control to adjust the mixture in cruis=
e.)


Steve Boese
RV6A 1986 13B NA RD1A EC2