richman@uxe.cso.uiuc.edu (11/21/87)
I'm new to this group, and this question has probably ben asked before:
What is the fastest bps modem commercially available under $1000. which
works over standard phone lines in a full-duplex mode?
What I'm looking for is something ~9600 - 19,200 which will interface
a terminal with a campus mainframe via Ma Bell. Any suggestions?
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Mike Richman
uucp: {ihnp4,seismo,puree,convex,uunet}!uiucdcs!uiucuxc!uiucuxe!richman
arpanet: richman%uiucuxe@a.cs.uiuc.edu bitnet: richman@uiucuxe
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=david@ms.uky.edu (David Herron -- Resident E-mail Hack) (11/23/87)
In article <48400001@uxe.cso.uiuc.edu> richman@uxe.cso.uiuc.edu writes: > What is the fastest bps modem commercially available under $1000. which > works over standard phone lines in a full-duplex mode? > >What I'm looking for is something ~9600 - 19,200 which will interface >a terminal with a campus mainframe via Ma Bell. Any suggestions? Well ... seeing that I'm doing just that right now I suppose I'll tell you what I've got. I'm testing Telebit TrailBlazer's to see if we want to get one/many. We almost certainly want to get one, but who knows about many. Anyway ... for the description. This modem uses a funky highly proprietary protocol. It cuts the telephone bandwidth into some large number of channels that can each do 0, 2, 4, or 6 bits of data. There's a 68010 in the modem that (among other things) examines the phone line occasionally to see how each of the channels are doing. If one is getting bad it'll drop the number of bits it's handling and if it's really bad it'll drop it completely. The baud rate will vary from something low to approx 18,000 baud in something like 300 baud increments. It also works as a normal 300/1200/2400 (+MNP) modem. It does all this over extremely normal phone lines. (I live in the student ghetto around campus ... the phone system leaves something to be desired ... the modem says it's doing around 15,000 baud right now). It's fun watching rn paint a screen faster than it does on the terminal in my office! :-) And "rain" is fun too ...! The cost? Well, it's a tad more than you wanted. The list price is $1300. Per modem. But Telebit has a deal going where they'll sell (up to two) modems to people who are in the Usenet maps for half price. ~$670 Some other features of the modem is that it can be told that to watch for packets from XMODEM/KERMIT/UUCP "g" protocols ... when it see's the packet, it accepts it and acks it ... then it passes the data on to the other modem using it's normal protocol ... at the other end the other guy creates the proper packet again and swallow the ack which'll come back. In other words .. you "sidestep" the problems you get with other fast modems where the protocol being used will cause problems with line turnaround to get acks back to the sender. I have only one reservation. It's a proprietary protocol which will only work with Trailblazers. There are some almost-standards for 9600 baud (at least I think there are some standards), and telebit may get shut out of the game. However, there are whole bunch of proprietary modems in this speed range, so maybe the standards haven't settled down yet. I think also that Telebit might be able to insert rom's and have the hardware emulate just about any of the protocols w/o too much problem. It seems likely that's the way they were able to also have the modem handle 300/1200/2400 baud ... there isn't a second set of modem chips in the box for instance. This special deal from Telebit has put a lot of Telebit's in the hands of a lot of people around the country. This will help to establish a good niche for telebit ... if only as a heavily loved part of the Usenet backbone :-). There have been reports of 9000 baud to 14000 baud throughput using these modems with UUCP transfers. This is real throughput as reported by uucp, and includes the protocol overhead. This is using long distance phone lines. etc etc. I LOVE THIS MODEM! (Note that I am in no way connected with either telebit or the vendor we will be buying through. I am merely a customer and an interested observer...) -- <---- David Herron -- Resident E-Mail Hack <david@ms.uky.edu> <---- or: {rutgers,uunet,cbosgd}!ukma!david, david@UKMA.BITNET <---- "The market doesn't drop hundreds of points on a normal day..." -- <---- Fidelity Investments Corporation
W8SDZ@SIMTEL20.ARPA (Keith Petersen) (11/23/87)
Mike, I have two US Robotics HST 9600 modems and am very pleased with
them. This is a full-duplex modem with 9600 in one direction and 300
in the other. That presents no problems since I can't type that fast.
The end with the greatest amount of data to transfer gets the 9600 bps
channel. It switches dynamically and transparent to the user.
I've been using these modems to transfer very large amounts of data
between mainframes and my microcomputer using the Xmodem-1k, Ymodem,
and Zmodem protocols, which all use 1k or greater block sizes.
Protocols using 128-byte blocks are not efficient through any
high-speed morem, especially one with built-in error checking.
I have seen the HST 9600 modem advertised in the "Computer Shopper"
for as little as $695.
--Keith Petersen
Arpa: W8SDZ@SIMTEL20.ARPA
Uucp: {bellcore,decwrl,harvard,lll-crg,ucbvax,uw-beaver}!simtel20.arpa!w8sdz
GEnie: W8SDZ
RCP/M Royal Oak: 313-759-6569 - 300, 1200, 2400 (V.22bis) or 9600 (USR HST)berger@clio.las.uiuc.edu (11/24/87)
There are two things unclear about your request. First, all of the
modems that I'm aware of that operate on phone lines at > 2400 baud
don't operate in a real full-duplex mode, where the reverse channel
has the same bandwidth as the forward channel, both operating
concurrently. If your normal use of the remote computer is as a
terminal communicating with a mainframe, that may not matter. If
you're sending a lot of data both ways, it probably does.
Secondly, the major consideration for you should be what high
speed modems are available on the computer you intend to dial up.
Most high speed modems don't work with other brands or types.
That consideration alone will probably limit your choices.
Mike Berger
Center for Advanced Study
University of Illinois
berger@clio.las.uiuc.edu
{ihnp4 | convex | pur-ee}!uiucuxc!clio!bergertim@ism780c.UUCP (Tim Smith) (12/02/87)
david@ms.uky.edu (David Herron -- Resident E-mail Hack) writes:
< I have only one reservation.
<
< It's a proprietary protocol which will only work with Trailblazers.
< There are some almost-standards for 9600 baud (at least I think there
< are some standards), and telebit may get shut out of the game.
I seem to recall that the proposed standards ( e.g., from ISO ) suck.
They do things like fall back from 9600 to 4800 if there is too much
line noise. The Trailblazer is much better.
--
Tim Smith, Knowledgian tim@ism780c.isc.comdavid@ms.uky.edu (David Herron -- Resident E-mail Hack) (12/07/87)
In article <8093@ism780c.UUCP> tim@ism780c.UUCP (Tim Smith) writes: >david@ms.uky.edu (David Herron -- Resident E-mail Hack) writes: >< I have only one reservation. >< >< It's a proprietary protocol which will only work with Trailblazers. >I seem to recall that the proposed standards ( e.g., from ISO ) suck. >They do things like fall back from 9600 to 4800 if there is too much >line noise. The Trailblazer is much better. I completely understand and agree. The trailblazer will fall back as well, but in 300 baud speed increments instead of 4800 baud increments. However, one version of reality is that the PTT's in various countries control sales of modems within their countries. Also .. those very same PTT's are what makes up ISO. (at least as far as communications standards are concerned). This gives modems adhering to the ISO protocols an advantage in built-in sales even though the Telebit protocol is superior in many many many ways. Another version of reality is that if Telebit makes enough of a market niche, then they'll become a defacto standard which the PTT's will have to pay attention to because of pressure from their customers. -- <---- David Herron -- The E-Mail guy <david@ms.uky.edu> <---- or: {rutgers,uunet,cbosgd}!ukma!david, david@UKMA.BITNET <---- <---- Classic beer bellies aren't born! They're made!
wunder@hpcea.UUCP (12/07/87)
> I seem to recall that the proposed standards ( e.g., from ISO ) suck. > They do things like fall back from 9600 to 4800 if there is too much > line noise. The Trailblazer is much better. > -- > Tim Smith, Knowledgian tim@ism780c.isc.com Well, you recall wrong. The V.32 standard is for full-duplex (no turnaround needed), synchronous traffic. This is not what you want for async login or UUCP, but it works great for IBM stuff, or HDLC between IP gateways. We use them for emergency backup between X.25 switches and gateways. They are also great for demos, trade shows, etc. HP had an IP connection from the trade show at Usenix last winter back to Palo Alto. We used a 56Kbit satellite link and a V.32 modem in parallel. We ran a similar link from Geneva for Telecom '87. The fall back is in increments of 2400 bits/sec, and doesn't cause any problem if your synchrous equipment uses external clock (the modem will clock you at 9600, 7200, whatever). Trailblazer falls back, too, but in smaller increments. So, V.32 doesn't suck, it is just different. Trailblazer (PEP) and V.32 use different (but equally effective) advanced singal processing magic, and they have comparable performance (PEP is 18Kbits in one direction, V.32 is 9600 in two directions). PEP is spread spectrum with packets and retransmission, while V.32 uses signal cancelling and forward error correction (Trellis encoding) without retransmission. Different modem karma, as I said. There is a CCITT 2400 baud standard that uses some of the V.32 tricks and is more noise-resistant than the 1200 baud modems, but it didn't catch on, alas. It cost more, and the market went for the more fragile 2400 baud modems that we all use. wunder
pete@octopus.UUCP (12/07/87)
In article <7826@g.ms.uky.edu> david@ms.uky.edu (David Herron -- Resident E-mail Hack) writes: >In article <8093@ism780c.UUCP> tim@ism780c.UUCP (Tim Smith) writes: >>I seem to recall that the proposed standards ( e.g., from ISO ) suck. >>They do things like fall back from 9600 to 4800 if there is too much >>line noise. The Trailblazer is much better. > >The trailblazer will fall back as well, but in 300 baud speed increments >instead of 4800 baud increments. > Unless I am mistaken, the trailblazer falls back in 17.5 bps increments; at least, that's what the transmit/receive bits-per-channel reports indicate. For those of you who haven't yet joined the ranks of happy users: the modem has several reports available if you inquire the value of certain S-registers. Two of these show, for each of 511 bands spaced 7.8 Hz apart, how many bits are being sent or received in each band. The number of bits/band adjusts as a value in the set (0,2,4,6) individually for each band. If you compare these numbers with the instantaneous transmit/receive rate (available in another S-register), you find that a change of 2 in a single band corresponds to approximately 17.5 bps change in the data rate. Note that since the modem performs buffered data rate translation, there is no need for the host CPU to know any of these details: on a bad line, the data throughput just goes down! In case you're wondering what kind of modem would have detailed reports like this... that's only the beginning. There's even a report that shows line noise level in each of the 512 bands in dBm to the nearest tenth! This thing could probably be reprogrammed to be a nice signal analyzer. Disclaimer: Just another (very) satisfied customer. -- OOO __| ___ Peter Holzmann, Octopus Enterprises OOOOOOO___/ _______ USPS: 19611 La Mar Court, Cupertino, CA 95014 OOOOO \___/ UUCP: {hpda,pyramid}!octopus!pete ___| \_____ Phone: 408/996-7746