Analog Devices ADRV9001 User Manual page 143

Preliminary Technical Data UG-1828
The digital gain block is controlled by the Rx gain table as mentioned earlier. Note different digital gain will be applied when configured
in gain correction or gain compensation mode. The Rx gain table has a unique front-end attenuator setting, with a corresponding
amount of digital gain, programmed at each index of the table, as shown in Table 47.
For the gain compensation mode, it can be used in either AGC or MGC mode. The digital gain allows for compensation of both the
internal analog attenuator and an external gain component (such as a DSA or LNA). After the digital gain compensation, the signal
power should only depend on the input signal power.
Around the end of the Rx datapath, Rx interface gain could be further applied by using a "Slicer" block for 2 major purposes. One is to
avoid digital saturation due to the bit-width limitation of the data port in gain compensation mode. The other one is to ensure the overall
SNR is limited only by analog noise and unaffected by quantization noise. When gain compensation mode is used, any analog
attenuation is compensated by a corresponding digital gain, such that the sum of the analog and digital gain is always equal to the
nominal Rx analog gain of 20dB. At the ADC input, the full scale input signal is approximately 8.6 dBm. This value translates to 0 dBFS
in the digital datapath for either the I or Q channel. As an example, assuming a 5dBm signal is applied at Rx input port, at the Rx output,
the signal power will be 5+20 = 25dBm or 25-8.6= 16.4dBFS. This will cause clipping in 16-bit output signal. Therefore, interface gain
(less than 0 in this case) could be applied to attenuate the signal to avoid clipping. On the other hand, for a very low signal level, at Rx
input, within the RF bandwidth of interest, it must be assured that the analog noise dominates the quantization noise. In the Rx datapath,
the final 16-bit quantizer could become the dominant noise source as a result of the final interface quantization. This quantization noise
as a result of the final 16-bit quantizer will be spread over a bandwidth equivalent to its output sampling frequency. For NB applications
where the output sampling frequency is low, the total quantization noise per Hz could be larger than the analog noise per Hz. By
applying interface gain (greater than 0 in this case), prior to the final quantizer, the signal level and analog noise level are both increased.
Therefore, the analog noise dominates over the quantization noise so that SNR is dominated by analog front-end noise in the RF
bandwidth of interest. For WB applications, since the sampling frequency is higher, the total quantization noise becomes much smaller.
In such a case, the analog noise could be way above the quantization noise, therefore, interface gain is not required.
Figure 119 is a block diagram of the digital gain control portion of the Rx chain, showing the locations of the various blocks in the
simplified datapath.
SIGNAL AFTER
WB/NB
HB FILTERING
DECIMATION
INTERFACE
DIGITAL
GAIN
GAIN CONTROL
(SLICER)
RSSI
API
Figure 119. Gain Control and Slicer Section of the Receiver Data Path
It can be seen from Figure 119 that digital gain control is performed in the WB/NB Decimation block. In NB and WB applications, the
digital gain control is actually performed at different stages of the Rx data chain to achieve optimal performance, which is simplified in
Figure 119. The slicer must be dependent on the desired signal power alone and must be done only when all the interfering signals have
been filtered out, i.e. close to the end of the datapath. The Slicer operation can either be controlled automatically by the device internally
or by user externally through API commands. When controlled internally the RSSI block is used to determine the amount of interface
gain.
The following sections describe four different digital gain control modes in the device.
Mode 1: No Digital Gain Compensation with Internal Interface Gain Control
In this mode the digital gain block is used for gain correction. It applies a small amount of digital gain/attenuation to provide consistent
gain steps in a gain table. The premise is that because the analog attenuator does not have consistent steps in dB across its range then the
digital gain block can be utilized to even out the steps for consistency (the default table utilizes the digital gain block to provide consistent
0.5dB steps).
With internal control, the device automatically applies the interface gain determined by RSSI, which measures the input signal power
right before the slicer. Note in the gain correction mode, interface gain less than 0 is not needed since the Rx output level should not
exceed 0dBFS through either AGC or MGC. When in NB applications, the interface gain range could be from 0 to 18dB in 6 dB step size
(0,6,12,18) for improving the sensitivity. In WB applications, as discussed earlier, the sensitivity is already satisfied by the high sampling
rate so the interface gain is always 0.
Rev. PrA | Page 143 of 253