设计我最赞的博客

结合了微控制器、传感器和无线功能，ZSTAR3评估套件能够成为您下一个项目伟大的起点。除了一个无线的加速度传感器外，它还包括一个USB即插即用无线集线器以及易于操作的计算机应用软件。

A Healthy Mix of MCUs, Sensors, and Wireless Technology

With a combination of MCUs, sensors, and wireless capabilities, the ZSTAR3 evaluation kit could be a great starting point for your next project. In addition to a wireless accelerometer, it includes a USB plug-in wireless hub and handy PCbased utility software.

What are three of the tastiest ingredients in an embedded designer’s pantry?

The first is that modern-age miracle-worker, the MCU, packing the intelligence needed to give formerly ho-hum applications a silicon enhancement. Thanks to the fact Moore’s law continues to deliver on the promise of more for less, ever better MCUs are finding their way into practically any gadget with moving electrons.

On another shelf we find ever-smarter sensors that give those fancy MCUs some realworld data to chew on. After all, it’s the sensors that single-handedly enable, or not, potential applications. It’s game over if you can’t sense, because then you can’t control (at least very well).

Thanks to these high-integration, easy-to-use MCUs and sensors, wiring up a new design is simple. It’s even simpler if you skip the wiring part altogether by taking advantage of lowcost radio chips that are proliferating like bunnies.

Combine all three ingredients and you can cook up an endless variety of innovative embedded applications. This month, let’s take a look at an excellent example of the trend courtesy of Freescale Semiconductor. Read on and I think you’ll agree that mixing MCUs, sensors, and wireless together, seasoned with a healthy dash of designer creativity, is the recipe for success.

XYZ
“Just ask the Axis, He knows everything…”

—The Jimi Hendrix Experience, “Bold as Love,” Axis: Bold as Love, MCA Records, 1968.

There is no better example of “It’s the sensors, stupid” than the MEMS accelerometer. From their humble high-g airbag roots, low-g accelerometers have emerged to single-handedly enable vast new applications (i.e., Nintendo Wii), as well as, pardon the pun, shaken up existing ones (i.e., Apple iPod Nano “shake to shuffle”). Indeed, low-g accelerometers have come full circle, now back on the road in active suspension systems reunited with their high-g airbag ancestors.

Figure 1—Add some e-motion to your designs with a Freescale Semiconductor MMA7456L smart, three-axis, low-g MEMS accelerometer.

I got my feet wet with a first-generation low-g MEMS accelerometer way back in 1995 (“A Saab Story: A Tale of Speed and Acceleration,” Circuit Cellar 57). Since then, it’s been the usual silicon story with the newest-generation parts, such as the Freescale MMA7456L (see Figure 1), offering better specs, more features, and, above all, a much lower price (just $2.87 in 1,000-unit quantities).

In the beginning, accelerometers measured a single axis (X), followed later with dual-axis (XY) versions. It is not surprising that the latest and greatest like the MMA7456L have upped the ante with full three-axis (XYZ) capability. Remember that even applications that can get by with a single- or dual-axis part can take advantage of an extra axis to enhance features, reliability, and ease of use.

For instance, in “A Saab Story,” I described how I used the accelerometer as the basis for a time/speed/distance display (i.e., using time and acceleration to calculate speed and distance). But the gadget would work only on a reasonably level road, lest a change in the Z-axis orientation be falsely interpreted as acceleration in the X-axis. Traveling at a fixed speed, going up hill would appear faster, and downhill slower, than on a level road. With a Z-axis, I could have dynamically compensated the speed calculations based on the pitch (i.e., heading up or down hill) information.

Figure 2—Measuring tilt with a single-axis accelerometer is complicated by the fact that sensitivity varies with the tilt angle, and is poor at the extreme. Using a second axis, and a bit of trigonometry, sensitivity is boosted and linearized across the tilt angle range.^[1]

Another more-is-better situation is using an accelerometer as an inclinometer to measure tilt, as in the “electronic levels” you’ll find in the tool department these days. Sure, you can measure tilt with a single-axis accelerometer, but the problem is that the g-output is quite nonlinear over the range of 0° to 90° (see Figure 2). Getting a feel for this phenomenon is as easy as dropping and giving me 10 pushups. (Consult your doctor first.) Now stand up and lean against a wall at a slight angle and give me 10 more. A lot easier, huh? By adding an extra axis, you can use whichever sensor is in its sweet spot for better resolution across the full range and especially near the extremes (i.e., 0° and 90°).

The accelerometer I used back in 1995 had an analog output, albeit thankfully signal-conditioned to provide a decently high-level signal. By contrast, the MMA7456L is fully digital with on-board 10-bit ADCs and a SPI. In my book, digital is generally preferred because it minimizes susceptibility to noise and enables “smart” features. However, there may be scenarios where an analog version still makes sense. An obvious one is when you’re connecting to an MCU that has unused ADC inputs. Some applications may require higher resolution and are willing to trade-off bandwidth and signal processing MIPS to get it. Also, an analog part can work stand-alone in hardwired applications that don’t need an MCU. Letting you have it your way, Freescale also offers the MMA7361L, a part that’s similar to the MMA7456L except it has analog outputs with a healthy 800-mV/g swing.

Figure 3—Configuring the MMA7456L single- and double-“shake” (i.e., pulse) detection feature is simply a matter of setting up registers that define the threshold and timing.

But it’s the digital smarts (offset correction and programmable threshold, motion, freefall, and single-/dualpulse detection) that set the MMA7456L apart. The pulse detection feature is particularly useful for user interface applications as the shaken-not-stirred equivalent of single- and double-click on a mouse. There is a set of registers to adjust the pulse threshold and timing like you would with the mouse control panel on your PC (see Figure 3). In applications that need it, pulse detection is a big time (and power) saver for the MCU, which would otherwise be burdened having to constantly be on the lookout for properly formed pulses.

Low-power is all the rage and the MMA7456L obliges by sipping a mere 0.5 mA during normal operation. The chip has separate digital and analog power rails, but you can run both from a single supply anywhere between 2.4 and 3.6 V. Thanks to the low power and wide voltage range, powering the MMA7456L from an MCU output pin is a viable option. There’s also a standby mode that slashes power consumption to just a few microamps. Do keep in mind there’s a bit of latency entering and exiting standby (up to 20 ms each way).

ON THE AIR
With a sensor in hand, now we need an MCU and a radio to pull it all together. Freescale makes that especially easy with their MC1321x, which combines an ’S08 flash memory MCU with their 802.15.4-compatible 2.4-GHz radio on a single chip (see Figure 4).

Figure 4—The two-die “System-in-Package” MC13213 looks like a typical ’S08 MCU, just one that happens to have a complete 2.4-GHz radio built-in.

The MC1321x line-up comprises three parts that differ only in the MCU’s flash memory/RAM capacity with the MC13211 offering 16/1 KB, the MC13212 32/2 KB, and the MC13213 at 60/4 KB. Besides addressing applications of different scope and complexity, each part is semi-tailored to fit three different wireless protocols.

At the low end, Freescale offers their own “Simple MAC” (SMAC) solution. Supporting basic point-topoint and star networks, SMAC easily fits in an MC13211. Stepping up a notch is a full IEEE 802.15.4 MAC that supports all that standard’s features, such as fancy topologies (e.g., mesh and tree) and bulletproof security (128-bit AES encryption). Finally, at the top of the stack so to speak, is Freescale’s “BeeStack” fully ZigBee-2006-compliant platform. Note that the SMAC source code is available for adding your own proprietary protocol tweaks while 802.15.4 and Zig-Bee are delivered only as object-code.

The MCU and radio work well together. For example, because the radio needs a precise clock, it uses a 16-MHz crystal and a trimmable oscillator. In turn, the radio’s clock can be output to a pin for use as the MCU clock, so only a single crystal is required.

Besides flash memory and RAM, the protocol software consumes some of the MCU I/O resources (e.g., using some of the timer channels for scheduling radio activity). That leaves plenty of MCU I/O resources (e.g., SPI, I²C, UART, 8 × 10-bit ADC, GPIO, and more) free for your application.

Just add a 16-MHz crystal, a single voltage (thanks to on-board regulators) 2- to 3.4-V power supply, an antenna, and a few discretes, and you’ve got a complete wireless gadget (aka “Wadget”). What could be easier?

WELL ROUNDED
What could be easier indeed? How about just heading over to the Freescale web site and ordering the ZSTAR3 evaluation kit (see Photo 1). In addition to the wireless accelerometer (versions are available with the digital MMA7456L or the analog MMA7361L), the kit includes a USB plug-in wireless hub and some cool PC-based utility software.

Photo 1—The ZSTAR3 evaluation kit provides a quick and easy way to taste-test Freescale’s wireless sensor recipe, and the price is right at just $99. The network supports up to 16 sensors. Additional sensor boards (using the digital MMA7456L or analog MMA7361L) are available for $59.

The ZSTAR3 wireless scheme supports up to 16 sensors simultaneously, each delivering data at 30 Hz. The protocol is based on the aforementioned SMAC upgraded with some new timing-related features for efficient scheduling (see Figure 5).

Figure 5—One advantage of centralized (i.e., hub) network control is the ability to schedule activity efficiently. Each ZSTAR3 node schedules radio activity (i.e., transmit/receive “windows”), which avoids interference and steadies data throughput. Scheduling also minimizes superfluous radio activity and allows the node to sleep most of the time, extending battery life.

I experimented with the original version of the ZSTAR software back in 2006 (“Three-Axis Foursome,” Circuit Cellar 191). It was fine for getting up and running and playing around a bit, but certainly nothing to write home about. But it’s apparent that the Freescale folks have been hacking their hearts out in the interim because the ZSTAR3 software has a lot of new features to exercise the hardware and show off applications ideas.

Not that the version “0.2.3.0” of the software didn’t come with some quirks and head scratching. I had to poke at it a bit to get the sensor linked in and bumped over some dubious error-message potholes from time to time, but nothing that was a showstopper.

Photo 2—The ZSTAR3 evaluation kit’s PC GUI software provides a complete set of MMA7456L hardware utilities and demo applications.

The splash screen aggregates the 16-sensor display with tabs at the top to exercise various features (see Photo 2). The “RF Overview and Control” tab enables you to manually or automatically select the RF channel frequency and shows the energy in each band. A dynamic RSSI display is useful for evaluating signal strength and range, which are well-known challenges for 2.4-GHz radios with teensy PCB antennas (see Photo 3).

Photo 3—Range for low-cost 2.4-GHz radios using passive PCB trace antennas can be limited. Even the device’s packaging (e.g., batteries) can block the signal. The ZSTAR3 RSSI utility makes it easy to monitor signal strength in real time and experiment with installation-specific alternatives.

“General Sensor Tasks” start with calibration (i.e., zeroing out XYZ offsets while the sensor is motionless) to compensate for variations, such as mechanical misalignment and temperature drift. Naturally, there’s a “Scope” display that traces acceleration in real time, and also offers the ability to log data to an Excel file.

Photo 4—A little signal processing goes a long way to filter noise, as demonstrated by this “electronic level” demo.

The “Tilt” tab demonstrates the technology at work in demo applications, such as PDA scrolling or switching a display automatically between portrait and landscape modes. There’s a “Filtered Tilt” demo that highlights the ability of some simple signal processing to boost accuracy (see Photo 4). Using raw accelerometer data there’s a noticeable jitter in the tilt reading on the order of 2° to 3°. But all it takes is a little bit of filtering (i.e., moving average) to cut the jitter to a fraction of a degree.

Likewise, the “Motion” tab demonstrates how you can use the threshold feature to give applications longer battery life (i.e., using motion as an on-off switch) and antitheft security. For instance, office equipment could be designed to sound an alarm if it’s moved in a suspicious way or at an unusual time.

Photo 5—Today, the “I’ve fallen, and I can’t get up” lady has to tell someone she’s in trouble verbally. Err, kind of hard to do if you’re unconscious. Tomorrow, an electronic fall detector will make the 911 call for her.

The “Freefall” tab exercises a feature that could prove useful in a variety of reliability, safety, and health applications. A neat feature is the software takes a decent stab at determining the distance of the fall based on the elapsed time (see Photo 5). Before you get carried away, keep in mind the 5,000-g maximum spec, which is less than it sounds (the “Drop Test” spec is 1.8 m beyond which permanent damage may occur). The demo works fine if you drop the gadget onto something soft, all the better to be safe than sorry.

Similarly, the “Shock” tab has demos of interest in blue-collar applications such as shipping and handling. For example, how many packages have you received with a “This Side Up” label? Wouldn’t it be interesting to actually know if (and when, and where) it got tipped upside down along the way? Finally, the “Digital” tab pops the hood on the chip so you can probe and configure all of the accelerometers, registers, and options.

ONE FROM COLUMN A …
So there you have it, the recipe for your next killer app. Like the menu in a Chinese restaurant, all you have to do is choose one item from column A, one from column B, and one from column C.

Column A is an MCU. There are so many delicious choices here you can’t go wrong, everything from 50-cent, 8-bit appetizers to MIPS-laden 32-bit entrees.

Column B is the radio. Here you have a choice of standard favorites like ZigBee, Wi-Fi, and IEEE 802.15.4, along with a variety of boutique alternatives. Options cover the spectrum from low-speed point-to-point “wire-replacement” to emerging “smart dust” mesh networks with hundreds or thousands of nodes. You can spice up your wireless offering with cellular broadband, RFID, or GPS for a new and improved taste.

Finish off with dessert, always my favorite part of the meal, by choosing a sensor from Column C. Or, for that matter, why not indulge with two or three of them. Thanks to the march of silicon, your designs can be both less filling and taste great.


Tom Cantrell has been working on chip, board, and systems design and marketing for several years. You may reach him by e-mail at tom.cantrell@circuitcellar.com.

REFERENCE
[1] K. Tuck, “AN3461: Tilt Sensing Using Linear Accelerometers,” Freescale Semiconductor, Inc., 2007.

SOURCE
MC13213 MCU and IEEE 802.15.4 Radio, MMA7361L accelerometer, MMA7456L accelerometer, and ZSTAR3 accelerometer evaluation kit Freescale Semiconductor, Inc. | www.freescale.com