Optimize instrumentation signal chains to reduce cost and improve performance

Efficient signal acquisition is critical for applications ranging from industrial process control and measurement to high-speed communications and imaging. In such a wide range of applications, it is necessary to match the appropriate components to the task in order to construct a signal chain that meets performance requirements at the lowest cost. With the expected proliferation of deeply embedded sensor systems serving the Internet of Things (IoT), this balance of cost and performance becomes even more important.

Efficient signal acquisition is critical for applications ranging from industrial process control and measurement to high-speed communications and imaging. In such a wide range of applications, it is necessary to match the appropriate components to the task in order to construct a signal chain that meets performance requirements at the lowest cost. With the expected proliferation of deeply embedded sensor systems serving the Internet of Things (IoT), this balance of cost and performance becomes even more important. Over time, the number of IoT devices is expected to reach tens of billions, and if costs can be reduced at each stage of the signal chain, the overall savings can be substantial.

For the designer, building an effective signal chain means balancing the specifications of the individual components at each stage to meet the target performance level of the entire signal chain. While some applications require extremely high-spec equipment (Figure 1), designers can often use more cost-effective components to achieve the required level of performance and functionality in a signal chain.

Optimize instrumentation signal chains to reduce cost and improve performance
Figure 1: A variety of high-performance analog components, including analog-to-digital converters and multiplexers, enable CERN’s Large Hadron Collider to measure magnetic fields with the highest possible performance.

In its most basic form, a signal acquisition circuit preferably contains only a single component, an analog-to-digital converter (ADC), that digitizes an analog input signal from a sensor or other source. However, for any practical application, this simple approach is not applicable to real-world signals, which places further demands on signal conditioning including amplification and filtering (Figure 2). For applications using active sensors, additional components such as digital-to-analog converters (DACs), precision voltage references, and amplifiers are required on the front end to provide the excitation current or voltage levels required by the sensor.

Optimize instrumentation signal chains to reduce cost and improve performance
Figure 2: Before data conversion, a typical analog signal chain requires conditioning to compensate for small signal inputs, signal skew, and other signal characteristics specific to various applications (Image source: Maxim Integrated)

signal conditioning

Typically, the signal produced by the sensor has a small amplitude. Without amplification, only a fraction of the ADC’s full dynamic range is available for these signals, which is likely to result in a loss of signal detail due to the limited ADC resolution and greater susceptibility to converter quantization errors. Therefore, designers often need to add an analog front-end (AFE) amplifier stage to increase the span of the input signal to match the ADC’s full dynamic range. The input amplifier also ensures that the sensor remains properly loaded, while also buffering the front-end load transients that occur when the signal is sampled, which is present at the input of some types of ADCs.

Engineers can choose from a wide variety of amplifiers that vary in functionality and performance. While there is a natural tendency to look for amplifiers with the highest possible performance specifications, engineers can significantly reduce design costs by thoroughly understanding the amplifier’s specifications and comparing them rigorously with the characteristics of the input signal and desired output resolution. For example, when an instrumentation amplifier (IA) needs to process a signal with a slow rate of change and an amplitude well above the noise floor, the pursuit of the fastest possible slew rate and the lowest possible noise may increase the Lots of unnecessary expenses. Likewise, an amplifier with a very good linearity specification will clearly end up performing better than an ADC with good accuracy, but it is entirely possible for the latter to meet the overall performance requirements of the signal chain even if the quantization error is significantly larger than the former.

Depending on signal characteristics and application requirements, engineers can choose from a variety of full-featured amplifiers such as high-precision IAs, low-noise amplifiers (LNAs), and programmable gain amplifiers (PGAs) when faced with more stringent requirements, but the performance of traditional operational amplifiers Features still work for most applications. For example, rail-to-rail input/output (RRIO) low-noise op amps, including products such as the Analog Devices AD850x, Maxim Integrated MAX963x, and the Texas Instruments OPA320 family, offer a relatively low-cost option for Maximize dynamic range and minimize noise in acquisition applications.

While traditional single-ended input amplifiers are sufficient for many applications, there are many signal acquisition applications where good common-mode rejection is a key requirement, requiring differential inputs. For example, applications using bridge sensors and designs operating in very noisy environments require amplifiers with fully differential inputs to provide high common-mode rejection. In fact, some differential amplifiers, such as the Analog Devices AD8476 and Texas Instruments THS4531, are specifically designed to meet differential signal conditioning requirements and also include features designed to simplify interfacing with ADCs. In addition to ADC interface options, features such as the integrated laser-trimmed resistors in the Analog Devices AD8476 help reduce component count and cost in signal chain designs (Figure 3).

Optimize instrumentation signal chains to reduce cost and improve performance
Figure 3: Differential amplifiers such as the Analog Devices AD8476 have integrated laser trimming and output capability to adjust to ADC interface requirements, helping to simplify the design of signal chains with differential input requirements.

While amplification extends the span of the input signal, it also exacerbates the noise characteristics of the signal, a problem that significantly limits the dynamic range. Therefore, the signal chain often needs to incorporate a filtering stage prior to conversion to limit the effects of noise at frequencies other than the signal to be processed.

For applications requiring high flexibility, engineers can use digital potentiometers such as the Maxim Integrated MAX540x and Texas Instruments TPL0102 series to build these filters and drive them with simple control logic or a host MCU. However, for applications with relatively stable signal characteristics, simple passive components are usually sufficient for the filtering requirements of the design.

In addition to noise bandwidth limitation issues, anti-aliasing filters are often required in the signal chain to reduce sampling artifacts that occur at frequencies above half the sampling rate. With the advent of oversampling conversion methods, the need for this stage has been greatly reduced.

signal conversion

The sole purpose of the signal conditioning circuit consisting of amplifiers and filters is to provide a clean signal to the input of the ADC. Therefore, the complexity and performance specifications of these front-end components are mainly determined by the characteristics and requirements of the ADC. If the signal chain only needs to perform low-resolution conversions of slowly changing signals, there is no need to spend unnecessary expenditures on expensive high-precision amplifiers.

In practice, the selection of the most suitable ADC and the required signal conditioning components, in turn, depends on careful analysis of the characteristics of the input signal and the overall functional requirements of the application. There is a big difference in the accuracy (and cost) of a signal chain that periodically measures changes in ambient temperature versus one that provides immediate feedback for mission-critical process control. In fact, the choice of ADC often depends on the requirements for signal conversion throughput and latency (the time from the start of signal acquisition until the ADC provides the corresponding data on its output).

Engineers can learn from ADC architectures that are designed to deliver vastly different levels of performance. Each architecture provides inherent capabilities for achieving high throughput and low latency, but also brings inherent limitations. For example, flash ADC architectures typically have very high throughput and very low latency, but are usually only cost-effective at lower bit resolutions. Flash ADCs, including the Analog Devices AD782x and Texas Instruments TLC0820, use a parallel configuration of conversion elements to perform high-speed conversions. Their high throughput and low latency make them ideal for situations where significant latency cannot be tolerated, such as in speech coding applications.

In contrast, successive approximation register (SAR) ADCs and sigma-delta (ΣΔ) ADCs offer cost-effective performance for a variety of requirements and applications. SAR ADCs have become the primary choice for most mid- to high-resolution applications. These devices complete the conversion in one cycle, making them suitable for data acquisition applications where latency is critical, such as control loops, power monitoring, and signal analysis.

Sigma-Delta ADCs often provide a low-cost option for high-resolution conversions due to their inherent oversampling architecture. On the other hand, the relatively slow settling time and sharp cutoff of the internal digital filter in traditional ΣΔ ADCs limit their use in certain applications. Therefore, designers sometimes choose a SAR ADC even when the performance of a ΣΔ ADC is sufficient. For example, SAR ADCs have been the preferred choice for control loop and multiplexing applications due to concerns over the performance of traditional sigma-delta converters.

Feedback delays can create instability in industrial processes, home appliances, or automotive control loops, so designers sometimes choose a SAR ADC rather than risk the longer delays of a ΣΔ ADC. However, in control applications with relatively slow signals, the predictable delay of the ΣΔ ADC has a practically negligible effect on the stability of the control loop.

In multi-channel applications, designers often multiplex multiple input channels into a single ADC to save cost, footprint, and overall component count. For these designs, engineers choose a SAR ADC based on the traditional concern that the ΣΔ ADC might not settle fast enough to complete the conversion before the next channel is multiplexed into the ADC input. However, in many sensing applications, the rate of change of the physical phenomenon being monitored is well below the settling time of a ΣΔ ADC, so a ΣΔ ADC can easily handle many multiplexed channels.

While traditional sigma-delta ADCs are ideal for applications with slow signal changes, newer sigma-delta ADCs such as the Texas Instruments ADS124x have more sophisticated features that largely eliminate traditional concerns (Figure 4). For example, more sophisticated filtering techniques in the new generation of devices enable their outputs to settle with zero cycle delay. Therefore, 24-bit ΣΔ ADCs such as the TI ADS124x can provide differential multiplexed inputs with output rates up to 2ksps.

Optimize instrumentation signal chains to reduce cost and improve performance
Figure 4: New technologies in 24-bit ΣΔ ADCs such as the TI ADS124x eliminate many of the traditional concerns when using ΣΔ ADCs in low-latency designs and with multiplexed differential inputs.

In addition to matching ADC specifications to application requirements, designers can further optimize the analog signal chain by considering the role of the voltage reference in the application. By providing a stable reference voltage, precision voltage references are essential to ensure absolute accuracy in signal conversion. These devices are generally considered important in applications such as battery-powered designs or energy-harvesting designs, where the supply voltage fluctuates as the battery discharge cycle ends or the harvested energy source periodically weakens.

For applications that do not require this level of absolute accuracy, designers can eliminate the need for a precision voltage reference by using a ratiometric conversion method (Figure 5). The result of the ratio conversion is the ratio of the reference voltage, which is usually the supply voltage or the excitation voltage. In this way, the ADC output remains ratiometric even if there are fluctuations in the power supply.

Optimize instrumentation signal chains to reduce cost and improve performance
Figure 5: ADCs such as the Maxim Integrated MAX1415 can operate in ratiometric mode, eliminating the need for precision voltage references in signal conversion.

digital domain

ADCs typically provide a standard I2C or SPI compatible serial interface for connecting the output of the analog signal chain to the MCU. As data points flow into the MCU, designers can use traditional filtering algorithms running in software or digital signal processing hardware to improve the signal-to-noise ratio for higher performance-intensive applications.

Digitally creating cutoff and notch filters with a high signal-to-noise ratio can greatly help relax requirements in the analog signal chain. For example, instead of sacrificing the design’s footprint to use more complex analog filtering components that may be required in a particular application, designers can choose to move the complexity of filtering to the digital domain. Of course, increased software complexity means higher demands on memory and MCU performance.

Epilogue

It is not difficult for designers to find among the many analog signal conditioning and conversion components with performance specifications that meet demanding data acquisition requirements. However, for many applications, the signal chain only needs to meet the application requirements and does not need to use all components with the best specifications in all aspects. By matching the ADC to the conversion requirements and the signal conditioning components to the ADC specifications, engineers can design signal chains that easily meet performance and cost goals.

The Links:   LTD104C11U LQ104V1DC31 7MBR50SB120-50