how to measure sensitivity of a CC1110?

The short answer is there is no test pin that  can measures sensitivity.

First sensitivity needs to be defined. This is typically done by measuring the input power at a desired packet error rate.   This must be done via a cable and not over the air.

A super calibrated Agilent communication test set is the way to go. The “poor man’s” approach I use works very well.  I’m working with the CC1110 at 915MHz but this will work with any of the Ti devices. 

I know at my bandwidth / data rate I can expect performance down to -80dBm to -100dBm.  The CC1110’s output is programmable from +10dBm to -30dBm.  I use a SMA cable from a CC1110EM with 3 20dB attenuators and a 10dB attenuator for a total of 70dB of attenuation that is then connected to the board I’m testing.  This gets me a power output out of the attenuators of -60dBm to -100dBm by programming the CC1110’s power output and is accurate to several dB.   Using Ti RF Studio to drive the CC1110EM I set the parameters I want and then use Packet Tx set to send 1000 packets.  Using a second Ti RF Studio set to the same parameters for the board being tested I set it to Packet Rx.

       RF Studio Tx  --->  CC1110EM ---> Cable  ---> Attenuator  ---> Board under test --- RF Studio Rx

On a recent CC1110 design when I first started I received 1000 of 1000 packets with the CC1110Em power set to +3dBm which is -67dBm after the attenuators which is pretty bad. After finding a wrong value inductor and playing with the parameters in RF Studio I measured 996 out of 1000 packets with the CC1110EM power out set to -21dBm which is -91dBm after the attenuators and at the input of my CC1110 board which is a pinch better than the Ti spec.  Set at -30 which is -100dBm at the attenuator output I received 175 packets out of the 1000 sent which is unusable for most applications. It is easy to vary the RFStudio parameters, supply voltage, bias resistor value, etc to see which has the most impact on PER as each 1000 point measure loop only takes a few seconds

SMA attenuators on E-Bay are $5 to $15 each and are usually very accurate.  If you have a power meter to check the power out of the CCxxxx device you can get accuracies of +/- 3 dB which for 20 bucks in test gear is a bargain.

Some ways this is better than a $60K test set because it includes the whole link including PER errors caused by errors and distortion on the Tx side.

This approach does not test the antenna but then again if the receiver is under performing the best antenna will not help it. Last, once you know how the receiver works replace the attenuators with antennas and some distance between the units. As you vary the antennas and watch the PER you can correlate the PER with a specific input power at the boards input connector and get a actual comparison, in dB, of the antennas you are testing.

BTW with the above set up you can accurately find the correct RSSI offset making the RSSI measurement much more accurate.