This is only a preview of the April 1999 issue of Silicon Chip. You can view 34 of the 96 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. Articles in this series:
Articles in this series:
Items relevant to "High-Power Electric Fence Controller":
Items relevant to "Programmable Thermostat/Thermometer":
Articles in this series:
Items relevant to "A Rev Limiter For Cars":
Purchase a printed copy of this issue for $10.00. |
Autopilots for
radio controlled
model aircraft
Everyone is familiar with the concept of
autopilots for aircraft. They take over control
of the aircraft and let the pilot have a rest
from the humdrum of normal flight. But the
concept of an autopilot for a radio controlled
model aircraft is quite different.
By BOB YOUNG
The difference between an autopilot for a jetliner and one for a model
aircraft is that while a full-scale aircraft always has a skilled human pilot
on hand to intervene, an autopilot
for a model aircraft lets an unskilled
operator do the flying. A real aircraft’s
autopilot controls all flight functions,
while a model aircraft’s autopilot
controls only aileron and elevator, as
we shall see.
The story of how autopilots became
a viable proposition harks back to
early Russian attempts at biological
control of crop pests. If that seems
like an odd precursor, read on.
The development of autopilots for
model aircraft goes back a long way
and probably began with the first R/C
models. However, the first serious
commercial attempt to create a viable
low cost autopilot would probably be
the device designed by the renowned
American R/C pioneer Maynard Hill.
Maynard set several altitude records
for R/C models in the early 1960s,
some reaching nearly 10,000 metres.
Now the problem with high altitude
flying is that of keeping the model in
the correct attitude to achieve best
rate of climb. At any sort of altitude
it becomes almost impossible to tell
if the model is climbing or diving, let
alone tell if it is at the best climb angle.
It is also very easy to tear the wings
off a model if the pilot is unaware of
speed build up in a dive. It is here
that an autopilot becomes invaluable.
Maynard’s device used radioactive
isotopes, such as those found in some
smoke detectors, in the sensors and
these were mounted on the wing
tips, nose and tail of the model. The
principle of operation of this device
was extremely clever and relied upon
the gradient of the electrostatic field
surrounding the Earth. The voltage
of this field diminishes with altitude
and so if the model raised or lowered
its wing tips or raised or lowered its
nose or tail, a voltage differential
was detected by the sensors. This
was amplified and used to apply the
appropriate corrections to the model
aircraft’s flight controls.
While this device worked very effectively, the radioactive components
caused concern and it never found its
way into popular usage.
Russian development
This photo shows the original crop spraying model aircraft which was fitted
with an optical autopilot so that unskilled users could fly it.
4 Silicon Chip
The autopilot described here had its
beginnings in 1975, in Russia, when
Igor Tsibizov, fresh from military
service, arrived at the SKB-AM (Stu-
This view of the crop sprayer shows the hole at the end of the wing spar through
which the paper balls were ejected.
dent’s design office for aeromodelling)
and began work there. Igor was soon
approached by A. S. Abashkin, the
chief of a mechanisation department
at the Kishinev Institute of Biological
Methods of Plant Protection. His brief
was in regard to the development of
model aircraft to scatter wasp larvae
over crop fields.
It appears that the USSR was
amongst the first countries in the
world to recognise that the large-scale
use of chemicals in farming was not
a wise practice. They therefore embarked on an extensive program of
biological methods of pest control and
this became very large in relation to
the rest of the world. In 1990 alone,
the USSR claimed to have treated
27.6 million hectares with a parasitic
wasp (Trichogramma) that lays its eggs
inside the larvae of crop pests.
Now the cost of delivering the wasp
larvae was, and still is, a serious concern. Normal methods of delivery include tractor, aircraft and helicopters,
with rates of treatment ranging from
100 to 250 hectares per hour. Divide
250 into 27.6 million and you get a
lot of hours.
It turns out that this Trichogramma
wasp is very tiny and this means that
the aircraft are flying with a very peculiar cargo, about 2kg of tiny paper
balls! There have been over 70 spe-
cies of Trichogramma used around
the world but of these only about 20
species have been mass-reared for
field use. And this in itself is a very
interesting story.
In the project that Igor worked on,
the biological plants cultivated the
larvae to the chrysalis phase, at which
point they were placed in darkness,
whereupon their development was
suspended. The transformation of the
chrysalis into an adult wasp can only
take place in the presence of light.
The chrysalises were then packed
into paper balls about 10mm in dia
meter, without any food. Still without light, the chrysalis remained in
suspended development. Just prior
to being dropped over the fields, the
paper balls were pierced with a sharp
instrument, thus letting in sufficient
light to allow the wasp to resume
development. Within 24 hours of
being dropped, the adult wasp would
emerge from the paper ball and immediately look for a suitable host for
its eggs. The eggs develop into larvae
which eventually kill the host, thus
achieving the pest control function.
Approximately 400 balls were
dropped per hectare and in tests
conducted in Moldavia and Krasnodarskiy Krai, the system worked well.
But clearly, the balls weigh practically
nothing and so a full size aeroplane is
flying almost empty, merely carrying
air in the balls. With aircraft and helicopters being very expensive to run,
it becomes obvious that there are great
savings to be had using model aircraft
to deliver the wasps.
However the real saving comes
about if the farmers can manage the
model themselves and it is here that
the autopilot is not a luxury but an
absolute necessity.
Solving the problems
The development of a suitable
model aircraft was a major project.
It was immediately apparent that
the highest level of automation was
required and all of this had to fit into
a model aircraft of modest dimensions. Remember here that all of this
took place from 1975 onwards. Miniaturisation was only just beginning
and the autopilots available in those
days were confined to Maynard Hill’s
electrostatic system and some small
military gyroscopes, which were far
too big and bulky.
Maynard’s device proved to be
unsuitable because the flying took
place at an altitude of no more than 3
metres and changing atmospheric and
Earth field conditions caused serious
instability. And the one thing you do
not need when cruising at 3 metres
and 100km/h is instability!
Igor went through an intense period
of trying all sorts of devices, ranging
from the simple to the exotic, before
April 1999 5
Flying only a few metres
above the crop, the plane
would release hundreds of
tiny paper balls over each
hectare. Each paper ball
was pierced at release so
that the developing wasp
inside could escape and
release its eggs.
finally settling on an optical system
of sensing.
In the optical system an array of four
photodiodes was arranged to “look”
in four directions, to the left, right,
front and rear. In effect, the diodes
“look” at the horizon and they sense
the line between the bright sky and
the darker ground.
Operating principle
The operating principle is quite
simple. As long as the outputs from
all four photodiodes are equal, the
output from the autopilot is zero. If
the model begins to drop its nose, for
example, the rear diode will “see” the
bright sky and the front diode will
“see” the darker ground. This will
develop an error voltage across the
sensor array and this voltage is fed
to the processor in the autopilot. The
processor then sends a correction to
the elevator servo which results in an
UP elevator correction being applied.
As the nose comes up, the error
voltage diminishes until equilibrium is re-established. The more
clearly defined the horizon is, the
better the system works. Two obvious
disadvantages to this system are that
no night flying is possible and snow
and haze can cause serious resolution
problems. However, for most conditions the system works well.
6 Silicon Chip
It is important to note that this
autopilot will not control altitude or
direction (yaw). It is merely a device
to maintain level (horizontal) flight.
However, this is the hard bit and
it leaves the pilot plenty of time to
concentrate on direction and height.
The original agricultural aircraft
was a rather unusual looking model
fitted with some very unusual mechanics. The fuse
lage was a basic
fibreglass shell with one former upon
which almost all of the mechanical
components, engine, scatter mechan
ism, main undercarriage and struts,
were mounted. The engine was a
standard 10cc 2-stroke, attached via
a shock-absorbing mount. The fuel
tank was under the engine and thus
used a fuel pump.
The wing was foam covered with
polyester film. The wing spars were
made of titanium pipe 18mm in dia
meter and also acted as spray ducts
for the paper balls. The scatter mechanism was driven from a small turbine
mounted in the air collector. It took
the balls, punched a hole in them and
then delivered the ball along with a
portion of air to the hollow wing spars.
Thus the balls shot from the wing
tips. An additional outlet shot balls
directly downwards. Up to 2000 balls
could be carried per flight.
In operation, the aircraft treated
approximately 100 hectares per hour
and was flown successfully by unskilled operators; all in all a significant achievement.
Present day autopilots
As a footnote to this biological control story, the patents to the autopilot
were sold overseas and form the basis
of the HAL-2100 and PA-1 autopilots
now available in most model shops. It
is also sold as the Graupner AP-2000.
Included in this article is a photo of
the latest version, soon to be marketed
by Silvertone Electronics.
As can be seen from the photo,
there is a mushroom-shaped module
and this houses the four photodiodes. The module is mounted under
the model in a very precise manner.
There a number of important points
in the installation. Briefly, they are
the alignment of the sensor head in
the correct sense; ie, the front diode
of the fore/aft pair is actually pointing
to the front of the model. To assist in
this, the housing is marked with two
small arrows, one for the “+” mode
and one for the “X” mode.
In the “+” mode, the diodes point
to the front, back and to the two sides.
In the “X” option (45° to the line of
flight), the photodiodes point to the
left/front, right/front, left/rear and
right/rear. This mode is available
because it sometimes helps eliminate shading from wing-mounted
undercarriages and mufflers on side
mounted engines. A DIP switch is
used to select this option.
The second point in the installation
is that the alignment of the diode
array must be perfectly horizontal
in relation to the tailplane. Finally, a
trainer installation should have 2-3°
pitch offset which will result in a
gentle climb.
It is not recommended that the
module be mounted on the top of the
model because bright sunlight can
cause serious problems. Often the
model will turn towards the Sun and
possibly enter a shallow dive. The
photodiodes in the array are set well
back in a small tube in the sensor
head moulding to provide additional
shielding from bright sunlight.
Before each flight it is important to
check that these holes have not been
blocked by dirt, grass or other debris.
A blocked or dirty hole will cause a
serious imbalance in the sensor input. Side mounted motors present a
This is the current model autopilot which is microprocessor controlled. The
mushroom-shaped module contains the four photodiodes which look at the
horizon. This model will be marketed by Silvertone Electronics.
particular problem here because the
oily exhaust gas is sprayed over the
sensor head; this installation is not
recommended.
Photodiode memory
One interesting point in regard to
the photodiodes is the problem of
memory. If the sensors have been left
in the dark for a length of time or the
model has been stored, initially they
may not work correctly. A simple
analogy would be if a person is kept
in the dark for several hours and then
brought out into bright daylight. It
then takes some time for that person’s
eyes to adapt to the higher light levels. Thus, it is recommended that the
module is left in daylight in a bright
t
Shop soiled bu
!
HALF PRICE
area for at least 12 hours after removal
from prolonged darkness. This allows
the module to adjust to the light levels
and balance itself. It is important to
ensure that all sides of the module
are exposed to the same light levels.
The output of the photodiode array
is fed into a microprocessor which
then applies the appropriate corrections to the two main flight controls,
aileron and elevator. In essence, this
is similar in action to the in-line mixer
published in the July 1997 issue of
SILICON CHIP, in that the output from
the receiver goes into the autopilot
and the servos plug into the auto
pilot, making it an in-line device.
However, in this case the mixing occurs between the light source inputs
and the channel inputs, not between
channels.
As the transmitter control sticks
are moved off-centre, the effects of
the autopilot corrections are reduced.
However, when the sticks are returned
to neutral, the full effect of the auto
pilot control corrections are applied
to the controls and the model returns
to horizontal flight.
Thus if a beginner is using the system on a model aircraft and he gets
into difficulties, then all he need do
is let go of the controls and the model
will return to horizontal flight auto
matically.
An additional channel is required
to adjust the gain (or sensitivity) from
the transmitter for the most successful
operation of the optical autopilot. The
gain control sets the amount of correction the autopilot will apply to the
flight controls and the gain may be set
14 Model Railway Projects
THE PROJECTS: LED Flasher; Railpower Walkaround Throttle; SteamSound Simulator; Diesel
Sound Generator; Fluorescent Light Simulator; IR Remote Controlled Throttle; Track Tester;
Single Chip Sound Recorder; Three Simple Projects (Train Controller, Traffic Lights Simulator
& Points Controller); Level Crossing Detector; Sound & Lights For Level Crossings; Diesel
Sound Simulator.
Our stocks of this book are now limited. All we have left are newsagents’ returns which means
that they may be slightly shop-soiled or have minor cover blemishes.
SPECIAL CLEARANCE PRICE: $3.95 + $3 P&P (Aust. & NZ)
Order by phoning (02) 9979 5644 & quoting your credit card number; or fax the details to (02) 9979 6503; or mail your
order with cheque or credit card details to Silicon Chip Publications, PO Box 139, Collaroy, NSW 2097.
April 1999 7
in flight using a proportional channel.
Gain control is needed to prevent the
model from “over controlling”, a situation where the autopilot applies too
much correction to the flight controls
and quiet literally shakes the model.
It is very useful in different flying
conditions.
The extreme low gain setting
switches the autopilot off completely.
Some autopilots feature a programming function that will allow a preset
reduction in gain when flying without
the extra gain control channel (4channel systems).
There is one very important final
point when setting up a model with
an autopilot. It makes good sense to
install a throttle fail-safe device such
as that published in the June 1997
issue of SILICON CHIP. Because the
model will now fly perfectly well
by itself, a radio failure becomes a
serious business. The model can fly
long distances under perfect control
from the auto pilot and could land
goodness knows where. The throttle
fail-safe will shut off the motor upon
loss of signal and the autopilot will
bring the model down safely in close
proximity to the field.
Learning to fly
Learning to fly with an autopilot
fitted is an interesting experience
and it certainly speeds up the process
remarkably as well as increasing the
life span of the models and improving
safety all round. In a model helicopter,
the autopilot would be used to control
the “cyclic pitch”, thus keeping the
rotor disc horizontal. Combined with
a tail rotor gyro, this device can take
most of the pain out of learning to fly
helicopters.
Acknowledgments
These cross-sectional drawings show the construction of the crop-spraying
model aircraft and the hollow tubing used as wing spars and spray outlets.
8 Silicon Chip
• My thanks to Dmitry Bernt,
Moscow, for bringing this story to
my attention and providing the translations.
• To Alan Westcott of the Elizabeth
Macarthur Agricultural Insti
t ute,
Menangle, NSW for his assistance in
providing information and advice.
• Worldwide Use of Trichogramma
for Biological Control on Different
Crops: A survey. Li-Ying Li. Guang
dong Entomological Institute, Guang
zhou. China.
• University of California, Riverside. http://insects.ucr.edu/tricho/
SC
tricho.html
|