What is the selectivity and how could it be measured?
Selectivity tells how many dBs above the wanted signal level the receiver can tolerate a blocker signal while still maintaining the minimum sensitivity criterion.
Solution to measure the selectivity: Typically selectivity can be measured with two generators. One of the generators that provides the wanted signal should be set to 3 dB above the sensitivity level (measured as the power level at which the BER is 0.1%). The second generator with an unmodulated signal is used as the interferer and combined with the primary signal using a power combiner. The second generator is placed at the desired frequency offset (red arrow) and then the power is increased until the BER degrades to 0.1%.
A narrowband selectivity curve in the 868 MHz band is shown on the next figure below. The x axis is frequency offset from the wanted channel in MHz, the y axis is Selectivity. This value is inverted on this graph for a more intuitive perception.
From the operation of the chip it follows that there is one offset frequency where the selectivity is worse, this frequency (at the little hump on above graph), is called the image frequency. The image frequency of the chip is at 2 x IF below the actual RF frequency. For example for the nominal crystal frequency of 26.0 MHz, the image frequency of the chip is at 2 x IF=812.5 kHz (2 x 406.25 kHz = 812.5 kHz) below the actual RF frequency. Let’s return to the first figure. The offset frequency –the big red arrow- generates a not wanted signal inband. The little red arrow represents this signal.
In order to receive the desired signal when the interferer is present, the desired signal must typically be 8–10 dB higher than the interferer. This is the co-channel rejection. As it can be seen on the Figure above, rejection is the ratio of the signal strength at the image frequency (interferer) to its counterpart at the interferer.
Solution to measure the selectivity: This method is easier than the first one, because only one generator is needed and RSSI values need to be read back from the chip. The RSSI is measured continuously while the chip is in RX mode. The RSSI value represents the moving average of the signal strength. On the next figure two selectivity curves are shown in the 868 MHz band, which were measured with the two different ways on the same board. The red curve represents the RSSI measurements.
As it can be seen on the figure, that there is a constant offset between the two curves. The explanation is, that the RSSI curve represents the rejection and not the selectivity. The RSSI method is only recommended to be used around the adjacent channels and at the image, because the RSSI has a minimum noise level, which is typically too high for measuring accurately at further out offsets.