[Asterisk-Users] Intel 536ep as a FXO?

[Bruce Ferrell]

As someone who used to adjust hybrids for a living a number of years 
ago, I can tell you, complex impedence matching is only a part of the 
equation.

The most important part is proper gain structure.  If that's wrong no 
there is no way to control echo.  No amount of tweaking of compensation 
networks will bring one into balance... No Convolution processing can 
control it.  On old style equipment i.e. stuff built by Tellabs, the 
gain structure had to be "right" within about .5 DBm0.

Alignment meant dialing up a milliwatt test signal, measuring that 
signal at the 2 wire point and adjusting pads on the module so that the 
4 wire transmit point was at a fixed and correct level.  If memory 
serves, on an analog microwave system, 0 DBm into a module was supposed 
to be -16 DBm on the 4 wire transmit point.  The "picture" below may 
help to clarify:

=======================================================================

       ------->            2 wire      TX            |
                              |                      |
         0DBm                 | /------o  -16DBm     |
                              |/                     |
C.O. milliwatt o-------------x       4 wire         |
                               \                     |
                                \------o  +7DBm      |
                                                     |
       <-------                         RX           |

=======================================================================

So... given that we know the C.O. milliwatt is 0 DBm we also know that 
the signal seen at the point marked 2 wire is the sum of 0DBm minus the 
line loss, usually around 3 to 4 DB.  When that signal passes through 
the hybrid and correctly adjusted associated attenuator it will appear 
as marked and discussed.  Conversly, +7 DBm is inserted at the 4 wire RX 
point and the associated attenuator adjusted so that sufficient signal 
is seen at the 2 wire point at 0 DBm.

The microwave system that connects to the 4 wire point has 23 DB of gain 
so that the layout above can be mirrored for a complete analog 2 wire/4 
wire/2 wire circuit with an overall loss of between 6 to 8 DB.

The old bell specifications called for minimum 12 DB longitudinal loss 
across the 4 wire points for a hybrid on a local circuit and 16 DB for 
long haul.  There were milage specifications, but I don't remember them 
anymore.  Just getting the gain structure right was usually enough to 
meet that requirement.  If not, then we got into a backend adjustment to 
impedance match the 2 wire circuit to the hybrid... Interestingly 
enough, on an in use circuit, the losses and impedances didn't tend to 
change much over a period of years.

I think this has gone on long enough... suffice to say, gain/levels are 
crucial to echo control... It you send is too hot, you WILL have echo 
and I don't care how good your card is.  These principals applied to 
channel banks that I adjusted in olden days as well... Mostly Northern 
Telecom DE4, but others as well.  We used special equipment to measure 
signal levels at the T1 point.  I have to presume E1 equipment is/was 
similar, but I have no experience there.

'nuff said

Steve Underwood wrote:
> Andrew Kohlsmith wrote:
> 
>> [...]
>>
>> Not at all.  Any of the channel banks I've tested have better echo and 
>> audio quality than the X100P.  I believe it comes down to the Part68 
>> interface being better able to accomodate different lines but YMMV.  I 
>> have never had decent results with an X100P.  All of the tricks and 
>> hacks you see on the wiki with it are proof that it's a substandard 
>> card, IMO.
>>  
>>
> If you are trying to do cellular, satellite, VoIP  or any telephony with 
> high latency and do not use echo cancellation you are on to a looser. 
> Sure, the problem is worse with some interfaces in combination with 
> certain lines (you can't separate the two), but echo performance will 
> always be lousy without proper echo cancellation. With echo cancellation 
> almost any FXO interface should work well. Every cell phone call to the 
> PSTN is echo cancelled. Every cheapo or expensive VoIP interface box 
> echo cancels. Hybrids of any design are really lousy, and do little more 
> than stop howling. You can hand tweak some of them connected to a 
> particular line and get great performance. However, they always drift, 
> the lines get altered, or in some other way they get screwed up again. 
> Echo cancellation is a requirement, not an option.
> 
> If the X100P's interface matches a line well it will work well. If it 
> matches it badly it will work badly. Same with the channel banks, or any 
> other analogue line interface. Almost all use a compromise line match. 
> Adaptive line matching is rare. The echo you get is the luck of the 
> draw, regardless of what FXO hardware you are using.
> 
> Regards,
> Steve
> 
> _______________________________________________
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users
> To UNSUBSCRIBE or update options visit:
>   http://lists.digium.com/mailman/listinfo/asterisk-users
>