CCWN 75:15                                                 

     balanced-modulator. By algebraically adding the two
balanced-modulator outputs we obtain a signal whose r.m.s. value
is always equal to the total energy received from the signal,
regardless of its phase with respect to the input switching
signal.      A practical filter has been built using CMOS analog
switches, operational amplifiers, and CMOS digital counters and
gates. By frequency division from a master frequency standard,
all required signals are obtained with the required extreme
accuracy. The pair of switching signals required for both the
input mixer and the output balanced modulators is obtained
digitally. The "dump" control signal for the integrator and the
"sample" signal for the sample/hold circuit are also obtained
digitally.      Now let us see how the CCW filter operating at a
code-speed of 12 w.p.m. can exhibit such narrow bandwidth. At 12
w.p.m., each dit, dah, and space is an exact multiple of 100
milliseconds, or 0.1 second. Therefore, the filter is set to
process the input signal of 100 milliseconds, 10 such blocks per
second. At what off zero-beat will the input mixer output go
through one in one processing interval? Clearly, at 10 Hz.
Therefore, a signal 10 Hz removed from center will produce no
output from the filter' (We assume a steady carrier.) Five Hz
from center the beat note goes exactly half a cycle in one
interval, and thus the response is down 6 dB at this point.
Figure 5 is a plot of the frequency response of a 12 w.p.m. CCW
filter. If this frequency response were to be placed inside the
response graph of a typical s.s.b. filter, you can quickly see
the contrast.                                                    

    in blocks                                                    

    frequency                                                    

    complete cycle                                               

                                                                 

       How does the filter receive a c.w. signal without ringing?
The incoming signal is presumed to have a very precise timing:
each dit and dah begins and ends at instants which are exact
multiples of 100 milliseconds. The processing intervals of the
filter are set so they begin and end at exactly these same
instants. Suppose a dit is sent. At exactly the same time that
the dit begins, a processing interval begins in the filter. Since
the frequency is, for practical purposes, exactly zero beat, one
or both of the input mixers will produce an output having a
non-zero average value. The integrator will accumulate this
signal power and at the end of the interval the sampler will read
the total and then the integrator will be "dumped" set to zero.
The audio output you hear is not the signal itself, but a steady
tone produced by the constant output of the sample/holds driving
the output balanced modulators. The dit itself stopped at the
same instant the integrator was set to zero. So for the next
processing interval the integrator see only the space following
the dit. Thus, when the time comes for the next sampling, the
result is zero--and consequently the audio output instantly drops
to zero. Result: clean output, no ringing.      If the timing of
the receiving filter is not synchronized with the transmitter,
then portions of one dit will appear in two adjacent processing
periods, and the result will sound like ringing or will be
unintelligible.