Orban OPTIMOD-FM 8600, Operating Manual

Page: 41 / 397

OPTIMOD-FM DIGITAL INTRODUCTION
1-13
Location of OPTIMOD-FM
Optimal Control of Peak Modulation Levels
The audio processing circuitry in OPTIMOD-FM produces a signal that is pre-
emphasized to either the 50
µ
s or 75
µ
s standard preemphasis curve. It is precisely and
absolutely high-frequency-controlled and peak-controlled to prevent over-
modulation and is filtered at 15 kHz to protect the 19 kHz pilot and prevent distor-
tion caused by aliasing-related non-linear crosstalk. If this signal is fed directly into a
stereo encoder, peak modulation levels on the air will be precisely controlled. How-
ever, if the audio processor’s signal is fed to the stereo encoder through any circuitry
with frequency response errors and/or non-constant group delay, the peaks will be
magnified. Peak modulation will increase, but average modulation will not. The
modulation level must consequently be reduced to accommodate the larger peaks.
Reduced average modulation level will cause reduced loudness and a poorer signal-
to-noise ratio at the receiver.
Landline equalizers, transformers, and 15 kHz low-pass filters and preemphasis net-
works in stereo encoders typically introduce frequency response errors and non-
constant group delay. There are three criteria for preservation of peak levels
through the audio system:
1) The system group delay must be essentially constant throughout the frequency
range containing significant energy (30-15,000Hz). If low-pass filters are present,
this may require the use of delay equalization. The deviation from linear-phase
must not exceed
±
1
0°
from 30-15,000Hz.
2) The low-frequency
−
3 dB point of the system must be placed at 0.15Hz or lower
(this is not a misprint!). This is necessary to ensure less than 1% overshoot in a
50Hz square wave and essentially constant group delay to 30Hz.
3) Any preemphasis used in the audio transmission system prior to the stereo en-
coder must be canceled by a precisely complementary de-emphasis: Every pole
and zero in the preemphasis filter must be complemented by a zero and pole of
identical complex frequency in the de-emphasis network. An all-pole de-
emphasis network (like the classic series resistor feeding a grounded capacitor) is
not appropriate.
In this example, the network could be fixed by adding a second resistor
between ground and the capacitor, which would introduce a zero.
Low-pass filters (including anti-aliasing filters in digital links), high-pass filters, trans-
formers, distribution amplifiers, and long transmission lines can all cause the above
criteria to be violated, and must be tested and qualified. It is clear that the above
criteria for optimal control of peak modulation levels are most easily met when the
audio processor directly feeds the stereo encoder. In the 8600, no circuit elements
that might distort the shape of the waveform are interposed between the audio
processor and the stereo encoder. We therefore recommend using the 8600 with its
built-in stereo encoder whenever practical.