Higher integration, lower cost requires deeper system understanding

Higher integration, lower cost requires deeper system understanding

Industry analysts agree that the future of systems will be mobile and portable, “green” and energy efficient, as well as integrating more sensors into end devices. This trend requires analog-to-digital (ADC) converters and digital-to-analog (DAC) converters with higher channel counts, higher speed, and performance, while also requiring lower power budgets, smaller size, and lower cost.

The major data converter manufacturers have responded positively to these demands by making more data converters that integrate other circuit components. Despite the large number of peripherals surrounding many microprocessor cores, some performance requirements are driving the development of many special analog front ends or other analog “companion” chips that work with a single processor.

For example, TI recently introduced the ADS1298, a complete front-end for electrocardiography (ECG) systems. It packs eight 24-bit ADCs with programmable gain amplifiers and extensive auxiliary circuitry into a single BGA or TQFP package. As data converters become part of a single-package integrated system, they tend to become more practical; the ADS1298 data sheet covers many specific functions and terminology that may not be familiar to some manufacturers outside the field of ECG equipment. Does this mean you can only use the ADS1298 for ECG applications?

To study these integrated devices and see how they can benefit your system, simply take them apart and see how they implement what is called a signal chain, as shown in Figure 1.

Figure 1 Signal chain structure diagram

The block diagram shown in Figure 1 is representative of all systems for signal processing. If it is a measurement or data acquisition system, the signal chain starts at the sensor, goes through the signal conditioning circuit, into an ADCr, and ends at the processor. If it’s a control system, an audio processing system, or a software-defined radio there may be some processor output that must be changed back to analog; this is shown in the right-hand half of the block diagram.

Regardless of the type of system you are designing, there is a good way to identify some of the components that implement your signal chain. Generally speaking, the processor is the first component to be selected first. This choice is generally based on familiarity with the device (which is a processor your company has used in previous designs), or on certain peripherals and the features they have. So you start at the center of the structure diagram shown in Figure 1 and work your way out.

This means that the next choice is the data converter, and it is logical to start with an analog circuit. Assuming we are designing a measurement system, we only need to deal with one ADC. Determining how high resolution and how fast your measurements need to be is an important decision. Of course, there are many other aspects to consider, but two important ones are speed and resolution. Note that I didn’t say anything like how many bits the data converter has – just how much resolution you need for your measurements, it’s some physical parameter. So it’s better to say that your measurements need at least 250ppm resolution rather than to choose a 12-bit converter.

If our design process is really inside-out, the next step is signal conditioning, but the purpose is to use all the signal the sensor provides and then match it to the input range of the data converter. So, we first have to understand what kind of signal the sensor is giving us. We assume that the sensor can output a maximum of 2V, then 2*250ppm = 0.5mV is what you want the sensor to measure.

Now you can think about how to measure a 0.5mV change. One way to get around this is to use an amplifier to gain the signal to match the full-scale range of your converter—let’s say 5V. After a gain of 2.5, the sensor’s 0.5mV becomes 1.25mV, so the converter needs to resolve 1.25mV from 5V, or 1/4000. So, a 12-bit converter will do the job. Alternatively, use a higher resolution converter that can measure 0.5mV directly, eliminating the need for signal conditioning. Which method to use depends on the power and size savings, and the cost savings, of removing the amplifier and using a high-resolution converter. There is another situation where the sensor impedance is so high that it cannot go directly into the converter, so removing the amplifier is not an option.

Understanding the system signal chain and understanding the requirements of each block can help you determine whether one of these highly integrated converters is really helpful in your design. You can certainly use the ADS1298 for systems other than ECGs, but its many benefits are only attractive if your signal chain requires all of the device’s internal modules.

In future articles, we’ll cover some of the basics of accurately acquiring a signal and representing it in the digital domain. Many of the rules of thumb or recommendations we take for granted require a case-by-case analysis in order to understand why they are given in order to help you understand how to apply them given specific system requirements.

The Links:   CM600DX-24T BSM35GD120DN2 BUY-IGBT