ST7021@SIUCVMB.BITNET (03/20/91)
I am curious to know what the 'real' connection speeds (minus compression, etc) of some of today's high speed modems are (Hayes, USR and Telebit, for example). I had an interesting coversation with an IBM telecomm guru about six years ago where he told me that 9600bps was about as high as normal phone lines can handle due to their quality. Is this still generally true? I realize that he had given me a number that was in a logical step (2400, 4800, 9600, ...) so is there a maxed-out ceiling speed, once again minus compression, etc? Lastly, on a similar topic, what is the most common technique used by modem manufacturers to deal with the frequency overlap (echo suppression, ping-pong, etc)? Thanks for all the help! Thomas Boehler
lstowell@pyrnova.pyramid.com (Lon Stowell) (03/21/91)
In article <9103200803.AA13245@ucbvax.Berkeley.EDU> ST7021@SIUCVMB.BITNET writes: >I am curious to know what the 'real' connection speeds (minus compression, etc) >of some of today's high speed modems are (Hayes, USR and Telebit, for example). >I had an interesting coversation with an IBM telecomm guru about six years ago >where he told me that 9600bps was about as high as normal phone lines can >handle due to their quality. Is this still generally true? I realize that he >had given me a number that was in a logical step (2400, 4800, 9600, ...) so >is there a maxed-out ceiling speed, once again minus compression, etc? Technology marches on. With a good V.32bis modem you can actually get 14,400 bits/second thru-put on most lines in North America. If you are using an async terminal, the thruput is slightly higher due to conversion in the modems... For sync protocols, 14,400 is available. If you choose to go to the added expense of data conditioning for your access loop, this speed is more typical than not.
tnixon@hayes.uucp (03/21/91)
In article <9103200803.AA13245@ucbvax.Berkeley.EDU>, ST7021@SIUCVMB.BITNET writes: > I am curious to know what the 'real' connection speeds (minus compression, etc) > of some of today's high speed modems are (Hayes, USR and Telebit, for example). The Hayes V-series Smartmodem 9600 proprietary ping-pong modulation scheme is true 9600 bits per second, as is the V.32 modulation scheme used in the Ultra 96. The USR HST modulation scheme is 14,400bps in one direction (raw bit rate) and 450 bps in the reverse direction (asymmetrical); V.32bis, as implemented in the Courier Dual Standard and Courier V.32bis, is true 14,400bps full duplex. The peak modulation rate of the Telebit DAMQAM modulation scheme is, to the best of my understanding, 18031bps. With V.32 and V.32bis, if you have a synchronous DTE, it actually sees the full rated throughput (9600 or 14400). With Hayes ping-pong, you have to subtract out the turnaround time, which takes, oh, about 40 milliseconds per turnaround (adding the two directions together). It's much more difficult to predict that actual data rate of a DAMQAM modem, since it varies in two-bit increments based on the phone line, and is also half duplex (and the turnaround time is a lot higher than the Hayes ping-pong). > I had an interesting coversation with an IBM telecomm guru about six years ago > where he told me that 9600bps was about as high as normal phone lines can > handle due to their quality. Is this still generally true? I realize that he > had given me a number that was in a logical step (2400, 4800, 9600, ...) so > is there a maxed-out ceiling speed, once again minus compression, etc? I'm surprised! "Gurus" always used to say the limit was 2400, and that 9600 was a pipe dream. Now they're saying 9600. Oh, well. The truth is, 14,400 works very well on average, everyday US phone lines! It doesn't work nearly as well on international calls, because of the 40kbit ADPCM encoding used in the DCME/PCME equipment on transoceanic trunks. But even on long-distance calls in the USA, V.32bis works well. CCITT Study Group XVII is currently studying "V.fast", which is intended to be the "ultimate modem". It will provide the optimal performance on whatever the phone circuit has to offer. On a good circuit (80% or more of the circuits in the USA), it should be able to achieve 24,000 bps -- raw throughput, without error control or compression. This uses multidemimensional coding techniques, sophisticated line probing and measurement, and various other newfangled pieces of technology understood by very few people, it seems. But it's been demonstrated to work! It will be 1993, probably, before the standard is released, but prototypes are working today. > Lastly, on a similar topic, what is the most common technique used by modem > manufacturers to deal with the frequency overlap (echo suppression, ping-pong, > etc)? The most popular standards (V.32, V.32bis) use echo cancellation (not supression). V.fast will be an echo cancelling modem, too. Hayes V-series, Microcom MNP6, and Telebit PEP are "ping-pong" techniques. USR HST, CompuCom, EVI Fastcomm are "asymmetrical" (one high speed channel with a low-speed reverse channel). Racal-Vadic 9600VP is "dynamic"; V.22bis during low volume, switching to V.29 when necessary to push through data in one direction. It's really hard to say what is the "most common" among the non-standard techniques; it varies based on the market and application. -- Toby Nixon, Principal Engineer | Voice +1-404-840-9200 Telex 151243420 Hayes Microcomputer Products Inc. | Fax +1-404-447-0178 CIS 70271,404 P.O. Box 105203 | UUCP uunet!hayes!tnixon AT&T !tnixon Atlanta, Georgia 30348 USA | Internet hayes!tnixon@uunet.uu.net
larry@nstar.rn.com (Larry Snyder) (03/21/91)
We run slip via a leased line between a pair of usr v.32/v.32bis/v.42bis modems and get throughput on text file (using ftp) around 3.1 kb/sec and around 1.7kb/sec sending compressed binary data. both ends are locked at 38400 and using hardware flow control and we are using the stock interactive slip drivers (we would like to try ppp once it is available for interactive unix release 2.21) larry -- Larry Snyder, NSTAR Public Access Unix 219-289-0287 (HST/PEP/V.32/v.42bis) regional UUCP mapping coordinator {larry@nstar.rn.com, ..!uunet!nstar.rn.com!larry}
rdippold@gdansk.qualcomm.com (Ron Dippold) (03/22/91)
In article <9103200803.AA13245@ucbvax.Berkeley.EDU> ST7021@SIUCVMB.BITNET writes: >I am curious to know what the 'real' connection speeds (minus compression, etc) >of some of today's high speed modems are (Hayes, USR and Telebit, for example). >I had an interesting coversation with an IBM telecomm guru about six years ago >where he told me that 9600bps was about as high as normal phone lines can >handle due to their quality. Is this still generally true? I realize that he >had given me a number that was in a logical step (2400, 4800, 9600, ...) so >is there a maxed-out ceiling speed, once again minus compression, etc? He might have been using IBM standard procedure when talking about quality and was doing what they usually do: huge safety margins. I regularly use my Dual Standard to connect at 14.4K from and to standard residential lines. Now granted, when I have error correction turned on I won't see any errors, but even with it turned off (just for kicks) it still works fine. Several BBSes in town use Telebit Trailblazer 19.2K modems with no trouble. So: I think there _is_ more noise at speeds above 2400, but the modems all have error correction protocols built in, so it really doesn't matter.
ef@tools.uucp (Edgar Fuss) (03/26/91)
>The most popular standards (V.32, V.32bis) use echo cancellation >(not supression). Sorry for the stupid question, but what's the difference?
tnixon@hayes.uucp (03/27/91)
In article <EF.91Mar26162802@fidel.tools.uucp>, ef@tools.uucp (Edgar Fuss) writes: >>The most popular standards (V.32, V.32bis) use echo cancellation >>(not supression). > Sorry for the stupid question, but what's the difference? An echo suppressor (used in the four-wire portion of long distance circuits, not in modems) complete mutes (turns off) the signal in the opposite direction when one direction is "talking". This works fine with half-duplex modems and normal speech, since only one side talks at once. The only problem is with the annoyance of sometimes having the first syllable of a word "clipped" because the supressor didn't switch directions fast enough. To eliminate this, networks have switched to using echo cancellation, which records the transmitted signal and subtracts echoes of it out of the received signal rather than completely supressing the received signal. V.32 modems also use echo cancellation to achieve two-way transmission (they couldn't use echo suppression, since echo suppression explicitly enforces one-way-at-a-time transmission). -- Toby Nixon, Principal Engineer | Voice +1-404-840-9200 Telex 151243420 Hayes Microcomputer Products Inc. | Fax +1-404-447-0178 CIS 70271,404 P.O. Box 105203 | UUCP uunet!hayes!tnixon AT&T !tnixon Atlanta, Georgia 30348 USA | Internet hayes!tnixon@uunet.uu.net