[Asterisk-Users] Anyone using * in a live production environment?

Stephen R. Besch sbesch at acsu.buffalo.edu
Thu Nov 6 12:44:22 MST 2003


Andrew Kohlsmith wrote:

>>    5) Attempt to balance the hybrid at the 2-line to 4 line interface.
>>    
>>
>
>This is _precisely_ why my rollouts are all strongly recommending using a 
>channel bank instead of the cheap X100P/TDM400P cards -- a lot of work has 
>been put into the hybrid circuitry to dynamically adjust to the line 
>impedance. 
>
God, I wish I could be so sure of that.  I've looked at the circuitry on 
some high end channel bank active hybrids. SPICE modeling predicts a 
maximum 26 dB attenuation of the returned echo, even with a balanced 
line.  A simple circuitry change which adds a balance adjustment should 
permit cancellation to approach the parctical limit of a good quality 
op-amp (60-100 dB).  I don't know what the Digium stuff looks like, so I 
can't comment on it.

> I've had no serious issues with the X100P/TDM400P in small 
>scale stuff but the echo cancel IMO should be done where it originates -- 
>at the hybrid.
>
>
>Having said that, I do have "echocancel=32" in my zapata.conf for the T100P 
>connected directly to an Adit600 FXS channel bank.
>
Did it help?

>>be the only real solution.  Part of the problem arises from the use of
>>lower impedance telephone wiring nowdays. The typical characteristic
>>impedance of Cat5 twisted pair is about 100 ohms and many line cards are
>>optimized for a 600 ohm line. This is made worse if the DC resistance of
>>the wiring to the CO switch is relatively low.  I haven't tried this
>>    
>>
>
>This is a neat idea; something I have not thought of.  However my ideal PSTN 
>termination is digital (PRI) ... something to eliminate the hybrid 
>altogether, at least on my end.  :-)   For deployments where I am simply 
>providing VOIP to an existing phone system, I am recommending installing a 
>T100P and a digital trunk for the existing KSU; again to eliminate the 
>hybrid mess, or at least push it off to someone else's problem.  :-)
>
I hope someone tries it and reports back. I may do it myself if I can 
get the time.  Hard to do when the system is up and running.

>>    6) Try messing with Tx and Rx gains.
>>    
>>
>
>Something I have noticed is that on the Adit600 FXS ports, I have had to set 
>its RX attenuation to -7dB!!  (TX to -3dB) If my math is correct, that 
>means I am attenuating 85% of my incoming signal!  Is this perhaps what you 
>are referring to with the super-low impedance?
>
I hate decibels!  Technically, dB is defined either for power gain 
(10log(P/Pr)) or Voltage gain (20log(V/Vr)). They are related via the 
proportionality between power and voltage (P prop to V squared).  When 
you move the square out of the log, it introduces the factor of 2 
difference. The problem with all this is that many people are very 
sloppy about specifying whether they mean power gain or voltage gain.  
It is especially problematical when one is talking about line impedance 
matching, which affects power transfer.  However, once you are past the 
hybrid and into the amplifiers, you are talkin about voltage gain and 
should use the '20log' formula.  In this case, -7dB = 
10^^(-7/20)=44.7%.  In otherwords, you are losing 55.3% of the signal, 
just over half.

The low impedance is another, related issue.  Theoretically, when 
everyone is operating at the characteristic impedance of the line (which 
is determined by the capacitance of the wiring, spacing of the wires, 
wire gauge, insulation material, degree if twist, etc), all of the power 
delivered to the line at the sending end will be transferred to the 
line, and all of the power at the receiving end will be accepted by the 
load.  Under any other conditions, some of the power will be reflected 
back onto the line.  Delay is only a few ns per foot, so this doesn't 
amount to much in an analog only system with wires that are only a few 
miles long.

In the present case, the real echo problem arises because of how the 
hybrid sees the line impedance.  In order for the hybrid to work, it 
must subtract an exact copy of the transmitted signal from the received 
signal.  Since there is inductance and capacitance involved, there are 
also phase shifts, so it isn't enough to just have a signal of the same 
amplitude.  The phase also has to be matched.  Mismatching the impedance 
throws off both the phase and the amplitude of the copy of the 
transmitted signal that is subtracted, resulting in imperfect 
cancellation and an echo.





More information about the asterisk-users mailing list