The Test
I followed the Azimuth-based test procedure to test each card with a Linksys WRT54G V5 802.11b/g wireless router. I chose this router because it is probably one of the most widely-used 11b/g routers, with typical 802.11g performance. I did make one key change to my normal test procedure.
Instead of making direct connection to each card, I used the Near-Field Antenna (NFA) to couple the cards under test into the Azimuth ACE. I did this because the Linksys WPC54G uses a unique mini-connector that didn't fit any of the adapter cables that Azimuth provided. So to ensure as much of an apples-to-apples comparison as possible, I used the NFA with all cards.
Figure 7 shows how the NFA was attached to notebook / Cardbus cards. Because plastic housing tends to vary in height, I attached the NFA to the bottom surface of each card's antenna area. Since the NFA is very sensitive to how far it is from a signal source (they really do mean near!), attaching it to the bottom provided a more consistent board-to-NFA distance.

Figure 7: NFA attached to bottom of Cardbus card
Figure 8 shows the NFA coupled to the dipole antenna of the EnGenius USB adapter. Can't get much closer than that!

Figure 8: NFA attached to dipole antenna
Testing the internal Intel 2915ABG in my Fujitsu P7120 notebook called for a little experimentation, since I didn't know where the antennas were. So I fired up IxChariot and moved the antenna around while watching throughput until I found the little buggers! Figure 9 shows where the NFA finally ended up.

Figure 9: NFA attached to notebook
Note that the WRT54G router was directly connected via a cable from one of its antenna connectors to the Azimuth system.






