     |
|
The original "Pulse
Width Tracking" model airplane servo was developed in 1962. Nothing much has
changed since then. The weak link in the servo is the feedback pot. It is
subject to wear due to use and vibration and the nonlinearity can make the
travel unequal in each direction. When servos are replaced they do not travel
the same amount as the old one due to the pot nonlinearities.

The answer to this problem is an optical feedback disc mounted on the output
shaft of the motor. The pictures above show the disc and the optical sensors on
the PC board. These sensors drive a PIC chip which decides how many revolutions
and the rotational direction of the motor. This data is remembered in the chip
and used to compare the 8 bit command byte sent to the servo. If there is a
difference the motor is driven to make the feedback and command data equal.
Due to the nature of the circuitry, each motor revolution is divided into 4
equal segments similar to a stepper motor with 90 degrees per step. This means
from 0 to 255 the motor turns 64 revolutions. Since most gear trains are in
the 250:1 range 64 revs will amount to plus or minus approximately 45 degrees of
output shaft revolution. This will result in ALL servos traveling exactly the
same amount. The linearity is perfect. There are 128 steps CW (128 t0 0) and 127
steps CCW (129 to 255) with 128 being neutral.
There is another disc on the pot shaft to encode neutral position, The left
photo shows the neutral detector and it's optical sensor. The neutral position
is updated during turn-on and each time the servo slews past the the neutral
position.
Obviously, this new servo does not work with the old pulse width tracking
transmitters and receiver decoders. I have designed a new data transmission
system based on the data format used in ASCII data transmission systems. The
format is shown in the diagram. Data is transmitted at 2400 baud
to keep the bandwidth to FCC specifications. It is transmitted as FSK (frequency
shift keyed) data. The data is stored and sent to each servo, during the sync
time, at 9600 baud. The resolution is 8 bits or 1 part in 256. Neutral is 128
counts. This equals the performance characteristics of most modern PWT (pulse
width tracking) systems. Some of the more expensive radios used in 1/4 scale
models can achieve better accuracy by having the servo hunt, or oscillate,
around neutral. This does increase the apparent resolution along with the
battery drain. Since PWT data is essentially a variable pulse width, the
resolution is potentially infinite. The very pots that are used to create and
feed back position, limit this resolution due to the pot noise and uncertainty.
|