In Part 1 of this entry (Is Cellular RF PA Integration Finally Here? Part 1), I discussed how the PA section of a cellular handset has evolved to need higher performance (higher linearity) to support newer standards, and that the level of integration was very low compared to the rest of the phone.
Qualcomm announced a new "RF Platform" at Mobile World Congress in Barcelona. Details are sketchy unless you are under NDA. In fact, other than the press release, there’s nothing to be found on the Qualcomm website. Zip. Nada. But it appears that Qualcomm is taking a system-level approach to the PA problem. The "RF360" solution starts with an integrated PA and switch… made in CMOS. That’s interesting, but not earth-shattering.
There is also a Dynamic Antenna Matching Tuner -- that’s pretty cool. It seems to cover all bands from 700MHz to 2.7GHz and tunes the antenna to compensate for impedance changes due to holding the phone in your hand, near your head, etc. I want to get more info on it to see if they have the tunable reactive components on-chip, or if that’s left to the user. Turns out that Qualcomm was recently issued some patents (8,363,380 and 8,362,853) for "MEMS varactors" and some earlier ones on tunable inductors. I'm unclear on whether these are being used in this platform, but a semiconductor-based tunable capacitor or inductor sure would be handy in a tunable-filter or impedance-matching application.
Then there is the third chip, which is the most interesting to me: an Envelope Power Tracker that is "modem-assisted." If this is as cool as it could be, the antenna tuner and this chip would mark the first time that a chipset manufacturer has gotten the baseband architects to consider the entire signal chain all the way to the antenna, and provide hooks to make the PA work better.
Envelope power tracking is not a new technique. In its simplest form, it dates back to the early days of AM broadcasting. The unmodulated RF carrier went into the "input" port of the final amplifier (a vacuum-tube’s grid), while the modulation (audio) was amplified and modulated the plate supply voltage of the tube by way of a big transformer (secondary connected in series with B+). The modulation, of course, had to be amplified to the same amount of power as the desired output RF carrier, but getting a lot of clean audio power is easier than getting a lot of clean RF gain. Modern implementations are a bit more complicated, since the low-level RF signal is usually multi-carrier or has phase modulation, and the extra oomph is only provided when it is needed by using the envelope to drive a higher supply voltage to the final transistor stage.
Looking over Qualcomm’s recent patents, one of them shows the determination of when additional power is needed in the handset transmitter to be derived way up in the system -- using the received signal strength from the base station as a sign that more transmit power is needed. This outer power control loop is a key part of most modern cellular systems, allowing the handset to transmit only as much power as is needed for the base station to hear and decode it while not swamping the other handset transmitters in the area.
Qualcomm has also received patents on new, non-system-level implementations of classical pre-distortion, using a sample of the PA output converted to baseband, compared to the desired signal, and creating a lookup table for subsequent transmission. If you can get this right without the cure being worse than the disease, it can reduce the power of a smartphone’s PA by 30 percent or more.
If Qualcomm (or anyone else) can get all of this together, we may in the next few years see the RF section shrink with the same level of integration as the digital section, with a single very smart component including bandswitching, antenna impedance matching, and linearization while reducing power, cost, and size.