High-Speed Modems on Voice Grade Telephone Lines - Troubleshooting connection problems. (this is a consolidation of info published previously by ASTEC, with the critical info now in this single document) With the introduction of the new 28,800 bps V.34 and V.FC type modems, there has been a lot of hype about how these modems wonÕt work very well on existing voice grade telephone circuits and facility wiring. While this is correct in some isolated cases, the truth is that with some careful examination and adjustment, the new technology that these new data modems represent, will work and work well on voice grade lines. This short Technical Note will point you in the direction that you should look, when troubleshooting high-speed modem connections. 1.) WRONG MODEM CABLE - We've written so much about this in the past three years that it has become second nature to us NOT TO TRUST any modem cable, unless it came in the box from the modem manufacturer. Unfortunately, very few do. In 9 out of 10 cases of modem operation problems, it usually is the cable and the way it is wired. Look for this first. We've also seen some inexpensive serial port rotary switches for A-B-C switching that crosstalk badly at 38.4k or 57.6k bps, but work okay at 19.2k bps or lower. The DataSpec push-button A-B-C switch doesn't have this problem. 2.) WRONG MODEM INITIALIZATION STRINGS - Second biggest problem is modem initialization strings. Sometimes they simply don't exist, or the end user has no idea how to deal with them. In some cases, the end user has tinkered with what originally were good strings and has broken them or written faulty strings as parameters into their modem, which get recalled with every power on cycle. 3.) WRONG MAC AND MODEM COMBINATION - The new high-speed modems (certainly those operating with a 57.6k bps DTE) require more CPU processing power than most users realize. Putting such a modem on a Mac Plus (Mac SE and Classic too) simply doesn't have the CPU power and speed to keep up with the demands of the modem, if any uncompressed files (text for example) are involved. Pentium PCÕs running Windows arenÕt any better. Windows chokes above 19.2kbps without making some changes to serial port hardware. 4.) BAD TELEPHONE CONNECTIONS - While most of the carriers are pretty up-to-date in terms of providing clean, noise free connections, there are locations and user facilities that will have problems. Partially damaged facility wiring, oxidized connections, AC interference and other such gremlin's will raise havoc with 14400 and 28800 bps connections, that were never evident when the user was running at 2400 bps. I carry a simple RadioShack push-button phone (about $9) that is known to be quiet (no internal circuit noise) and simply hook it to the clients line, dial "1" and listen. This simple test has revealed quite a few problems, such as hum, static, crosstalk, etc. Here are specific things to look for: A. Oxidized connections on the telephone punch-down block. A major problem in older office buildings with the punch-down block in a damp basement! Also look for old style lightening arrester's and replace them. They can act like a high resistance shunt. B. Copper plated (not pure copper!) steel telephone station wiring, which creates a high resistance connection. I don't know where this cheap copper platted stuff comes from, but in long cable runs it is a big problem. C. Station wiring draped over the top of fluorescent light ballast transformers in the ceiling. (typically drop ceilings, where its easy to string cables and people have done so.) D. Unterminated wire runs off the punch-down block to who knows where? These wire runs act like an antenna and pickup stuff along their way, that the modem has to deal with and over come, if it can. Also, have your phone company sweep the line between your exchange and the exchange you are calling. The test is free and can resolve problems you'll never find otherwise. 5.) TOO MANY INSTRUMENTS ON THE LINE - If you've got too many extension phones (don't forget an answering machine and Fax machine) on the same line as your modem, it can drag down the signal level and cause problems. Why some people will never see 28.8k connects with V.34 or V.FC modems: V.34 (previously known as V.Fast) was designed to adapt itself to virtually any channel, running as fast as possible on that channel with acceptable symbol-error rates. V.34 is designed to run at maximum rate: over a digital (PCM) channel without robbed bit signalling (used for call supervision in some circuits); in which the allowed signal bandwidth is 200-3700 Hz; with very low levels of thermal noise, harmonic and intermodulation distortion, no frequency shift; and little amplitude or envelope shaping by the two subscriber line circuits. People who are physically a long distance (5-30 miles) from their telephone central office with older subscriber-loop wiring will see a drop-off in speed because of the narrower bandwidth caused by the long subscriber loop. City dwellers in high-density housing may see dropoff because the local telephone company uses digital speech compression to carry more traffic over existing cables. Compression introduces phase distortion, and varies depending on the particular compression algorithm used. People connected to older electronic switches will experience dropoff because the subscriber line card doesn't have the extended passband of the newer cards. Excessive bit errors on T1 trunks connecting PBX systems to the central office may limit the connection speed for businesses. Finally, line faults which introduce noise onto the line can limit connection speeds. We hope you find this information helpful. Please call us if you have any comments or questions. Highlander uses 28.8kbps modems daily and has since 1993, with expert help and products from our friends at Hayes Microcomputer and Practical Peripherals.