How We Test

For these preliminary tests, the author calculated speeds based on hand-timed drag-and-drop transfers of a single 87MB file from a server (wired via 10/100 ethernet to the test router) to a client laptop outfitted with the test Wi-Fi PC Card.

All tests were conducted in a suburban home in which a spectrum analysis showed average signal levels near zero over an extended period, indicating a very low level of interference that we deemed acceptable for a preliminary test.

The author tested each router and PC Card at short, medium, and long range. In the close-range tests, the router was set up in a home office with the client approximately 5 feet away in the same room. In the medium-range tests, the router remained in the office while the client was moved two rooms away, or about 25 feet. Bookcases, cabinets, and three walls stood in the path of the signal. In the long-range test, the router remained in the office while the client was moved upstairs to the other side of the house, through several walls and one ceiling, for a total distance of about 50 feet.

Our server test bed was a Dell Latitude D600 notebook; our client test bed was an HP Pavilion dv8000 notebook. The author disabled internal Wi-Fi in both systems.

On the routers, the author turned on WPA2 encryption and selected both auto channel and auto mode, if available. "Auto mode" means the router automatically switches between single-channel and channel-bonding modes as required. "Auto channel" means the router automatically operates on the best available channel (namely, the one with the least interference). Other settings were left at defaults.

The author performed three tests at each distance and averaged the results. All tests were performed within a 5-hour time frame to minimize differences based on atmospheric change.

Back to "First Draft-N Wi-Fi Gear Disappointing."