Rene,

The DAA calculations I use take into consideration the effect of the
transformer (and CO shunt) on a set of 16 values of loop impedance
(Rline+Xline), over which the local transmitter echo (side tone) is to be
minimized.  These values were selected from scatter plots of data from
AT&T 1977 and 1983/84 Loop surveys using the following rationale.  Most of
the data was taken at 1 , 3 and 3.6 KHz.  In the past we considered the 1
KHz data to be most important, but the advent of V.34 caused us to soften
that concept and place a little more consideration at the high end (3.6
KHz) of the channel.  In the beginning the 1 KHz was favored more (10 of
the 16 values.  The 1 KHz data also favored non-loaded loops, which are
both a majority (80%) and statisically more tightly grouped than loaded
loops.  All chosen points are capacitive and that 600=j0omega is not
included. This the reason it does not work well with 600 ohms reflected
impedance. 

There are a number of engineers out there that doubt the old AT&T local
loop surveys are still relevant.  But, my experience says they are and
these hybrids keep proving it. 

To insure we could produce some semblance of balance in the hybrid we did
a least squares fit of the 16 points.  This provided a curve we could used
in one leg of the hybrid bridge to in sure balance over a large percentage
of real loop lengths.  The circuit in that leg is actually a simple RC
network. 

The calculations have a limitation and that is: New calculations must be 
made every time a different transformer is used.  In the beginning I 
considered this to be pure drudgery.  Now I don't mind.

I hope this gives you some idea how this hybrid functions

John Bingham is no longer at Telebit.  He is with Amati.

Amities,

Duane

Reply via email to