56.6K modems
Dave Liquorice
In article <33b438be...@newnews.dial.pipex.com>
wa...@dial.pipex.com "Walt Davidson" writes:
> N.B. This modem uses the US Robotics X2 protocol, which is not the
> same as the Rockwell 56.6 Kflex syatem employed in Motorola modems. Most
> UK ISPs who offer 56.6K access are currently using the US Robotics
> flavour.
This is true but don't get misled. 57600 is the top speed available,
looking at comments from others elsewhere 48000 is more common in
reality.
Also it appears that some people think that two X2 modems will talk X2
at to each other. This is not the case and X2 modem will only talk X2 to
an X2 server. Two X2 modems will talk V34 to each other, maximum speed
33600.
Cheers
Dave.
Brian Gibbs
In article <33b438be...@newnews.dial.pipex.com>, Walt Davidson
<wa...@dial.pipex.com> writes
>Not strictly "amateur radio" .... but if anyone is thinking of getting
>a 56.6K modem, US Robotics (UK) Ltd are offering a cut-price trade-in
>deal for users of *any* modem 14,400K or better against the new
>Sportster Flash 56.6K External Modem. The cost is £99 if your old
>modem is a US Robotics one, or £125 if it's any other make.
>
>N.B. This modem uses the US Robotics X2 protocol, which is not the
>same as the Rockwell 56.6 Kflex syatem employed in Motorola modems.
>Most UK ISPs who offer 56.6K access are currently using the US
>Robotics flavour.
>
>
>73 de G3NYY
>
THis is indeed good news....thanks Walt. Guess what modem I am using at
present (Tee Hee!!!)
73....Brian
Brian Gibbs Email: br...@brimar.demon.co.uk
WWW: http://www.brimar.demon.co.uk
Nr Corsham, Wiltshire Packet: G3MBN @ GB7SDN #49.GBR.EU
Dave Liquorice
In article <867092...@kirsta.demon.co.uk>
Jo...@kirsta.demon.co.uk "John Morris" writes:
>> This is true but don't get misled. 57600 is the top speed available,
>
> A little less, due to constraints on the average signal level (at least
> in the US). 52k is the best achieved speed I've seen reported so far.
I was refering to the top speed available not what you will get in
reality. Interestingly looking inside the UK X2 SDL it has result codes
up to 64000 and the supplied windows INF file has them up to 61333,
which doesn't appear in the SDL. The SDL goes directly from 57333 >
64000. I wonder what happened to 58666, 60000, 61333 and 62666?
Cheers
Dave.
Geoff Blake
Walt Davidson (wa...@dial.pipex.com) wrote:
: Yes, indeed! I parcelled up my non-upgradeable USR Sportster 28,800
: yesterday and sent it back for the trade-in. The Sportster 56.6 which
: USR are offering in exchange is "Flash upgradeable", so hopefully if
: the eventual standard turns out to be k56flex - which is quite
: probable - it will be an easy job to convert it.
: I guess you have a 14,400 modem, Brian. I won't tell you how much I
: paid for my Sportster 28,800 a couple or three years ago ... it brings
: tears to my eyes! (But then again, I paid over 400 quid for a Quattro
: 2400 modem back in 1985! It still works, too .... any offers?)
In my infinite humility, Walt, I would have paid the extra 26 quid and
traded the Quattro. After all, it may well be better to fall back to the
28K8 rather than the 2440 modem.
Geoff
--
--
Geoff Blake ge...@palaemon.demon.co.uk linux 2.0.29
Chelmsford g8...@g8gnz.ampr.org i586
Intel create faster processors - Microsoft create slower processes
Michael Gathergood
In message <33b15e04...@newnews.dial.pipex.com>
wa...@dial.pipex.com (Walt Davidson) writes:
> Unfortunately they wouldn't wear that one, Geoff :-( The modem
> traded in must be 14,400 bps or better. (Otherwise people would be
> sending them old Pace 1200/75 Prestel modems ... ROTFL).
Damn! Foiled again! Anybody want a GPO modem 13B? :-)
PS - it's 56.0 Kilobits, not 56.6
73
Mike
G4KFK
IO91PK
Geoff Blake
Walt Davidson (wa...@dial.pipex.com) wrote:
: On Tue, 24 Jun 1997 19:18:52 GMT, Geoff Blake
: <ge...@palaemon.demon.co.uk> wrote:
: >In my infinite humility, Walt, I would have paid the extra 26 quid and
: >traded the Quattro. After all, it may well be better to fall back to the
: >28K8 rather than the 2440 modem.
: Unfortunately they wouldn't wear that one, Geoff :-( The modem
: traded in must be 14,400 bps or better. (Otherwise people would be
: sending them old Pace 1200/75 Prestel modems ... ROTFL).
Damn, I better put it in the bin then :-)
Still the power suppy is useful.
Geoff Blake
Michael Gathergood (g4...@zetnet.co.uk) wrote:
: In message <33b15e04...@newnews.dial.pipex.com>
: wa...@dial.pipex.com (Walt Davidson) writes:
: > Unfortunately they wouldn't wear that one, Geoff :-( The modem
: > traded in must be 14,400 bps or better. (Otherwise people would be
: > sending them old Pace 1200/75 Prestel modems ... ROTFL).
: Damn! Foiled again! Anybody want a GPO modem 13B? :-)
: PS - it's 56.0 Kilobits, not 56.6
PPS Mike, it's 55.6K (or slower)
Roll on cable modems..
Colin Wright
Michael Gathergood <g4...@zetnet.co.uk> wrote:
<snip>
>PS - it's 56.0 Kilobits, not 56.6
......but only on a good day!
>73
>Mike
>G4KFK
>IO91PK
from
Colin Wright
Michael Gathergood
In message <5otu08$i40$2...@trog.dra.hmg.gb>
C.Wr...@dra.hmg.gb (Colin Wright) writes:
> Michael Gathergood <g4...@zetnet.co.uk> wrote:
> <snip>
> >PS - it's 56.0 Kilobits, not 56.6
> ......but only on a good day!
The main source of impairment to K56 or X2 is on the local loop.
Since the local loop to any given subscriber never changes, the
best thing to do is borrow a 56K modem once they become available.
If you get 56K reliably, then buy one. If you don't, then save
your money and stick with V.34
BTW - has anybody seen the press coverage of Transend's new
modulation scheme for which they claim a raw data rate (IOW,
before compression) of 67 Kbps?
73
Mike
G4KFK
IO91PK
Michael Gathergood
In message <ECCn0...@palaemon.demon.co.uk>
Geoff Blake <ge...@palaemon.demon.co.uk> writes:
> Michael Gathergood (g4...@zetnet.co.uk) wrote:
> : PS - it's 56.0 Kilobits, not 56.6
> PPS Mike, it's 55.6K (or slower)
I can't imagine where this figure came from. X2 and K56Plus both
specify 128 levels (which yields 56.0 kbps*). The fallbacks vary,
but neither scheme offers 55.6
* 128 = 2^7, so 7 data bits per sample.
8000 samples per second
8000 x 7 = 56000 bits per second
73
Mike
G4KFK
IO91PK
Yves Remedios
>same as the Rockwell 56.6 Kflex syatem employed in Motorola modems.
>Most UK ISPs who offer 56.6K access are currently using the US
>Robotics flavour.
The last Demon magazine had an interesting article about the 56k modems.
This said that the majority of the ISPs are using Ascend modems which
are based on the Rockwell chipset. These modems are upgradable to the
Rockwell 56K Flex system which ***is not compatible*** with the USR X2
protocol.
If you ISP uses Ascend modems (most of the serious ISPs do), then look
at the Demon article before following Walt's advice.
Interestingly, Demon anticipate problems with the 56k protocols and
would prefer ISDN. However, ISDN (in the UK) is currently not sensibly
priced for the domestic market (see the German market in comparison).
The problem with 56k is that it requires very high line quality so users
are likely to be disappointed with the realistically achievable speed.
Initial UK tests conducted by USR with their modems showed that it was
not possible to achieve better than 48k. It would seem that USR have
decided to "flood" the UK market with their modems in the hope that X2
will become the accepted standard.
Additionally, 56k is really pitched at the US market where consumers get
free local phonecalls. It is unlikely that free ISP calls will be
generally available in Europe so ISDN has to be a better option. It is
possible to establish an Internet connection using ISDN in under 5
seconds whereas with a modem it takes 20 seconds. So with ISDN it is
possible to set up your link to disconnect more often when idle because
it is so fast to re-establish.
Having read the Demon article, I shall not be upgrading to 56k. Instead
I am lobbying OFTEL for a sensibly priced domestic ISDN service (like
the Germans have).
--
Yves Remedios
When replying by e-mail, please replace the word nospam with demon
Michael Gathergood
In message <ECH5I...@palaemon.demon.co.uk>
Geoff Blake <ge...@palaemon.demon.co.uk> writes:
> I read somwhere - in a trade mag - that a trial of 56K technology over a
> "perfect" (eg an exchange simulator set to "ideal" settings) link to an
> appropriate modem yielded less than perfect results, a few percent of
> connections dropping back to a lower speed or failing to connect at 56K for
> no identifiable reason. For some reason, the average speed of the (inbound)
> connection at 55.6K stuck in my mind.
The exchange will not significantly affect X2 or K56 connections. It's the
local loop (in other words the twisted pair from your house to the exchange)
that will limit the speed. Hence my earlier posting suggesting that you
borrow an X2 / K56 modem - if it doesn't give you the speed you're looking
for, don't buy one unless you're prepared to move house.
73
Mike
G4KFK
IO91PK
Dave Liquorice
In article <33b6d05c...@newnews.dial.pipex.com>
wa...@dial.pipex.com "Walt Davidson" writes:
> The theoretical maximum achievable speed from the host to the terminal
> on an X2 link, allowing for full compression, etc, is actually 57.6K.
So why does the SDL have results codes up to 64000, which will be a raw
data rate.
V42bis compression is applied to the data incoming data stream to
squeeze it down the 56kbps/64kbps raw data link, this could effectively
double the throughput, to 128kbps or higher (depending on the data of
course).
Cheers
Dave.
John Morris
In article <867618664s...@nexus.demon.co.uk>
ne...@nexus.demon.co.uk "Dave Liquorice" writes:
> In article <33b6d05c...@newnews.dial.pipex.com>
> wa...@dial.pipex.com "Walt Davidson" writes:
>
> > The theoretical maximum achievable speed from the host to the terminal
> > on an X2 link, allowing for full compression, etc, is actually 57.6K.
>
> So why does the SDL have results codes up to 64000, which will be a raw
> data rate.
They are probably thinking ahead.
This needs some background - bear with me, there's quite a lot you need
to know to understand the answer.
Almost universally, the unit of telecoms digital communications is
the B channel, which is 8 bits sent 8,000 times per second - 64,000
bits per second.
That's the very most you can get (without compression) through a phone
call if any part of it is digital. And these days, 99.something%
of all calls are digital (at least in the developed world) for at least
part of their path.
In Europe, if you order a B channel (actually two - that's what ISDN
is) you get the full 64k in each B channel. There's a different, slower
channel (the D channel) for signalling. The same applies throughout the
network. A standard voice call is carried over a B channel (multiplexed
with lots of others, of course), as is a data call. But in all cases
you get all eight bits for the data.
In much of the USA the same does not apply. There, if you order ISDN -
ie two B channels - you are quite likely to get 56k per channel. The B
channel itself is still 64k, but bits - typically the low order bit of
each byte - are stolen for signalling. That only leaves 7 bits for you,
and 7 bits * 8,000 samples per second is 56,000 bits per second. That's
where the 56k comes from. They even call it "robbed bit" signaling in
some contexts!
So how do these 56k modems work?
In most cases, the telecoms network is entirely digital, apart from
the few km of wire between you and the local exchange. That's analogue.
But at the exchange, D to A and A to D converters interface your old
fashioned analogue (voice or modem) signal into the single B data
channel that gets chucked around the phone network.
The various 56k schemes work by representing each 7 bits by one of 128
different voltage levels. This is just a D to A converter. And that D
to A converter is the one in the exchange down the road - the one
that's there to cater for your analogue phone line attached to the
otherwise all digital network. A corresponding A to D converter in the
56k modem allows the 7 bits to be recovered - in principle.
In other words, the 56k scheme works by directly encoding digital
values as analogue levels, instead of modulating (frequency, phase and
amplitude) various audio carriers, as the traditional modems do.
(Remember that "modem" is itself derived from "modulate/demodulate").
Now it starts to get interesting...
The use of direct D to A conversion, instead of modulation, places
great demands on the quality of the D to A and A to D converters. If
they differ by a bit, you've got an error. A bit out in a voice call
doesn't matter. In data, it's a catastrophe.
It happens that the D to A conversion used is not linear, but vaguely
logarithmic. (Something called "A" law is used in Europe, and Mu law in
the USA. The details are irrelevant here.) That makes eminent sense
for voice - which is what all this gear was originally deisgned for -
but it does mean that the low analogue levels (corresponding to small
digital values) are very close together. Which means it is very hard
for the A to D converter in the modem to reliably distinguish (say)
2 and 3, but quite easy for it to distinguish 126 and 127.
That's why the speed drops as the line quality goes down: a few of the
lower levels are not used, to make it easier for the A to D in your
modem to distinguish them more reliably. So instead of seven bits per
sample, you only get maybe 6.5 (yes, fractional bits are used), and the
speed drops accordingly.
A further complication is that the FCC rules have limits on the average
signal levels allowed on a phone line. If the modem happened to have to
send all ones for a while, these rules would be broken. So that means
some of the higher codes cannot be used either - which is why nobody
actually gets even the 56k.
Assuming a perfect line, and with full eight bit transmission (as in
Eu, and in parts of USA) then 64k is theoretically possible. In
practice, the potential loss of 1 bit in 8 makes it wise for them to
claim only 56k - these are American companies, who focus 99.9%
about the US market. Dropping low level codes (line quality) and high
level codes (FCC power limits) inevitably reduces this.
So my guess at why the result code is there is that it is in case a
full European version, working with all 8 bits per sample, on a perfect
line, and with no FCC to worry about, is produced.
Incidentally, the fact that a B channel is 64k bits per second through
the network is reason to be very skeptical about the report I saw here
that some company is claiming 67k (?) raw bit rate - ie with no
compression.
> V42bis compression is applied to the data incoming data stream to
> squeeze it down the 56kbps/64kbps raw data link, this could effectively
> double the throughput, to 128kbps or higher (depending on the data of
> course).
Yes. In fact compression changes the picture in two ways.
First, it makes the achieved rate data dependent. Random data won't
compress, bitmap pictures usually compress very well. Hand wavingly,
for the type of data I tend to transfer (mostly text and code) a factor
of two is usually not too far out.
Second, it uses quite a lot of CPU horsepower, especially at the
compress side (decompression is a bit easier). That's a whole 'nother
topic, that I won't go into: suffice to say that the more processor you
throw in, the better compression you can get.
Which gives a hint for anyone transferring large files (and this
applies over packet even more so!). If you want to minimise the time
spent connected, then compress the file off-line. That allows plenty
of CPU to be applied to get the best compression. Relying on the
modem's own compression will usually give poorer results, as it has a
CPU specified for a price, and only has a few microseconds to deal with
each byte - it can't fall behind, or else the compression is pointless.
If it does start to fall behind, it will give up, and just throw the
data out uncompressed until it catches up again.
73, John, GM4ANB
--
John Morris Wanted: Hardware and Software engineers - all
Jo...@kirsta.demon.co.uk levels - for challenging work in international
GM4ANB@GB7EDN.#77.GBR.EU networking company. Development site in Edinburgh.
Great place - I work there! Mail me if interested.
Matthew Haigh
John Morris said...
>Which gives a hint for anyone transferring large files (and this
>applies over packet even more so!). If you want to minimise the time
>spent connected, then compress the file off-line. That allows plenty
>of CPU to be applied to get the best compression. Relying on the
>modem's own compression will usually give poorer results, as it has a
>CPU specified for a price, and only has a few microseconds to deal with
>each byte - it can't fall behind, or else the compression is pointless.
>If it does start to fall behind, it will give up, and just throw the
>data out uncompressed until it catches up again.
>
>73, John, GM4ANB
>
if you have a lot of small files that you want to send (typically up
to a couple of Kb), then add them into an archive with no compression.
Then, add that archive into another one using compression. You have
to do two unarcs to get the data back, but you can typically save
25 percent on your final archive size (depending on your data, of
course).
73, Matt G7GCR
--
Matthew Haigh --m...@haigh1.demon.co.uk--
SMSMaster V2.02--send text messages to digital mobile phones using
a PC and a modem. Can now forward email direct to your phone!
Download from http://www.haigh1.demon.co.uk
John Morris
In article <5p80ml$mjc$2...@miranda.gmrc.gecm.com>
richard...@gecm.com "Richard Herring" writes:
> John Morris (Jo...@kirsta.demon.co.uk) wrote:
> >In article <33bc9f4a...@newnews.dial.pipex.com>
> > y...@127.0.0.1 "Walt Davidson" writes:
>
> >> I know the "56K" speed on X2 applies only in the "host -> terminal"
> >> direction. The "terminal -> host" traffic is limited to a lower
> >> speed.
> >>
> >> Two questions:
> >>
> >> (1) Does the same thing apply to k56flex?
>
> >Yes.
>
> >I hear industry rumours that there is work going on to make "56K" work in
> >the other direction (there's no intrinsic reason why it shouldn't) but
> >nobody has announced it yet.
>
> I'd have thought (though I'm open to correction - this is no more than
> thinking aloud) that it was a little tricky matching the transmitting
> modem's clock to the telco's fixed sampling rate - at 56k it's at the
> Nyquist limit, so lack of sync between the clocks will cause artefacts.
> Presumably to do this the modem has to extract the clock from the
> received signal on the assumption that the telco's sending and receiving
> clocks are locked, and then do some precision training?
Damned good question! I've wondered about that myself. I do have access
to the Rockwell source code, but under all sorts of non-disclosure
agreements - and it's sufficiently undocumented that it would be a
major exercise to reverse engineer it. Extracting the clock from the
received data looks plausible though - spot the transitions, and use
them to lock a PLL.
The send and receive clocks of the digital parts of the network are
apparently very well synchronised throughout the network. This came
up in conversation at work recently, in a completely different context,
and I was told that throughout the US all the clocks are not only
frequency locked, but phase locked to better than 1 microsecond
(at 8 kHz). (The position in Europe wasn't mentioned.) I was a bit
taken aback when I heard this, so missed the chance to ask just how
they do this.
> >> (2) What is the maximum speed from terminal to host? I have seen it
> >> quoted as 28.8K and 33.6K in two different journals and don't know
> >> which to believe.
>
> >33.6.
>
> I assume the confusion is because 28.8 is part of a formal standard
> (V.34) but 33.6 is a common extension, not formally ratified.
I assume so.
J.
Yves Remedios
In article <33bc346b...@newnews.dial.pipex.com>, Walt Davidson
<y...@127.0.0.1> writes
>>If you ISP uses Ascend modems (most of the serious ISPs do), then look
>>at the Demon article before following Walt's advice.
>
>UU-Net-Pipex is not?
>
>Bwaaaahahahahaha!
Stop being childish Walt!
>
>Actually, Pipex already support X2 and they are shortly going to
>support K56flex too (it's being beta-tested at the moment). Now that
>*is* being serious! :-)
That's fine for you. But it would be better to check before paying out
money for an upgrade. You have not said whether Pipex use Ascend
modems...but then we can't really expect you to fully grasp technical
issues :-)
Matthew Haigh
Mike Willis said...
>Walt Davidson wrote:
>
>> BTW, I received my X2 Sportster Flash this morning and it's most
>> impressive! I haven't succeeded in connecting at 56K yet, but all
>> connections so far to Pipex and Virgin have been at 45,000 bps or
>> more, which is very pleasing. Web browsing and ftp downloading are
>> spectacularly improved!
>>
>
>I fear you will find more of what those of us with fast links already
>know. It is not the modem that is the bottleneck any more for
>anything but local downloads.
Yes, but this will only really affect you if you spend hours browsing
the web or only doing an FTP. My connection is usually used for news,
ftp, email and web simultaneously, meaning that whilst any one of the
machines I am connected to may not be sending me data quickly enough to
fill my modems bandwidth, but between them they do.
Its just a matter of using the available resource to it's best
advantage.
73, Matt G7GCR
--
Matthew Haigh --m...@haigh1.demon.co.uk--
SMSMaster V2.02--send text messages to digital mobile phones
UK repeater listings V1.3, maps and information in Windows Help format.
Download from http://www.haigh1.demon.co.uk
Mike Willis
Stephen Oxlade wrote:
>
> On Thu, 3 Jul 1997 11:24:14 +0100, Matthew Haigh
> <repl...@address.in.sig> wrote:
>
> >Mike Willis said...
> >>Walt Davidson wrote:
> >>
> >>> BTW, I received my X2 Sportster Flash this morning and it's most
> >>> impressive!
>
> of the modem - is it really worth it ?urely a 28.8 is fast enough ?
> I would be grateful for views ?
>
> Stephen
> --
> Stephen
> sox...@aol.com
> sox...@clara.net
Lets see.
Phone charges for local tarrif are 1p / minute. The extra cost
of the modem is 99.95 pounds. Lets make that £100, so you get
10 000 minutes or 600 000 seconds. At 28.8kb/s extra throughput
it is just over 2 Gb.
So if you download less than 2 Gb extra during the life of the modem it
is not worth it purely from a cost basis.
It is quicker and that may be important.
Matthew Haigh
Mike Willis said...
<snip is it worth the extra cash for 56.6K?>
>Lets see.
>
>Phone charges for local tarrif are 1p / minute. The extra cost
>of the modem is 99.95 pounds. Lets make that £100, so you get
>10 000 minutes or 600 000 seconds. At 28.8kb/s extra throughput
>it is just over 2 Gb.
>
>So if you download less than 2 Gb extra during the life of the modem it
>is not worth it purely from a cost basis.
>
>It is quicker and that may be important.
an hour or two, it probably isn't worth it, unless you factor in
the possible extra enjoyment for faster response times (if the
site you are visiting has enough bandwidth).
If you are using it for business, time becomes far more valuable than
a penny a minute.
73, Matt G7GCR
--
Matthew Haigh --m...@haigh1.demon.co.uk--
Mike Willis
Matthew Haigh wrote:
>
> If you are using it for business, time becomes far more valuable than
> a penny a minute.
>
But this is uk.radio.amateur - for business, ISDN as a minimum.
Roger Barker
In article <33BCA4...@rcru.rl.ac.uk>, Mike Willis <"mjw
"@rcru.rl.ac.uk> writes
[snip]
>Lets see.
>
>Phone charges for local tarrif are 1p / minute.
No they aren't, or at least according to my latest UK call prices booket
they aren't.
During the daytime it is 3.95p/min, during the evening it is 1.65p/min,
only at weekends is it 1p/min. There is of course, always a minimum
charge of 5p per call, which complicates the maths somewhat!
--
Roger Barker, G4IDE ro...@peaksys.demon.co.uk
Boston, UK http://www.peaksys.demon.co.uk
"Be tolerant of everything, except for intolerance!"
James Dunnett
Walt wrote:
> BTW, I received my X2 Sportster Flash this morning and it's most
> impressive! I haven't succeeded in connecting at 56K yet, but all
> connections so far to Pipex and Virgin have been at 45,000 bps or
> more, which is very pleasing. Web browsing and ftp downloading are
> spectacularly improved!
Delighted to hear it. Think I'll swap my Sportster 36K6 for one
of those Flash things. When Prestel get their fingers out, of
course.
73
--
Jim, g4rga | In language, the ignorant have
jay...@cixz.co.uk | prescribed laws to the learned.
g4...@jimdee.prestelz.co.uk |
1066...@compuservez.com | - Richard Duppa. 1770 - 1831.
Rallies Info etc: http://www2.prestel.co.uk/jimdee
+ Remove 'z' from return address to avoid Private Enterprise SPAM +
Steve
I've just recently tuned into this thread so I'm probably not up to
"speed" yet. Here in the US the telephone companies do not guarantee
the phone line for above 14.4kbs although I am now using a regular line
according to Windows 95 I am connected at 115kbs using my 33.3kbs
modem. This is due to the agreement in protocalls between my modem and
my internet provider. However, during high traffic times when many
people are logged in my top speed is 26.4kbs. This indicated to me that
even with a 56.6kbs modem there are conditions wher that speed wont be
achieved.
By the way, we have a flat monthly rate for unlimited local phone
useage. Then there is another flat rate for unlimited time with my
internet provider. However, the US telephone companies are trying to
get regulations passed which will allow them to add a surcharge on
internet connections. So we in the US may also have to start calculating
out per minute charges in the not too distant future.
73 Steve , N4EY
Martin Trump
In article <atjfgnstkyk...@nexus.demon.co.uk>, Dave Liquorice
<new...@nexus.demon.co.uk> writes
>> Here in the US the telephone companies do not guarantee the phone line for
>> above 14.4kbs
>
Half a mo. I think "Bauds" are being confused here with "bits per
second". I'm no expert (enlightenment would be welcome!) but I believe
that a 14.4 bits/sec modem uses 2,400 Baud.
As I understand it "Bauds" are signalling units per second while
"bits/sec" are *b*inary dig*its* per second and one Baud can transmit
many bits/sec.
Any experts on hand?
Regards,
--
Martin Trump
Steve
> Half a mo. I think "Bauds" are being confused here with "bits per
> second". I'm no expert (enlightenment would be welcome!) but I believe
> that a 14.4 bits/sec modem uses 2,400 Baud.
>
> --
> Martin Trump
My modem is 33600 bits per sec using v.34 for data. It is software
upgradable to 56K per the manufacturer but I haven't been convinced of
the need to do so at this point. It's also rated at 14400 bps for fax.
It uses V.34 data standard for the 33.6K and V.32 for the fax.
Terminal to modem (DTE) data rates are specified in the owners manual as
115200/57600/38400/19200/14400/9600/7200/4800/2400/1200/300.
The only readout of speed I have is from Win95. On Compuserve the max
connection i could get was 14400. With the service I use now, Infinet,
I usually connect at 115200.
Hope this additional info helps!
Steve, N4EY
Steve
Steve wrote:
>
> > Half a mo. I think "Bauds" are being confused here with "bits per
> > second". I'm no expert (enlightenment would be welcome!) but I believe
> > that a 14.4 bits/sec modem uses 2,400 Baud.
> >
> Regards,
> > --
> > Martin Trump
>
"Baud rate: the actual carrier frequency being used to transmit data.
Often used incorrectly as equivalent to bits per second (bps). For
example, both V.32 (9600 bps) and V.22bis (2400 bps) modems transmit
data at 2400 baud, but V.32 modems squeeze more data into each 2400 Hz
cycle."
"Bits per second (bps): the speed at which a modem sends or receives
information. For example, if your modem is operating at 9600 bps, it
can transfer 9600 binary digits every second."
73 Steve, N4EY
Dave Liquorice
On Sat, 05 Jul 1997 23:31:30 -0400, Steve wrote:
>> No, it's Windows 95 reporting the DTE rate (speed between your computer
>> and modem) not the DCE (speed between the modems). It looks good but like
>> Windows that's all.
>
> Why would my computer's ability to communicate with my modem vary?
> Why would this speed (which is internal according to your comment) vary
> with different internet providers and times of day?
Diferrent set up strings for your programs communicatimg with different
providers. See my other post re: .INF files and Win95.
--
Cheers new...@nexus.demon.co.uk
Dave. Remove "pam" for valid email.
Dave Liquorice
On Sat, 5 Jul 1997 23:16:18 +0100, Martin Trump wrote:
>>Bit better than BT who say 2400 baud or "What's data?"
>
> Half a mo. I think "Bauds" are being confused here with "bits per
> second". I'm no expert (enlightenment would be welcome!) but I believe
> that a 14.4 bits/sec modem uses 2,400 Baud.
>
> As I understand it "Bauds" are signalling units per second while
> "bits/sec" are *b*inary dig*its* per second and one Baud can transmit
> many bits/sec.
This is true I put 2400 baud in to see if anyone was watching. B-)
AFAIK all current modems actually commincate at 2400 baud. They use phase
and amlitude modulation to squeeze more than on data bit onto a single
transition.
Dave Liquorice
On Sat, 05 Jul 1997 20:47:47 -0400, Steve wrote:
> The only readout of speed I have is from Win95. On Compuserve the max
> connection i could get was 14400. With the service I use now, Infinet,
> I usually connect at 115200.
No, Win 95 is reporting the DTE not DCE. The speed between the modems with
your non-upgraded modem will be a maximum of 33600bps. You need to talk to
your modem at at least twice this rate because of the V.42bis data
compression applied to the data stream.
With a 33600bps modem you need to talk to it at at least 57600bps, 115200bps
is better, at least then the modem will never have to wait for the terminal
or be short of data to send. With a 57600bps DTE rate and compressable data
the modem might be able to receive faster than it can send to the terminal
or the modem might be able to send faster than it is receiving from the
terminal.
The reason you now get 115200 reported is a configuration problem within
Win95 or your modem. I suspect the former, Win95 defaults to reporting the
DTE rate if the .INF file for the modem is not absolutly correct or the
modem isn't installed to Win95 correctly.
Steve
Dave Liquorice wrote:
> Diferrent set up strings for your programs communicatimg with different
> providers. See my other post re: .INF files and Win95.
>
> --
> Cheers new...@nexus.demon.co.uk
> Dave. Remove "pam" for valid email.
Thanks for taking the time to reply. I will be investigating this and
get back to you with my findings.
73 Steve, N4EY
Steve
Thanks to Dave and James for the information. After checking with the
techs at my ISP and checking some reference material I've found that I
was wrong and these two gentlemen are correct.
Unfortunately this means that my origional posting quoting numbers that
Win95 reads out was invalid. Sorry for the confusion!
There is supposed to be a way for the connect speed shown by Win95 to
represent the modem to modem rate. This is supposedly determined by
modem initialiation commands. I am researching this now.
Thanks again for your help!
73 Steve, N4EY
James Dunnett
Martin Trump wrote:
> Half a mo. I think "Bauds" are being confused here with "bits per
> second". I'm no expert (enlightenment would be welcome!) but I believe
> that a 14.4 bits/sec modem uses 2,400 Baud.
You're quite right. Mustn't confuse bauds with bytes. 1Kbs is 1000
bits per second (not 1024). How many bauds it is depends on how
many bits are in your byte (usually 7) (and how many start, stop
and parity bits are added I suppose).
James Dunnett
N4EY wrote:
> according to Windows 95 I am connected at 115kbs using my 33.3kbs
> modem. This is due to the agreement in protocalls between my modem and
> my internet provider.
I get 115.2 Kbs here also! Steve, the 115 Kbs speed is the speed
between your computer/terminal and the modem, not the actual line
speed. (The DTE-DCE speed).
> However, during high traffic times when many people are logged
> in my top speed is 26.4kbs.
That's more like it.
> By the way, we have a flat monthly rate for unlimited local phone
> useage.
Many people over here want that as well, but it's a double-edged
sword. With free local calls, everyone and his cat will be on
the Net, slowing it down more and more as time goes on.
Richard Herring
>My modem is 33600 bits per sec using v.34 for data. It is software
>upgradable to 56K per the manufacturer but I haven't been convinced of
>the need to do so at this point.
Pointless until (a) your ISP's modems also support 56k, (b)
both modems speak the same kind of 56k.
>It's also rated at 14400 bps for fax.
>It uses V.34 data standard for the 33.6K and V.32 for the fax.
That's clever ;-) as V.34 only goes up to 28.8. (but to be fair most
manufacturers have agreed on the same extensions to go up to 33.6.)
>Terminal to modem (DTE) data rates are specified in the owners manual as
>115200/57600/38400/19200/14400/9600/7200/4800/2400/1200/300.
>The only readout of speed I have is from Win95. On Compuserve the max
>connection i could get was 14400. With the service I use now, Infinet,
>I usually connect at 115200.
That's the DTE speed - the speed of the serial link between modem and
computer. The DCE (modem-to-modem) speed won't be more than 33.6.
What may be confusing you is that some modems report DCE, some report DTE,
some can do either depending on how they are configured. I suspect that the
Compu$erve and Infinet software configure your modem differently.
Why would you want DTE greater than DCE? Compression.
--
Richard Herring | richard...@gecm.com | Speaking for myself
GEC-Marconi Research Centre |
James Dunnett
N4EY writes:
> There is supposed to be a way for the connect speed shown by Win95 to
> represent the modem to modem rate. This is supposedly determined by
> modem initialiation commands. I am researching this now.
I'm also trying to find out how.
If you find how to do it, please post or EMail.
James Dunnett
G3NYY wrote:
> ...They go on to say that by setting the DTE speed to 430 kbps,
> they managed to get a throughput of over 250 kbps from their 28.8 kbps
> modem.
> I cannot understand how they achieve this, because the maximum
> configurable DTE speed on all the telephone modems I have ever seen is
> 115,200 bps. Therefore, even if you set the port capability to 430K,
> there is no way the modem can communicate with it at a faster speed
> than 115,200 bps.
In Windows '95 you can set the DTE-DCE speed up to 921,600 bps.
Nonetheless I would be sceptical of 250 kbps. The new Transend
modem will only do a mere 67 kbps, and needs the bandwidth of
two telephone lines to do it. I also understand that Hayes and
USR plan a 100 kbps modem by the end of the year, so 250 kbps
seems to be somewhat "creative".
> ...Do the new 56K modems offer higher DTE speeds than 115.2K?
Wouldn't have thought so. I believe the rule is set DTE-DCE to at
least twice the anticipated maximum line speed. I'm still waiting
for my upgrade.
(67 kbps machine details: http://transendmodems.com)
John Innes
Steve <sem...@nr.infi.net> wrote:
>Thanks to Dave and James for the information. After checking with the
>techs at my ISP and checking some reference material I've found that I
>was wrong and these two gentlemen are correct.
>
>Unfortunately this means that my origional posting quoting numbers that
>Win95 reads out was invalid. Sorry for the confusion!
>
>There is supposed to be a way for the connect speed shown by Win95 to
>represent the modem to modem rate. This is supposedly determined by
>modem initialiation commands. I am researching this now.
>
>Thanks again for your help!
>
>73 Steve, N4EY
Hi Steve,
as someone else mentioned, the problem is likely to be your modem.inf
file, there may not be any entries for speeds above 28.8k, under those
conditions windows95 defaults to reporting the DTE speed.
The solution is to contact your modem manufacturer and see if a more
up to date .inf file is available. Alternatively, you could try a
generic one for your modem chipset (the Rockwell ones from Zoltrix for
instance) which report correctly.
Hope this helps
John
--
Edinburgh, UK.
Packet Radio GM7OLQ @ GB7EDN
Internet http://www.btinternet.com/~john.innes/
David Wood
In article <33c2ecdc...@news.dial.pipex.com>, Walt Davidson
<wa...@dial.pipex.com> writes
>On 7 Jul 1997 09:50:56 GMT, r...@gmrc.gecm.com (Richard Herring) wrote:
>
>>That's the DTE speed - the speed of the serial link between modem and
>>computer.
>
>retailer who sent me the new modem enclosed a leaflet offering me
>something called a "Lava Link-650". It appears to be a 16550-based
>serial port on an ISA card, but they say the UART buffer has been
>increased in size to 32 bytes and the clock speed has been quadrupled.
>It claims "The real-world result is significantly faster operation
>from a high-speed modem for only a modest investment" ... i.e. 40
>quid. They go on to say that by setting the DTE speed to 430 kbps,
>they managed to get a throughput of over 250 kbps from their 28.8 kbps
>modem.
I think the Lava card is a 16650, not a 16550. AFAIR, the buffer size on
the 16650 is 32 bytes (double the 16550, IIRC), and the clock speed is
such that 4 x 115200 is possible (roughly 430kbps) - this seems to
confirm my theory. There is also a 16750 around somewhere (TI?), though
it may still be vapourware! I've got a hunch that the 16750 is meant to
have a 64 byte buffer, and be capable of 8 x 115200 on the line!
The word that has to be remembered is *compression* when discussing
modems. I have seen Netscape once report a data transfer of 10.3K a
second on a highly compressible web page (on a 115200 DTE, 36k6 modem
link).
Honestly, though, how many people get such massive compression as 10
times? The data has to be incredibly compressible - V42bis is usually
reckoned to be about 2-3 times at best. I wonder if they were
transferring a file of entirely zero bytes by FTP?
Walt, I doubt that you'll get much improvement with your shining 56k
modem if you run the DTE faster than 115200 (assuming that the modem can
take such a high DTE rate - which I doubt somewhat on a 'consumer grade'
modem). Maybe once in a blue moon, the faster serial card will stop an
overrun, but I doubt it's worth the expense.
The 16650 has greatest application with external ISDN terminal adaptors.
If you're bonding (aggregating, or whatever the current buzzword is now)
two channels together, then you've got 128kbps on the telco side, and
only 115k2 on the serial side. Rather daft...!
>I cannot understand how they achieve this, because the maximum
>configurable DTE speed on all the telephone modems I have ever seen is
>115,200 bps. Therefore, even if you set the port capability to 430K,
>there is no way the modem can communicate with it at a faster speed
>than 115,200 bps.
Quite probably it can't cope with DTE > 115k2.
>Is this a con-trick .... or am I missing something? Do the new 56K
>modems offer higher DTE speeds than 115.2K? I wasn't aware of it.
I'd quiz the dealer hard. Those 16650 cards are jolly expensive now by
serial card standards, and I don't think that it'll do a lot to your
performance, unless your PC is dead slow (486/25, 4MB RAM sort of slow),
causing an UART RX buffer overrun with 16 bytes. Anyway, I'm not sure
whether Windows is able to use the 32 byte buffer - has anyone written
proper 16650 drivers for 95, NT (or 3.1? - though some incarnations of
Win 3.1 have really broken serial drivers, IIRC)? I've got a feeling
that the latest Linux 2.1 kernels support the 16650.
Thinking of things serial, has anyone seen much advance on the USB
peripherals front? I can't remember much about USB, but I seem to
remember that it's jolly fast.
Cheers,
David
--
David Wood, G0WZA
da...@wood2.demon.co.uk
James Dunnett
GM7OLQ wrote:
> as someone else mentioned, the problem is likely to be your modem.inf
> file...
This seems most likely. I must contact the USR BBS for
one for mine. Thanks for the tip, folks.