- Analog AGC is mainly used to increase the signal level to match the ADC sensitivity? Is Analog AGC also used to attenuate the signal?.
- Digital AGC is the first block used after ADC. Is the main purpose of digital AGC is to provide constant level to the signal processing receiver chain( RRC, CIC, demod....)
- So analog AGC output is range of signal levels but digital AGC always gives out a fixed level?
Analog AGC is mainly used to increase the signal level to match the ADC sensitivity? Is Analog AGC also used to attenuate the signal?.
For "Analog AGC" we use analog control components (typically voltage variable amplifiers and voltage variable attenuators) to adjust the receiver gain, and we use an analog power detector to measure the total band power (within the bandwidth of the filtering up to the detector). This measured power is compared to a target which generates an error signal (often with an op-amp in an analog circuit) and the error is integrated in a loop filter (also often with the same op-amp) which then integrates to the control voltage needed to make the error go to zero (and the loop filter can be further modified for higher order loop control). In the end the band power AS MEASURED AT THE DETECTOR will be constant except for dynamic conditions during power transitions. Depending on what is used in the receiver for gain control, the AGC can amplify or attenuate all the signals in band as allowed.
Digital AGC is the first block used after ADC. Is the main purpose of digital AGC is to provide constant level to the signal processing receiver chain( RRC, CIC, demod....)
Digital AGC is not necessarily the first block after the ADC, or at least the level detector part of the AGC loop. A significant benefit of doing a subsequent AGC digitally is that you can measure the level (which sets the level of the overall waveform within the bandwidth at the detector) after more thorough filtering of the channel signal of interest and rejecting out of band interference (which could include other channels in a multiband receiver). This interference would reduce the signal of interest in band when AGC'd (power measured) earlier in the chain. Consider the above case earlier for the analog AGC where we have a very strong interference signal near the channel bandwidth of interest, so close that filtering in the analog would be costly. Assume too that we have a high dynamic range ADC with excess dynamic range allowing us to shift some of this filtering challenge to the digital. The analog AGC's purpose is to avoid excessive clipping at the ADC input (or other similar level sensitive components in the receiver chain), so with the strong interference our signal of interest will effectively be significantly attenuated, give the interference will set the full scale level at the ADC input.
So the Digital AGC's purpose is to further bring this signal back to the proper full scale level (and minimize subsequent receiver processing) by selecting the bandwidth of interest and removing the remaining interference digitally, and then once removed, bring the channel signal back to the proper full scale level (some determined back-off to minimize clipping in the waveform but also reduce the subsequent operations required assuming a fixed point design, meaning minimize datapath requirements).
Also we must be careful to consider all components between the adjusted signal level and the detector (assuming filtering prior to the detector means that the actual power in the signal affecting the components in line can be much higher leading to saturation, clipping and gain compression.
So analog AGC output is range of signal levels but digital AGC always gives out a fixed level?
Not quite. Both analog and digital AGC provide a fixed level of channel power, as measured by the relevant power/signal detector used. What may more likely occur in the analog is strong interference controlling the AGC, so the AGC sets its level based on the interference and not our signal of interest. The AGC only measures power and does not recognize signal from interference: we help that with filter design.
There are also systems where the level is measured digitally and the front-end gain is controlled but it is important to monitor the total signal level at the input the ADC to optimize the trade between quantization noise and clipping, maximizing the SNR of the signal with the proper AGC set-point at the ADC input. (Some clipping is often actually better than backing off too far and introducing more quantization noise in the process).
The AGC loop consists of a detector and a way to adjust the power along with an appropriate loop filter as a control loop. The power adjustment can be done as gain or attenuation or both, depending on what is available in the design. The detection and gain adjust and loop filter can all be done digitally or analog or both and as suggested above it can be a multi-loop design, and the gain can also include stepped gain control.
For more info see these posts: How to adjust receiver gains to avoid saturation and quantization noise to optimise post digital processing?