Test Specs

We took a multipronged approach to pick the best telephony devices. First, we ran our traditional modem performance benchmarks that measure data throughput for file transfers. We compiled performance, implementation, and technology scores and then computed a best overall score by assigning a weight to each element: performance (40 percent), implementation (50 percent, which is further divided into 45 percent usability, 45 percent features, and 10 percent value), and technology (10 percent).

Performance

Since the core of the telephony devices is their on-board modem, we ran tests that measured how fast they could transmit and receive files over simulated telephone lines. Testers connected like pairs of modems to a Telecom Analysis Systems ((908) 544-8700) Series II modem tester. Th e machine emulates line-impairment conditions with mild white noise, near and far echo, and a short satellite delay, so the modems had to clear some interference. The loop emulator generates an EEIA1at both ends of the connection. These conditions represent a direct telephone connection at 2000 feet and is the shortest local loop.

During the test cycle, the modems transfer three types of files (compressed, graphics, and text) that are representative of what an on-line browser might encounter. Lab technicians measured each pair for one-way and two-way transmissions with the telephony devices configured to receive data from the computer at the fastest rate they support. Testers installed serial communications enhancement software that took advantage of the 16550 and overcame the limitations of standard Windows communications drivers. Testers enable V.42bis data compression and V.42 error correction, even if the modem's default settings specify other protocols. Technicians also set up the modems to use har dware (Ready to Send) flow control, rather than XON/XOFF. For data compression and error control parameters, we used the default window and dictionary sizes.

Under our one-way transmission tests, modem A calls modem B. Modem A then sends a file and hangs up. Modem A then repeats the process three more times. During a one-way transmission, modem B only receives data; it does not send anything back to modem A. During the two-way transmission tests, modem A still makes four calls to modem B; however, when modem B answers, both modems simultaneously send files to one another. The modems from Boca Research, Creative Labs, Connectware, and Zoom Telephonics failed some of the two-way tests, which ultimately hurt their overall performance score.

Usability/Features

To judge these products' level of usability, we used them to send faxes. Our objective was to see how easy it is to install the communications software and how easy it is to use once it was fired up. We took a long look at voi ce-mail capabilities to see, for instance, how many messages the products supported and if you could easily access messages from remote locations. We studied the communications capabilities of each product. We checked to see if the documentation was clear and technically correct. And we weighted and scored the important features that we believe a telephony product should have.

---