This is only a preview of the March 1991 issue of Silicon Chip. You can view 43 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:
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
The development of digital
proportional servos
Last month, we discussed the difficulties that
arose in the very early development of
proportional control and examined some of the
primitive systems which preceded the modern
digital R/C system. The most primitive was the
galloping ghost system which ultimately led to
the early analog systems.
It is difficult for modern modellers,
who see only digital sets which are
virtual clones of each other (even
down to interchangable servos), to
comprehend just how many different
types of systems were in operation in
the early sixties. A trip to the flying
field in those days was really interesting, for you never knew just what
would show up next.
This was enhanced by the large
numbers of modellers who scratchbuilt their own equipment. Thus it
was possible, on one day, to see singlechannel valve sets operating alongside tuned reed, tuned filter, galloping ghost, Walter Good, analog proportional and digital proportional
CLOCK
RESET
CH1 OUT
CH2 OUT
17. . ._____________.I
J7_________________
CH3 OUT
CH4 OUT
Fig.1: serial to parallel conversion is performed in the remote control
receiver. In this process, a series of pulses from the transmitter are
converted by the receiver to servo control pulses.
74
SILICON CHIP
sets, both home made and commercial.
All of these early attempts were
useful but far from satisfactory, lacking accuracy, speed and power in the
control actuators. They also lacked
reliabilty, simultaneous operation and
the required number of channels.
Those days have long gone and while
I do not miss the unreliability, I do
miss the enthus iasm generated from
our quest for perfection.
However, in the very early 1960s
there occurred one of those quantum
leaps in technology that result in one
system being adopted as an industry
standard, due to the fact that it delivers exactly what the application calls
for. Such was the case with the development of the digital proportional
system as we now know it. It cured
all of the above problems in one fell
swoop and gave unlimited numbers
of truly proportional and simultaneous controls, coupled with unheard
of reliability. Thus did the new age of
R/C modelling sweep away the old.
At the very heart of this revolutionary discovery - and believe me it was
truely revolutionary, breaking away
completely from all lines of development to that date - was the proportional servo as we now know it.
So comp lete was the development
by Doug Spreng and Don Mathers
(USA) in the early 1960s, that the
concept remains virtually unchanged
to this very day.
Our article traces the changes in
technology that have improved the
operation and reliability of the concept, but the Spreng and Mathers
touch is still easily identified in the
most modern PCM equipment available today.
The system they developed was the
Pulse Position Modulation (PPM)
system in which a series of pulses are
transmitted in serial form and converted in the receiver (Rx) via a serial
to parallel decoder (see Fig.1). This
results in a number of output pulses,
usually from 2 to 8, with the pulse
width directly related to the width of
the transmitted pulse. This in turn is
directly related to the position of the
control stick on the transmitter.
Usually, pulse widths run fully
counterclockwise (CCW), 1.0ms; neutral 1.5ms; and fully clockwise (CW)
2ms. Full control is available over
this range in steps determined by the
sensitivity of the servo amplifier. This
minimum step is termed the "minimum impuls e" and is typically
.005ms.
Thus , approximately 100 steps are
available each side of neutral (centre)
on a good quality servo. These steps
are so small that to all practical purposes they cannot be felt, and the
servo appears to be slaved directly to
the transmitter stick. This is the concept of "proportional control".
Herein lies the secret of success of
the digital system: fantastic servo
performance. Fast, powerful and extremely accurate, it fulfilled all of our
dreams. Here was the most clever
development in model electronics and
it has yet to be surpassed.
For this reason, I have chosen to
present the servo first in the following series of articles. Whilst it is traditional to start with the transmitter, I
feel that delivering the control pulse
to the servo input is the easy part.
What takes place in the servo is where
the magic resides.
The question which must be dealt
with first is just how do these servo
amplifiers work? How do they convert the pulse width information into
a clockwise (CW) or counter-clockwise) CCW instruction?
i-•--5
5
0
I
---6
28
3
.•~rl
11:-, •,~rlO
---+-- - -1
0
4
1,5,24,26,27 screw
2 output wheel
3 output arm
4,6 case, top
7 right rack gear
8 left rack gear
9 pot drive gear
10 second intermediate gear
11 pot shaft
12 drive gear
13 first intermediate gear
14, 15 gear pins
16 case, centre
17 motor, 1OW
18 pot wiper contact
19 pot element, 1.5kQ
20 fibre washers
21 decoder PC board
22 grommet
23 connector assembly
25 case, bottom
28 output shaft cap
1
13
~
.
15 - -- ---,----::----
·
- - -- - - 14
~~~
16 -------i
'
----17
~-®·------+-1--20
20------t--~.
4
27
27---__,__; : _ _-i-----·
e
,
22----•o
23✓
F
24
24
Closed loop feedback
Essentially, the modern R/C servo
is a closed loop feedback servo in
which a pulse is fed to the servo
amplifier and compared to the output
pulse from a reference generator on
the servo amplifier board. This reference pulse is controlled by the position of the servo output arm via a
potentiometer. The input pulse and
the reference pulse are applied to a
summing junction and the resultant
26---&
KPS11-11A
Fig.2: this exploded diagram shows all the parts used in a typical servo control.
The key elements include the decoder PC board (21), the motor (17), a servo
feedback pot (18,19), various gears & the output wheel or arm (2,3). Modern
servos are built around dedicated IC servo chips (eg, the NE544 from Signetics)
& are very compact & reliable.
MARCH 1991
75
,---------- -- ----- --- ----- - -- ---,
\
\
I
10011
\
+5V
'
I
I
I
22k
2.2k
GNO
I
I
I
I
l
.,.
.,.
Rl
4.7k
22k
.,.
Cl
Fig.3: an early servo decoder and drive amplifier, using
discrete components throughout. Effectively, the motor
drives the pot (VRl) until the pulses produced by Ql and
Q2 match those at the input.
error signal is then available to control the power and direction of the
servo motor's rotation.
The motor drive circuitry is arranged in such a manner that the servo
always attempts to cancel any error
(zero output to the summing junction), at which point the servo comes
to rest until another error appears.
Fig.2 shows an exploded diagram of a
typical servo, in this case one using
linear and rotary output arms.
100U
As we have control over the input
pulse width from the transmitter, we
therefore have control over the position of the servo output arm. The accuracy of the servo is dependent upon
the servo amplifier sensitivity which
is termed the minimum impulse
power. As noted previously, a good
servo will deliver up to 100 steps
each side of neutral and so we have
complete control over the servo output arm. It is slaved precisely to the
REFERENCE
GENERATOR
OUT
PULSE
IN
SUM APPLIED
TO PULSE
STRETCHER
____
n'-------------
..,
MINIMUM
IMPULSE
NEGATIVE SUM
WHEN INPUT
SHORTER THAN
REFERENCE
GENERATOR
Li
Fig.4: this diagram shows what happens in the decoder circuitry of
Fig.3. Pulses from the input and reference generator are compared
to generate an error pulse which is applied to the servo motor.
76
SILICON CHIP
.,.
transmitter control stick, hence the
name proportional control.
Servo circuit
Fig.3 is a circuit diagram of a very
early American servo, the Orbit PS4D, manufactured by one of the pioneers, Bob Dunham. Orbit lead the
way for many years in high quality
radio control systems and Bob Dunham was a top contest flyer in his
early years.
Transistors Q1 & Q2 form a one
shot multivibrator which is triggered
by the leading edge of the incoming
pulse from the decoder output. Interestingly enough, this servo worked
on a negative input pulse, whereas
the industry standard is now positive. This one shot will generate a
pulse of opposite polarity to the incoming pulse from the decoder and is
called the "reference generator".
The width of this pulse is controlled by the position of servo feedback potentiometer VR1, which is in
turn related to the position of the
output arm on the servo. This pot is
usually driven by the output gear of
the servo mechanism.
Both of these pulses are applied to
the summing junction R1, R2 . Fig.4
shows the effect at this summing junction. Briefly the output of the sum-
INFRA RED NIGHT VIEWER
A limited purchase of some 6032A tubes which were
removed from new, and near new equipment allows us
to offer this Incredibly priced IR NIGHT VIEWER KIT
BARGAIN! You could payover$2000foraviewer which
uses a similar tube. All the tubes are "AS NEW", and are
GUARANTEED not to have any blemishes!
THE PRICE OF THIS UNREPEATABLE
BARGAIN??:
ONLY $259.00
Fig.5: Several servos are shown in this photo, with the one on the right being a
currently available model.
ming junction will be a pulse of either negative or positive polarity,
depending upon which pulse is the
longer.
This pulse is then applied to the
bases of Q3 & Q4, a PNP/NPN pair.
Depending upon polarity, one of these
transistors triggers and the output is
applied to the following pulse
stretcher/Schmitt trigger.
The output of this network is ap plied to motor drive transistors Q7 &
QlO which in turn drive the motor
CW or CCW, depending upon the
polarity of the longer pulse at the
summing junction.
The motor is driven until the potentiometer causes the Ql/Q2 one
shot to deliver a pulse to the summing junction of equal length to the
incoming pulse from the decoder. At
this point, the motor will switch off
and the servo will take up its new
position until there is a change in the
width of the incoming pulse, whereupon the process will start all over
again.
Diode pair D3 & D4 prevent both
sides of the servo amplifier from
switching on simultaneously which
would instantly destroy the output
transistors.
C2 is a noise suppression filter. Notice that one side of the armature is
connected to the motor case for additional shielding.
Feedback resistor R3 changes the
pulse width of the reference generator order to shut the motor down
ahead of time, so that the servo does
not overshoot and go into oscillation
about neutral.
This is a critical function, for if
there is too much damping, the servo
shuts down early and the centring
accuracy is badly affected; too little
damping, and the servo hunts or oscillates and servo current consumption shoots up and the output transistors start to run hot.
The ideal result is called "dead
beat" damping in which the servo
runs to the point and stops instantly.
In practice, this is very difficult to
achieve and I have found that it is
best for centring accuracy if the servo
overshoots and moves back just once.
This gives a very accurate neutral.
Another problem with closed loop
servo amplifiers is that a certain
amount of deadband must be introduced into the system if the servo is
ever to come to rest and not sit there
oscillating.
Capacitor Cl across the summing
junction performs this function. The
value of this capacitor is critical in
establishing the centring accuracy of
the servo. If it is too big, the deadband is too wide and the servo will be
sloppy around neutral. If Cl is too
small, the servo will jitter, causing
excessive servo current drain, and
probably damaging the output transistors and motor.
As you can see from Fig.3, this
amplifier had quite a large component count and could only be fitted
into fairly large servos. Several such
servos are shown in the photo of Fig.5.
The photo of Fig. 7 shows a servo made
by Silvertone Electronics.
Note the double deck PC board in
the Silvertone unit. This was quite a
...... includes a 6032A tube (as per sketch) electronics
kit, ample plastics for the case and a 75mm round I.R.
filter: Can be cut to suit your torch. All you need to finish
off this kit is a good torch, and any old camera lens,
small magnifying glass or an eye piece: The lenses can
be obtained from camera repairers, camera shops, or
can be recovered from old cameras.
VISIBLE LASER DIODE
POINTER - KIT
Based on a "State of the art" 3mW Visible Laser diode, and a matching heatsink/collimatorassembly.
The circuit even has provision for digital switching:
Communications, security, etc. This complete kit
includes everything you need to make the pointer
illustrated except for the batteries (3 AA cells). Our
SPECIAL price for the VISIBLE LASER POINTER
KIT??
ONLY 239.00
Also available is a kit with the same PCB
and all onboard .components, but using an
Infra-Red laser diode and it's matching
heatsink/ collimator assembly: This can
be used for communications, security etc!!
ONLY$99.00
LOOK AT THESE BARGAINS
Some of these are in limited
supply so be quick. All are NEW.
Hall Effect IC's. data supplied ............ 1o for
$20
Stepper Motors ............ ... ... ........... 2 for
$20
Small 3 12VDC motors..... ................. 2 for
$ 5
20VDC to 1SKVDC (S00uA) converters .......$30each
0.BmW HE-NE Laser tubes ............. .... ...... $120each
Stereo (Dual) VU meters .........................•.•...$4each
150ohm 25W wirewound pots............ 2 for
$ 5
[Z][B
OATLEY ELECTRONICS
PO BOX 89, OATLEY, NSW 2223
Telephone: (02) 579 4985
Fax No:
(02) 570 7910
Certified p&p: $5 inAusl. NZ (Airmail):$1 D
Fax orders are accepted with credit card
payments.
MARCH 1991
77
,--- -- - --- - -- - -- ---- ~I
I
I
I
I
I
330!l
I
I
10k
-
1
Q6
AT188
I
VR1
+
2.2k
.,.
01
1N914
+4.BV
I
I
100!l
0.1
I
I
I
I
\
I
I
+GND
0.1
3.9k
47k100k
27k
.,.
.,.
...
.,.
1.5I
BP
* ADJUST TO SET DAMPING
t
ADJUST TO SET DEADBANO
problem to produce and service.
To compound rny problems, I always used an emitter follower stage
on the input to buffer the servo from
the decoder, giving a transistor count
of 11 compared to the 10-transistor
Orbit servo .
The relentless demand for lower
cost and smaller servos eventually
forced us to use the circuit in Fig.6
which is simpler in construction. It
was used by many manufacturers but
it never worked as well as the Schmitt
trigger amplifier of Fig.3.
It was prone to several problems,
amongst which were non-linearity
and changes of deadband and damping with servo position.
One thing both of these discrete
amplifiers shared in cornrnon how-
ever was the fact that both required a
centre tap on the battery pack and
this type of servo was cornrnonly referred to as a "4-wire servo". Motor
resistance was typically 3Q as against
the 11Q motors used in the modern
IC servo. The four wires were signal,
usually a colour; positive 4.8V, usually red or orange; centre tap +2.4V,
usually white; and ground or zero
volts, usually black or brown.
Note that reversing the direction of
one of these servos requires the two
end leads on the feedback pot to be
reversed, and the two armature wires
on the motor to be reversed. Do not
reverse the red and black wires, for
all that will achieve is a burnt out
servo. One point here is that 3-wire
servos can be used in 4-wire systems
Fig.7: this photo shows a servo made by Silvertone Electronics. Note the doubledeck PC board with the parts crammed in to save space. Also visible is the drive
motor and feedback pot.
78
SILICON CHIP
Fig.6: this was a later and simpler
proportional control receiver which
was the first to use a DTL (diodetransistor-logic) to reduce the
component count.
but not the other way around.
Technology finally came to the rescue of the servo manufacturer in the
form of the IC servo amplifier. There
were many early versions of these
chips and most of them suffered serious defects of one kind or the other.
Sarne of these were voltage instability, output drive latch-up in which
the chip just simply melted down,
non linearity and a host of other small
problems, not the least of which was
loss of drive voltage across the output
transistors.
Time and perseverence finally rid
us of these little challenges and the
present generation of IC servo amplifiers are immaculate in their operation. Reliable, accurate and extremely
small, especially when made in surface mount form, these amplifiers
have given the modeller true proportional control. I wonder how many
truly appreciate the incredible cleverness of the human minds that conceived these devices?
One very popular version of the IC
servo is the Signetics NE544. Here in
one small package are all of the features that we dreamed of for many,
many years. If you ever have occasion
to use one of these little devices,
please take the time to marvel at the
wonder of it all and spare a small
thought for Don Mathers and Doug
Spreng, two people who helped to
make it all possible.
SC
|