Monthly Archives: August 2013

Examining the Enduring Appeal of the 8051

By: Jonah McLeod, Kilopass Technology Inc.

The widespread deployment of MEMS in smart phones has ancillary benefits that I thought worth looking into. One of these was breathing new life into 8-bit processors.  I began my search by calling Hal Barbour. He’s the CEO of CAST, Inc. a company that has made a business offering a range of popular and standards-based IP cores including the venerable 8051.  His COO, Nikos Zervas, joined the call as well.

The question I had for the two of them was the roll that the 8051 had in the development of the smart sensors going into smart phones, automobiles and Internet of Things. I picked the 8051 because of my past association with the processor while at my old company ARC International. One of the growing markets for ARC was in 8051 replacements and one of the company’s successes was displacing the 8051 in USB drive controllers.

The old saying that as one door closes another one opens appeared to be working for the 8051.  As the CPU was designed out of the USB applications it began being integrated with smart sensors in a multichip package.  The automotive tire pressure sensor was an example. Mandated by law to be included on new vehicles, all car models produced after September 2007 came with the tire pressure sensor. Many of the sensors came with the 8051, for example the Texas Instruments TPIC82000 Series.  The electric utility power meter was another application adopting the 8051.

Hal pointed out that many of the sensor designs that began adopting the 8051 were being fabricated in the older processes, 130nm and larger.  At these process nodes, gate count matters and the relative small size of the 8051 is a desired feature.

The continuing attraction to this simple 8-bit processor is borne out by market research data.  The last market research on 8-bit processors I was able to find was published 2008 and it showed the 8051 with a declining 19 percent of the embedded processor market, still the largest share of all the 8-bit processors featured in the report.  A recent IC Insights research bulletin published August 13 showed embedded processor in general growing; accounting for 11 percent of MPU sales in 2013 (versus 9 percent previously).

Hal and Nikos had a laundry list of reasons why the 8051 has remained popular since its formal introduction in 1980 beginning with cost. Unlike 32-bit processors that come with a license fee and royalty stream, the 8051 can be had for one upfront charge.  And like the popular 32-bit architectures, the 8051 has a wide and deep ecosystem of software, programmer familiarity, and design expertise that make it and ideal solution for a wide range of embedded computing tasks.

Next, the two cited the fact that most of the engineers building sensors are analog and mixed signal designers.  Their designs need more than a state machine to control analog mixed signal circuits and the 8051 fits the bill. The 8-bit microcontroller core also comes with the interfaces—Philips’ I2C, Motorola’s SPI, Bosch’s CAN buses—to connect analog-to-digital and digital-to-analog converters as well as other peripherals. To support remote sensors, the 8051 comes with circuits and stacks for wireless communications protocols:  Zigbee, WiFi, BlueTooth, as well as Ethernet.

I asked Hal when the 8051 would reach the end of the road. He said he had thought the 8-bit workhorse was reaching its end six or seven years ago.  In anticipation, he had begun adding 32-bit CPU IP to CAST’s product offering.  But as history has demonstrated, the 8051 kept on going. Today though there are signs it may be loosing out to 32-bit alternatives.

Nikos pointed out that the applications that are migrating to more functionality are leaving the 8051 behind. If the data being processed comes in 8- or 16-bit resolution, simple sensors detecting on or off conditions or non-critical temperature and pressure readings, for example, then the area and power is competitive with a 32-bit solution, which might be overkill. If the data requires 32-bit resolution, such as being demanded in smart phone applications with their 6-, 9-, and now 10-degrees of freedom inertial measurement units, then 32-bit processors win hands down.

However, just as new applications came in to save the venerable 8-bit processor before, Internet of Things may yet hold some salvation for the 8051 going forward. It may find service as a controller in home monitoring systems providing WiFi, BlueTooth, and Zigbee protocol processing for communications to a central controller. Nikos says that these are the applications that the 8051 perform well. The growth of disposable sensors in medical and home health care applications could be another area of growth for the 8-bit engine.


How the smart phone is driving the Internet-of-things

By: Jonah McLeod, Kilopass Technology Inc.

The Internet of things is coming thanks to the technology being developed for the smart phone. MEMS accelerometers in automobiles that ship in the 10s of millions of units worldwide each year are in smart phones which Credit Suisse predicts will be over a billion units in 2014. Yole Développement says this will be a huge market driver for MEMS devices. Worldwide, smart phone and tables created a MEMS market worth $2.2 billion in 2012 and is set to be worth about $2.7 billion in 2013, the Lyon, France-based market research firm predicts. Furthermore, the average selling price of the MEMS devices will decline over the period thus opening up applications in the realm of Internet of things.

But smart phones not only come with three-axis accelerometers, but gyroscope and magnetometer as well as compass and barometer (pressure sensor).  Dan Brown, CEO of Sensor Platforms, Inc. a San Jose, Calif.-based supplier of software that makes sensor fusion, user context awareness, and pedestrian dead reckoning possible in smart phones and tablets, says a three-axis accelerometer is all that is needed to determine the context of a phone:  in a pocket, on a table, in a moving vehicle, etc.  The next major step is indoor navigation, where GPS tracking isn’t available and where GPS would not be not effective determining location in a multi-story building.

That’s where the barometer and magnetometer come in. The former provides the signature for elevation, what floor in the building. The magnetometer provides the signature for magnetic anomalies that can distort the direction of true north, essential for accurately determining the direction of the user moving about indoors. Today, getting rid of this distortion requires the gyroscope, although many solutions request the user to move the phone in a series of figure eights to recalibrate the compass back to north, which even if done does not improve the situation.

The software Brown’s company supplies processes the electrical stimulus from the array of sensors in a smart device to determine the location and orientation of the user’s phone at any given point in time.  More important, however, is the software achieves this result while reducing power use to the absolute minimum to preserve a phone’s battery life. Some sensors are more frugal with power than others. For example, the accelerometer uses the least power while the gyroscope uses much more. Thus, by minimizing the operating time of the power-hungry sensors, the software can achieve its purpose while minimizing battery consumption.

Brown believes that the state-of-the-art in sensors, smart phone computing power, software, and applications are converging to make augmented reality possible. By way of example, in the next 12 to 18 months it may be possible for smart devices with the appropriate apps loaded to determine the location of a user in a building and what he or she is viewing. Furthermore, the app will be able to describe to the user what he or she is viewing:  for example, a piece of art in a museum, the artist, the subject, and background on the creation of the piece.  Knowing a users’ location, activity, and interests will enable Location Based Services (LBS) such as instantaneously providing a coupon for a discount in the museum gift shop targeted to the specific user.

The software development continues to progress assured that the hardware platform that will run the code will be available when it’s ready to ship. The hardware meanwhile continues to evolve as CMOS process node advances make it possible to include more functionality on chip.  Among sensor suppliers the drive is to integrate more functions on chip. For example, the Bosch Sensortec BMA355 integrates a three-axis accelerometer with a first nine-axis inertial measurement unit (IMU) including a gyroscope and magnetometer along with a 32-bit microcontroller, thus creating what the market has termed the sensor hub.

The debate is whether the phone hardware platform will evolve to one that includes a sensor hub or whether the software effecting the function will be distributed among the existing processors already in the phone:  applications processor, power management IC (PMIC), GPU, etc. or a combination of hub and one of the existing processors. The economics of packaging disparate devices—electro-mechanical MEMS that favor larger process nodes and non-standard materials with digital devices that are best done in standard CMOS and smaller process nodes may ultimate make the determination.

The proliferation of sensors driven by the connectivity afforded by the phone represents the leading edge of the Internet of things. These include wearable sensor, comprising everything from athletic performance monitors like the Nike Fuelband to health aids like the activity monitors for an aging population. Today, these wearable sensors lack the reliability of those in smart phones and tablets. Furthermore, the limited-function hardware platform found in these devices lack the refinement that exists in hardware platforms for mobile phones, which affords a widely varied functionality.

Architected to operate on battery power for long periods, these wearable designs will require the lowest power CPU in combination with small amounts of program storage and data uploaded to the cloud and not stored locally.  Just as the mobile phone created a huge market for the ARM processor and flash storage for program storage, these wearable devices will create their own hardware platforms.  They will be driven by a highly constrained power budget.  The budget will dictate CPUs that sip power and non-volatile program storage besides flash, either ROM or one-time programmable memory or a combination of the two that use no power to retain memory and little power during memory access.