This is only a preview of the February 1991 issue of Silicon Chip. You can view 47 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. Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
|
REMOTE CONTROL
BY BOB YOUNG
Galloping ghost - the missing link in
the evolution of proportional control
These days, modellers take proportional
control systems for granted, without realising
just how much better they are than the systems
they made obsolete. This month, we take a look
at a system which was popular before digital
electronics took over.
. Over the past few months, we have
discussed much in relation to the
modern radio control system but we
have yet to answer some vital questions: "what exactly is digital proportional control; why is it so good; and
how did it evolve?"
All of these questions will be answered in the following series of articles.
However, before progressing further, I would suggest that the reader
refresh his memory with the first article in this series, published in the
October 1989 issue, for much of the
history of the developments leading
up to this system was covered there.
For those who missed that article,
it is sufficient to state that prior to
proportional control, the typical
multi-channel R/C system was usu-
1 •
ally some sort of audio tone system
using either filters or tuned reeds for
decoding. Tuned reeds were by far
the most popular and successful system, certainly in this country and in
America.
Now the point here is that these
systems only gave two to three simultaneous controls (usually two) but,
most important of all, they only gave
neutral, full clockwise and full anticlockwise control positions. No intermediate positions were available.
This was not really satisfactory for
any form of modelling but despite
these shortcomings, the best flyers
could give demonstrations which
would be hard to distinguish from
those given by flyers today using ·
modern proportional control equipment.
01-lsec
,
I
~-A ~ ~ ~
~
DOWN
MAIN AERODYNAMIC FORCE= NEUTRAL
Fig.1: the inertia of the model plus the elasticity of the air will
average out a rapidly oscillating control to give smooth flight.
If the control oscillates symmetrically about the neutral
position as shown here, the result will be neutral control.
82
SILICON CHIP
We learned to adapt and we learned
all the tricks. A separate servo was
used for elevator trim , one which
could be nudged into the correct position and left there until retrimming
was required. A similar type of servo
was used for throttle as well.
Pulsed controls
Controls could be pulsed to average out the aerodynamic effects and
short pulses only gave small control
throws. Long pulses gave about half
throw and the lever held full on gave
full throw of course. The pulses never
showed up in the manoeuvres and all
looked perfectly smooth - in the hands
of an expert that is.
The average flyer, which included
myself, never really knew what to
make of all of these shortcomings and
thus we never really learned to feel
comfortable with this system. Believe
me, I crashed many a good model
trying to master this system.
Something had to be done, and we
all knew what it was. But the burning
question was how?
This problem, like all problems
when solved, turned out to be easy so easy as a matter of fact, that I was
stunned at the time at just how simple
and effective the system really was.
But let me tell you that the problem
occupied many of the finest minds in
the electronics world and baffled most
of them until the early 1960s when
two Americans, Don Mathers and
Doug Spreng, developed the proportional system as we now know it.
So complete was their approach
that to this day little has been changed
except the technology applied to solving the practical problems. However,
we are now a little ahead of our story,
I
TONE OR
CARRIER ONI
CONTROL POTS
REPETITION RATE
~1
J
ONI
I
0.1-1sec
#I
J
I
50% MARK/S PACE RATIO
_.,.
B1
1
!.i...
-T
B2 ,
!.i...
Fig.2: the "Galloping Ghost" system used mark-space/pulse rate encoding &
decoding. In this scheme, the decoded pulses controlled a relay & this in turn
was used to rapidly switch a small electric motor first in one direction & then
the other. The model thus responded to the mean aerodynamic position of the
control surface.
having missed that important step
mentioned earlier.
The Galloping Ghost
What on earth was the Galloping
Ghost? I can hear the readers now:
"this lad has finally flipped for real
and is now seeing apparitions". Sorry
to disappoint you, but "Galloping
Ghost" is the name given to that missing step in the proportional control
story.
Earlier, I mentioned that the inertia
of the model and the elasticity of air
will average out rapidly oscillating
controls and give a smooth result
around the averaged control positions,
as shown in Fig.1.
Now this phenomenon has two very
important uses:
(1). If interference or weak signal
areas are encountered, then random
noise intruding into the servo amplifier will strike some sort of average
which may be anywhere from full
down to full up. This average is changing rapidly but, combined with the
occasional snatch of uncorrupted
control data, will often keep a model
flying until full control is restored.
This is the factor that made PPM such
an effective system and which renders the concept of "fail-safe" (neutralised flying controls) invalid.
(2). It's also useful in the Galloping
Ghost type systems of constantly flapping servos, in which the rapid flutter is deliberately introduced and
averaged out to a mean aerodynamic
value by virtue of a controlled encoder. This is our missing link in the
chain of development leading to true
digital proportional control.
The basis of the Galloping Ghost
system was the concept of markspace, pulse-rate decoding. In this
system, a small electric motor was
connected directly to the terminals of
a relay which was switched by an
encoder. This encoder was controlled.
through gimballed joysticks driving
potentiometers, suitably arranged to
vary the mark-space, pulse-rate ratios
of the relay switching.
Oscillating control surfaces
The name "Galloping Ghost" came
from the franti c appearance of the
constantly thrashing control surfac es.
To hear a model gliding overhead
controlled by this system was a real
experience - the noise was unbelievable. Of course, the motor armatures
and brushes didn't take too well to
this sort of treatment.
Nevertheless, the system was remarkably reliable considering the
strain on the motors and relays. The
block diagrams in Fig.2 show the
theory behind this truly incredible
and very primitive system. Fig.3 gives
the timing diagrams for various combinations of mark-space ratio and
pulse-rate.
For a modern modeller, versed in
the art of digital electronics, the following system will seem almost incomprehensible, yet it formed a major link in the chain of development
and some really satisfying flying was
done using this system. It also serves
to indicate just how difficult the prob-
lem of achieving proportional control
really was and just how powerful the
techniques of digital electronics are
today
The electronics revolution has
largely been possible due to digital
concepts an d it is difficult for modern readers to realise that in 1960
very few of us had even heard of the
term , let alone had any knowledge of
the techniques involved.
Fig.4 shows the way the two separate control outputs are taken from
the servo drive disc, one for the elevator and one for the rudder. The series of diagrams in Fig.5 shows the
way in which the mean aerodynam ic
force is generated for the two controls; in other words, the decoded
output. The system worked as follows :
·
The transmitter was a conventional
single-channel type using either tone
modulation or straight carrier switching. A mark-space, pulse-rate encoder
controlled the transmitter and delivered to the receiver a coded signal
carrying two streams of data. One
stream cons isted of a signal with a
mark-space ratio varying from 20-80%
to 80-20%
Is Your Product
Or Service
Getting The
Exposure It
Deserves?
25,000 consumers will
read this page. They
need to see your
advertisement if you
want them to buy your
product.
Contact
Paul (018) 28 5532 or
Ian (03) 696 5411 to
reserve this space.
FEBRUARY1991
83
50-50 0.5sec
50-50 0.25sec
50-50 1sec
n
n
25-75 0.5sec
LJ
LJ
75-25 0.5sec
40-60 1sec
Fig.3: these timing diagrams show
various combinations of mark-space
ratio and pulse-rate for a proportional
control system.
The second stream was carried in
the coded signal generated by the
variation of speed at which this markspace signal was repeated, commonly
termed the pulse rate. This rate was
usually varied from approximately
0.ls to 0.8s. The exact rate depended
upon the transit time of the servo
motor.
Both these signals were of course
superimposed upon one another and
needed to be separated at the receiver
end. Now the logical and modern
thing to do would be to use , solid
state, mark-space, pulse-rate detectors
and this was eventually done in some
systems, but once this constantly variable DC voltage is generated, what do
you do with it? Remember, this is in
the days long before ICs and those
nice little op amps.
This was in the electronic dark ages
and electronic solutions were prohibitively expensive. I can remember
buying my first transistor around this
time; it cost me 25 shillings and I
bled for a week. However, we are
ahead of our story again.
Motor decoding
Modellers, being by nature a tight
fisted lot, came up with a primitive
84
SILICON CHIP
but effective system using a single
electric motor for decoding.
Fig.4 shows the mechanical arrangement of the basic mark-space,
pulse- rate decoder. The mark-space
decoder is simply a slotted yoke
mounted and pivoted at a suitable
point to allow the control pin to accomplish almost one complete revolution (80% of a revolution to be exact). This yoke is usually connected
to the rudder of the model, neutral
being at the 12 o'clock position.
If the motor rotates clockwise, the
rudder will turn right, for example,
and if the motor rotates anticlockwise,
it turns left. In practice, the relay is
switching constantly between open
and closed at a mark-space ratio determined by the transmitter encoder.
Thus, we now have a yoke which
will deliver to the control surface a
deflection which is rapidly and
equally varying around neutral if the
mark-space ratio is exactly 50%. If
the mark-space ratio is varied, then
the avtJrage position will shift from
neutral in an amount directly proportional to the mark-space variation
(Fig.5d & 5e) .
Fig.4 also shows the arrangement
for the pulse rate detector, which is
simply another yoke set at right angles
to the first and again pivoted suitably
to all9w the control pin to rotate freely
over 80% of a revolution.
If the pulse rate is taken to its highest, the motor oscillates quickly about
the 12 o'clock position and the averTO
ELEVATOR
DRIVE PIN
age deflection is considered to be "up"
elevator. (The rate control was usually connected to the elevator in the
model). As the rate was slowed, the
motor had more time to rotate and the
average position moved towards the
"down" position.
At the slowest rate, the elevator was
in the full "down" position. Note here
that, in effect, full down is really neutral elevator. Fig.5c shows the timing
of the arcs involved. However, the
dwell time at servo reversal added to
the down elevator effect and the trimming of the model biased the system
to account for the rest.
In this way, we achieved two proportional controls from a single channel receiver - quite a step forward but
well short of the requirements for a
model aircraft, in which four simultaneous controls are tL~ minimum
required for full control.
Note also that the decoding was far
from perfect, with a great deal of
mixing between controls occuring
(Fig.5 illustrates this quite clearly).
Theoretically, the transmitter joystick
should have been positioned in the
centre of a square hole, but in order to
overcome some of the problems of
the control mixing, which occured
mainly at control extremes, there were
some strange configurations at the
borders of this no longer square hole.
In fact, you had to learn to fly not
to the spring centre of the sticks but
to the feel of the model. If it was
going where you wanted it, then it
really did not matter much where the
stick was. You never had time look at
that anyway. Believe me , it was all
you could manage to keep the model
airborne.
It was ingenious, simple, cheap and
diabolically primitive. However it
worked, and worked well and gave
many of us a taste for a much more
professional system, if only we could
work out how to do it, that is.
New developments
RUO.OER
NEUTRAL
DRIVE DISK ON MOTOR OUTPUT SHAFT
Fig.4: this diagram shows the
arrangement of a mechanical markspace/pulse rate decoder. The markspace decoder consisted of a slotted
yoke connected to the rudder, while
the pulse-rate detector used a second
yoke (connected at right angles to the
first) to control the elevator.
There were several important developments that followed on from this
basic system.
One was the system marketed by
Dr Walter Good in which two audio
tones were transmitted simultaneously, each modulated for mark-space,
pulse rate. This gave four simultaneous proportional (galloping) channels.
Motor control was achieved through
a trimable (positionable) servo driven
~ t ,_
0
"
~
™
O
o~: +~J, - ~:,H"
"""'""'""'""''"
01.~i ~ ~
ol~. 1
~
0
RUDDER
{/
\) I.~
0.25sec
~h\/
EFFECTIVE
--Fi~,;::;,;:;,_,_EL_EV-AT_O_R
't'.f"
~
,Jj -
~,---
lsec
NEUTRAL
MIXING OR~:RFECT
DECODING EFFECT
Fig.5: this series of diagrams shows the way in which the mean
aerodynamic force is generated for the two controls; in other words,
the decoded output. Note how the rudder moves away from the
neutral position for mark-space ratios other than 50:50, while the
elevator position varies according to the pulse rate.
from a pulse om1ss10n detector on
each of the two tone channels - one
for high throttle and one for low.
This was a very sophisticated and
complex system for its time and held
us all in awe whenever the very occasional example showed up on one of
our fields. Here was a taste of what
was soon to come. The feel of the
transmitter was marvellous and those
twin proportional sticks felt just right,
compared with the five cumbersome
lever switches on our reed transmitters. Never mind that the performance of the rest of the set was as primitive as it is possible to imagine.
A model fitted with three flapping
controls was a sight to behold but at
the time, we thought it was great.
They were exciting days and it is
impossible for modern modellers to
comprehend the degree of yearning
that the dream of true, simultaneous
control generated in us.
One must remember that at that
time, a transistorised anything was
rare on our fields, for many of us
were still flying single channel valve
sets. As a matter of fact, I did not stop
producing valve super-regenerative
receivers until around 1969, long after I began producing a solid state
superhet 6-channel digital system.
Even then, the only thing that stopped
me was the difficulty in obtaining
valves and 22.5V batteries.
The demand for a valve system was
always there. In this respect, a trip to
the flying field was always very interesting, for one would see operating,
side by side, valve super-regen single
channel systems, superhet reed, 8channel audio tone, Galloping Ghost,
Walter Good dual Galloping Ghost,
and a host of home-made systems
using many and varied approaches.
The final step in the mark-space
systems came with the development
of solid state mark-space, pulse rate
decoders. These delivered a DC output proportional to the input. From
here the difficult part began. Just what
could you do with this sort of output?
These days, a DC-coupled, closed loop
feedback servo would be possible but
this was well before op amps.
The answer in those days was to
feed this power into a magnetic actuator. This device consisted of a circular magnet inside a coil wound at
right angles to the magnetic field. This
was lightly spring-loaded for return
to neutral. Thus, with the polarity
aligned in one direction, the magnet
would begin to deflect clockwise and
with the current flow reversed, in the
counter clockwise direction.
The greater the power fed into the
coil, the more deflection obtained.
They were not very powerful, giving
about 3-5 oun,ces on a spring balance,
but this was enough for a lightlyloaded, slow-flying aircraft.
They were quite accurate and gave
true and smooth proportional control,
free of the gallop and decoding mixing associated with the "galloping
ghost" systems.
They were, however, not the answer we were looking for. That had to
wait until two very gifted Americans
made their contribution to the field
of radio control and that story will be
told in next month's issue. See you
then.
SC
FEBRUARY1991
85
|