17
Hybrid Technical Description 4
4 Hybrid Technical Description
The hybrid has to be very flexible to maintain its operation for the many
different countries’ telephony systems. There are 3 main parts to consider:
- the electrical interface; the impedance presented by the central exchange;
and the actual echo cancellation techniques and algorithms. The country
code presets will use a theoretical optimum set of values for the various
internal parameters of the unit. However in practice in may be necessary to
tune some of the parameters to achieve the best possible results.
Electrical Interface
The telephone system is essentially similar from country to country, but
there are small yet significant changes around the globe: e.g. max signal
levels, line current limits, ringing conditions etc. These values are preset
by the country code selection and can be adjusted via the serial port pa-
rameters.
Impedance Matching
The line to the central exchange will present certain impedance to the
hybrid, which will mean that the echo response to any signal will be modi-
fied according to the frequency and level of that signal. By impedance
matching the signal we are able to predict, and therefore subtract, a simple
estimation of the echoed signal. The AC impedance and DC impedance
parameters can also be both set via the serial port parameters.
Echo Cancellation
The echo part of the returned signal can be expressed in terms of delayed
sent signal.
The hybrid coefficients are a measure of the amount of signal that is
expected to be echoed from each delay slot (TAP). There are 2 echo
cancella-tion routines running on this unit. The first is a simple 8 TAP unit
running in the telephone interface ICs. It allows the data presented to the
DSP to be easily filtered into a signal with caller input and a signal with just
sent signal echo. The second filter is a variable length TAP with dynamic
adapta-tion running on the on-board 24bit DSP. The length of this TAP is
dependent of the expected delay in any echoes from the central exchange.
The default length is 24ms, but the unit is capable of calculating delays up
to 250ms.
The adaptation routines constantly adjust the internal DSP coefficients to
achieve the best echo cancellation. However the dynamic performance
of this algorithm (i.e. the speed that the echo cancellation is optimised) is
inversely dependant on the maximum delay (number of TAPs) allowed, so
increasing the delay will result in a slightly poorer dynamic performance.
To ensure the best performance after a power on we recommend saving
the cancellation coefficients following a successful call. These coefficients
will be loaded during the power up cycle and will improve the ‘settling’ time
for the echo cancellation algorithm.
Updating Firmware
The preferable method of communicating with the DHY-04 is to use the
Webserver embedded within the DHY-04 Firmware.
The DHY-04 operating software is stored in flash memory and can be
overwritten as new code is published. The flash memory is partitioned into
boot code, firmware, and saved parameters. Only the firmware partitions
are erased so that even in the unlikely occurrence of a power fail during the
flash update, the system can still be recovered. If no Firmware file is found,
the unit will automatically start the Bootstrap code that contains a minimal
webserver allowing the unit to have new code uploaded.
If a firmware file is corrupted, but still seen by the system as valid code
then the unit will, most likely, not be able to connect to the integrated
