Cirrus Logic CS6422 User Manual

Technical Brief

UPGRADING FROM THE CS6420 TO THE CS6422

CS6422

This document contains information that is useful
in transitioning a system from the CS6420 to the
CS6422.

Performance Differences

The analog and echo cancelling performance of the
CS6422 is identical to the CS6420. The CS6422 in­corporates a super-set of features and enhanced
configuration options.

Board-level Changes

The CS6420 and CS6422 are 100% pin compatible.
No board-level changes are necessary (or recom­mended) for migrating to the CS6422. However,
any circuitry added to a CS6420 system to imple­ment an external sidetone can be eliminated (or re­main unpopulated) if the internal programmable
sidetones on the CS6422 are used.

Register Definition Changes

The following table summarizes the bit fields that
have changed:

CS6420

Reg

HD HDD same same same (Half

RVol RVol -60dB to

TVol TVol -60dB to

THDet THDet 3,5,6dB 6,9,12dB 6dB
RHDet RHDet 3,5,6dB 6,9,12dB 6dB

RSThd RSThd 3,5,6dB 6,9,12dB 6dB

Taps Taps 55.5/--- ms 55.5/8 ms same

NseRmp NseRmp 1.5,3,6dB/s 3,6,12dB/s same

TSThd TSThd 9,12,15dB 9,12,15,

CS6422

Reg

CS6420

Values

+30dB

+30dB

CS6422

Values

-12dB to
+30dB

-12dB to
+30dB

18dB

CS6422

Default

duplex

allowed)

same

(+18dB)

same
(0dB)

(39.5/24 ms)

(3dB/s)

same

(15dB)

New Register/Feature Additions:

A number of CS6420 register bit fields have been
renamed and some of the definitions and values
have been modified in the CS6422. The register
mapping (location and decoding of registers) is
completely new.

The CS6422 register map provides for 4 bits of ad­dress decoding (compared to 3 bits in the case of the
CS6420), or 8 independent addresses (bit 0 is al­ways ‘0’), 6 of which are implemented. Each regis­ter contains 12 bits of configuration information.

Furthermore, all but one of the bit fields reside on
nibble boundaries, making conversions to and from
hexadecimal very convenient.

http://www.cirrus.com

Copyright © Cirrus Logic, Inc. 2006

(All Rights Reserved)

1) Explicit disable controls AECD and NECD
have been added to disable the acoustic and net­work echo cancellers independently. In the
CS6420, the network canceller is disabled by
allocating all the Taps to the AEC. In the
CS6422, the NEC is disabled by setting NECD.
The Taps parameter can be set to any allowed
value.

2) Auto network path detect. The CS6422 is able
to detect when a network path is present or ab­sent and can automatically enable or disable the
network echo canceller as appropriate, even
during a call in progress. To enable this feature,
NFNse should be set to a non-zero value (this
allows the NEC to be disabled if a network path

MAR ‘06

DS295TB2

CS6422

is not detected), and AuNECD should be set to
‘0’ (this allows the NEC to be enabled after it
has been disabled by NFNse).

3) Transmit Noise Guard. Transmit noise guard is
a noise squelch feature that introduces attenua­tion (set by TSAtt) in the transmit channel
when the near-end is silent. This prevents un­wanted noise (from a car or other noisy envi­ronment) from being transmitted to the far-end
listener. Noise guard is enabled by default, and
is disabled by setting TSMde to a ‘1’.

4) Double-talk Attenuation. Double-talk Attenua­tion introduces a programmable amount of loss
into the transmit and/or the receive path during
full-duplex double-talk. This feature enhances
system stability in hands-free full-duplex to
hands-free full-duplex conversations by limit­ing the gain in the acoustic coupling loop, al­lowing for higher volume levels at both ends of
the connection.

Two parameters (TDbtS and RDbtS) control
the amount of attenuation that is added to the
transmit and receive channels, respectively.
Double-talk attenuation is disabled by default
(both attenuators are set to 0 dB).

The decay rate of the attenuation is config­urable through TDSRmp and RDSRmp for the
transmit and receive channels respectively.
Both of these parameters default to the ‘slow’
setting which provides a decay rate of about 1
second. Most implementations should use the

'normal' setting, which provides a decay rate of
about 250 ms.

5) Programmable Sidetones. The parameters ASdt
and NSdt allow for a digital path to be created
between AO and APO (ASdt) or NO and NI
(NSdt). These digital paths are superimposed
on the real acoustic or network paths and are
canceled by the appropriate echo canceller. In
certain environments, most notably the car en­vironment where the acoustic path is weak, in­consistent, or distorted, the presence of a strong
linear path can enhance system performance.
Both acoustic and network sidetones are dis­abled by default.

6) Test features. The CS6422 contains a number
of features that are useful in the design and de­bug stages of a system. Separate disable con­trols (AECD and NECD) allow the acoustic
path and the network path to be debugged and
tested independently. HwlD and TD allow the
howl detector and tone detector to be disabled.
APCD and NPCD allow the acoustic and net­work path change detectors to be disabled, forc­ing each canceller to train out of a path change
rather than training from a cleared state on a
major path change. APFD and NPFD allow the
pre-emphasis filters on the acoustic and net­work echo cancellers to be disabled. This en­hances results when testing ERLE performance
with white noise.

REVISION HISTORY

Revision Date Changes

TB1 JAN 1999 Initial release.
TB2 MAR 2006 Update company contact information.

2 DS295TB2