vimeo.com/.../379152025

- Now that we've got our MAF scaling
dialled in, we can actually begin the process

of optimising our tune.

Before we get stuck in though, there are
a few aspects to the Cobb Accessport

that it's worth touching on so that you've
got a better understanding of how the

Accessport works as well as how the basic
Subaru ECU structure and logic works.

We're going to start by talking about how the
ignition control in the Subaru ECU works

and particularly if you are new to the Subaru
ECU, this is quite unusual, if you don't

understand how it works, it can be really
confusing and you can easily come

unstuck here.

Rather than just running off one base
ignition table, there are several ignition

tables and they interact depending on
feedback from the knock control strategy.

And this is part of Subaru's dynamic advance
multiplier system.

So let's talk about how that works.

Let's have a look at the software for a start
and if we go into our real time tables,

the main table that we're going to be doing
our ignition tuning on is found here.

This is the primary ignition table, TGV's
open main.

So let's have a look at that.

So this is our table here, we can see
three dimensions, nothing particularly

surprising here.

We've got our load in grams per revolution
on the horizontal axis and we've got our

engine RPM on the vertical axis.

It's important here because we are dealing
with grams per revolution, which is a

channel or parameter that's calculated based
off the mass airflow and engine RPM.

It's important to log this parameter so that
we know exactly where we are operating

inside of this table.

Now we also have a graphical representation
of this table of course down here at the

bottom and just like any ignition table,
we should expect in a properly tuned table,

for this to look relatively smooth.

However you may notice, particularly in the
higher load areas of the table where we're

likely to be running under wide open throttle,
full boost, that there is very low ignition

timing, particularly through the peak torque
area we've got negative ignition timing

values, the ignition starting after top dead
centre.

Now the reason for this is that this
table works in conjunction with a couple

of other tables which can add timing based
on the knock control strategy.

So let's have a look at those.

In order to get to those we're going to
need to come down in our menu structure

to the ignition tables, let's open up the
ignition tables here.

And what we're looking for here is advance
dynamic.

So if we open up the advance dynamic,
there are two options of adding timing here.

There's the dynamic advance adder tables.

Now in stock form, with the base calibration
from Cobb, these are actually zeroed out

just to make things a little bit easier to
understand so we know where abouts

our timing is coming from.

Instead the Cobb calibration will work on
these tables here which are our dynamic

advance base tables.

So let's have a look at the dynamic advance
base TGVs open table.

So we've got exactly the same table
structure here as our base ignition table,

our primary ignition table.

We've got the same load axis, the same
RPM axis and the numbers in this table

essentially can be added over on top
of those primary ignition timing table

values.

So if for example we take a value here,
6400 RPM, 2.46 grams per revolution,

you can see that this table has the ability
to add 5.27 degrees of ignition timing.

So to see what that means, let's go back
to our primary ignition table.

We'll find the same point here at 6400 RPM
and 2.46 grams per revolution.

You can see that we've got 3.5.

So what this means is that if the ECU is
adding all of our dynamic advance base

table over on top of our primary table,
we'd actually end up with a final ignition

advance being delivered of about 8.8
degrees at that point.

So the next obvious question is how does
the ECU decide how much of our dynamic

advance base table to add on top of our
primary ignition timing?

And this comes down to another parameter
inside of the ECU called the dynamic

advance multiplier D.A.M. or DAM
depending on how you want to pronounce it.

Now the dynamic advance multiplier is
a number between zero and 1.0 that's

based on the feedback from the knock control
system.

Essentially depends on how much activity the
knock sensor is seeing.

And if there is no knock occurring, that
number is going to increment up to 1.0

And if it's at 1.0, it's a simple multiplier
so it means that all of our dynamic

advance base table is added in.

If we come back to our dynamic advance base
table here, the unit or the cell that we were

looking at before, that would mean 5.27
degrees multiplied by 1.0 and of course is

5.27 degrees.

All of that table value is going to be added
on top of our primary ignition table.

Now if on the other hand some knock
activity is occurring, let's say for example

the ignition timing is overadvanced causing
knock or the fuel that the car is running on

is a lower octane that's leading to knock then
what's going to happen is that over time

that dynamic advance multiplier is going to
decrement down from 1.0.

Let's say it gets to a point where the
dynamic advance multiplier is sitting

at 0.5, again this is a multiplier for our
dynamic advance base table so if we

take our multiplier, our DAM value at
0.5, and we use that same value of

5.27 degrees, of course we're going to be
adding half of that, 0.5 multiplied by 5.27,

we're looking at about 2.6, 2.7 degrees
added in.

So this is how the ignition system works.

And it's important to understand that,
and it's also important to understand as

we'll see as we go through this part of the
worked example, what parameters we need

to log so that we can really understand how
the ignition system is working and what

that dynamic advance multiplier is sitting
at.

Now since we've just been talking about
the dynamic advance multiplier,

it's also worth just having a look at the
parameter for the dynamic advance

multiplier starting points.

So that's down here, let's click on that
and we can see that in the Cobb base map,

that's set to 1.0.

So in a stock calibration we'll actually find
that that's set to 0.7.

Meaning that when we first start the car,
the dynamic advance multiplier is not

going to be adding all of the available
ignition timing in, meaning that we're

actually going to start with a slightly
detuned engine over the first few minutes

of running, provided no knock is occurring,
that dynamic advance multiplier is going

to increment back up to 1.0, adding in
some additional timing.

So by setting this to a starting point of 1.0
we're actually starting with a slightly more

aggressive tune immediately after startup.

Now the activity from the knock control
system is going to constantly be changing

that dynamic advance multiplier so it's a
really good parameter to have a look at

in our logging to give you a quick snapshot
of what's happening, whether the engine

is suffering from knock.

And essentially any time we're seeing the
dynamic advance multiplier sitting anywhere

but 1.0, this should be a bit of a red flag that
perhaps the ignition timing is a little bit

aggressive.

Now working in conjunction with the ignition
control system of course is the knock control

strategy and again we need to understand
how that works and which parameters we

need to be looking at.

Now there are two parameters here that
we need to be logging.

If we look at our gauge list here,
we've got those two listed over here.

We've got our feedback knock correction
and our fine knock learning.

And essentially the feedback knock correction
is an immediate response to a knock event.

So if we're doing a ramp run on the dyno or
we're running in steady state and the engine

starts to suffer from knock, we're going to
immediately see that feedback knock

correction value drop into the negatives.

Meaning that the ECU is retarding timing in
response to a knock event.

So we obviously want to be logging that
and take note of where abouts that's

occurring, that's an indication that we're
potentially running a little bit too much

timing in that particular area of our
primary ignition timing table.

Now this also works in conjunction though
with the fine knock learning parameter

and as its name implies this is a knock
learning strategy where if knock is occurring,

there is a table inside of the ECU where the
learned knock value or knock retard

required to prevent detonation is stored.

What this means is that the next time the
ECU runs through the same area of the map,

the ignition timing will automatically be
retarded as a preemptive measure to stop

knock from occurring.

Now over time, if no knock is occurring then
the numbers in the fine knock learning table

will be reduced back to zero, basically
adding timing back in.

So this is a fluid strategy where the timing
is constantly changing in response to knock

in order to try and prevent knock occurring.

Now in the Cobb Accesstuner software, we do
have the ability to see what's going on in

our fine knock learning table, we can see that
here under our real time tables.

If we open up the fine knock learning
table, we can see we've got a relatively

small three dimensional table there,
we've got load in grams per revolution

versus engine RPM.

You can see at the moment this table
is all zeroed out which is what we hope

to see.

This means there is no fine knock learning
going on while we've been running the car.

The reason it's important to take note of
this table is that if you are performing

ramp runs and you're a little bit aggressive
with the ignition timing causing some knock,

we can end up with that knock retard coming
through into the fine knock learning table.

If you're not aware of this it can cause a
little bit of run to run variation because

the engine may not be receiving as much
ignition timing as you're expecting.

Another aspect that's worth touching on
before we get into the tuning is the

boost control strategy in the Subaru ECU.

And we're going to do some basic boost
tuning through this particular worked

example.

As I've already mentioned here, because
we have got a three port solenoid fitted,

the existing boost control duty cycle tables
are going to need some manipulation

to get our boost onto our target.

But let's have a look at how that all works.

First of all if we go to our boost targets
under our real time tuning table,

we can see the existing boost targets,
these are displayed in bar however you can

display them in imperial units if you
prefer.

We've got our load axis as requested torque
here and this is an important aspect to note

here because we do want to be logging our
requested torque parameter so that we

know exactly where abouts in the table
the ECU is referencing.

It happens that for this particular calibration
under wide open throttle operation,

we're constantly referencing the 350
newton metre row here, or column here.

We've obviously also got our RPM axis
there on our vertical axis.

So it's important to make sure that these
boost targets obviously are realistic

and safe as our first starting point.

We've also got our boost limits here so we
can make sure that the boost limit is set

to something that's going to be safe.

Now in order to get us onto our boost
targets we've got two tables here,

our wastegate duty cycles high and low.

We'll just talk about how these interact.

Let's bring up our high duty cycle table.

We've got the same axis as our boost
control table with requested torque

on the horizontal axis, RPM on the vertical
axis.

And this is simply the maximum duty
cycle that can be delivered to the

wastegate solenoid.

So regardless of the proportional and
integral gains, as part of the closed loop

control strategy, nothing can ever exceed
the numbers in this particular table.

So for our solenoid we need to start by
essentially doubling the numbers in

this table.

We'll get onto that a little bit later though.

Now we've also got a second table here,
replicates that first table,

this is our wastegate duty cycle low table.

And the idea with this table is that it
should be set somewhere around about

8% duty cycle lower than our high duty
cycle table.

How this works is that if we perform a
ramp run and we come up onto boost,

the ECU will use the high duty cycle
table as an open loop duty cycle

table or a feed forward table for
the starting position before it begins

closed loop control.

However if we were only working off that
high duty cycle table, on a transient or

a gear shift we're very likely to see an
overboost.

So in this situation the Subaru ECU on a
gearshift will revert to the low duty cycle

table, once the engine is stable again
and back at full throttle,

it will then instigate closed loop control
to get back onto the targets.

This helps prevent that overboost.

And the suggestion here is to keep about an
8% duty cycle difference between high and low.

Now this also works in conjunction with
the proportional and integral gains system.

We're going to see how that works as we
go through the tuning.

And another aspect that's important to note
here, if we cycle down to our boost control

tables which we'll see in our menu structure
here, let's open that up.

And if we look at our wastegate duty cycle
here or wastegate duty and open that up,

we're also going to have the ability to
adjust our wastegate solenoid frequency.

You'll see that in the stock calibration
for the factory Subaru boost control solenoid

this is set to 10 hertz.

Now this is quite low for a lot of the three
port solenoids that we use in the aftermarket.

Typically I'd be running somewhere between
about 20 and 30 hertz for the typical MAC

three port solenoid.

Now a little trick with the Subaru ECU is
it appears the maximum duty cycle,

it's able to use here is 15 hertz.

So we're going to set that to 15 hertz and
if you do try and set this high,

you're going to run into problems where
you're going to end up with very poor

or no boost control.

So make sure that you're aware of that and
keep your boost control frequency at 15 hertz

maximum.

The next topic to talk about here is Cobb's
MAF simple mode.

And this is a little bit complex to explain
in really simple terms but the Subaru

ECU while it primarily uses the mass airflow
sensor as its load input, deciding on fuel

and ignition based on the MAF signal,
under transient conditions, sharp throttle

inputs or gear changes, the MAF sensor
can lag a little bit.

So the stock Subaru ECU logic actually
uses a volumetric efficiency or speed

density subsystem that it can revert to
on these throttle transients.

Now in stock form, Cobb believe that there
can be as many as 200 different parameters

that can affect the final calculation during
these transients.

They've highlighted the two key ones and
if we cycle down to sensor calibrations,

we can see we've got MAF corrected
volumetric efficiency input,

TGVs open and closed, let's have a look
at one of those tables and this is

essentially a volumetric efficiency table
where we've got manifold absolute pressure

on the load axis this time.

Now the problem for us as tuners is that
there's no easy way for us to make the Subaru

ECU run continuously on the volumetric
efficiency tables so that we can calibrate

these accurately.

So this makes it really tricky for us to tune
these tables and get good results.

So as an alternative Cobb have instigated
their MAF simple mode which basically

eliminates the volumetric efficiency
subsystem and runs continuously off

the mass airflow sensor.

And in simple terms, it does work really
well.

So it's a good idea to understand that that
option is available.

We'll close down our sensor calibrations
and we will find the MAF simple mode

functionality under our Cobb custom features
and we can see that under the MAF simple

structure here.

Now we don't really need to make too many
adjustments to this, what we do want to do

though is activate it here so if we click
on our MAF simple mode activation

threshold, we can see that at the moment
this is set to 1.5 grams per revolution.

In the base calibration you're likely to find
that this is set to 3.2 grams per revolution

which is a little high for it to really do any
good.

So Cobb's recommendation is to set this
activation point between about 1.5 and 1.7

grams per revolution.

We do need to be wary, if you set this too
low, you're likely to run into drivability

problems.

It seems that the ECU does work best if we
set this around about that 1.5 to 1.7

grams per revolution mark.

So that's going to give you a really
simple way of eliminating that subsystem

which can be problematic for us to tune
in the aftermarket.

Another aspect I want to touch on here is
some of the compensations available inside

of the Subaru ECU.

In particular we'll focus here on our ignition
trim for intake air temperature.

So let's have a look at that, we're going
to find that under our ignition tables

and then if we move down to compensation
and then finally to intake temp.

Now there are two tables here that interact
so we'll start here with our ignition timing

compensation.

Now this is a simple two dimensional table
based on our intake air temperature.

Now it is possible here to choose either
the intake air temperature sensor

pre turbocharger or in the inlet manifold
based on your own personal preferences here.

And we can see we've got our trim here
and you can see that does look like it might

be a little bit excessive, particularly at
70 and 80°C, we've got -6.5 and -7.5

degrees of ignition retard.

It's important to understand that this does
work in conjunction with our activation

table below.

And essentially this is a three dimensional
table with our load in grams per revolution

on the horizontal axis and RPM on our
vertical axis of course.

Now the way this table works is that it
gives the ability to control the magnitude

or weighting of how much of that trim will
be applied.

So we can see that in particular down in
the idle areas we've got -100% so

essentially there's going to be no trim
applied down in those idle areas.

However under high load here and high
RPM, we're applying about 50% of that

trim so instead of the full magnitude of
the trim from our table here,

we can see that we're only applying
half of that so for example if we were

at 70°C with a trim of 0%, we would see
all of that applied, so 6.5 degrees of trim

minus 50%, we would only see half of that,
so 3.25 degrees applied.

So you are able to manipulate that as you
see fit based on the intake air temperatures

you're seeing and how sensitive to knock
your particular engine is.

Next we can take a look at the injection
timing control.

Now being that this 2015 WRX does run
a direct injected engine, the engine timing

tends to be a little bit more sensitive than
on a port injected engine,

but it's still worth having a look at how
this is controlled.

So we can find this under our fuel
tables and then if we cycle down to

injectors and finally injector timing.

The table that we're going to primarily
be using here is our fuel injector,

start of injection, homogenous main
beta table.

We can see here we've got our three
dimensional table of our injection

timing values.

Now this is a stock table, doesn't actually
really benefit from modification where

we're running on pump gas here.

But it is important to just point out where
you'll find that if you do want to make

some adjustments to this and see the
effect of those changes.

Because it is a real time tuning table
you can make these adjustments while

the engine is running and see the effects
instantly.

Now that we've had a tour of the software,
we've talked about the interaction of some

of the more important tables, let's get
started and lay down a base run here

on our Dynajet dyno so we can see what
our starting point is before we start making

adjustments to any of those tables.

Alright we've got our baseline run laid down
and we've actually got a slightly lower

figure there than we were working with after
our MAF scaling at 248.9 horsepower

at the wheels.

Now I am not too worried about that at
this point, we're going to be making some

pretty dramatic changes from here anyway
and we are also on the dyno on a different

day than when we finished our MAF scaling so
we're likely to have some day to day

variations creep in there.

Let's see how we progress as we begin
optimising that tune.

Before we make any changes though,
let's jump into our MegaLogViewer HD

software and we'll have a look at what's
going on in the software, what happened

during that ramp run.

So for a start, let's have a quick look at
our MAF scaling just to confirm that that's

still on point.

So we can see that as I click through,
we are within about 1% or 2% positive

and negative of our zero so we know
our trims are pretty good there,

we know that our MAF scaling is still good
from when we finished that up exactly

as we'd expect.

What we want to do now is we'll get rid
of that particular graph, just so we've got

a little bit less going on because we
shouldn't need to worry about our MAF

scaling from here on in.

What we're going to do is start by looking
at our boost control.

So we've got some parameters set up here
that we're going to be looking at,

we've got our wastegate duty cycle,
we've got our manifold relative pressure

and then we've got a parameter called
turbo dynamics boost error extended.

Essentially this is just showing us how far
away we are from our boost target.

So our yellow line there in this plot is our
boost pressure that we're actually

achieving in the manifold.

So we can see that our peak there,
sitting around about 0.95 so remembering

that our boost targets were sitting at about
1.43, 1.45, we're a little way off that and

that is shown by our turbo dynamics boost
error showing at the moment 0.522.

So we're a little bit lower, the reason for
this is our red line here that we can see

in our plot is our wastegate duty cycle and
we can see that that is basically flatlined

at 32% so remember this comes from our
wastegate max duty cycle.

So the first thing we're going to do is
head back into our Accesstuner software

and we're going to double our wastegate
duty cycle tables.

Before we do that though, one other aspect
that we always want to be looking at here

is our knock control.

So on this set of graphs here, we've got
our ignition timing, we've got our feedback

knock correction and our fine knock
learning as well as our dynamic advance

multiplier.

So really what we want to always see is
both our feedback knock correction and our

final knock learning sitting at zero,
exactly what we've got, as well as our

dynamic advance multiplier sitting at
1.0 so we know that currently we've got

no knock, exactly where we'd expect to be.

So let's head back to our Accesstuner
software and what we'll do is we'll start

with a really simply modification here,
doubling our wastegate duty cycle.

We can do that quickly by holding down
the shift key and the ctrl key, right arrow

will highlight an entire row, down arrow
will highlight an entire column,

we'll press the M key for multiply and
we're simply going to double that table.

Now what I'm going to do is press ctrl C,
that will copy that table to the clipboard,

we can now head across to our low duty cycle
table, ctrl V will copy that onto our low duty

cycle but remember we need a spread here,
we want to reduce this so what I'm going

to do is hold down the shift and minus
key and we're going to reduce that so we've

got about a eight degree spread from our
high table to our low table.

So that's a very basic change there.

Obviously when we are making changes like
this, we do want to monitor our boost

and make sure that it's not becoming
excessive.

So with that simple change done there,
let's do another run and see the effect

of that result.

So we can see with that one simple change
to our boost control we've definitely got a

healthier result there on the dyno with a
jump up to 288 horsepower,

321 pound foot of torque.

Let's have a look at our logging through and
we'll see how all of that panned out.

Now of course the first thing we always
want to check is our knock strategy and we

can see in this group of graphs here,
we've still got our dynamic advance multiplier

sitting at one, we've got no knock feedback,
no knock correction, no knock learning,

so that's good.

We've got no knock occurring so we know that
the engine is safe.

Let's have a look at our boost control though
and see how that panned out.

So if we head down and we have a look at
our yellow plot for our manifold pressure,

we can see that our peak number now
looks about here, 1.53 bar so we're

actually a little bit above our target
there, 1.43 bar.

And we can see what our boost control
system is doing.

So looking at our group with our turbo
dynamics integral and our turbo dynamics

proportional, what we're looking for here
is what the proportional gain is doing.

So we can see that at the start of our
ramp run here as the turbo is coming up

onto boost our proportional gain is
essentially maxed out at positive 10%.

So the problem with this is we get what's
referred to as integral wind up

which can result in overboost as the turbo
comes up on boost.

We can see that that's exactly what's
happened here.

At this point here where we've hit
1.53 bar, we've still got a positive

integral there of about 5.5%.

So what this generally means is that we're
targeting a boost pressure down at low RPM,

that the engine can't really quite achieve.

So we're seeing 3000 to maybe 3500 RPM
our boost targets are a little bit optimistic.

Once we're actually up on boost though,
we can see that our control is not too bad,

our integral gain here is moving around a
little bit, we want to try and get our

integral gain through the ramp run to be
within sort of 1-2% of zero ideally,

giving it a little bit of head room to add
or subtract to our boost control system.

We can see that particularly at higher RPM,
the integral gain is pulling quite a lot of

duty cycle out so we want to address that
here.

So we're going to make two changes here,
first of all what we're going to do is below

around about 3500 RPM, we're going to
reduce our boost targets to something

a little bit more realistic.

And then above around about 6000 RPM,
we're going to start reducing our wastegate

duty cycle so let's head across and do that
now.

So first of all let's head up to our boost
target table.

And we know that we are running at a target
here of 350 newton metres of requested torque,

let's just quickly head back and we'll see
where we can find that value.

So I've got our requested torque added into
our top set of graphs in white and we can see,

or you can't actually see it doing anything
because we're at full throttle right through

this ramp run, but we can see that that
value there is sitting constantly wherever

I click at 350 newton metres.

So that's our requested torque so we'll
head back across to our Accesstuner.

So what we want to do is below around
about 3600 RPM, we're going to reduce

our boost target.

So I'm actually going to extend that down
a little bit down to 300 newton metres

and we'll take it down to in this case
2000 RPM.

So what we can do now is just reduce our
boost targets a little bit, make them just

a little bit more realistic.

We do always want to make sure that our
boost targets are relatively smooth so that

the ECU can track them.

So what we're going to do here is highlight
the cells that I've just adjusted here

and we can press V for vertical interpolation
and this will just smooth those values across.

Alright so that's our first change there,
we've reduced our boost targets,

hopefully that's going to be a little bit more
realistic.

Now we'll head across to our wastegate
duty cycle high table here and just

heading back to have another look at
our data here.

It's around about 6000 RPM and above that
we want to make that change to our duty

cycle so let's head to that zone there.

So again we'll just extend the change that
we're making down here all the way to

300 newton metres and I'm highlighting
6400 RPM and above and what we'll do is

just shift and minus a couple of times just
to remove a little bit of duty cycle there,

I'm also going to make a further change
at 6800 RPM.

Once we've done that we can simply copy
and paste that table back across to our

low duty cycle rather than making the
same change, I just find this a little

bit easier.

And then again we're going to just reduce that
to make sure that we've got around about

that 6% spread.

So now that we're ready for another run,
let's check that out and see what our

results are.

Alright looking at our results from that
particular run we can see we've picked up

a little bit of peak power there,
we're up to 293.9 horsepower.

We've got 322 pound foot of torque
at the wheels.

More interesting though is our data so
let's have a look at that.

And we can see if we look at our boost,
we've still got this spike here in our boost

and that's coming up to a peak of 1.54 bar.

However we have got better control there.

What we want to do is again have a look at
our integral gains.

So we can see that we're still probably
requesting a little bit much boost and in

the early part of that ramp run we've got the
integral basically tapped out at positive 10%.

Now that's causing that overboost here because
where we see that overboost we're still,

we've got a positive integral there of about
7.5%.

So we've got a little bit more work to do
there in terms of our duty cycle as well

as our boost aims, just getting those in line
with what the engine can actually produce.

Now this is a common problem with boost
control where tuners try and achieve a boost

set point that's simply unachievable.

This causes havoc with the closed loop
control strategy, trying to drive the turbo

harder to achieve a boost that it simply
can't get to and once the turbo gets to

a point where it can produce boost,
it then creates this overboost which is

exactly what we're seeing.

However you'll also recall that we reduced
our wastegate duty cycles at higher RPM

and that's worked pretty well, we can see at
high RPM now we've got a integral of -2.5,

which is pretty close to where I'd want it
to be.

So this is an iterative process just like our
MAF scaling so you can simply go through

this process, adjusting your boost targets
and adjusting your wastegate duty cycle until

you've got something where your integral
is under control and you've got nice smooth

control.

So let's jump ahead and we'll perform another
ramp run with our final boost control strategy

dialled in and see what that looks like.

With our third run up on the dyno screen
now, we can see that we've picked up

a little bit of peak torque up to 330 pound
foot of torque.

We've actually lost very slightly in terms of
peak power but that's more to do with a

very slight spike that we were having in
our boost in our previous runs.

So by actually smoothing out that boost
control and getting it where it should be,

we've actually sacrificed a little bit of peak
power that is genuinely just across a very

very small RPM range.

We're not too worried about that right now
because we've still got some timing to add

into this.

More importantly though, let's jump into
our log file and we'll have a look at what's

going on.

So what we can see here is our peak boost
here, we're sitting at about 1.5 bar so

slightly above our target and we can see
that from that point on, about 4000 RPM

and to the right there, all the way through
to the rev limit, we can see that our red

trace here for our turbo dymanics integral
is moving around reasonably close to our

zero point.

We've got positive integral in that from that
point onwards.

But as I've mentioned we're trying to get that
so it's relatively close to zero,

we don't want it maxed out and that's what
it's doing.

Now what we've actually ended up doing
here is artificially limiting the maxixmum

duty cycle as we're coming up onto boost.

And this can be beneficial in some ways to
just prevent overboost.

So what we're doing is just limiting that
maximum duty cycle in the max duty

cycle table up to a point where we've
reached peak boost and then we're

bringing that max duty cycle to a point
where the integral and proportional

gains can take over.

So at this point I'm comfortable with the
amount of boost that we're running,

I'm comfortable with our boost control,
so what we're going to do now is move on

and we're going to try adding some timing.

So let's have a look at how we can go
about doing that.

Before we head back to our Accesstuner
software, it's a good idea to take note

of what maximum load we are actually
seeing, so in other words where abouts

we're actually operating within that ignition
table.

So we've got our calculated load channel in
red up here in our top set of graphs.

And what we can see is that our peak values
here are sitting around about 3.1, 3.2

grams per revolution.

At higher RPM as we get up towards our
rev limiter, we're dropping back down to

about 2.3 grams per revolution.

So let's head across and have a look at
our ignition table.

Now the first thing we can see here,
straight away is we've got a bit of an

issue in that the maximum load site in
this ignition table at the moment from

the base file is 2.84 grams per revolution.

So straight away even with the base file
and the boost pressure targets that we're

achieving we're exceeding that limit.

Now that's not necessarily to say that it's
a problem however what will happen is that

our ignition timing is going to essentially
remain stagnant from 2.84 grams per

revolution and above.

Not necessarily what we want and it's
generally considered best practice to

at least make sure that we're spanning out
our load cells so that we're always operating

within the map.

So what we're going to need to do here is
rejig our break points for our load axis

so that we're spanning that out for the
full range.

So I'm going to go out to 3.3 grams per
revolution out here in our far right cell.

Now because we have got a limited number
of break points in this particular table,

we do want to be a little mindful of the
difference or gap between these sites.

So what we don't want to do for example is
just take our last cell here at 2.84 grams

per revolution and change that to 3.3.

That's going to give us a really big jump
between 2.59 and 3.3.

So what I'm going to do here is sort of split
the difference, we're going to start by

copying our 2.59 gram per revolution
column here, we're going to copy those

values there, or ctrl C and we're going to
move those across to the left.

So that's now moved them across to the
2.46 gram per revolution break point site.

So of course I'm going to now change that
to 2.59.

Now what this means is I know that my
values there are going to be acccurate

when I get to 2.59 grams per revolution.

We're going to now repeat that process,
we'll ctrl C for our 2.84 gram per revolution

column, copy that across and again we'll
change our break point here to 2.84.

Lastly what we're going to do is change our
final break point here to 3.3.

Now we've got the same numbers here that
we were previously running.

So we already know that we were seeing
no knock with those numbers so at the

moment I'm going to leave them as they are.

You can get a little bit more fussy with these
break points if you want and start

interpolating between these, we've got
2.59, 2.84 and then quite a large jump up

there to 3.3.

So what we could do there is choose to
split the difference, if we highlight those

and use the horizontal interpolate, that's
going to give us a nice even set of

break points.

So this should give us essentially the same
ignition timing that we were seeing

previously.

What I'm going to do to start with for our
next run is we're just going to highlight

essentially the entire area that we're
operating in under wide open throttle

and we're going to go from 2000 RPM,
2.07 grams per revolution and above

and what I'm going to do is simply add
two degrees to all of those sites.

So this is just going to give us a good
starting point, we're going to see what the

effect of that two degree change to our
timing is.

So let's get another run underway and
we'll see what our results are.

Alright so we've got our run up there on the
dyno and we can see that we've got a nice

little gain there, we've picked up to 303
horsepower at the wheels, 346 pound foot

of torque at the wheels.

So that's a good gain for just two degrees
of ignition advance.

However to get the full picture of what's
going on, let's jump in and we'll have a look

at our log file.

So first of all, we obviously want to be
mindful of any level of knock that's

occurring so the first thing we want to
check is our feedback knock correction

and our fine knock learning, along with our
dynamic advance multiplier.

Good news there, no activity from our
knock control system which means that

we've got no detonation occurring so our
tune is still nice and safe.

It's a good point to mention here that during
the ramp runs we obviously want to be

mindful any time we're advancing the timing
of knock.

So it's a good idea while we are ramping the
car up, if you're not audibly listening for

detonation using audio knock detection,
check out your gauges while you're ramping

the car up and keep an eye on your knock
control feedback,

just making sure that if any knock is
occurring, you can abort that run,

get out of the throttle and address that
particular issue.

Now one thing we have seen here though is
now that we've changed our timing a little

bit, we've actually got a slight overboost
here that's occurring at about 4300 RPM.

So this is just going to require a little bit of
fettling to our closed loop boost control

tables and this is an iterative process because
as we progress through the tune,

we may find that the ignition timing,
because it effects the amount of energy

in terms of exhaust gas heat delivered to
the turbocharger, it can affect our boost

control strategy as well so this can be a
back and forth process between changing our

timing and coming back and affecting our
boost control strategy as well.

So that spike there to 1.57 bar, is probably
the main reason why we've seen that

increase in our peak torque.

Not to worry though, what we'll do is
go back to our boost control tables

and I'm simply going to make a change
to our maximum duty cycle at that point

so that we can ride that maximum duty
cycle table and bring down that peak spike.

However what we're going to do now is
make a further change to our ignition table.

So because we've got a gain in power and
torque everywhere with that change that

we've made, we'll head back to our
Accesstuner software, head back to our

primary ignition table and what we're going
to do is make a further change.

So I'm going to smooth this change that
I'm making down further into the load.

So we'll go down to 1.8 grams per
revolution now and we're going to make

that change from 2400 RPM and up.

So at this point again because we've got
no knock, we're going to try another two

degrees, it may prove that this is a little
bit optimistic but we'll see what the

effect of that change is.

When we're making these sort of block
changes, we do want to be a little bit

mindful of making sure that we keep a
smooth trace or smooth shape to the

curve and in particular here we can see
that at 1.55 grams per revolution,

we've now got a little bit of a step in our
ignition timing.

We move from 6.5 degrees to 1.5 and
then back up to two.

Likewise we move from eight degrees to
three degrees and then back up to 3.5.

Now we can address this quite easily
just by selecting the cells that we're

interested in there and we can use the
horizontal interpolate there just making

sure that we've got a nice smooth shape
to the curve.

Obviously when we are doing this,
we want to be a little mindful of how

far we extend our timing changes beyond
the area that we're actually operating in.

In the perfect world what we'd actually do
is creep up on this, starting at low boost,

optimising our timing at that particular
load area and then we'd increase the

boost and work just solely in the area that
we're operating in.

So for our demonstration today I'm cheating
this a little bit and just speeding up

the process.

Let's get another run underway and we'll
check out the results of that additional

timing.

With our last run complete there we can
see we've picked up another good

improvement in our power, we're up to
313 horsepower at the wheels.

Maximum torque there of 355 pound
foot at the wheels.

Now we can see we've got a good gain
everywhere over our last run and I've just

removed a few of the other runs just to
make it a little bit clearer where we're

going there on the dyno.

Let's jump into our data and we'll have a
look at our datalog.

Now straight away what we can see here
though is we are now seeing a little

bit of knock creep in.

And in particular we've got around about
1.5 degrees of feedback knock correction

and that's occurring at around about 4200 RPM
and around about 3.2 grams per revolution.

So we're right on the limit there obviously
of knock.

That is only some fairly light knock so I'm
not going to necessarily want to remove

all of that timing that we just added in
because we did see a good gain there.

So what we can do is make a specific change
to just that area of that map.

So taking into account 3.2 grams per
revolution, 4200 RPM, let's head back

and we'll have a look at our ignition table.

So we are spanned out not to 3.3 so this
is why it's important now, we can make

quite specific changes to just the area that
we are having problems with.

So what I'll do is we don't obviously have
a break point at 4200 so what I'm going to

do is remove some timing at 4000 and
4400 and just to safeguard the engine

as well I'm going to bring this change back
a little bit down in the load,

down to 2.59 grams per revolution as
well.

So what we'll do there is we'll just start
by taking 1.5 degrees out.

And I'm just going to smooth that change
as well.

So at 3200 RPM I'm going to remove one
degree and then again at 4800 RPM,

I'm going to remove one degree.

So this just gives a nice smooth shape
to the curve.

Now despite that little bit of knock that
we had there at 4200 RPM,

at higher RPM once the knock retard had gone
away we can see that we didn't have any

knock occurring and we've got rid of that
knock retard by around about 5600 RPM.

You can see the jagged shape there as
the knock retard basically increments back

up to zero.

So what we can do there, I know we're
probably pretty close to the limit,

is we can try adding a little bit of timing,
both below and above that point.

So let's look at 5600 RPM and we're at
2.6 grams per revolution and we do

drop down to 2.3 so let's have a look at that
area there.

So from 5000, I'm going to make it from
6000 RPM there just to get us a little

further away from that area of peak
cylinder pressure.

And we're going to make this change down
to 2.07 grams per revolution and above

so I'm just going to add this time,
one more degree so I'm just going to be

a little bit more gentle with the amount of
timing that I'm adding in.

Let's have a look at our low RPM range,
and our knock didn't occur until 4000 RPM,

we know we've removed a little bit of
timing below that.

So from about 3400 RPM and below we've
got no knock occurring and we're operating

between 1.69 grams per revolution and 3.1
so let's have a look at that area of our curve.

So what we'll do is we'll choose to add a
little bit of timing here, from 2400,

let's add that from, let's go from 1.8 grams
per revolution there and we'll make it

up to 3200 RPM.

So again I'm just going to add one degree
there, just a small change to our timing.

So let's get another run underway and see
the effect of that small change and hopefully

we've got rid of that knock in the mid range.

Alright with our last run complete there,
we can see that our peak power's actually

reading very slightly lower, 312.66 but
essentially it's there or thereabouts

compared to our last run.

What we can actually see though if we
compare the two runs is we have picked

up or smoothed our power very slightly
above that peak power level.

And we've seen another small boost variation
which has affected our peak torque so that's

why we're down to 334 pound foot compared
to our last run at 355.

So essentially the small gain that we've made
in timing there at low RPM and high RPM,

has shown us a very small improvement,
although in the lower RPM that's been

masked by the boost moving around.

So this again is why it's so important to make
iterative changes to your timing,

your fuelling, as well as your boost,
making sure that when you're comparing

two runs that you are actually comparing
apples with apples and that the boost

in particular was the same from one run
to the next.

Let's have a look at our logged data
though and we'll see how our knock

worked out.

So straight away we can see that our knock
samples now, our knock logging

is nice and clean, we've got no activity there
through that run.

So at this point we're probably pretty close
to where we're going to want our timing.

We could do another run there, creep in
another degree through the rev range

and just see how that affects everything,
although given the very very small change

that we've just seen there from that last
run and the fact that we know that we're

right on the brink of detonation or knock
right through that mid range around

peak torque, I'd say there's little to be
gained and we also end up risking having

a much more on edge tune.

So personally I would prefer to err on
the side of caution here and leave the

timing where it is.

There's one last aspect we want to look at
here with our tune optimisation and that

is our air/fuel ratio.

So far through our tune we've simply left
our air/fuel ratio targets exactly where

they were from the Cobb base file and as
I've already noted, this is quite rich,

particularly in the top end, however it's
been found that this particular FA20 DIT

engine does like to run a little bit richer
than what we'd conventionally see.

However for the sake of experimentation,
let's test and see how the engine

responds if we lean that air/fuel ratio out
a little bit.

So let's head back to our Accesstuner software
and what we'll do is we'll go to our closed

loop fuelling target base table and we know
that we are running up around this area

of our table.

And in this instance I'm not going to
change our break points for our load axis.

The reason for this is there's no need for me
to change our air/fuel ratio targets as we

move to higher load than 2.84 so we can
make all of our changes in this area

of the table.

Now what I'm going to do is just make this
change all the way down to 2.08 grams

per revolution.

The reason for this is that we are running
down to that area at higher RPM.

So what I'm going to do is just try for
our first test, leaning out our air/fuel

ratio targets.

What I'll actually do here is make that
change from 2000 RPM and above.

What we'll do is lean out our lambda target
there by 0.02.

Let's do another run and see the effect of that
change.

Now while we are doing this, because we've
leaned out the air/fuel ratio we don't

have the benefit of some additional fuelling
for combustion cooling there.

We would want to be very mindful of
knock during this next run.

Let's get our run underway and see the
effect of that change.

So with our run complete there we can see
that essentially our power hasn't been

affected at all.

We're sitting at 311.5 so we've actually
lost a touch of power but obviously we've

got some run to run variation in there.

Likewise our torque, we've picked up a little
bit to 338 but again that's within our

expectations of run to run variation.

Let's have a look at our datalogging though
for a full picture of what was going on.

So what I've done here is I've added in our
measured air/fuel ratio and our closed loop

target and while we can't compare these
directly now to our previous runs,

we know that we have leaned out that
target by 0.02 lambda so we can

see through the low RPM area we're
sitting with a target of 0.80 lambda,

at higher RPM that's sitting at 0.75 where
we were at 0.74 thereabouts.

So we haven't really seen any major
difference in our power.

What we are interested in of course is
our knock and we can see from our

group of data there on our knock control,
that we haven't actually ended up with any

knock feedback so still no detonation
occurring with that leaner mixture.

However in this instance I would tend to err
again on the side of caution.

I'd probably leave my lambda targets a little
bit rich.

We haven't seen any dramatic improvement
in power as I've leaned that lambda target out

and it's a benefit here with this particular
engine, particularly when we are pushing

that turbo very hard, out on the street I'd
prefer to have the benefit of that additional

fuel to help cool that combustion charge and
protect us.

We're obviously only doing short ramp runs
here on the dyno and what we may see

out on the road is the benefit of that richer
air/fuel ratio is going to actually help

protect our engine when we're running the
car through the gears and we're under

sustained high load at high RPM and high
speed.

Now we've got our new calibration flashed
into the ECU, it's also saved onto our laptop

as well as to our Accessport, so our job of
tuning is complete.

Normally at this point we would move onto
step six of our six step process which is to

confirm our calibration out on the road or
the racetrack.

Unfortunately due to our restrictions here,
relying on Cobb's facility and their own cars,

we're not able to do that.

However the process is exactly as you've seen
in the body of our course, you're simply

going to take the car out onto the road.

Using our datalogging we're going to confirm
first of all under cruise conditions that our

closed loop trims are under control,
meaning that our mass airflow sensor

calibration is still sound.

We're also going to be checking to make sure
that we're not seeing any knock under cruise

or transient conditions.

Once we're comfortable with that,
we can also test under some wide open

throttle pulls through the gears, again
datalogging to confirm that our boost

is under control, our air/fuel ratio and our
closed loop trims are under control and

most importantly making sure that we're
not seeing any knock activity.

If we are seeing any knock activity in
particular, we can use our datalog to

then highlight exactly where in the
calibration that's occurring so we can

pinpoint that area and remove timing just
from the areas where it's necessary.

So this brings us to the end of our worked
example and if you do have any questions on

this worked example, please ask them in the
forum and I'll be happy to answer them there.