Pancho

I think we are mostly in agreement. There are a couple of potential places you can apply the amplitude reconstruction to a polar transmitter. One possibility is to apply the amplitude reconstruction at a low level, so that most of the power gain and level control takes place after the amplitude modulator. This is not so helpful from a current savings point of view, as your TX linearity requirements are very similar as with a quadrature modulator.

The other possibility is to have the amplitude modulation in the PA, where the PA gain is amplitude modulated to achieve the fully modulated signal. This is the case where you may save some current in the transmit chain, as most of the gain in the transmit chain is constant envelope, so you could run compressed amps and get high PAE.

However, this means you are amplitude modulating the last stage in the transmitter. Any spurious or noise that the AM process adds is difficult to filter out. In addition, the exercise of amplitude modulating a PA with a high Peak to Average signal, achieving high PAE and good modulation quality is very challenging. I believe this is the approach RFMD used for their Polaris chipset, but I'm not sure anyone else has made that work. Happy to be corrected here.  To achieve good ACPR with a large signal Polar approach, I suspect you'd need to a predistortion system. Simple predistortion based on table lookup isn't so involved, but if you want active predistortion, then you need a pretty good feedback receiver somewhere in the system.

I'm seeing a fair amount of publications on dynamically biasing the PA so that the bias is adjusted as the output power is controlled, which may get you most of the current savings without as much work, but that' s probably not applicable to Base stations.

Dave