Summary:	ASTERISK-01877: tdm fxo echo when calling from sip phone
Reporter:	radamson (radamson)	Labels:
Date Opened:	2004-06-23 12:20:58	Date Closed:	2004-09-25 02:40:24
Priority:	Minor	Regression?	No
Status:	Closed/Complete	Components:	Core/General
Versions:		Frequency of Occurrence
Related Issues:
Environment:		Attachments:
Description:	Incoming pstn calls produces a disturbing amount of echo to a C7960 or Snom 200 sip phone during the first 10 to 15 seconds. The amount of echo varies from one call to another using the exact same analog phone to call into *. Three pstn lines from two Central Offices have exactly the same echo issues. Replaced two x100p's with a tdm04b to get rid of the echo problems, but still exactly the same. Mediatrix 1204 sip gateway does not have any echo at all on exact same pstn lines, therefore highly doubtfull the problem relates to the pstn lines. ****** ADDITIONAL INFORMATION ****** RHv9, CVS-HEAD-06/17/04, zapata.conf includes: signalling=fxs_ks echotraining=yes usecallerid=yes hidecallerid=no callwaiting=yes callwaitingcallerid=no threewaycalling=no echocancel=yes echocancelwhenbridged=yes immediate=no callprogress=no Have tried every combination suggested via IRC and user list over several months, including: tip-ring reversal, MMX compile options in zconfig.h, changing line impedence settings via Silicon Labs 3050 tech notes and fxstest.c within zaptel Make, etc, etc. Same echo whether using C7960 or Snom 200 on 100meg full-duplex local net. Incoming pstn calls are worse then outgoing (presumably due to echo canceller training period prior to voice). tdm card on its own IRQ. X has been disabled. CPU running <1% (2.2ghz) and a single call through *. Aggressive cancel somewhat better then without, but still objectionable to "non-technical" sip phone users. A number of people on the user list have the same issue; seems one either has echo or non at all, and gut feeling suggests it has something to do "certain" systems that have not yet been uncovered. My * does have a registered IP and can be used for testing, however I too have to work for a living (consultant) and can only be available at prearranged times (which can include full day). I'm not a programmer. If there is "anything" more that I can provide for documentation, email is best to radamson@routers.com
Comments:	By: Mark Spencer (markster) 2004-06-23 14:10:48 Just find me on IRC and we'll discuss a few other things to try. By: bfranks (bfranks) 2004-06-23 16:58:52 Mark, can you list things in the bug tracker or in some document somewhere as to other ideas and suggestions to try for the rest of us? We are experiencing the same issues with 2 X100P's and a T100P connected to a Adtran TA 750 Channel Bank (2 FXO interfaces). Thanks By: Mark Spencer (markster) 2004-06-23 18:16:18 Added option for echo training delay to be specified.  Rich was able to eliminate echo by increasing it to 800 from the default of 400.  With this evening's CVS you can now specify not only "echotraining=yes" but "echotraining=800". bfranks: if you haven't already, please contact technical support.  If you have and are still having trouble, find me on IRC.