Well, I can tell you how I do it. I usually work at 900 MHz, so I don't know how well this method will scale with frequency, though it should if you use the right cables.
First of all, you need a vector network analyzer. The 8753D or later model from HP/Agilent is a nice choice. The important thing when using a VNA is to calibrate the ports. More on that later.
On your board, you need a way to disconnect the DUT (device under test) from the rest of the circuitry. Normally, there is a DC blocking cap in line with most parts of an RF circuit, so the easiest thing to do is to remove the DC blocking capacitor. Next, take a short section of RG174 cable with a SMA connector on one end and strip back the other end to expose the shield and center conductor. You will also need a complete SMA RG-174 cable with SMA connectors on both ends for calibration.
Solder the shield of the test cable to a ground point very close to the antenna, filter, or whatever you are testing. Solder the center conductor to the pad of the DC block you just removed so that the cable is now electrically connected to the DUT.
Now, connect the other cable (the one we are using for calibration) to one of the ports of the VNA; Lets use port 1 for this example. Using the instructions for the VNA, perform a 1-port calibration on port 1. THis will involved you connected calibration standard loads to the end of the cable; there will be a 50-ohm, a short, and an open standard. Once complete, the VNA is now calibrated, including the length of cable you are using with the DUT.
You should now be able to connect the cable soldered to the DUT to port 1 of the VNA and perform your measurements.
You can do the same trick with 2 cables if you need to do 2 port analysis for filters.
Please post any questions you have if this isn't clear enough.
- Steve Montgomery
The RF Troubleshooterwww.rftroubleshooter.com