[comp.bugs.4bsd] tip on ventel modems under 4.2bsd

news@sci.UUCP (news) (11/12/86)

 I have some problems with ventel modems on dec and sun equipment that
the community at large may help with. My ventel modems seem to not communicate
with tip or uucp correctly. The following was done to see what exactly was
happening with the modem.

1) We placed a tee cable to see what the system was sending to the modem,
   and then reversed the cable to see what the modem was replying with.
   The following are the results.

	a) Under tip. The computer was sending two cr's to wake up the modem
	waiting for a "$" then it would issue "k", waiting for "DIAL:", then
	send the number, and wait for the connect.

	The problem was this The modem doesn't respond because the sleep
	between the two cr'r is longer than the modem will accept as valid.
	The sleep is 1 sec. in the code. Maybe this should be removed.

	The other problem was that the number was dialed before the modem
	picked up the line and had a dial tone. Ventel's fix was to place
	two %% before the number to wait for the dial tone before dialing.
	My personal view is for ventel to redo the logic. The wait is a
	design flaw.

	Then next the modem didn't get the number from the system when
	it redialed the second time on. And the code does'nt support
	redial with the redial command of the ventel.

	Last but not least the modem hangs up prematurly before the called
	system can answer the line and send a tone.

	b) Ditto for uucp my ratio is 9/1 against a connection, and my
	L.sys shows wait characters all over it.



If there is a patch I have a valid source license and need to know who to 
contact. If not I may just have to add -DHACKVEN to aculib.

Also my feeling is that a modemcap file to support new modems is a valid
consideration.


					Vance Turner
				Silicon Compilers, San Jose, Ca.
				(408)377-2900 decwrl!sci!vance@midas


"Spock..You are such a clutz..." Bloom County

dhp@ihlpa.UUCP (Douglas H. Price) (11/13/86)

>  I have some problems with ventel modems . . .
> 	a) Under tip. The computer was sending two cr's to wake up the modem
> 	waiting for a "$" . . .
> 	The problem was this The modem doesn't respond because the sleep
> 	between the two cr'r is longer than the modem will accept as valid.

My company has a Plexus P40 (Z8000 System III) box,
and the way I fixed this problem was to find an empty port,
open it CLOCAL at 300 baud, and write a bunch of characters to it
before sending the second carriage return.  A real hack, but it
does work for those individuals who don't have a timer resolution
better than 1 second.

-- 
						Douglas H. Price
						Analysts International Corp.
						@ AT&T Bell Laboratories
						..!ihnp4!ihlpa!dhp