Cirrus Logic AN168 - Manual (Page 24)

AN168
24 AN168REV2
reg 2: 0a14
reg 3: a046
reg 4: 5008
reg 5: 018a
For the EC Convergence Test, we also disable
Half-Duplex mode and the Transmit Suppressor by
setting HDD = TSD = ‘1’.
Start the test by setting ACC to ‘cleared’ and then
back to ‘normal’. This resets the adaptive filter to a
cleared state.
Initiate the call and have the far-end talker begin
counting while listening for his or her echo. The
level of the echo should lessen with each count
until it reaches a point at which it will attenuate no
further. The test is simply to find out if the echo
level drops on successive counts or not. If it doesn’t
attenuate, then there’s a problem in the system,
usually caused by too much distortion in the path.
The test can be repeated by setting ACC to
‘cleared’ and then back to ‘normal’.
1.5.4.1.2 Subtest B, Half-Duplex to Full-Du-
plex transition time
Subtest B measures the time it takes for the CS6422
to train sufficiently to move from Half-Duplex
mode to Full-Duplex mode.
The CS6422 is loaded with the standard
recommended coefficient set. At the start of the
test, ACC is set to ‘cleared’ then back to ‘normal’.
This resets the adaptive filter to its cleared state and
prepares it to train to the acoustic path.
After ACC has been set to ‘normal’, the far-end
talker begins counting and listening for echo. The
number at which echo first appears indicates the
transition time to full-duplex.
If it is difficult to determine when the echo appears,
the transmit suppressor can be disengaged for this
test by setting TSD to ‘1’. This removes the
supplementary echo suppression.
1.5.4.1.3 Subtest C, Transmit Suppression test
This test verifies that the TSThd parameter is set
correctly, which controls the engagement of the
Transmit Suppressor. If TSThd is set too high, the
suppressor will not engage reliably, and the far-end
listener will hear residual echo during single-talk.
In this test, the CS6422 is loaded with the standard
coefficient set, with TSMde set to ‘1’, Noise Guard
disabled. The far-end talker counts until the device
transitions to full-duplex. Once in full-duplex, the
transmit suppressor is disengaged by setting TSD
to ‘1’. The far-end talker resumes counting. If there
is no noticeable reduction in the echo level, then the
transmit suppressor is not engaging, and the TSThd
value should be reduced.
1.5.4.2 Double-talk
In the Double-talk test, both the far-end talker and
the near-end talker speak simultaneously. In the
Double-talk test, the CS6422 is loaded with the
recommended set of coefficients. The AEC trains
when the far-end talker speaks while the near-end
talker remains silent. Once the far-end talker hears
the system move to full-duplex, Double-talk testing
can begin.
During the testing, we recommend that one end of
the link count numbers while the other end names
letters of the alphabet, days of the week, months of
the year, or any other syllabic progression that is
easy to generate and easy to detect dropped words.
The goals are to detect how “stable” the system is
in full-duplex by seeing how long the double-talk
can persist without the system dropping to
half-duplex and to test the operation of the transmit
suppressor.
The transmit suppressor should disengage when the
near-end talker speaks. If it doesn’t disengage
properly, then the near-end talker’s speech will be
severely attenuated while the far-end is speaking.
If this happens, adjust the TSBias control to a lower
value (15 dB) and test again.