Integration, Lower Costs Drive The Need For System Understanding

Dec. 14, 2010
TI signal chain applications engineering manager Rick Downs discusses tradeoffs between using separate parts (amp + ADC) in the signal chain and using an integrated part designed for the application. He uses an EEG as a design example.

1 of Enlarge image
 

Signal chain

Industry analysts have identified the system trends of portability, “green” energy reduction, and more sensors in end equipment. These trends drive the need for higher channel count, higher speed, and higher performance from analog-to-digital converters (ADCs) and digital-to-analog converters (DACs), while still demanding lower power budgets, smaller size, and lower cost.

Data converter manufacturers are responding to these demands by creating more data converters that are integrated with other circuit components. While many microcontroller cores are surrounded by a rich set of peripherals, some performance demands are driving the creation of specific analog front ends or other analog “companion” chips that work together with a separate processor.

For example, the Texas Instruments ADS1298 is a complete front end for electrocardiogram (ECG) systems. It packs eight 24-bit ADCs with programmable gain amplifiers (PGAs) and a host of ancillary circuitry into a single ball-grid array (BGA) or thin quad flat pack (TQFP) package.

As data converters become part of an integrated system in a single package, they tend to become more application-specific. The datasheet for the ADS1298 refers to many specific functions and terminology that manufacturers outside the ECG equipment space may not be familiar with. Does that mean that you can only use the ADS1298 for ECG applications?

Links In The Signal Chain
Looking at these integrated devices, as well as how they can benefit your system, is simply a matter of breaking them down and seeing how they implement what is called the signal chain (see the figure). In fact, the block diagram could represent just about any system that processes a signal.

If it’s a measurement or data acquisition system, then the chain starts at the sensor, proceeds through signal conditioning circuitry, into an ADC, and then ends with the processor. If it’s a control system, an audio processing system, or even a software-defined radio, then there is likely some output from the processor that must be turned back into an analog signal, as shown in the right half of the diagram.

Regardless of the type of system you need to design, there’s a good approach to deciding on the components that realize your signal chain. Generally, the processor is the first component selected. This selection usually is made based on familiarity with the device (it’s one your company has used for previous designs) or for a certain set of peripherals and capabilities that it offers. Thus, you begin at the center of the diagram and work your way outward.

This would imply that the data converter is the next choice, and it’s a logical place to start with the analog circuitry. Let’s assume we’re designing a measurement system, so we’re only dealing with an ADC. The big decisions here are how much resolution you need for your measurement and how fast you need to measure things.

There’s a number of other things to consider, of course, but the big two are speed and resolution. Notice that I haven’t said anything about how many bits the data converter has yet—just how much, in physical parameters, you need to resolve in your measurement. At this point it’s better to say that your measurement system needs to resolve at least 250 ppm, rather than decide on a 12-bit converter.

Continue on next page

If we were truly working outward, the signal conditioning would come next, but its purpose is to take whatever signal the sensor provides and match it up to the input range of the data converter. So first we must understand what kind of signal the sensor is providing us. Let’s say that the sensor can output 2 V at its maximum, so 2 * 250 ppm = 0.5 mV is what you hope to be able to measure at the sensor.

Now, you can consider how to measure that 0.5-mV change. One way to approach this is to use an amplifier to gain the signal up to match the full-scale range of your converter—let’s say that’s 5 V. With a gain of 2.5, the 0.5 mV on the sensor becomes 1.25 mV, so the converter needs to resolve 1.25 mV out of 5 V, or one count out of 4000. So, a 12-bit converter would work here.

Another approach would be to use a higher-resolution converter that could measure 0.5 mV directly and not need the signal conditioning at all. Which approach to choose would depend upon how much power, size, and cost would be saved by eliminating the amplifier but using a converter with more resolution. It may also be that the impedance of the sensor is such that it couldn’t go directly into a converter, so eliminating the amplifier may not be an option.

Understanding the system signal chain, and what is needed from each block, can help you decide if one of these highly integrated converters really will help your design. You could certainly use that ADS1298 for systems other than ECG, but the benefits it could bring only become attractive if your signal chain needs all of the blocks inside the device.

About the Author

Rick Downs

Rick Downs is Director of Applications at Maxim Integrated. He has authored over 50 articles and application notes and has prepared and delivered many seminars on data acquisition. Before joining Maxim, he held various positions at Burr-Brown, Dallas Semiconductor, and Texas Instruments. He received his BSEE from the University of Arizona and holds four patents. 

Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!