[Asterisk-Users] Re: [Asterisk-Dev] Openloop disconnect?

[Rich Adamson]

>    I tried to found documentation about openloop disconnect on 
> Asterisk/Zaptel.  And up to now, I didn't find anything.  Is openloop 
> disconnect supported by zaptel/wcfxo drivers?

Yes, it works for me and have verified by watching a voltmeter placed
across the pstn line and noting a correspondance between when the
open loop occurs and a sip call is dropped.

However, the open loop disconnect is not fed back to the voicemail
system and hasn't for a looooog time. Not sure why this is, but you'll
find postings relative to this going way back. The typical response
has been to use maxsilence.

>    Other question: maxsilence and silencethreshold don't seem to have 
> any effect on voicemail.  Could it be possible that in certain situation 
> it just wont work?  I tried adding some verbose in app.c to have the 
> totalsilent variable printed out and it remains a 0 forever.

Yes, its possible. Since the maxsilence approach watches the incoming
pcm stream (from the pstn as an example), noise on the pstn line could
be interpreted as incomng audio, setting the rxgain to high could 
cause it, busy tone coming from the CO after it disconnects the call
has been known to cause it, etc.

I'm using the following on four solid pstn lines:
; How many seconds of silence before we end the recording
maxsilence=10
; Silence threshold (what we consider silence, the lower, the more sensitive)
silencethreshold=128
 
>    Finaly, does callprogress=yes should be working down here in 
> Canada?   Actually, everything seems to be broken (the wildcard x100p or 
> the zaptel drivers plus a combination of many libraries/etc).

Don't know as I use callprogress=no on all TDM-fxo ports (and used it
with x100p cards prior) with US telco lines. For the most part, there
isn't any significant difference between Canadian and US telco specs.
Its my understanding (from previous postings only) that callprogress
attempts to detect busy signals from the pstn, etc, and supposedly
isn't all that effective.