Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

28.8k bps Modem

284 views
Skip to first unread message

Eric Nelson

unread,
Feb 1, 1995, 10:31:10 PM2/1/95
to
A 28.8 kbps will transmit 28.8 kbps if the line can support the
modulation used to get that 28.8k rate. Data compression can be used
on uncompressed files to increase the throughput, but this has little
value when transferring compressed files. Additionally, the other end
must have a 28.8k capable modem.

My internet provider does have 28.8k connection.

John Dearing

unread,
Feb 2, 1995, 12:25:43 AM2/2/95
to
Victor Hu (v...@AGSM.UCLA.EDU) wrote:

> 1. Is the bps across the twisted pair wire actually running at 28.8 or
> 14.4 when 28.8 is invoked? Or is it just data compression?

The modem-to-modem (across the wire) communications rate will be up to
28.8Kbps assuming a clean connection.

> 2. What kinds of host supports 28.8K? I only connect up to my
> university's computer which only runs at 9.6K max.

Let me get this straight ... you only use the modem to connect to the
University computer system which only has 9600bps modems. Then why buy
a 28.8Kbps modem if the other end can't do 28.8Kbps?? Even a 14.4Kbps
modem would have been overkill.

A 28.8Kbps modem will only connect at 28.8Kbps with another 28.8Kbps
modem. Unless/until the University decides to upgrade their modems,
you won't see any improvement.

: 3. What is the speed of fax machines?

Almost all real fax machines are 9600 bps. There are a few fax
machines that support fax at 14.4. If you connect to another fax modem
that also supports 14.4K then both faxmodems will negotiate a higher
speed. Expect most of your faxes to go through at 9600 bps.


John Dearing jdea...@netaxs.com

Alan Shen

unread,
Feb 2, 1995, 3:26:40 AM2/2/95
to
On Wed, 25 Jan 1995, Victor Hu wrote:

> 1. Is the bps across the twisted pair wire actually running at 28.8 or
> 14.4 when 28.8 is invoked? Or is it just data compression?

28,800 bps uncompressed. RAW SPEED.

> 2. What kinds of host supports 28.8K? I only connect up to my
> university's computer which only runs at 9.6K max.

It's not dependent on the host, but if the other modem supports that
speed. To connect at 28.8Kbps, the other modem must support it too.
Here at our university's computer, we can only connect at 14.4, but
supposedly, they are installing 28.8 V.34's by this summer.

> 3. What is the speed of fax machines?

99% of the stand-alone fax machines out there (the non-fax/modems)
only support speeds of up to 9600bps. And actually, that's plenty of
power for Group III. Others, very few though, can go up to 14,400bps
on the standard V.17. Most fax-modems support that speed.

> My impression of my new modem:

> 1. The Supra has a nice display (external version for the PC) that
> shows the mode of transmission.

I have a PM14400FXSA with a 12-character LCD. Sure beats LEDs doesn't it!

> 2. However, I found that it required a different initialization string
> than that suggested as default for modems that are Hayes compatible.

Some modems work okay with just a standard init string and a simple
ATZ. I finally figured out one that works with 300bps - 14,400bps. Do
a little tweaking and some experimenting. You'll figure one out too soon!

If you have any more question, feel free to e-mail me.


(ker...@u.washington.edu) Daniel Kao

David Hayes

unread,
Feb 2, 1995, 12:41:21 PM2/2/95
to

> 1. Is the bps across the twisted pair wire actually running at 28.8 or
> 14.4 when 28.8 is invoked? Or is it just data compression?

The modem is capable of sending 28.8k without compression. On top of
this, the standard V.42bis compression scheme can achieve up to 4:1
compression, so you could theoretically get up to 115k bits/second. I
observe 2:1 to be more common, though, so expect about the same
throughput as an uncompressed 56kbps leased line.

> 2. What kinds of host supports 28.8K? I only connect up to my
> university's computer which only runs at 9.6K max.

Anything that's got a fast async serial port. Your university's
computer probably could handle it, but the modems they use limit you
to 9.6 (standard V.32). This is a common situation.

To test your modem, try some of the bulletin board systems in your
area. Many of them upgrade more quickly than large university sites,
because they have fewer modems to upgrade. Note that you can even try
BBS's on which you do not have a login. All you need to see is that
your modem will connect.

>3. What is the speed of fax machines?

CCITT (now renamed ITU.T) standard Group III fax machines run at 9600
bps.


David Hayes dha...@onramp.net

David Sacerdote

unread,
Feb 2, 1995, 1:10:49 PM2/2/95
to
If you purchased a modem which supports the v.34 standard AND are
using a computer to modem communications speed which is faster than
28800bps it will actually travel across the wire at 28800bps, assuming
no line noise, no error correction, and no compression. I am also
assuming that you are connecting to another modem which supports the
V.34 standard, or whatever proprietary standard your modem supports.


David Sacerdote

Scott Lorditch

unread,
Feb 2, 1995, 2:08:03 PM2/2/95
to
You should set your PC's serial port to at least 56K when using a 28.8
modem to get the effect of compression as well. I often set mine to
115 kbps.

> 2. What kinds of host supports 28.8K? I only connect up to my
> university's computer which only runs at 9.6K max.

Many Internet service providers offer SLIP and PPP service using 28.8
modems attached to their terminal servers.

Every modem manufacturer uses a slightly different superset of the
"standard" Hayes command set.


Scott Lorditch Senior Network Architect
Pepsi Cola North America Sco...@Pepsi.Com

Steve Midgley

unread,
Feb 2, 1995, 3:09:02 PM2/2/95
to
With sheepish apologies to the moderator and readers, I amend my
previous post. I must have sleeping sitting down :-)

V.32 is not the protocol spec for 14.4 modems. It's V.42. Apologies,
apologies.


Steve Midgley Tailored Solutions

Paul Robinson

unread,
Feb 6, 1995, 10:54:58 AM2/6/95
to
Victor Hu <v...@AGSM.UCLA.EDU>, writes:

> I just purchased a 28.8 K modem with the brand "Supra". I paid extra
> to get the 28.8 K instead of the 14.4 K.

Probably about twice as much. 14K modems are down to $50 or so. 28Ks
are probably in the $100 range. I purchased a 14.4 two years ago and
it cost me $215. Consider yourself fortunate. Mine still works quite
well -- I am using it to enter this message -- and like most modems,
will be obsolete long before it wears out. If my Internet provider
decides to upgrade to 28.8K, I will probably get one. Otherwise,
since I can't use one yet, I'm not going to bother.

> 1. Is the bps across the twisted pair wire actually running at 28.8 or
> 14.4 when 28.8 is invoked? Or is it just data compression?

The raw data rate for a modem will be from 110 to 28,800 baud (or
14,400 baud) depending on what the other side agrees on. The rate
will be the lowest of whatever the two modems agree on. If you call up
a service that has only 14.4 modems, or 9600 baud modems, or even
2400, you will only get 14.4 or 9600 or 2400 even though your modem
can do more. If both modems are 28.8 and both have their highest
speed enabled, you should see 28,800 baud before any compression
occurs.

The data is not sent at 28,800 bits per second, however. Typically
the modem will divide up the telephone line into six or more channels,
and run each channel at 2400 to 4800 bits per second. By multiplexing
six channels at 2400 baud, you get 14,400 baud, etc.

> 2. What kinds of host supports 28.8K? I only connect up to my
> university's computer which only runs at 9.6K max.

Your university may have so much load they can't run faster than 9600,
or the terminal controller might not be able to handle it. Many
mainframes can't do I/O faster than some otherwise slow speed by
comparison. When a IBM Sierra mainframe came out, a 9600 baud modem
probably cost as much as an ISDN BRI interface does now; hundreds or
thousands of dollars, so the port controller was probably set up with
that as the maximum. The school might not have money in the budget to
upgrade modems, or the hardware might not be able to support those
kind of speeds, or it could be the administration was waiting until
the 28,800 speed was standardized by ITU, as some modems used
proprietary methods to communicate about 14.4 and thus you might need
the same brand at both ends. You might ask them if they plan to
upgrade to 28.8 now that the ITU has standardized the method of
delivering 28,800 baud.

> 3. What is the speed of fax machines?

There are two speeds for transmissions. First, when the connection is
being set up, each side will send an identifier sequence. I call it
the "answerback" after the similar sequence sent by a telex machine.
This identifier sequence is called a TTI or CSI. One of these will
typically appear in the log that the fax machine prints after 20-40
transmissions indicating the identifying machine. The other is the
telephone number or other identifier that appears in the display
window. The two items may be different. This information is
transmitted by each machine at 300 baud, which is okay since it is
typically no more than 60 characters for each side. The sending
machine then increases its speed and the transmission takes place in
the equivalent of "half duplex" mode, except that the recipient
machine typically acknowledges the end of each page and end of
transmission.

The ITU standard for fax machine transmissions supports 4800, 9600,
12000, and 14400 baud, but typically a fax machine that does printing
will do 9600 tops, and can be downgraded to 4800 if line conditions
are bad. 12000 and 14400 are typically for fax modems in computers.

> 1. The Supra has a nice display (external version for the PC) that
> shows the mode of transmission.

> 2. However, I found that it required a different initialization string

> than that suggested as default for modems that are Hayes compatible.

The strings for each modem are because they all do different things,
and thus, to enable those features you have to set certain values.
For example, you can do a feature called "port locking". Currently,
my terminal program sends and receives data to and from my modem at a
"locked" speed of 19200 baud. The modem will transfer data to and
from the computer at 19200 baud, whether the connection at the other
end is 110 all the way to 14400.

The typical rule is to indicate every connection is at 19200. I can
set a switch register and the modem will also report the actual
connection speed. I can set a switch and have it enable or disable
data compression, and I can enable or disable error correction. And I
can set switches so that it tells me whether the other side allows or
does not allow compression or error correction, or force error
correction or compression.

For example, when playing DOOM over a modem, the modem will run at
9600 baud (later revisions of the driver support 14,400). The modems
must not use data compression or error correction (because the extra
time to do this can lose synchronization between the two computers),
and the setup string for my modem will specifically disable these
features. At the end of the game, the driver will issue an AT Z HO to
disconnect the line and reset the modem to the default settings.

If I have caller ID service on a line, I can enable the modem to send
the data after the "RING" message. I can enable the data in hexadecimal
display digits, or I can enable it as ASCII text. I can also tell the
modem to resend the last Caller-ID string it got, and I can tell it to
resend it in hex or ASCII, even if I had already received it using the
other mode. I can also turn off reception of Caller ID data.

If I want the modem to send or receive facsimile data, the program
must use the AT&F prefix with certain commands to tell the modem to
either place a fax call or receive a fax. Every feature in a modem
requires controls on it in order to enable or disable them as needed
in a particular instance or application.

Oh, yes, one more thing. The alleged claims of data compression
giving throughput rates of 50+K and 110K on 14.4 and 28.8 modems is
sheer fantasy. If you were sending a 100,000 byte file consisting of
all spaces, or all the same character, you might see those kind of
rates.

On my 14.4K modem, on ASCII text files, if I use a locked port at
38,800 baud, and enable compression I can see transfer rates of as
much as 3000 cps in rare cases, and typically around 1800 cps if the
data compresses well. For binary and compressed ZIP archive files, I
have generally seen average transfer rates in the 1600-1620 cps.

For a 28.8K modem, you can expect to probably see rates around
3200-3400 cps, depending on the content of the material, if the other
side uses a 28.8K and can stuff it fast enough to keep the line
loaded.

MCI Mail supports 14,400 baud on their dialins, but I typically see
rates in the 700-1000 cps rate, probably because their VAX machines
are heavily loaded. Occasionally I'll see rates as high as 1200 cps.

Marc A Randolph

unread,
Feb 7, 1995, 9:09:29 PM2/7/95
to
In article <telecom...@eecs.nwu.edu>, Paul Robinson <pa...@tdr.com>
wrote:

>> 1. Is the bps across the twisted pair wire actually running at 28.8 or
>> 14.4 when 28.8 is invoked? Or is it just data compression?

> The data is not sent at 28,800 bits per second, however. Typically


> the modem will divide up the telephone line into six or more channels,
> and run each channel at 2400 to 4800 bits per second. By multiplexing
> six channels at 2400 baud, you get 14,400 baud, etc.

Sorry, this is not correct. The data IS sent at 28,800 bits per
second. The line is not divided into channels either; a phone line
can only handle something in the range of 2900 baud (+/- 500 baud or
so). Baud means symbols per second, not necessarily characters per
second.

Here is part of a message from U.S. Robotics that explains it ...

[if someone could send me the whole article I would appreciate it, I
deleted everything except this extraction - mr...@tamu.edu]

From: SUP...@usr.com (Robert Chechi)
Newsgroups: comp.dcom.modems
Subject: V.34 in a Nutshell
Date: Fri, 23 Sep 1994 12:23:34
Organization: U.S. Robotics, Inc.

V.34: The next-generation modem
by Dale Walsh, U.S. Robotics vice president, advanced development

[... lots of background deleted...]

A V.22 bis, or 2,400 bit/sec, modem sends four bits per symbol and 600
symbols per second to achieve its speed of 2,400 bit/sec. The V32
modem sends four bits per symbol and 2,400 symbols per second to reach
its speed of 9,600 bit/sec. V.32 bis sends six bits per symbol and
2,400 symbols per second to achieve 14.4K bit/sec. But V.34 will send
up to nine bits per symbol, and 3,200 symbols per second, to achieve
28,800 bit/sec, twice as fast as V.32 bis.


[... lots of detailed v.34 info deleted... ]

-----------------

Marc Randolph mr...@tamu.edu -or- mar...@tamu.edu
PGP keyID: 4C95994D ...!{uunet,gatech}!tamu.edu!mrand

James Carlson

unread,
Feb 9, 1995, 1:02:04 PM2/9/95
to
In article <telecom...@eecs.nwu.edu>, Paul Robinson <pa...@tdr.com>
writes:

>> 1. Is the bps across the twisted pair wire actually running at 28.8 or
>> 14.4 when 28.8 is invoked? Or is it just data compression?

> The raw data rate for a modem will be from 110 to 28,800 baud (or
> 14,400 baud) depending on what the other side agrees on. The rate
> will be the lowest of whatever the two modems agree on. If you call up
> a service that has only 14.4 modems, or 9600 baud modems, or even
> 2400, you will only get 14.4 or 9600 or 2400 even though your modem
> can do more. If both modems are 28.8 and both have their highest
> speed enabled, you should see 28,800 baud before any compression
> occurs.

> The data is not sent at 28,800 bits per second, however. Typically


> the modem will divide up the telephone line into six or more channels,
> and run each channel at 2400 to 4800 bits per second. By multiplexing
> six channels at 2400 baud, you get 14,400 baud, etc.

One or two minor nits: the data are sent at 28,800 bits per second, but not
at 28,800 baud. The difference is that a bit is a binary digit (a single one
or zero) while a baud is a signal-element-per-second. The signal elements
sent by the modem each represent several bits (actually, with 28.8Kbps, it's
a variable amount), thus with about 3200 baud and 9 bits per baud you get
28,800.

This is a synchronous data rate, so async framing conversion data and
data compression run on top of this 28.8Kbps pipe.

Unfortunately, too many sales and marketing folks have confused the bps
versus baud issue, and the terms have lost much of their original
meaning. The language is all the poorer for this.


James Carlson <car...@xylogics.com> Tel: +1 617 272 8140
Annex Software Support / Xylogics, Inc. +1 800 225 3317
53 Third Avenue / Burlington MA 01803-4491 Fax: +1 617 272 2618

Stephen [kiwin] PALM

unread,
Feb 9, 1995, 7:20:03 AM2/9/95
to
Steve Midgley <tail...@netcom.com> wrote:

> With sheepish apologies to the moderator and readers, I amend my
> previous post. I must have sleeping sitting down :-)

> V.32 is not the protocol spec for 14.4 modems. It's V.42. Apologies,
apologies.

Actually, you still have it wrong.

V.32bis is where the 14.4kbps full duplex modulation is defined.

V.42 is an error correction procedure (than can be used with several
different modulations including V.32bis).

Paul Robinson <pa...@tdr.com> wrote:

> The raw data rate for a modem will be from 110 to 28,800 baud (or
> 14,400 baud) depending on what the other side agrees on.

Actually, it would be 110 to 28,800 bits ber second.

A V.34 28,800 modem can use one of 6 baud (or symbol) rates at a time:
2400, 2743, 2800, 3000, 3200, 3249

> The data is not sent at 28,800 bits per second, however.
> Typically the modem will divide up the telephone line into six
> or more channels, and run each channel at 2400 to 4800 bits per
> second. By multiplexing six channels at 2400 baud, you get
> 14,400 baud, etc.

This is incorrect for V.34 (28,800) modems. A V.34 modem only uses a
single "channel". During negotiation, the line is characterized by a
process refered to as "line probing". Based on that information, one
of the 6 symbol rates listed above is selected. To achieve multiple
bits per baud, Trellis Coding is used.

[stuff deleted... now discussing FAX]

> There are two speeds for transmissions. First, when the connection is
> being set up, each side will send an identifier sequence. I call it
> the "answerback" after the similar sequence sent by a telex machine.
> This identifier sequence is called a TTI or CSI. One of these will
> typically appear in the log that the fax machine prints after 20-40
> transmissions indicating the identifying machine. The other is the
> telephone number or other identifier that appears in the display
> window. The two items may be different. This information is
> transmitted by each machine at 300 baud, which is okay since it is
> typically no more than 60 characters for each side.

This 300 baud (which is also 300 bits per second in this case) modulation
is the V.21 high channel. Several other pieces of information (such
as machine capabilities, page width, etc) are also transfered in a
protocol defined in T.30.

> The sending machine then increases its speed and the transmission
> takes place in the equivalent of "half duplex" mode, except that the
> recipient machine typically acknowledges the end of each page and end
> of transmission.

Image transmission is done by V.17, V.29, or V.27ter which are all
Half Duplex only modulations.

> The ITU standard for fax machine transmissions supports 4800, 9600,
> 12000, and 14400 baud, but typically a fax machine that does printing
> will do 9600 tops, and can be downgraded to 4800 if line conditions
> are bad.

The bottom speed is 2400 bits per second for really crummy lines.

> 12000 and 14400 are typically for fax modems in computers.

But many expensive FAX machines support 14,400 (V.17) too. And you
should see a lot more "cheaper" FAX machines supporting V.17 in the
coming year.


Regards,

Stephen [kiwin] Palm TEL (Voice mail): +81-3-5371-1564
Rockwell - Digital Communications Division COMNET: 930-1564
Japan Engineering Design Center (JST=PST+17hours) FAX: +81-3-5371-1507
pa...@tokyo.rockwell.com s.p...@ieee.org sp...@cmu.edu pa...@itu.ch

Ken Culbert

unread,
Feb 9, 1995, 7:35:10 AM2/9/95
to
In article <telecom...@eecs.nwu.edu>, tail...@netcom.com (Steve
Midgley) wrote:

> With sheepish apologies to the moderator and readers, I amend my
> previous post. I must have sleeping sitting down :-)

> V.32 is not the protocol spec for 14.4 modems. It's V.42. Apologies,
> apologies.

Wrong again. V.32bis is the modulation protocol spec for 14.4 kbaud;
v.42 is the reliability spec; v.42bis is the compression standard;
v.34 is the modulation protocol for 28.8 kbaud.

Not too confusing, eh? ;)


Ken Culbert k...@funk.com
Funk Software, Inc. http://www.funk.com
222 Third Street voice: 617 497-6339
Cambridge, MA 02142 fax: 617 547-1031

John Lundgren

unread,
Feb 9, 1995, 5:39:49 AM2/9/95
to
David Sacerdote (DSac...@world.std.com) wrote:

What is 'it' in 'it will travel..' above. I think that the above
isn't telling much of the story. The link between modems may be at
28,800 BPS, but the bytes are being transmitted as octets,
synchronously. They are not 10 bit asynchronous bytes as they are
between the PC and modem. Also, there are other things done between
the two modems, such as error detection and correction, and
compression. So what is going on between the PC and modem has little
relationship to what the modems are doing on the link.


John Lundgren - Elec Tech - Info Tech Svcs
Rancho Santiago Community College District
17th St. at Bristol \ Santa Ana, CA 92706
jlun...@pop.rancho.cc.ca.us\jlun...@kn.pacbell.com

Juan C Amaya

unread,
Feb 9, 1995, 5:22:49 PM2/9/95
to
k...@funk.com (Ken Culbert) writes:

> In article <telecom...@eecs.nwu.edu>, tail...@netcom.com (Steve
> Midgley) wrote:

>> With sheepish apologies to the moderator and readers, I amend my
>> previous post. I must have sleeping sitting down :-)

>> V.32 is not the protocol spec for 14.4 modems. It's V.42. Apologies,
>> apologies.

> Wrong again. V.32bis is the modulation protocol spec for 14.4 kbaud;
> v.42 is the reliability spec; v.42bis is the compression standard;
> v.34 is the modulation protocol for 28.8 kbaud.

> Not too confusing, eh? ;)

Wrong again. You got the units wrong, it's not kbaud but kbs!

he he ... did we finally get it right?


Regards,

juan
Elec. & Comp. Engineering | Email: ama...@cmu.edu Phone: 412-862-2752
Carnegie Mellon University | For_more_info: finger ja...@andrew.cmu.edu

0 new messages