mats@forbrk.UUCP (02/12/87)
Okay, we have had lots of "this modem costs $314.15927" and
"that modem did well in the Byte Review". I want to turn
the topic in a different direction: useability on my system.
At this particular office, we only have one data line; the
voice lines are on a Merlin so I don't want to mess with them.
Thus, I need my line to be bidirectional - both incoming and
outgoing calls. The system is one of our own boxes, running
V.2 (but with the uucp/cu code out of V.3, i.e. HoneyDanBer).
When we opened the office, we only had an old Cermetek available.
Lacking the time and inclination to fiddle with the internal
strap settings to synthesize the right signals to make an
outgoing connection, we operated as answer-only. The company
just got the Hayes 2400 baud modems they ordered (against my
recommendation), and sent one up here. Now we are originate-only
(in other words, now it all goes on my phone bill).
Why? The Hayes "Smartmodem" has "a few" problems (besides just
the fact that it is a Hayes).
Problem 1: the modem has a result code sequence which is activated
for answered calls as well as originated calls. This is
wrong - the result codes should only come back when I
originate a call. I don't want to be sent "RING" and
"CONNECT" when someone calls me.
Problem 2: the modem does automatic baud rate detection on both
call originate and call answer. This is not what I want;
I want the receiving modem to match my originate speed if
possible. If not possible, I want to terminate the connection
and try again at another speed (actually, I want my
Systems file to determine which speed to try first, and
whether to try again at a different speed). When *I*
am the receiving modem, I want to select the same speed
as the modem the called me; the guy on the other end can
then cycle through my gettydefs entries until he finds
the right baud rate uugetty. Simple, right?
This is *not* a request for help (I can solve 1) without too much
headache and don't yet have an answer for 2) ). Instead, I want
to stimulate some discussion of whether a particular modem
works the way the user needs it to. I turn thumbs down on the
Hayes because it does not fit my need (although I can't swap it
out for something different, unfortunately).
How do other 2400 baud modems handle the interesting question of
baud rate detection?
Mats Wichmann
Fortune Systemsschumann@puff.UUCP (02/12/87)
In article <189@forbrk.UUCP>, mats@forbrk.UUCP (Mats Wichmann) writes: >Problem 2: the modem does automatic baud rate detection on both >call originate and call answer. This is not what I want; >I want the receiving modem to match my originate speed if >possible. This is what is normally done. >If not possible, I want to terminate the connection >and try again at another speed (....) Most answering modems will try all of thier speeds, and the originating modem will catch on to one of these, unless you are trying to comminucate at a rate faster than the answering modem can go, in which case most modems would simply not connect. It appears that Hayes believes this is a feature. >When *I* >am the receiving modem, I want to select the same speed >as the modem the called me; It looks as this is fine (Your modem does this). Just some extra text *********************************************** "It... It's your hand, Buckaroo" -- Akita Chris Schumann schumann@puff.wisc.edu
berger@clio.UUCP (02/16/87)
I sympathize, and I don't like Hayes type modems for the reasons you
mentioned, among others.
I had a problem a while back with a new machine that had a design flaw,
which resulted in a 1.7% baud rate error at 1200 baud. Western
Electric 212A specs allow a maximum error of -2.4% to +1%. Our high
quality commercial grade modems didn't work properly, but the cheap
plastic hobby modems worked better. I really prefer to know when the
hardware is out of spec, since I have to support many types of
equipment working together. Wider error specs aren't necessarily
better.
I order NEC modems for several reasons:
* educational institutional discount prices are excellent
* extraordinarily high quality
* constant improvement and new features, while prices
decrease
* outstanding factory support
* 100% Western Electric standard compatibility
(plug into same rack, meet specs precisely. The newer
NEC modems have a switch for WE specs or sloppy specs,
so you can have it both ways).
* Substantially better specifications
(filters are 10 dB better than Hayes and others)
I prefer the menu-driven proprietary NEC auto dialing system to the
more obscure Hayes AT-command system, though both are available.
I don't like funny messages, and I definitely don't want the modem
to hang up just because my error-free file transfer program sent
some combination of ascii bytes over the data line. I need all
the rs232 handshaking signals. Many of the cheap plastic modems
just tie them high (and ignore the incoming lines)
Limiting is adjustable by internal switches, and the modem's
dialing directory and parameter memory is backed up by a lithium
battery. The built-in autodial feature means that your modem
dials the same way (and contains one copy of your directory)
completely independent of which communications software package
you use. This is a great benefit to people who aren't sophisticated
computer users.
Finally, NEC offers a 5 year warranty.
Mike Berger
Center for Advanced Study
University of Illinois
{ihnp4|convex|pur-ee}!uiucuxc!clio!bergerben@catnip.UUCP (02/17/87)
In article <189@forbrk.UUCP> mats@forbrk.UUCP (Mats Wichmann) writes: >Thus, I need my line to be bidirectional - both incoming and >outgoing calls. The system is one of our own boxes, running >V.2 (but with the uucp/cu code out of V.3, i.e. HoneyDanBer). > ... >just got the Hayes 2400 baud modems they ordered (against my >recommendation), and sent one up here. Now we are originate-only >(in other words, now it all goes on my phone bill). > >Why? The Hayes "Smartmodem" has "a few" problems (besides just >the fact that it is a Hayes). >Problem 1: the modem has a result code sequence which is activated > for answered calls as well as originated calls. This is > wrong - the result codes should only come back when I > originate a call. I don't want to be sent "RING" and > "CONNECT" when someone calls me. This should not cause a problem on Unix machines if the modem is correctly cabled and configured. Explanation: Hayes modems send the "RING" and "CONNECT" messages *before* they raise the "CD" signal on pin 8 of the serial port. Most Unix systems will ignore input on ports with modem control if this signal is held low, so getty will never see "RING" or "CONNECT." Make sure that the carrier detect signal is not strapped high, either with wire or dip switches/cmos. And make sure that modem control is enabled on all ports used for modem logins. (See the info on CLOCAL in TERMIO(7), you want modem control on for incoming lines, off for outgoing lines. I don't have a system with uugetty, but I'd be surprised if it doesn't do this right) >Problem 2: the modem does automatic baud rate detection on both > call originate and call answer. This is not what I want; This is a bit misleading. When used in originate mode, the Hayes adjusts to the speed of *your* system. It will not change rates if your dial in to a machine that wishes to use a different rate. When used in answer mode, the Hayes autobauds to the speed of the calling modem. On a Unix system, the caller is then free to cycle through the speeds in gettydefs, on a non-unix machine (such as a PC), the communications software can detect speed from the "CONNECT speed" message, which is sent in the original baud rate, prior to switching to the new speed. > I want the receiving modem to match my originate speed if > possible. If not possible, I want to terminate the connection > and try again at another speed (actually, I want my > Systems file to determine which speed to try first, and > whether to try again at a different speed). When *I* > am the receiving modem, I want to select the same speed > as the modem the called me; the guy on the other end can > then cycle through my gettydefs entries until he finds > the right baud rate uugetty. Simple, right? The Hayes does *exactly* what you describe above. If yours doesn't, it is broken. >This is *not* a request for help (I can solve 1) without too much >headache and don't yet have an answer for 2) ). Instead, I want >to stimulate some discussion of whether a particular modem >works the way the user needs it to. I turn thumbs down on the >Hayes because it does not fit my need (although I can't swap it >out for something different, unfortunately). I have been using Hayes and Hayes compatible modems on Unix boxes (including this one) for several years now. I think most of your complaints stem from not knowing how to use a Hayes with your system (not to mention an unjustified prejudice against this make), rather than an actual deficiency in the modem. -- Ben Broder {ihnp4,decvax} !hjuxa!catnip!ben {houxm,clyde}/