W8SDZ@SIMTEL20.ARPA.UUCP (03/13/87)
There is great interest in the new 9600 bps modems now available.
This review was recently posted to FidoNet News and is presented here
for its informational value. Appologies to anyone who has already
seen this.
--Keith Petersen <Info-Modems-Request@SIMTEL20.ARPA>
---------------
From: hoptoad.UUCP!pozar@cgl.ucsf.edu (Tim Pozar)
Newsgroups: mod.mag.fidonet
Subject: FidoNET Newsletter, Volume 4, # 10
Date: 9 Mar 87 23:04:44 GMT
Copyright (C) 1987, by the International FidoNet Association.
All rights reserved. Duplication and/or distribution permitted
for noncommercial purposes only. For use in other circumstances,
please contact IFNA.
Fidonews Page 2 9 Mar 1987
=================================================================
ARTICLES
=================================================================
What is the Story on 9600 Baud Modems
by
Bob Hartman, Sysop 132/101
So, you've heard all of the stories about 9600 baud modems,
and you've seen all the ads, now you want to buy one, but the
problem is which one to buy? Before you make the leap, there are
some things which you should know:
First of all, the International FidoNet Association (IFNA),
has created a special committee to study the different 9600 baud
(and even higher speed) modems to find out which ones are best
suited for our uses with Fido/Opus/SEAdog/etc. The reason for
this is simple - there is no standard (that is affordable) for
having 9600 baud modems work over standard voice grade phone
lines. At 2400 baud there was a well defined standard (V.22bis),
but at 9600 baud the only standard is V.32, and modems using that
technology tend to cost $3000 or more! This price tag is out of
reach for most consumers, so modem manufacturers have taken it
upon themselves to try and manufacture modems that work over
voice grade lines, and then try to get their technology
recognized as the new standard. With over 1000 Sysops world
wide, and a user base many hundreds of times that size, the
FidoNet community is in a position which is fairly unique - we
can help a standard along by deciding that a certain modem is one
that does the job for us. In fact, we MUST do this because if we
do not, we will all end up with different 9600 baud modems that
will not communicate with each other. Yes, it is true, two
modems from two different manufacturers will not talk to each
other at 9600 baud. Think about it - if you go out and buy right
now, and the rest of the network decides on a different modem,
then you will not be able to communicate at 9600 baud. Is it
worth taking that chance?
The second thing you should know about is that all of the
manufacturers of high speed modems are working with our committee
to try and make us choose their modem. They are all offering
discounts of 50% or more to Fido Sysops (they all learned from
what USRobotics did at 2400 baud). They are all working very
closely with me personally, trying to get my standard SEAdog/Opus
system to run on their modems. This obviously is important to us
as Sysops, since if it doesn't work with our software, then we
can't use the modem no matter how fast it runs. I must say that
all three companies involved right now - USRobotics (USR Courier
HST), Microcom (Microcom AX9624c), and Telebit (Telebit
TrailBlazer) have been EXTREMELY helpful (although as you will
see later, not necessarily successful).
The final thing you should know before I get on to the
results of my testing, is that this report is VERY PRELIMINARY,
and should be taken as such. My testing is by no means complete,
and as you will see, many technical problems still need to be
Fidonews Page 3 9 Mar 1987
overcome.
Some Preliminary Results:
This section contains my thoughts about each modem so far,
and do not represent anything except my own views based upon my
exposure to each of the modems.
Let me start off with Microcom, since they are the easiest.
They have so far shipped me three modems, of which two have
arrived DOA. Without two working modems it is impossible to do
any 9600 baud testing, so only 2400 baud testing was done. At
the lower baud rates this modem will not work properly with
SEAdog 4.0. The modem does not react quickly enough to toggling
DTR, and as a result there are many times when SEAdog will send a
command to the modem which will not be executed. SEAdog will
then print an error message and try again - this time the command
will generally work properly. SEAdog 4.0 also will not properly
"fall back" to a lower baud rate when making a call out. For
example, you attempt to call a system at 2400 baud, but he is
actually running a 1200 baud modem - SEAdog will connect, but
will attempt to communicate at 2400 baud rather than 1200 baud.
This is due to the timing the Microcom modem uses in first
raising the DCD (carrier detect) signal, and then outputting the
CONNECT 1200 message. Hayes compatible modems first output the
CONNECT 1200 message, and then raise the DCD signal (which is
what SEAdog expects). This modem also only accepts upper case
'AT' commands, while most Hayes compatible modems will accept
either 'at' or 'AT'. Having been manufactured by Microcom, this
modem obviously supports MNP protocol in hardware. It also
supports the Microcom SX command set as well as the Hayes AT
command set. The modem is quite confusing to configure due to it
not being just a Hayes clone. The modem also has two sets of dip
switches, one on the front, and one on the back of the modem -
making for even more problems. It is also a half-duplex 9600
baud modem - meaning that data can only be sent in one direction
at a time.
I guess that I will talk about the USRobotics Courier HST
next. As some of you (anybody that reads SYSOP echomail) know, I
have been unable to get two USR HST modems to talk to each other
over a local phone company connection. I have been able to send
two long distance messages to the USR Opus system (115/500 Sit
UBU Sit), but those two messages took several hundred attempts -
not because of busy signals, but because of ARQ (MNP) disconnects
(read on). First some technical information about the modem. It
is VERY Hayes/Courier 2400 compatible. In fact, in the 2400 baud
testing (and below) it behaves exactly like a Courier 2400 baud
modem. I have found no differences between the two when I run
the modem in non-ARQ mode. The modem also has MNP (which
USRobotics calls ARQ) protocol that can be enabled. It is
necessary at 9600 baud to use this option. The modem works as a
sort-of full-duplex modem. What does this mean? Well, the
modems that are connected determine which side is sending more
data (don't ask me how), and that side gets the 9600 baud
channel. The other side then has a 300 baud channel for sending
Fidonews Page 4 9 Mar 1987
data the other direction. This is a very nice feature to have
for interactive operation since the modems do not have to
constantly turn the line around for each character that is typed.
Unfortunately, this method takes up all of the available
bandwidth of the telephone connection, and therefore requires a
reasonable signal to noise ratio over the entire bandwidth. On
local connections this should be no problem, but on long distance
connections with the many games that AT&T plays on long distance,
it could be a problem a lot more often. I saw this happen myself
when connecting to 115/500 when every time I connected (except
for twice), I got disconnected almost immediately because the ARQ
(MNP) retry count was too high. In other words, the connection
was so bad that the two systems took more than 5 tries to send
one packet of data, and therefore they disconnected. Since I can
call 115/500 and send mail perfectly at 2400 baud (no errors, or
very rare), it seems that the USR HST modem may be trying to push
the outer limits of the bandwidth a bit too far. What are the
advantages of the USR HST? It will work with my SEAdog/Opus
setup without any software changes (other than my modem init
string). It is also very Hayes compatible, and as simple to set
up as the Courier 2400 baud modem.
Finally, let me discuss the Telebit TrailBlazer modem. This
modem does not use any sort of conventional technology to get its
high speed. It uses what they call PEP transmission mode, and
basically what it does is break down the entire phone line
bandwidth (0Hz-4000Hz) into 512 channels each 7.8 Hz apart. When
the two modems sync up to start, they each do analysis of the
line quality on each of the 512 frequencies. Then only the best
frequencies are used for the actual data transmission. They
modulate each carrier at 7.5 baud (that is the actual speed of
the modem -7.5 baud), and encode 2, 4, or 6 (almost always 6)
bits of data on each frequency. This can give throughputs of
18,000 bps and up! In a normal phone line, the usable bandwidth
is something like 300Hz-3000Hz which would give about 350 usable
channels, which corresponds to 15750 bps. On my system I was
constantly seeing throughputs of 15000-16500 bps. The advantage
of this method is that it should work on ANY phone line,
regardless of how bad it is. Even if only 100 channels are
usable (which seems silly since that is only a usable bandwidth
of 780 Hz, and voice grade really requires at least 2000 Hz),
then you would have a rate of 4500 bps, which is still twice as
fast as our current 2400 bps modems. In addition, the
TrailBlazer does automatic error correction when in PEP mode, and
because they choose the best frequencies to start with, there are
almost never any retransmissions. In my testing I was seeing
about 1 retransmitted packet in every 5000 packets sent. Well,
how did this modem stack up? First of all, it was the only modem
that so far has been able to transmit data at 9600 baud or higher
on my local phone connections. This in itself is a major feat
which neither USR or Microcom has been able to achieve. It also
acts as a normal 2400 baud modem, with a few minor
incompatibilities with the normal Hayes way of doing things.
Everything (and I mean everything except the on/off switch) is
controlled by AT commands in this modem. There are no external
sliders for volume, no dip switches, nothing. It is all done
Fidonews Page 5 9 Mar 1987
with AT commands (once you get used to it, it is rather nice).
Setup with this modem (for that reason) is a pain in the rear
end, but once done, the settings can be stored and never have to
be worried about again. Unfortunately, this modem raises DCD,
and outputs the connect message just like the Microcom modem
does, so it does not work for making outgoing calls with SEAdog.
I have spoken to the technical support department at Telebit, and
they are going to change this so that it will work with the
current software available for FidoNet sysops. The modem is also
half duplex, and typing a character and waiting for the echo from
the BBS can be time consuming, but again, the technical support
department has a fix for this which I will supposedly be allowed
to try out under a beta test arrangement.
Well, now that I have said all of this, what does it mean?
First of all, it means that going out within the next month and
buying a 9600 baud modem is probably not a good idea. While
Microcom and Telebit have been selling their modems for over a
year, they are currently not quite compatible with the software
that we have available to us, and USR simply has not worked the
kinks out of their modem yet (after all, it just started shipping
out last week). It also is not clear to me that the USR modem
will work on most phone lines. They have not been able to get
their modem working on my phone lines, and they have been working
with me since day one. In being fair to Microcom, they simply
have not been able to get me two modems that work properly, but
they also seem the least interested in getting our business.
Now on to some specific questions:
If I HAD TO buy a modem today, and it had to work with
Fido/Opus/SEAdog, which one would I buy? The answer to that is
simple - the USR modem is the only one that works with SEAdog/etc
right now. Unfortunately, their modem is very new, still has
problems working on my phone system, and has not proven itself in
the field.
What are the prices going to be? From what I know right
now, the USR modem lists for $995, and they will sell it to Fido
Sysops for $495. The Microcom price has not been set yet. The
Telebit TrailBlazer lists for $1345, and they will give us 50%
off for a price of $675. Included for no extra charge is the
2400 baud option which normally would add $100 to their price.
Which modem do other people seem to like? This is a hard
one. I have seen many people choose Microcom, because until
recently Telebit did not agressively market their modem, and it
was also marketed under the name FastLink by another company.
That is all changing now, and Telebit is committed to being more
competetive (hence their offer for FidoNet Sysops). The
TrailBlazer has recently been named the PC Magazine Product of
the year, and has been reviewed very favorably by Digital Review,
Mini-Micro Systems, Popular Science, Network World, PC-Week, and
Infoworld. Both USR and Telebit have applied to CCITT to have
their technology recognized as the standard for 9600 baud data
transmission over voice grade phone lines.
Fidonews Page 6 9 Mar 1987
If I was going to buy a 9600 baud modem for high speed file
transfers, rather than BBS use, which would it be? In this case
I would undoubtedly choose the Telebit TrailBlazer. It was
designed for that purpose, and it is the only modem that can send
data at baud rates exceeding 9600 baud. Actually, the Microcom
can do some data compression to get up above 9600 baud, but I
have yet to see two of them work, so I can't really comment on
that. The TrailBlazer will only be faster than the USR (again,
assuming the USR will work on the phone lines in question) when
using a transfer protocol that does not require full-duplex
transmissions. This rules out XMODEM, but Ymodem-g, and Zmodem
both work nicely, and SEAlink with large window sizes (20 or more
blocks) also performs admirably.
If I was going to buy a modem today, which would it be,
given no other restrictions? I would not buy any of these modems
today!!! USR has not proven their technology, Microcom has not
proven to me that their modems work either, and the TrailBlazer
does not currently work with Fido/Opus/SEAdog.
If I was going to buy a modem in six months, which would I
buy? At that point a lot could have changed, but I would
probably go with the Telebit TrailBlazer. By then, the small
incompatibilities with our software will have been fixed, the
modem is definitely going to work on any phone lines, and it is
the fastest of the group. The price is $180 higher than USR, but
right now I would think that paying the extra $180 to get a modem
which will definitely work is worth the higher price. Who knows,
perhaps six months from now USR will have improved their modem to
the point that it too will work on any phone lines, and at that
point, the 300 baud reverse channel becomes a factor in their
favor. With that 300 baud channel, file transfer protocols like
XMODEM, and more importantly SEAlink will work just fine.
Which standard - USR or Telebit - will the CCITT approve?
Your guess is as good as mine - yet another good reason to wait
before buying a 9600 baud modem.
Which modem has the best error correction? Well, consider
that the Telebit TrailBlazer can continue a file transfer even if
you are talking on the same line. It won't just retransmit the
data that is in error, it will actually send a large number of
packets that won't have any errors in them anyway! Call it the
"whistle test" if you will, but imaging seeing your file transfer
continue at about 50% of its normal speed while you are whistling
into the phone. Enough said?
Anything else I should know about these modems? Well, there
is a lot more to know, but I just can't provide the answers right
now. Some things I do know that might be important are:
1. USR is giving a 30 day money back guarantee to Sysops who
purchase the modem - please, make sure that if your modem (if
you have already ordered one) arrives and does not seem to
work, be sure to return it for a full refund! Don't be
bashful about this, because it is your money. The IFNA Board
Fidonews Page 7 9 Mar 1987
of Directors, and Ken Kaplan in particular argued long and
hard for this provision, because of the problems that I have
been having with the modems that USR sent to me for testing.
2. The people to contact about these modems are:
USR - Wes Cowell - (800) 342-5877
Telebit - Bruce Blain - (617) 263-9449
Microcom - Victor Forgetta - (800) 822-8224
3. USR and Telebit will both be advertising heavily in the coming
months. Try to read the ads, and note important figures. Ask
questions. I will answer any questions sent to my board, and
if I don't know the answer, then I will find out from someone
who does know the answer.
Well, that about covers it for this issue of Fidonews. As I
said at the start of this article, this is by no means a final
evaluation, just a first step. It was prompted by the fact that
people were blindly buying 9600 baud modems without having all of
the facts (not that you do now, but you hopefully are at least a
little bit more informed).
- Bob Hartman -
Sysop, the UN*X Gateway
SEAdog/Opus Node 132/101
From Bob Tue Mar 10 10:24:14 1987
Path: hoptoad!Bob
From: Bob Hartman
Newsgroups: mod.mag.fidonet
Subject: FidoNews Replies
Message-ID: <1887@hoptoad.uucp>
Date: 10 Mar 87 18:24:14 GMT
Sender: pozar@hoptoad.uucp
Organization: Syncstream (San Francisco)
Lines: 42
Approved: pozar@hoptoad.UUCP
Date: Mon 9 Mar 87 15:49
From: Bob Hartman, Net 1014 Node 1, Spark Software, Nashua NH
To: All, Net 125 Node 406, KLOK-FM BBS, San Francisco CA
Subj: 9600 bps modem article
Well, it seems that Thom and I did not coordinate this too
well. The article which appeared in this week's FidoNews was
not the final one that I wrote. The only sections that were
incorrect dealt with the Telebit TrailBlazer modem.
In the article that was published, it mentions that this modem
"raises DCD, then outputs the connect message", and that this
is incompatible with SEAdog. The engineering department at
Telebit heard of this, and sent me new firmware that changed
that part of the modem, and with this new firmware, the modem
did in fact work correctly with SEAdog. That of course also
changes the answers to a couple of questions that are at the
end of the article. In particular, the question about which
modem should be bought if one had to be bought right now. The
article states that the only viable contender is USR, but that
is no longer true. The Telebit TrailBlazer will also work
with SEAdog, and the next release of Opus will also have
support for the CONNECT FAST message that the TrailBlazer
generates. The parts of the article that deal with
incompatibility with the Telebit TrailBlazer and current
network software should be revised to read that the modems
currently shipped exhibit these flaws, but new firmware is
developed and has been tested and it does work with the
current network software.
Also left out of the article were network addresses for two of
the vendors:
USRobotics - Noah Gregoropolous 115/500
Telebit - Bruce Blain 132/101
Microcom - not reachable via FidoNet
I am very sorry that the incorrect version of the article was
sent out, but mistakes get made by everyone.
- Bob -caf@omen.UUCP (Chuck Forsberg WA7KGX) (03/16/87)
: favor. With that 300 baud channel, file transfer protocols like : XMODEM, and more importantly SEAlink will work just fine. The 300 bps reverse channel isn't the same as that on a 103 modem, which uses a relatively wide bandwidth in the prime segment of the phone line's bandpass to acheive a low error rate. The reverse channel is used for management of the error correction protocol used by the modems, and is not immediately available to send user's data as it is with a 103 modem. The end result (with repect to turnaround delays) would resemble an MNP modem operating at 450 bps, and would not give the expected throughput. So don't throw away your ZMODEM source code. Chuck Forsberg WA7KGX Author of Pro-YAM communications Tools for PCDOS and Unix ...!tektronix!reed!omen!caf Omen Technology Inc "The High Reliability Software" 17505-V Northwest Sauvie Island Road Portland OR 97231 Voice: 503-621-3406 TeleGodzilla BBS: 621-3746 2400/1200 CIS:70007,2304 Genie:CAF Source:TCE022 omen Any ACU 1200 1-503-621-3746 se:--se: link ord: Giznoid in:--in: uucp omen!/usr/spool/uucppublic/FILES lists all uucp-able files, updated hourly