Smarter analogue, designed faster

17 November 2014

By integrating sensors into electronic devices, we can go beyond the basics of analogue interfacing.

Smart analogue emerged alongside the development of microcontrollers (MCUs). The focus for MCUs was initially on sampling of common parameters, using thermocouples, strain gages, pressure transducers, and similar sensors. Issues of variation, linearity, offset, and calibration were straightforward, and resolution beyond 8bits was a luxury.

But sensor technology has made a huge leap in the last decade. Accelerometers, gyroscopes and other advanced sensors are now common in devices like smartphones, fitness bands, and quadcopters, as well as industrial instrumentation. 

Signal processing demands supporting these sensors are on the rise. Features like hardware multipliers, DSP extensions, and floating point units make for fast, high precision computational ability. Adaptive algorithms such as Kalman filtering and Rao-Blackwellised particle filtering are running as part of real-time sensor fusion schemes.

“Always-on” sensor-based systems must use energy wisely; techniques such as clock gating and integrated sleep modes support this. Improved memory accesses and features such as bit-field manipulation, add up to a 32-bit processor core being more efficient for smarter analogue.

Devices are becoming situation-aware, understanding the context of sensor readings and other information. Smarter analogue offers the capability to analyse and adapt in real-time.
Attitude and heading reference systems (AHRS) combine the elements of orientation, acceleration, magnetic heading, and location via GNSS. Road grade estimation takes a simplistic 2D map, and adds topographic information and input from vehicle dynamics which can be applied to look-ahead cruise control or emissions reduction. Alternatively, the learning and alerting functions can be used in digital health to detect a sudden weight gain indicating congestive heart failure, or correct seating posture throughout a day. Extended to the Internet of Things and “big data,” context helps judge acuity and notifying parties, without false alarms.


Reducing errors and noise


Real-world conditions are usually suboptimal, and even the most advanced sensor can produce questionable readings unless signals are extracted from noise and sources of error are dealt with.

MEMS sensors avoid some pitfalls of their mechanical counterparts, but still have errors requiring compensation. Typical MEMS accelerometers, magnetometers and gyroscopes contain bias errors that vary with environmental or mechanical anomalies.

Calibration is an important first step. Simply cutting resistance values in or out with analog switches plays an important role in trimming. Many designers are turning to the configurable analog front end (AFE), compensating for gain and offset parameters close to the sensor using D/A converters. This approach is particularly effective for environmental adjustments, tidying up performance variation over temperature, location, and time.

The motivation for smarter analogue is evident: sensor performance can be greatly enhanced using a combination of analog and digital domain techniques, providing results a system can rely on.


Designing smarter analogue silicon


In a smarter analogue scenario, expertise in both analogue and digital comes to bear in designing sensor-based systems.

Advances in EDA tools, fabrication processes, and IP blocks have made smarter analogue design a straightforward proposition. Designers can address issues such as signal integrity, power management and connectivity in one design flow, quickly producing reliable chips with tailored capability. The Cadence Virtuoso mixed-signal design suite, for example, offers a complete analogue design environment, ranging from layout and SPICE block-level simulation to a complete system-level mixed-signal simulation.

ARM Cortex-M processor core IP can be incorporated into a variety of analogue-friendly process nodes. Analogue designers usually opt for more mature processes; advanced geometries present isolation, leakage, and other issues. Smarter analogue is likely to be implemented in 90, 65, 55, and 40nm CMOS processes for general purpose designs. Thick Gate Oxide (TGO) cell libraries can reduce leakage current significantly. 

This approach also helps make chip respins less costly, and enables better reuse of design IP for subsequent designs. Rather than concentrating only on analogue domain stages, configurable designs with analog and digital IP blocks can adapt quickly to accommodate multi-sourcing of MEMS sensors, changes in sensor fusion algorithms, process changes for yield or reliability improvement, cost reduction, or deployment in new applications.


ARM Cortex-M processors


Cortex-M processors provide energy-efficient, easy to use 32-bit engines with better code density and advanced algorithmic capability. Variants include the small and inexpensive Cortex-M0, the ultra-low power Cortex-M0+, the Cortex-M3 which blends performance with efficiency and the top-of the-range Cortex-M7 processor, which incorporates highly efficient signal processing features for digital signal control, as well as accelerated SIMD (Single Instruction, Multiple Data) operation.

The task for smarter analogue designers becomes how to quickly integrate and debug a solution crafted from IP blocks. Rather than starting from transistors, analogue IP blocks are now widely available as third-party reusable functions that designers can grab and implement quickly.
 
To help designers get started with ARM Cortex-M processors based designs, ARM has packaged elements of bus infrastructure (AMBA) and common peripherals, with design examples, scripts, and more into the Cortex-M system design kit (CMSDK). When ready for fabrication, the ARM artisan physical IP product range provides solutions including standard cell libraries for an assortment of foundries and process nodes.  

The ARM-developed mbed platform offers open source hardware development kits for ARM Cortex-M microcontrollers, plus many off-the-shelf implementations to get ideas prototyped quickly. A free online mbed compiler and software libraries plus over 5000 contributed C/C++ code examples means anyone with a web browser can easily and quickly write and share programs. 

Middleware is also an important piece of the solution. For sensor fusion, ARM Cortex-M designers can reach into the ARM CMSIS-DSP software library whose functions run in software on all ARM Cortex-M cores. Sensor hubs, ARM and Sensor Platforms have teamed up on the open sensor platform, simplifying integration and providing a framework for sensor data. 


Putting it all together

The view of analogue as simply a peripheral to a microcontroller has been replaced by a holistic approach, where sensors are shaped into context. As we have seen, the sophistication of system-level algorithms coupling sensors, an analogue subsystem, and a 32-bit engine for calibration and computation leads to new possibilities.

What may have been true about the difficulty of analogue design only a few years ago has been blown away by progress in technology. Designers equipped with the right EDA tools, IP blocks, and software can now engage smarter analogue designs with confidence – producing better results in less time.


Contact Details and Archive...

Print this page | E-mail this page






This website uses cookies primarily for visitor analytics. Certain pages will ask you to fill in contact details to receive additional information. On these pages you have the option of having the site log your details for future visits. Indicating you want the site to remember your details will place a cookie on your device. To view our full cookie policy, please click here. You can also view it at any time by going to our Contact Us page.