Unless you design EDGE or 3G/UMTS cell phone handsets, you may not have heard of polar modulation. As you know, there are only a few basic types of modulation, but they have been used in a variety of combinations over the years. This one is unique and may actually show up a great deal in the newer 3G handsets.
EDGE (enhanced data rate for global evolution), of course, is the higher-speed packet data addition to GSM/GPRS cell phones. Most GSM carriers, like Cingular and T-Mobile, have already converted most of their base stations to handle EDGE because it is primarily a software adjustment. Some EDGE handsets are now available, and more are on the way.
EDGE uses the same time-division-multiplexed slots as GRPS but adds 3π/8-8PSK modulation to get the higher transmission speed up to about 128 kbits/s. This is quite a bit faster than the 48 kbits/s or so possible with most GPRS arrangements. It certainly makes e-mail and Internet access on a cell phone far more useable.
The big problem is that EDGE does not use the constant envelope GMSK (Gaussian filtered minimum shift keying) modulation that is used in GSM/GPRS. The 3π/8-8PSK modulation has both a phase and amplitude component that requires a linear power amplifier (PA). And that, as you know, means higher power dissipation, lower efficiency, more heat, and shorter battery life. With just the GSM/GPRS GMSK circuitry, the power amplifier can be a more efficient class C, D, E, or F type. But these amplifiers clip off all the amplitude variations and therefore will not handle EDGE. Up to now, most vendors of EDGE phones have gone with a linear amplifier and accepted the lower efficiency as inherent in the design. Yet several other companies have been working to solve this problem. The result is polar modulation.
In polar modulation, the normal I and Q rectangular baseband outputs to the transmitter are first converted to a polar format with an amplitude phasor and a phase angle component. The phase signal is usually fed directly to a DAC whose output is the input phase-locked loop (PLL) that’s used as a phase/frequency modulator. The PLL-VCO output signal is then fed directly to a PA that operates near the saturation/clipping level. It can use any one of the more efficient non-linear designs.
A DAC converts the amplitude phasor into a signal that can now amplitude modulate the PA. The method is similar to that used in older high-level AM plate (vacuum tube) modulated amplifiers or collector modulators. This signal is typically used to control a device in series with the PA’s collector or drain supply to actually do the amplitude variation. In older AM transmitters, the final PA operated efficiently as a class C amplifier while the modulation was applied with a transformer in the plate supply line. It worked great. In contrast, a low-level AM scheme created the AM on a small-signal basis, then required a final, linear PA. Anyway, the result with this polar modulation scheme is an efficient PA but with the necessary amplitude components for 3π/8-shifted 8PSK. There are other schemes to produce polar modulation, including some that are low-level designs.
One of the first to offer a polar modulation EDGE chipset was RF Micro Devices (www.rfmd.com). The RFMD Polaris 2 Total Radio chipset is made up of the RF2723 quad band RF receiver, the RF6007 fractional-N PLL, modulator, digital signal processor and PA ramp controller, and their RF3186 PA.
Another recent polar modulation vendor is Tropian Inc. (www.tropian.com). Their Timestar TR15431N radio transceiver IC implements their own brand of polar modulation using any one of many standard GSM/GPRS PAs with power control. And it works with wideband CDMA (WCDMA as in 3G).
A third vendor worth looking at is Sequoia Communications (www.sequoiacommunications.com). Their new SEQ-5400 single-chip transceiver uses polar modulation to implement EDGE, but it also works with WCDMA as well as HSDPA (high speed downlink packet access).
But not all vendors are going the polar modulation route. As Patrick Morgan of Silicon Laboratories (www.silabs.com) told me recently, they voted not to use polar modulation on their new Aero IIe EDGE chipset. He said that polar modulation is only necessary when a fast EDGE data transmission is needed. Since most phone usage is voice, the operating mode is GSM and the standard saturated-type PA gives the best efficiency. He says that if you design the phone correctly, you can actually use a saturated PA for EDGE by backing off on the drive or changing the bias to just put the PA into the linear region.
You will no doubt see more of polar modulation. This is especially true for future WCDMA designs because efficiency is needed due to always-on, duplex nature of the technology.