Programmer 7500/MAX_PROGRAMMERS.iso/PROGRAMS/UTILS/MODEMS/HSMODEM2.ZIP/BULL2

Text File | 1991-01-29 | 14.0 KB | 249 lines

7

┌─────┬───────────────────────────────────────────────────────────────────────╖
│  7  │          Understand *ALL* the Esoteric Modem Standards                ║
╘═════╧═══════════════════════════════════════════════════════════════════════╝

Confused by all the terms? V.22, V.22bis, V.32, V.32bis, V.42, V.42bis and MNP?


These are all technical names given to communications "standards" as
recommended by the CCITT (Oh no! Another acronym!).  V.22bis,
pronounced, "VEE-DOT TWENTY-TWO BISS" is the technical name for what we
commonly know as a 2400 bps dial-up modem.  V.22 was the official
designation for the 1200 bps modems and the "bis" suffix (bis meaning
"second") was added when the 2400 bps standard was settled upon.

CCITT is an acronym for a long French name which when given literal
translation means the International Telephone and Telegraph Consultative
Committee.  The CCITT, through Study Groups which meet every few years,
promulgates standards and recommendations for standards to provide
compatible worldwide telecommunications circuits and devices. If it were
not for these standards, it would be difficult or impossible to
communicate between differing brands of equipment. So we have it that
V.22 defines a 1200 bps modem, V.22bis defines a 2400 bps modem, V.32
defines a 9600 bps modem and V.32bis is a 14,400 bps modem. Any modem
conforming to one of these standards will be able to communicate with
any other modem which conforms to the same standard.

                          V.42 and V.42bis

V.42 is not another type of modem, but rather a new standard for
error-correction.  V.42bis (second) is a new standard for data
compression.  V.42 and V.42bis are actually improvements on the older
MNP (Microcom Networking Protocol) error correction and compression
methods around for the past several years.  V.42 and V.42bis are
generally available on many newer modems (circa 1990 & later) which
operate at 2400 bps or higher.

Many of us, through our personal experiences with data compression
programs like ARC and ZIP and the like, have discovered that certain
types of files compress better than do others.  The same holds true for
MNP level 5 and V.42bis compression.

And just as Phil Katz's ZIP utilities and Yoshi's LHarc generally
provide tighter compression than the older ARC standard, the new V.42bis
provides a considerable improvement over MNP-5.

MNP levels 2, 3, and 4 and v.42 all refer to "hardware" error-correction
techniques built into some of the more expensive modems. MNP levels 3
and 4, and v.42, as part of the error-correction process, strip the
start and stop bits from your data, thus making the data  "appear" to
transfer at a higher cps throughput.  The modem speed doesn't change,
but since all the start and stop bits are stripped out, there's nearly
**20% LESS** data to send so it appears, on the surface at least, that
the modem speed has increased. The only reason the file appears to
travel faster is the simple fact that there's some 20% less of it to
send! Your comm program will likely report that the "cps transfer rate"
has increased dramatically,  when in reality, the only thing that really
*DID* happen was the file shrank. Of course, the modem on the other end
reinserts the start and stop bits on the fly, so the file arrives intact
and at the same size as when it was sent.  I'm sorry if I just burst
everyone's mythical balloon on file transfer rates, but it's important
for you to understand that the modem speed *DOES NOT* increase. MNP level
4 and v.42 is the fastest way to send already compressed files such as
.ZIP, .ARC, and .LZH files.

MNP level 5 introduces data compression, but here it's important that
you understand that MNP-5 *CANNOT* compress an already compressed file
and in fact will many times actually make the file bigger.  This is why
you should never attempt to transfer already compressed files with MNP-5
turned on.  The magic of MNP-5 only works with text files and data
files.  MNP-5 can, depending on how well the file compresses, nearly
double your effective throughput.  Just remember, MNP-5 does not work
well for transferring files which are already compressed.

V.42bis, the latest wrinkle in data compression, can effectively triple
or quadruple your effective transfer rate, but most often results are
only seen with files that are not already compressed. However, V.42bis,
unlike MNP-5, has the ability to determine file "compressability" on the
fly and sometimes can make minor throughput gains with files which are
already compressed. Unless you have v.42bis capabilities, MNP-4 is still
the best and fastest way to transfer these already compressed files.

                        High Speed Modem Magic

Lets burst the bubble on another myth! Would you believe me if I told
you the new 9600 and 14,400 bps modems actually communicate at only 2400
baud?  What?  It's a fact, they do! Okay, granted, there's quite a bit
of technological smoke and mirrors involved behind the scenes, but the
honest-to-goodness, old-fashioned  "baud" rate on the phone line is
2400!

Voice-grade telephone circuits, the kind we have in our homes, have a
usable frequency bandwidth of about 3100 Hz (from roughly 300 Hz to
3400 Hz). If you have ever played a musical instrument, this range
translates to E-flat above middle-C on the low end to about A-sharp
3-1/2 octaves higher. In other words, not a very broad range.

In theory, in a bandwidth of 3100 Hz we can change the state of a
signal on the line 3100 times per second, and still have it be passed
recognizably through the various analog-to-digital converters, filters,
amplifiers, and other pieces of equipment in the phone system that are
all designed to carry VOICE signals.  In reality, due to the need for
shaping of waveforms and dealing with line impairments (like delay,
attenuation and distortion at the outer edges of the band), only about
2800 Hz of the band is really usable.  A 9600 bps V.32 modem works
nicely in this range (it has an 1800Hz center "carrier" and operates
between 600Hz and 3000Hz).

The standards community doesn't like the term "baud" any more, since it
is so widely misused and misunderstood.  The preferred term today is
"bits per second" (bps), indicating the number of discrete signal elements
that can be transferred on the phone line per second.  By varying the
amplitude and phase of the signal, one signal element (baud) can indicate
several bits. An example of this is called Quadrature Amplitude
Modulation, or QAM.

For example, V.22bis (2400 bps) operates at 600 signal elements per
second, and uses 16 different combinations of phase and amplitude. This
means that 4 bits can be indicated per "baud", for a total of 2400 bits
per second. V.32 (9600 bps) operates at 2400 signal elements per second,
and supports 32 different combinations of phase and amplitude.  It
actually uses five bits per symbol, but one of the bits is a redundant
encoding of the other four to provide the receiver with more information
to properly decipher the received signal.  This is called Trellis-Coded
Modulation, or TCM, which is also sometimes referred to as "forward
error correction". Since only four of the bits actually contain user
data, V.32 can only operate at a maximum of 9600bps.

V.32bis (second) also sends 2400 symbols per second, so to achieve
14,400bps we use Trellis Coded Modulation  with 128 different
combinations of phase and amplitude for each symbol. While 14,400bps is
only 50% faster than 9600bps, the modulation pattern is 4 times as
complex.

US Robotics' famous "Courier HST" uses 14,400bps modulation in one
direction (2400 symbols per second, 128 possible states per symbol,
similar to V.32bis), but only has a 450bps carrier in the reverse
direction (150 symbols per second, 8 possible states giving 3 bits per
symbol).  The Courier "HST" is capable of transferring data at 14,400bps,
but only to another identical HST. The Courier HST can only transfer
data at a maximum speed of 2400bps to non-HST modems.

US Robotics also manufactures a model of modem called the "Dual
Standard." As the name implies, the USR "dual" can communicate at full
speed with either the "HST" at 14,400bps or the newer V.32 modems at
9600bps.  Dual Standard models manufactured after December of 1990
also include the new v.32bis modulation.

V.32bis has just been announced.  It is backward compatible with V.32.
In addition to V.32's speeds of 4800 and 9600, V.32bis also supports
7200, 12000, and 14400bps, *[both directions, simultaneously.]* It also
supports a "Rapid Rate Renegotiation" feature, which allows the modems
to quickly modify the data rate when necessary to respond to changing
line conditions (under 100 milliseconds, rather than the 5-10 seconds it
takes V.32 to change speeds and the 250 milliseconds it takes the older
USR HST to change speeds.)

V.32bis was technically agreed to by experts in Study Group 17 at the
April, 1990, CCITT meeting in Geneva.  At the next meeting, in October,
it was submitted to the CCITT's "accelerated approval procedure". In
order to become a standard under this procedure, it had to receive 100%
approval of the countries attending the meeting.  It was then translated
into the official CCITT languages, and sent by mail to all of the
countries in the United Nations.  They had three months to return their
final vote; of those returning a ballot, 70% had to vote to approve the
standard for it to be accepted.  By mid-February of 1991, V.32bis will
be an "official" CCITT Recommendation.

V.32bis (14,400 baud full duplex) modems were already in production and
on the store shelves even before the official blessing of the CCITT was
received however. Since the actual technical standards were already
agreed upon, nothing prohibited modem manufacturers from going ahead
with development.

US Robotics' entry in the V.32bis market is a completely new and
redesigned unit, including a new chassis and circuit board. The new
model is backwards compatible, though owners of earlier models were
terribly disappointed to learn that no retrofit or upgrade kit would be
available. Much to the dismay of the very large installed user base,
this makes it impossible to upgrade  previous units.  USR is offering a
"trade-in, trade-up program" but their trade-in allowance is so
ridiculously low that a person would be foolish to take the deal.

As we speak, still another, as yet to be designated "standard" is
being studied. Design engineers are experimenting with 256 combinations
of phase and amplitude in conjunction with Trellis coding to achieve
19,200 bps over a standard telephone line. No CCITT designation has yet
been given however and the protocol is likely several years away from
becoming a reality.

                              UARTs

What is a UART?  The word UART is yet another acronym meaning "Universal
Asynchronous Receiver-Transmitter."   In layman's terms, the UART is the
very heart and soul of your serial I/O port.  A UART is an integrated
circuit, or "chip" that handles the actual sending and receiving of data
between your PC's central processing unit (CPU) and your modem.  Most
PC's come equipped with a standard 8250 or 16450 UART.  There is a
supercharged, high performance replacement chip available under the part
number 16550. The most recent version of the 16550 is available from
National Semiconductor Corp., their part number  NS16550AFN.

Do I really need the 16550 UART chip in my serial port?  Maybe and maybe
not. One thing's for sure, it can't hurt.

There are some circumstances which you should consider:  if you are
multitasking, then you definitely need the buffering capabilities of the
NS16550AFN chip.  For example, suppose you want to run MS-Windows with
Qmodem downloading a file in one window while Lotus hammers out a
spreadsheet in another window. With both applications going at the same
time, the CPU now has only half the time that it used to have to respond
to interrupts.   The NS16550AFN chip helps alleviate this burden by
providing a larger HARDWARE serial buffer.  If the serial buffer of your
8250 or 16450 is overrun, data blocks will have to be retransmitted thus
slowing down your effective transfer rate. However, this does not
automatically mean that file transfer rates will be accelerated beyond
their normal speed.  You are simply increasing the ability of your CPU
to handle OTHER tasks successfully while serial activity is taking
place.  These activities include writing to the screen, writing to the
disk, handling timer interrupts, handling other DOS windows, etc.

I know this can get to be an almost religious issue nowadays, especially
among high speed modem users! You can't really go wrong in putting a
16550 chip in, but the question is whether you'll really see any
performance improvement. In most cases if you're working with a
dedicated machine and transferring files at 2400 bps or slower, you
shouldn't have any problems with an 8250 or a 16450. If you run under a
network or a multitasker or if you have a TSR or two loaded, the
NS16550AFN UART is strongly recommended.

Things don't magically improve with the mere installation of the 16550
UART however.  The chip's main feature, its 16-byte FIFO (first in first
out) buffer must first be enabled by your communications software before
it can do its thing.  Not every communications program is capable of
talking to the 16550.  Qmodem, Telix, {Commo}, Telemate, GT Powercom,
Zcomm, Pro-Yam and DSZ are a few programs that do support the 16550.
There are others as well.  Procomm and previous versions of Procomm
Plus, though quite popular with novice users, do not support the 16550
and neither of the Procomms is recommended for use with any of the new
breed of high speed modems. Datastorm Technology has announced a new
ProComm Plus version 2.0, but the jury is still deliberating it's value.

01/26/91 -  Prepared by Bill Garfield with the help of Sands BBS
            and the Sands system users.

Sands BBS [phone] - CoSysop Bill Garfield