Path: utzoo!mnetor!tmsoft!torsqnt!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!hayes!tnixon
From: tnixon@hayes.uucp
Newsgroups: comp.dcom.modems
Subject: Re: Why does disconnect cause garbage before NO CARRIER?
Message-ID: <3773.27b7e1b6@hayes.uucp>
Date: 12 Feb 91 12:38:14 GMT
References: <653@twg.bc.ca>
Organization: Hayes Microcomputer Products, Norcross, GA
Lines: 52

In article <653@twg.bc.ca>, bill@twg.bc.ca (Bill Irwin) asked about 
why he sometimes sees garbage on the screen when he logs off.  
Michael Ho has already given a very good response: that 
non-error-control modems will often spew garbage during the time 
between when the carrier is lost and when they actually disconnect 
the line, and that error-control modems generally hide this from the 
user.  There's not much you can do about it on non-error-control 
calls, except make the modem more sensitive to carrier loss by 
reducing the S10 time (but even this won't eliminate it entirely).  
Another choice is to simply hang up the phone yourself (+++, then 
ATH), but this leaves your session active on the remote until IT 
detects the carrier loss, which may be a security breach and cause 
other problems.

I did want to comment further on one point:  this problem is MUCH 
worse with V.32 modems than with V.22bis.  Why?  Becuase V.32 modems 
use echo cancellation, and thus have a much more difficult time 
determining when the remote modem has actually disconnected from the 
line.  When the remote disconnects, it changes the echo 
characteristics of the circuit, confusing the echo canceller.  Many 
modems, but not all, detect this situation and hang up, but it is by 
no means instantaneous (you don't want the modem hanging up on just 
a burst of line noise).  There are ways to take help, such as 
sending a period of unmodulated carrier first rather than simply 
hanging up the line, so that the other modem's echo canceller 
continues to work properly and so that the remote modem can detect 
the loss of signal properly.  But the best way is to use an 
error-control protocol, since all such protocols include a 
"disconnect" function in the protocol itself, which is used to 
signal the remote modem the desire to terminate the connection.  

Fortunately, most V.32 modems being sold today do have error control 
built in; smooth disconnections are just another reason not to buy 
non-error-control V.32 modem for async applications.

Bill did mention that he has some proprietary (non-V.32) 9600 
modems.  I should point out that _every_ proprietary 9600 modulation 
scheme in common use (PEP, HST, MNP6, Hayes, Vadic, etc.) all 
_require_ the use of error control; you can't use the modulation 
scheme without error control enabled, at least in async operation 
(Hayes does allow synchronous non-error-control ping-pong 
operation).  Therefore, you should never see garbage on 
disconnection from any of the proprietary schemes.

	-- Toby

----------------------------------+-----------------------------------------
Toby Nixon, Principal Engineer    | Voice   +1-404-449-8791  Telex 151243420
Hayes Microcomputer Products Inc. | Fax     +1-404-447-0178  CIS   70271,404
P.O. Box 105203                   | UUCP uunet!hayes!tnixon  AT&T    !tnixon
Atlanta, Georgia  30348  USA      | Internet       hayes!tnixon@uunet.uu.net
----------------------------------+-----------------------------------------