Engineers sometimes joke that a cable is a source of potential trouble connecting two other sources of potential trouble. Now, with low-cost, easy-to-use RF systems on chips, you can eliminate that intervening cable in many limited-distance applications. These applications include situations in which it would just be nice to avoid a cable, such as between a handheld control box and the unit it is managing, and those in which a cable would defeat the application, such as with a remote-keyless-entry (RKE) lock for a car or house. In many ways, though, cheap and easy RF links are enabling new applications that are sometimes possible, but usually impractical, with wired links.

Several forces drive wireless links. First, there's end-user convenience. Mass-market applications, such as ubiquitous infrared TV remote control, have created expectations that are hard to dismiss once you have been spoiled by the cable-free freedom that such applications provide. At the same time, IC vendors have pushed their processes, products, and clocks into the multimegahertz and even the gigahertz range, so the frequency bands you need for RF are a natural part of their components' operating range. These ICs can often function without special external circuit elements for frequency upconversion and downconversion. You need no longer assemble your RF link from a handful of discrete—and sometimes pesky—active and passive components. Even better, these IC-based links consume little power and operate in bands for which no individual licensing and few regulatory requirements exist. As long as end users stay within the designated frequencies and power levels, they have no paperwork to deal with. But don't think that effectively using these latest generation RF ICs for dedicated, point-to-point links is trivial. You need to study numerous issues if you make a commitment, because the various RF ICs differ in many related parameters, such as frequency, completeness, and overall performance. Reach out and touch the receiver Your first decision involves the range you need to reliably reach. At the milliwatt power levels of these ICs, which include the final-stage power amplifier, you can reach at least several meters and as much as 100m or more under favorable conditions. If you need longer range, you could add an external booster RF stage. However, adding this stage introduces problems, including regulatory issues, power-consumption concerns, additional components and cost, and signal overload at the receiver when you move closer. (And you could use a receiver with a wide-ranging automatic gain, which adds more complexity.) Another alternative is to use a directional antenna. Again, this simple approach is often impractical for several reasons. It increases effective radiated power (ERP) in the target direction, which you usually must keep below some regulatory limit. You can, though, use a directional antenna at the receiver without regulatory concerns, but then you restrict the user's freedom of movement . You also need to look at the trade-off between a simple, unidirectional, simplex link and a bidirectional, half-duplex link. Many applications, such as RF identification (RFID) and RKE, need only a simplex link but benefit from a two-way signal path. A simplex link gives you no confirmation that the system received the signal. If you're locking or unlocking your car, the car horn gives you audible feedback, and you're all set. But if you are setting an alarm system as you leave a building, you may want either confirmation that the system properly received and implemented your arming signal or feedback about why it did, such as when an alarm zone isn't set properly. The one-way link also restricts some security algorithms (see sidebar "Proper security lets you rest better"). Although a half-duplex system requires more components than a simplex one, it luckily involves little extra design complexity on your part in these low-power links. The only common system element is the antenna, and, because the link is not full-duplex, the receiver's front end need not to be active while the transmitter is on. Therefore, your design need not filter and separate the receiving and transmitting bands, and the transmitter power-on state doesn't saturate the receiver front end. Most of these low-power, short-distance RF links operate with low or sporadic duty cycles and short messages, so you calculate power consumption at the transmitter differently from at the receiver, which must always be at least in a standby, listening, or polling mode. You use a typical RKE unit several times a day but only for a few milliseconds each time, so battery life of one to two years is practical. Although each application is different, most target an operating life of at least two to three years from a battery. Note that if your application involves periodic messages, such as an entry point in an alarmed system that needs to tickle its central controller to confirm that all is still OK, then transmitter power consumption and operating voltage are more critical. You have to weigh that aspect of your design and perhaps go to larger capacity cells or higher operating voltages. Also, be sure to consider any microcontroller's power needs in your power budget. The inherent nature of wireless links is that you use them in casual, nonfixed-location applications; thus they are subject to temperature extremes, rough handling, and abuse. In these cases, the highly integrated approach gives you a distinct advantage, because every component you eliminate not only contributes to greater mean time between failures, but also keeps the link operating range at its nominal value. By their nature, RF circuits are subject to detuning, drift, and other sensitivities, and it's likely that the range you can guarantee over the long term will decrease dramatically with the amount of inevitable user abuse and environmental stress. You need to life-test your product under realistic operating conditions so you don't have irate users calling you in 12 to 18 months to complain that their garage-door-opener range is drastically shrinking or that the link operates erratically. The design and implementation of your receiver are more difficult than those of the transmitter for several reasons. The transmitter has a deterministic task: taking a known signal, encoding and modulating it onto a known carrier only on demand by the user, and providing enough RF power. In sharp contrast, the receiver has the unenviable signal-processing task of always being alert or polling while tuning for a distorted, noise-laden signal in the spectrum. It needs to extricate this signal despite unfavorable SNR, interference, and component-parameter drift. For these reasons and others, many vendors of RF ICs are concentrating on the receiver end of the link, because that's where they add the most value, implement circuit techniques that are most beneficial to you, and reduce most headaches for design engineers. This difference between transmitter and receiver, though, shouldn't lull you into complacency about the transmitter implementation. Unless you are confident of your ability to develop and qualify your own design, you still need to determine whether the receiver vendor offers a matching transmitter or a reference design that you can use if you want to minimize your risk. You should also consider the frequency that fits your application and market. In general, lower frequency designs are easier to lay out and debug, but these designs have several drawbacks. They require larger passive components (some passives are inevitable), and they operate in more crowded bands. The trend for RF links in the last few years has been to migrate from the 200- to 300-MHz spectrum to the 300- to 400-MHz region and even as high as 900 MHz. Units are even available for the 2.4-GHz ISM (industrial, scientific, and medical) band. Note that the virtue of 2.4-GHz operation is regulatory as much as technical. Its operating frequencies and constraints are nearly uniform throughout the world, so you can use one design—or minor variations of that design—everywhere. Diverse architectures to the rescue In the world of receivers, the venerable superheterodyne architecture that EH Armstrong developed in the 1930s is the predominant design for most applications. The superhet converts any received carrier to a fixed IF, which the receiver then demodulates and processes as needed. This excellent architecture, which engineers have refined over the many years of its existence, simultaneously solves several conflicting problems. (Only in recent years has a direct-conversion, zero-IF architecture become practical for some applications.) Unfortunately, the full superhet is too much of a good thing for low-cost, low-power receivers, which target a single-channel, single-modulation signal. RF-IC vendors thus strip whatever they don't need from the receiver, based on the fairly limited flexibility that the receiver needs in this limited-function application. In some cases, they have even abandoned the superhet for simpler designs, again with limited functions and optimized for one task only. Ironically, one of these designs is the super-regenerative receiver, which Armstrong also developed but before he did the superhet. (He soon found the operating limitations of the super-regen, despite its simplicity, too severe for general-purpose use.) As with most components and systems, you must be aware of some element of specmanship. Especially true of receiver-sensitivity specifications, many legitimate and difficult-to-compare ways exist to specify the maximum data rate, such as determining with which bit-error-rate (BER) value, input bandwidth, modulation factors, and duty cycle you are taking the measurements. You need to question vendors about the test setup they use if you are running near the maximum potential of the link.

A good example of what you can get today is the Micrel MICRF receiver family, which comprises 418- to 433-MHz- and 900-MHz-band devices. The lower band, 16-pin MICRF002 and eight-pin MICRF022 and higher band, but otherwise similar, 16-pin MICRF003 and eight-pin MICRF033 use an architecture that eliminates the need for manual tuning of each unit ( 1). These superhet receivers, which target on/off keying, or amplitude shift keying (ASK), require few external components: a 47-nF capacitor, a 4.7-µF capacitor, and an inexpensive, 6- to 7-MHz ceramic resonator. These ICs need no filters or inductors.