Re: [hylafax-users] multitech modem dropping DTMF digit

[Lee Howard <faxguy@xxxxxxxxxxxxxxxx>]

Christopher wrote:
> minicom output:
> 1: info
> OK                                                                              
> ATI                                                                             
> MT5656ZDXV                                                                      
>                                                                                 
> OK                                                                              
> AT
> +FCLASS=?                                                                     
> 0,1,1.0,2,8,10                 
>
>
> 2: Broken DTMF via minicom
> RING                                                                            
> AT+FCLASS=8;
> +VLS=1                                                              
> OK                                                                              
> ./.4.9.~./.8.~
>
>
> 3: Good DTMF from other modem
> RING                                                                         
> AT+FCLASS=8;
> +VLS=1                                                           
> OK                                      
> ./.4.~./.9.~./.8.~     
>   

This latter one is how it should be, marked with / and ~.  The "broken" 
version is missing some markers.  This is a bug in the modem.  But 
HylaFAX could be made to cope with it in this particular case.  (See the 
attached patch).

In the IS-101 specification there is a provision whereby the modem can 
inform the DTE (software) how long the DTMF was present.  It does this 
by repeating the DLE+digit within the / and ~ markers.

In this particular case HylaFAX could be made to cope because the 
"repeated" digit is not the same.  It changes from 4 to 9, and so it's 
easy to see the modem's mistake here.

However, if the DTMF digits had been 448 instead, and if the same error 
happened, then there's no way to know if the modem was trying to say 
that it heard one 4 and one 8, but that the 4 was present for twice as 
long as the 8.

I don't know if that makes sense.

So the untested patch (attached) will work fine as long as the DTMF 
digits are different from each other.  If the DTMF digits are "444" then 
it won't do much good.

Thanks,

Lee.

--- hylafax.orig/faxd/ClassModem.c++	2010-02-15 12:08:17.000000000 -0800
+++ hylafax/faxd/ClassModem.c++	2010-03-23 21:37:01.608828392 -0700
@@ -1488,25 +1488,26 @@
 		for (u_int j = 0 ; j < conf.idConfig.length(); j++) {
 		    if (conf.idConfig[j].pattern == "SHIELDED_DTMF") {	// retrieve DID, e.g. via voice DTMF
 			ringstart = Sys::now();
-			bool marked = false, gotdigit = false;
+			bool marked = false;
+			int gotdigit = 0;
 			do {
 			    int c = server.getModemChar(10000);
 			    if (c == 0x10) c = server.getModemChar(10000);
 			    if (c == 0x23 || c == 0x2A || (c >= 0x30 && c <= 0x39)) {
 				// a DTMF digit was received...
-				if (!marked || (marked && !gotdigit)) {
+				if (!marked || !gotdigit || (c != gotdigit)) {
 				    protoTrace("MODEM HEARD DTMF: %c", c);
 				    callid[j].append(fxStr::format("%c", c));
-				    gotdigit = true;
+				    gotdigit = c;
 				}
 			    } else if (c == 0x2F) {
 				// got IS-101 DTMF lead marker
 				marked = true;
-				gotdigit = false;
+				gotdigit = 0;
 			    } else if (c == 0x7E) {
 				// got IS-101 DTMF end marker
 				marked = false;
-				gotdigit = false;
+				gotdigit = 0;
 			    } else if (c == 0x73) {
 				// got silence, keep waiting
 				protoTrace("MODEM HEARD SILENCE");