well, it is a basic way to select how your antenna is connected to the chip. When DC current is flowing in the two pin diodes, then the PA is connected to the antenna, and the LNA input is short circuited to ground via the capacitor to ground right next to the second diode. If any of the PA output power goes down that line to the LNA, it sees the RF short circuit, and no big transmit voltage can develop to hurt the LNA input.
For any RF energy going down the line (with both PIN diodes forward biased), it has a 90 degree phase shift, hits the pin diode RF-shorted to ground and bounces back with 180 degrees, and travels back to the common junction with another 90 degrees. So, the reflection coefficient looking down the line toward the LNA is p=1 with an angle of 360 degrees, or equivalently p=1 with an angle of 0 degrees. That is also the definition of an "open circuit". So any RF energy looking down that line sees an open circuit, and instead goes toward the antenna.
Now the reality is that you can not make the line be 90 degrees in electrical length from 2400 to 2500 MHz--the best you can do is to make the line 90 degrees at midband (2450 MHz). Also, there are other complications. The PIN diode has some series inductance, which causes phase shift. The capacitor to ground, and the ground via it uses, also have some inductance. So you really do not want 90 degrees at 2450 MHz, but something slightly different from 90 degrees to compensate for it.
The proper way to do the design is to use a RF simulator program, with models of the diode, capacitor, via hole, and microstrip transmission lines, and tweak the transmission line length to center the insertion loss from PA to antenna at 2450 MHz.
A rough guess, if you keep the pin diode, capacitor, and via hole very close together, is that they might look like perhaps a 1 nH inductance to ground when the PIN diode is forward biased. That is J15 ohms of reactance. Using a smith chart, you can see that you need a line length of 0.453 wavelength for the line in question.
You can calculate how long this is if you know the dielectric constant of the board you are using, the width needed to make it 50 ohm, and the effective dielectric constant. You can probably find a free program online somewhere to calculate those things, maybe apcad, etc. In FR4 board, 14 mils thick, the line would be around 25 mils wide and 547 mils long.
Maguffin Microwave wireless design consulting
