In a message dated 11/21/2002, you write:
> Using default config ATZ I can never connect higher than 31200 > V42bis, as reported by the connection string reported onscreen > in the terminal software. But I think I should be able to connect at > V34 or V90 at higher throughput. > Hi Jeffrey: I've seen some of the other responses you got, and they point out some important considerations. I will just add a few comments to reinforce what I believe to be the important points: 1) The maximum possible modem-to-modem connect speed through the PSTN is V.34 33,600 bps. To get V.90 speeds you must have a V.90 modem connected to a "digital modem" at the central office. In other words, there is only one analog hop in a V.90 connection. V.90 works fine for internet access, but V.90 is not possible for two analog modems connected through the PSTN. 2) With many modems it can be difficult to determine the actual connect speed, especially if V.42bis (error correction plus data compression) is enabled. I suggest that you turn this feature off for your tests. 3) The maximum 33,600 bps rate for a V.34 modem requires an extremely clean channel. For example, this rate can only be achieved on about 5% of the connections through the North American PSTN. On the other hand, a good V.34 modem will achieve rates of at least 26 K on over 90% of the connections through the PSTN. 4) Connections through a typical PBX usually introduce enough impairments to keep the connect speed below 26 K on a port-to-port call. If you are calling through both a PBX and the PSTN, the connect speed will probably be even worse. 5) Most of the low cost line simulators, intended for analog telephones, also limit maximum connect speeds. To properly evaluate high speed modems you need a line simulator that is designed for that purpose, such as the ones available from TAS/Spirent. 6) If your goal is to properly evaluate modem performance in the presence of real-world impairments, the testing should be performed according to TIA/EIA TSB37A or the newer TSB3700, in combination with TSB38. This is a very sophisticated network model and well thought out evaluation process. TAS/Spirent sells test equipment to do this testing, and a few test labs have the equipment. 7) Many people have a simplistic view of what impairments affect modem performance, thinking that it is simply a matter of noise and bandwidth. In reality, there are several other important parameters. The TSB3700 network model does a good job of simulating just about all of them. Joe Randolph Telecom Design Consultant Randolph Telecom, Inc. 781-721-2848 http://www.randolph-telecom.com
