Rockwell RFX144V24-S23, Designer'S Manual, Page: 171 / 198

RFX144V24-S23 and RFX96V24-S23 Modem Designer’s Guide
1070 12-1
12. FULL-DUPLEX SPEAKERPHONE
12.1 INTRODUCTION
A full-duplex (FDX) speakerphone is a hands-free telephone which allows incoming and outgoing voice to flow spontaneously
without noticeable clipping and choppy sound effects. A separate microphone and separate loudspeaker are used instead of
a handset.
Since the microphone and speaker are located away from the user’s head, a gain loss in voice pick-up is introduced. In any
speakerphone design, a microphone gain must be added in order to transmit the outgoing voice at a desirable level and a
speaker gain must be added for listening to the incoming voice comfortably. It is very important to maintain "balanced" gain
settings between the microphone and speaker channels to keep both channels open in a FDX speakerphone design. (Half-
duplex speakerphones alternate the gain, only allowing voice transmission in one direction at a time and therefore are
unconcerned about balancing gain.) The total amount of gain must be limited to prevent feedback instability (ringing). This
limitation makes a FDX speakerphone design impossible without the application of echo cancellation technology.
A speakerphone must deal with two echo paths, one caused by the impedance mismatch in the electronic hybrid (line echo),
and one caused by the acoustic coupling from the loudspeaker to the microphone (room echo). Echo in the system can
distort the voice signal as well as make the system unstable. Speakerphone supported devices have both an acoustic echo
canceller (AEC) and a line echo canceller (LEC) to minimize the echo levels so that voice data is kept clean and system gain
is optimized.
Automatic gain control (AGC) for both the microphone and the loudspeaker circuits are important speakerphone design
elements. Microphone AGC helps keep the transmit output at a desired level when the talker moves around the room. The
speaker AGC copes with variations in attenuation characteristics in the telephone system to minimize the user’s need to
adjust the speaker volume.
The FDX speakerphone's internal operation is controlled by two speech detectors (SDs) which determine the adaptation
process of the echo cancellers and the gain balancing between microphone and speaker channels.
In addition to the voice functions, a speakerphone has to be able to dial DTMF digits, generate variety of ring-tones, and
detect some call-progress signaling tones such as dial tone. Two programmable dual-tone transmitters and one tone
detector are supplied for these purposes.
12.2 SPEAKERPHONE DESIGN
A speakerphone design includes the speakerphone hardware design and the software control to make the speakerphone
functional. As explained earlier, the FDX speakerphone is "gain-balanced", i.e., echo cancellers are used to provide
necessary microphone and speaker gain without feedback. Therefore, the following conditions must be satisfied in the real
speakerphone operating environment. Note that the speakerphone will not become full-duplex until the convergence process
of its echo cancellers is in progress.
1. The line echo canceller requires that hybrid echo return gain, measured from TXA1 or TXA2 to RIN pin at all pass-band
frequencies, is at most +6 dB (See Figure 14-1, Figure 14-2 and Figure 14-3).
2. The acoustic echo canceller requires that, after the placements of the microphone and speaker are fixed in the product
under design, the acoustic echo return gain, measured from TXA1 (SPKHI) or TXA2 (SPKLO) to RIN (MICOUT) pin at
all pass-band frequencies with maximum speaker volume control setting, is less than +12 dB (see Figure 14-1, Figure
14-4 and Figure 14-5).
3. A directional electret condenser microphone that provides more acoustic decoupling from speaker to microphone, better
noise immunity, and flatter spectrum characteristics, is strongly recommended in the design. Better acoustic decoupling
allows higher maximum speaker volume. Moreover, the analog codecs must not be overshot or overloaded by echoes in
order to maintain system linearity. This is very important for the echo cancellers to function properly.
A unique feature of the FDX speakerphone is that gain balancing is fully automatic. Depending on the design and real-time
operating conditions, the speakerphone can automatically fall back to pseudo-duplex operation. When the host reconfigures
the speakerphone or changes any parameters in real time, the speakerphone algorithm readjusts internal variables to deliver
the best performance. These capabilities make the speakerphone product design much easier.
12.2.1 Microphone/Speaker Placements
In most product designs, orienting the microphone and speaker in opposite directions and maximizing the distance in
between them will help minimize acoustic coupling from the speaker to the microphone. For example, the microphone can be
placed in the right-front corner facing forward and speaker can be placed at the left-back corner facing up or backward. A
sophisticated placement configuration can increase the maximum allowable speaker loudness by 10-20 dB.