TI pushes deeper into ZigBee territory with integrated radio SoCs

A lot of ZigBee applications these days are headed for very low price-points: wireless remote controls, thermostats, electric and gas meter wireless interfaces, and even light fixtures. The implication for SoC designers is that this is a very low cost application, but at the same time a complex one, since a ZigBee radio is not a trivial circuit. The situation calls for an integrated SoC radio solution. And that plays right into the hands of Texas Instruments, which has been sharpening its integrated RF design skills on cell phone handset chips for several generations now. An SoC that can integrate a ZigBee RF front-end, ZigBee baseband, CPU cluster, and the peripherals necessary to cover a range of applications would be just the ticket.

And that’s what TI has been aiming for with their growing family of ZigBee SoCs, the latest of which—the CC2530—was introduced today. TI is attempting to cover two main application areas with the chip: the growing range of RF4CE-compliant consumer remote control devices, and the potentially explosive—thanks to one of those market-distorting government investment programs—market for Smart Energy devices covered by the ZigBee PRO Smart Energy Profile.

To that end, the 2530 is a catalog of low-end SoC components. There’s a second-generation 2.4 GHz radio with excellent characteristics, including +4.5 dB transmitter power, a 101 dB link budget, and 49 dB adjacent-channel rejection. There is an 8051 MCU core and a thorough complement of peripheral blocks, an audio ADC, 8 KBytes of RAM and up to 256 KBytes of Flash. There will be a version of the chip, the 2531, with a full-speed USB Device interface as well. To keep the computing tasks required of a ZigBee device within the range of this relatively anemic processor, there is an AES Crypto engine. And to keep system costs down, there are a power management unit, local oscillators, local voltage regulator, and a Flash control block—presumably to control write timing—on the die as well. All in all, it’s a pretty complete system.

The problem is, the application space is growing faster than TI can create SoCs to span it. For instance, in the ZigBee PRO space, systems designers are thinking about a lot more than just connecting a thermostat to a furnace control board. For an example of what is coming in home and building management, you can actually link into TI’s demonstration San Diego Large Network—a 100-node ZigBee lashup that spans part of a building. for such applications, where the variables of walls, distances, and moving bodies can mean substantial path attenuation for some nodes, +4.5 dB– good for about 400 meters line-of-sight–is not going to be the whole answer. Right now, TI’s solution is a separate power RF front end chip to use beside the 2530. One suspects, however, that the design team is also looking at how to increase the power range of an integrated radio without increasing the power consumption at lower transmit power, or running up the cost of the chip.

One more example. RF4CE has designers so excited that they are discussing not just universal wireless remotes, but two-way, moderate-bandwidth communications between said remote and the devices it is controlling. That would make it possible, for example, for one of those numbingly complex home-theater receivers to provide a GUI on the screen of your remote control, rather than expecting you to remember the sequence of 11 keystrokes that gives you control over the volume of the sub-woofer channel.

The two-way communications capability is well within the reach of the current radio. But it’s not in the cards for the little 8051. There will need to be a more substantial processor—one thinks automatically of an ARM Cortex—much more memory, and probably a graphics controller of some sort. And these additions will have to fit within the change-the-batteries-when-you-change-your-clocks power budget of a remote control.

The challenges of expanding applications—and we haven’t even speculated about what a big infusion of borrowed federal dollars is going to do in the Smart Energy space—is going to put a lot of pressure on TI and other ZigBee SoC vendors either to expand their product lines rapidly—which can be ruinously expensive—or to find ways to offer a single, highly-featured die at a variety of price points. And that is going to force continued, aggressive process migration. This could be a market that ends up helping TSMC quite a bit.