smith@umbc3.UMBC.EDU (Fred Smith) (03/07/89)
I have a question that I hope someone will be able answer for me.
I realize that the effective "maximum" distance recommended for RS-232
under proper conditions, is 50 ft..
Although I have exceeded that distance many times, for the sake of
argument, I would like to know is if there exists an effective "maximum"
or standard distance for running CCITT V.35?
thanks in advance
fred
--
Frederick W. Smith | University of Md. At Balto.
| 610 W. Lombard St. Rm. 210
smith@umbc3.umbc.edu | Balto. Md. 21201 1-301-328-6143
goodloe@ingr.com (Tony Goodloe) (03/09/89)
In article <1766@umbc3.UMBC.EDU>, smith@umbc3.UMBC.EDU (Fred Smith) writes: > ... I would like to know is if there exists an effective "maximum" > or standard distance for running CCITT V.35? If somebody can point me to the page number in (one of the several standards associated with) the V.35 spec that talks about distance, I sure would appreciate. All we have been able to find is (paraphrased) "Don't run it a distance that it won't work at. This distance will be shorter as the data rate increases." Big help. I love standards :) tony goodloe
csg@pyramid.pyramid.com (Carl S. Gutekunst) (03/09/89)
Followups to comp.dcom.modems ONLY. In article <1766@umbc3.UMBC.EDU> smith@umbc3.UMBC.EDU (Fred Smith) writes: >I realize that the effective "maximum" distance recommended for RS-232 >under proper conditions, is 50 ft.... What RS-232-C actually says, is: Section 1.3 This standard is applicable for use at data signalling rates in the range from zero to a nominal 20,000 bits per second. Section 3.1 The use of short cables (each less than approximately 50 feet or 15 meters) is recommended; however, longer cables are permissible, provided that the resulting load capacitance, measured at the interface point and including the signal terminator, does not exceed 2500 picofarads. Typically you'll pick up anywhere from 300 to 600 picofarads (pf) before you ever plug in the cable. Cable capacitance is in the neighborhood of 25pf per foot, so that means something like 80 feet before you are exceeding the RS-232-C standard. Not that it won't *work*; you'll just be outside what the standard requires, and into the domain of implementation. You can also spend more money, and buy cable with a lower capacitance. >I would like to know is if there exists an effective "maximum" or standard >distance for running CCITT V.35? In summary, it is exactly the same: 2500 picofarads. The "50 foot" suggestion isn't made at all; just the limit due to capacitance. Here's the long story: V.35 is a mess. What Recommendation V.35 specifies is an encoding scheme at 48Kbps for leased lines. What has popularly become referred to as "V.35" is just the "Interface" portions of V.35, contained in Section 10 and Appendix II of the Recommendation. This in turn falls back to Recommendation V.28 for everything that isn't specified in V.35. V.28 is only an electrical specifi- cation; if you actually want to know what the circuits *do* you have to look at V.24. And the big M34 connector typically used by V.35 interfaces isn't a CCITT Recommendation at all. (Or at least, I've never found it.) The V.35 Interface sections define a low-impedance balanced circuit, kinda but not quite what us old Teletype hands called current loop. :-) The maxi- mum length is "for further study," although clearly you could go hundreds of feet without exceeding the standard's impedance requirements. (BTW, the phrase "for further study" is a magic cookie that means the committee was unable to agree on something, and let the issue drop for another four years.) Still with me? The problem is that V.35 specifies *only* the Transmit Data, Receive Data, and two clock circuits. The remaining circuits -- carrier detect, terminal ready, request to send, clear to send, and so on, are covered by V.28. And guess what V.28 is? Yup. The electrical portion of RS-232-C, hidden under a CCITT label. 2500pf is all you get. <csg>
kaufman@polya.Stanford.EDU (Marc T. Kaufman) (03/10/89)
In article <4287@ingr.com> goodloe@ingr.com (Tony Goodloe) writes: >In article <1766@umbc3.UMBC.EDU>, smith@umbc3.UMBC.EDU (Fred Smith) writes: .> ... I would like to know is if there exists an effective "maximum" .> or standard distance for running CCITT V.35? >If somebody can point me to the page number in (one of the several >standards associated with) the V.35 spec that talks about distance, I >sure would appreciate. All we have been able to find is (paraphrased) >"Don't run it a distance that it won't work at. This distance will be >shorter as the data rate increases." Big help. I love standards :) Nevertheless, that is the best statement of maximum distance I can think of. For instance, with RS-232, the "50 foot" limit is an absolute worst case based on 20KB data rate, recommended rise-time limiting, use of the recommended current limiting drivers, high capacitance cables, and +/- 3 volt (minimum) swings. Now you just can't buy components that are that lousy in all respects! The real answer is that you can run as far as you can receive good waveforms. Balanced circuits typically have a little more margin than unbalanced. I can't imagine why you care, if you can get the circuit to work, unless you are expecting someone to rewire with zip-cord when you aren't looking. Marc Kaufman (kaufman@polya.stanford.edu)
csg@pyramid.pyramid.com (Carl S. Gutekunst) (03/10/89)
In article <4287@ingr.com> goodloe@ingr.com (Tony Goodloe) writes: >If somebody can point me to the page number in (one of the several >standards associated with) the V.35 spec that talks about distance, I >sure would appreciate. The CCITT recommendations are very specific about what the cable has to look like; see Recommendation V.35, Appendix II and Recommendation V.28, chapter 3. The problem is that the "cable length" is defined in terms of impedance and capacitive load, since these are qualities that vary tremendously based on the type of cable being used. For example, Black Box sells what they call "V.35 cable" that is specified at 12pf and 0.03 ohms per foot. Given that cable and Pyramid's present V.35 Line Interface Module that has active slew rate limiting (don't ask :-)), I could run 200 feet of cable and be in compliance. On the other hand, if I went down to Radio Shack, bought a Centronics printer cable, clipped off the ends, and put M34 connectors on it, the maximum might be very close to 50 feet. I am well aware that what *works* is an entirely different issue from what's standard. But I just don't accept the "if it works, it must be OK" philosophy. Someone wanted to know, "why do you care what the standard says?" The reason *anyone* cares is realibility. If I push the margins to the edge to make a brand new installation work, what happens when the cables and equipment age? No, the standards are very specific just for this very reason, so that you can install something and expect it to work not only now, but 20 years from now, and when you switch out an old modem or an old length of cable and replace it with something else. And the standards *are* very specific; they just aren't specified in language that is useful to the average system administrator. Which is what keeps people like me employed. :-) <csg>
bobk@boulder.Colorado.EDU (Bob Kinne) (03/14/89)
In article <4287@ingr.com> goodloe@ingr.com (Tony Goodloe) writes: >In article <1766@umbc3.UMBC.EDU>, smith@umbc3.UMBC.EDU (Fred Smith) writes: >> ... I would like to know is if there exists an effective "maximum" >> or standard distance for running CCITT V.35? > >If somebody can point me to the page number in (one of the several >standards associated with) the V.35 spec that talks about distance, I The max distance and the baud rate are interdependent. In McNamara's book "Technical Aspects of Data Communications" on page 240 there is an analysis with graphs. Briefly, at 1200, 3000ft shielded or un, at 4800, 1000ft shielded, 250ft unshielded, at 9000, 250ft either way.
jim@stream.UUCP (Mr. Jim's Own Logon) (03/16/89)
In article <7369@boulder.Colorado.EDU>, bobk@boulder.Colorado.EDU (Bob Kinne) writes: > The max distance and the baud rate are interdependent. In McNamara's book > "Technical Aspects of Data Communications" on page 240 there is an analysis > with graphs. Briefly, at 1200, 3000ft shielded or un, at 4800, 1000ft > shielded, 250ft unshielded, at 9000, 250ft either way. While I hold the McNamara bible in high esteem, some of the graphs are misleading and incorrect. This is such a case. For almost any transmission medium, there is a speed distance product. The simplest way to prove this is the Capacitive Reactance of the cable: X(c) = 1/(2 (Pi) F C ) This is the AC resistance to ground. As the capacitance goes up (longer cable length) the resistance to ground goes down (less signal). The frequency (F) plays the same role, as F increases, the resistance to ground decreases. Depending on the cable that you have, X(c) may or may not be a dominant factor at the speed and distance you are considering. RS232, V35, fiber optics, the same principals apply. McNamara is just showing what 90% of the people are satisfied with 90% of the time. Just hope that you have 90% of their environment.... -Jim Wall Bell Technologies
csg@pyramid.pyramid.com (Carl S. Gutekunst) (03/16/89)
In article <7369@boulder.Colorado.EDU>, bobk@boulder.Colorado.EDU (Bob Kinne) writes: >For almost any transmission medium, there is a speed distance product. >The simplest way to prove this is the Capacitive Reactance of the cable: >X(c) = 1/(2 (Pi) F C ) As long as we're splitting hairs.... The aspect that throws this off is we're talking about square waves, not sine; reactance is higher for the higher har- monics, thus "rounding" the theoretically square digital signal. Which is why both RS-232-C and V.35 specify "cable length" only in terms of C, with an assumed constant F (20kHz and 48kHz, respectively), plus a mess of require- ments regarding limited slew rates and zero crossing times. (Don't tell me about F being half the 20000bps or 48000bps bit rate. The data signals are 10kHz and 24kHz, but the clock signals are 20kHz and 48kHz. And anybody who doesn't think those clocks have to be squeaky clean has never tried building a large SNA or X.25 network. :-)) Whew! Have we sufficiently beaten this subject to death now? :-) <csg>