Spread-spectrum method identifies audio path

Bob Dougherty, Nielsen Media Research -- EDN, 4/13/2000

An unusual method of audio spread spectrum can identify the audio path that's currently in use through a consumer-electronic device. The design in Figure 1a uses an injected audio spread-spectrum tag signal because a listener does not notice the low-level noise in the audio bandpass, whereas the listener would hear a steady-tone tag signal. The design injects the tag into the various inputs of the possible paths and detects the tag at the output, or speaker.

A direct-sequence noise generator running at 6 kHz furnishes the local oscillator, and a 9-kHz square-wave, lowpass filter that is filtered for its fundamental furnishes the sine-wave carrier. The circuit derives both of these signals from an 18-kHz clock. A PIC12C508A, running at 4 MHz, generates the local oscillator and the carrier. You can download the firmware by clicking here.

The local oscillator appears as sinx/x noise in the frequency domain, which the circuit then mixes with a clean, 9-kHz audio sine wave in a double-balanced mixer (DBM). Transformer and diode DBMs are unsuitable for the audio range. This circuit's DBM is simply an analog conditional complementor, and the local oscillator flips the output. The output of the DBM is a suppressed-carrier, double-sideband amplitude-modulated audio signal. The circuit bandpass-filters this output to 3 to 15 kHz to bandlimit the audio spread-spectrum signal (Figure 1b). Most of the audio "talk power" is lower than 3 kHz. The circuit then sets the level and buffers the output, which is the inject-tag signal. Note that, if the local-oscillator input to the DBMs is held either high or low, the DBMs become simple amplifiers and the tag signal is simply a 9-kHz tone.

Detection of a direct-sequence spread spectrum requires knowing that the signal, which sounds like noise, is present and that a synchronized-sequence local oscillator that's identical to the modulator side is present and requires a synchronized-sequence local oscillator that's identical to the modulator to demodulate the signal. In this case, the modulator and demodulator are in the same box, so the circuit knows when the signal is being generated, and it uses the same direct-sequence noise generator for both the injector and the detector. Thus, you eliminate the two biggest problems with spread spectrum.

The detector uses the same type of local oscillator and DBM, an analog conditional complementor, as does the injector. The detector then narrowly bandpass-filters the demodulated signal at the frequency of the modulated tag signal, or 9 kHz. This scheme separates the tag signal from the much louder audio. The audio has sporadic components in this narrow bandpass, so averaging is necessary to verify that the tag signal is present. (DI #2513)