This is only a preview of the December 1996 issue of Silicon Chip. You can view 28 of the 104 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to "Build A Sound Level Meter":
Items relevant to "Build An 8-Channel Stereo Mixer; Pt.2":
|
Engine braking from an automatic!
Mitsubishi’s int
automatic tran
Mitsubishi’s smart new automatic
transmission adjusts its shift points to
suit the driver’s style. What's more, it
shifts gear in a much more intelligent
manner than previous automatic
transmissions. Here's how it works.
By JULIAN EDGAR
20 Silicon Chip
Many automatic transmissions are
now at least partially electronically
controlled. Some use a hybrid system
of electronic and hydraulic control,
while others are fully electronic.
The latest innovation is the adaptive
“self-learning” au
tomatic transmission, as fitted to the new TE Magna
from Mitsu
bishi. In this system, a
transmission control unit is used to
constantly monitor the driver’s style.
Depending on what it “learns”, it
then adjusts its control behaviour
accordingly.
With this type of system, the economy/power switch fitted to some transmissions is made redundant. Drive
the car hard and the gear changes will
occur at higher engine speeds; gently
toddle along and the changes will slur
through early. However, there’s more
to adaptive shift control technology
than changing gear shift points, as we
shall see.
Common problems
In most modern automatic transmissions, gear selection is based mainly
on throttle position and vehicle speed.
However, there are many situations
where the gear selected is not appro
priate for the driving conditions – the
control system literally selects the
“wrong” gear.
A good example of this occurs when
driving an automatic car uphill along
a winding road. In this situation, a
slight easing of the throttle prior to
each corner can result in an up-shift,
the transmission then down-shifting
again after the corner has been negotiated.
Obviously, if driving a manual car,
the driver would not change into a
higher gear prior to entering a corner.
Because the automatic transmission
does, it losses engine braking and so
some degree of control is lost.
Fig.1: Mitsubishi’s earlier “Fuzzy Shift Scheduling” system allowed the
transmission control unit to use engine braking and hill-climbing modes.
Self-learning was not incorporated into the system, however.
Downhill driving in a conventional
automatic also results in a lack of
engine braking, unless the driver
manually selects a lower gear. (Incidentally, it is good driving practice
to manually lock a conventional auto
into a single appropriate gear in both
of the above scenarios – lazy drivers
take note!)
Getting back to the TE Magna, Mitsubishi’s research indi
cated that a
conventional automatic transmission
could be in the “wrong” gear for a
given situation up to 60% of the time
telligent
nsmission
– an extraordinarily high figure and
perhaps only possible if the car were
being driven on a racetrack! However,
there are certainly times when the
control system needs more brains.
For Mitsubishi, the first step in
overcoming these problems involved
the development of “Fuzzy Shift
Scheduling” – see Fig.1. In this system,
additional inputs are used to allow
the system to select from three shift
modes: engine braking, standard and
uphill. The current Mitsubishi system
is an extension of this design.
Fig.2 shows the basis of the new
system. The engine Electronic Control Unit (ECU) and the Transmission
Control Unit (TCU) are linked, and
exchange data on engine speed and airflow rate (ie, engine load). In addition,
the TCU receives additional inputs on
the throttle position, brake operation,
steering angle and the transmission
shaft speeds. The throttle opening is
derived from a throttle position sensor,
the frequency and/or duration of brake
operation by monitoring the brake
Fig.2: the new Mitsubishi
adaptive system accepts
additional inputs,
including steering angle
and the frequency and/
or duration of brake
application. This allows
the system to better
calculate appropriate
shift behaviour during
downhill coasting and
to match the style of the
driver.
December 1996 21
Fig.3 (above): if the TCU selects a gear which is too low and thus provides
excessive engine braking, the action of the driver applying throttle will
cause a correction to the downshift. Conversely, applying the brakes
excessively when coasting down a hill – as in Fig.4 (right) – will cause the
TCU to shift to a lower gear, thus increasing engine braking.
light switch, and the steering angle by
a dedicated sensor.
The speed of both the input and
output shafts of the auto transmission
is also measured. This allows the TCU
to monitor road speed and to calculate
the amount slippage occurring through
the transmission. It can then use these
inputs as a feedback mechanism to
reduce “shift shock”. In some versions
of the system, longitudinal and lateral
accelerometers are also employed.
Engine braking
Engine braking is achieved by
calculating an index called “engine
brake applicability”. This is carried
out by a so-called “neural network”
which links together the road gradient,
vehicle speed, braking frequency and
steering angle with varying degrees of
importance.
The influence of these various
factors depends on empirical data
originally gathered by monitoring the
gear-shifting behaviour of experienced
drivers.
The aim here is to approximate the
decision-making process adopted by
a driver in a manual car.
However, while downshift timing
is primarily controlled by the “neural
network” from empirically-collected
data, the calculat
ed timing is not
appropriate for all drivers because
of their individual preferences and
driving styles. A feedback mechanism
dubbed “Learning Control” has therefore been added. This judges the driver’s dissatisfaction with the amount
of engine braking being provided by
the TCU and corrects the downshift
condition until the driver’s preference
is reached.
This judgement is carried out by
monitoring the frequency and/or duration of brake use and by monitoring
throttle variations when the vehicle is
coasting downhill. If throttle needs to
be applied (Fig.3) then the transmission is in too low a gear. Conversely,
if the brake is applied (Fig.4) the gear
selected is too high.
Fig.5 shows how the learning
system changes the ease with which
up-shifts and down-shifts occur in
response to brake and accelerator
movement during downhill coasting.
Variable shift patterns
Fig.5: the self-learning behaviour of the system can be seen here,
where the ease of selection of either a downshift or upshift varies
with the driver’s preference – as sensed by the TCU through brake
or accelerator application while driving downhill.
22 Silicon Chip
A conventionally-controlled transmission has a shift pattern similar to
that shown in Fig.6. An upshift from
second to third, for example, occurs
at a certain combination of throttle
position and vehicle speed. Similarly,
at another precise mix of speed and
throttle, the downshift from third to
second will occur.
It’s this fixed approach which
causes the problem of upshifts before
corners when climbing a hill. To
avoid this, it is necessary to move the
upshift lines to a higher speed range.
Just how much the upshift points are
moved depends on the road gradient,
as derived from the TCU sensors.
Variations in individual driving
styles also require changes to the shift
points. For example, a “sporty” (or
aggressive) style means that the lower
gears need to be held to higher engine
speeds and also selected more readily.
The driving style is evaluated by a
variable that Mitsubishi’s engineers
call the “Sporty Driving Index”.
The “Sporty Driving Index” is calculated by selecting the larger of two
input factors – either the engine load
index or the tyre load index.
On some Mitsubishis (but not on the
Australian Magna), the tyre load index
is calculated by comparing the actual
lateral and longitudinal accelerations
with the maxima of which the tyre is
capable. Similarly, the engine load
index is calculated by measuring the
actual acceleration and comparing
this with the maximum possible acceleration.
How hard the car is being cornered
or accelerated varies the “Sporty Driving Index”, with the shift point maps
then moved as a result. Fig.7 shows a
schematic summary of the complete
TCU system.
Does it work?
According to Mitsubishi, the new
system selects the correct gear for
80% of the time. This represents a
considerable improvement on the 40%
of a conventional auto transmission
and 55% for their second-generation
fuzzy system.
In Australian Government AS2877
fuel economy tests, the V6 automatic
transmission Magna has equal econ
omy to its manual equivalent on the
highway cycle and is only 5% worse
Fig.6: in a conventional shift control system the up and down changes
always occur at a precise combination of speed and throttle position.
in the city cycle. By comparison, the
current model automatic Commodore
is 6% worse than the manual version
on the highway and 9.5% worse on the
city cycle. It would certainly appear
that the more sophisti
cated transmission control system of the Magna
yields economy benefits!
So what’s it like to drive? We took
an automatic TE Magna sedan for a
run to find out.
On the road we found that the
electronic control system had some
noticeable advantages over more traditional transmission control systems.
Most obvious was the transmission
down-changing to provide engine
braking when slowing for a red traffic
light, for example. And on country
roads, the downhill engine braking
was also noticeable.
However, many of the traditional
disadvantages of an automatic transmission appeared to remain. Any
demand for instant power (eg, when
overtaking on a country road) still
results in a relatively slow response,
there being approximately a 1-second
time lag for the transmission to “think”
and then change down a gear.
A quicker response was possible
by manually changing down. Jerky
changes also occurred in some situations – for example, when accelerating
hard away from a standstill and then
suddenly lifting the throttle.
That said, Mitsubishi’s new adaptive control system represents a real
improvement in automatic transmission technology. It matches the
shift points to suit the driver and it
provides superior shift patterns in
certain driving situations. And it does
this unobtrusively.
Certainly it never occurred to me
that the transmission was pre-empting
my decisions or matching its change
SC
patterns to my driving style!
Fig.7: block diagram of the TCU. The shift pattern is calculated from a range of input data.
December 1996 23
|