CMOS RF Power Amplifiers for Mobile Communications
It somehow seems that the integration of RF power amplifiers (PAs) in CMOS is not very obvious, especially not when compared to the successful integration of RF transceivers in CMOS. The first research efforts to integrated transceivers in CMOS were, indeed, focused on the receiver part. Given the demanding specifications on sensitivity, phase noise, and blocker levels, this is, of course, not surprising. Designing circuits that meet these stringent system-level specifications is definitely not obvious. But from the same system-level point of view, the transmiter part looks pretty simple. A direct up-conversion mixer followed by a power amplifier, and that’s it, right? That PA can be modeled by AM-AM and AM-PM distortion curves or by a Taylor series expansion, and one has to be careful about carrier leakage and noise transmitted in the receiver band, but what other problems might there be from a system point of view?
As always with circuits that make the interface to the outside world, the true trade-off picture is less clear. Whereas, for example, a down-conversion mixer is nicely embedded between an LNA and an ADC, the PA and the LNA see the outside world, so they also meet another environment and other technologies (SAW filters, duplexers, baluns, and many other discrete components mounted on the PCB). And where the LNA meets the limits ...