Filters and Flexibility: TI’s New Delta-Sigma ADC Gives Designers Both
With a range of filter options, a new delta-sigma A/C converter can help designers reap tailored, high-performance data from industrial systems.
Texas Instruments recently introduced the ADS127L11 delta-sigma analog-to-digital (A/D) converter. The new device is a compact 24-bit, 400-kSPS ADC with selectable wideband and low-latency digital filters. Let’s see how the filter options of this new device can help designers achieve tailored, high-performance data acquisition in a broad range of industrial systems.
How Does a Delta-Sigma ADC Work?
As shown below, a delta-sigma ADC consists of a delta-sigma modulator and a digital decimating filter.
The basic block diagram of a sigma-delta ADC. Image used courtesy of Texas Instruments
The delta-sigma modulator converts the analog input signal to a high-speed stream of single-bit values. The density of 1s in the data stream produced by the modulator is proportional to the amplitude of the analog input. A typical output waveform for a sinusoidal input is shown below.
Image used courtesy of Texas Instruments
Since the density of 1s is proportional to the amplitude of the input signal, averaging a certain number of the consecutive bits from the data stream gives us a multi-bit value, which can be considered a digital representation of the analog input value. Delta-sigma ADCs commonly use a finite impulse response (FIR) filter to implement the averaging function. These filters are inherently stable and exhibit a linear-phase response.
_filter.jpg)
A finite impulse response (FIR) filter. Image used courtesy of Analog Devices
Another function of the digital decimating filter is throwing away the redundant output samples. Sigma-delta ADCs oversample the input to produce the high-speed data stream at the modulator output and hence, the redundant samples should be discarded at the output to have a more efficient system.
Frequency Domain Point of View
From a frequency domain point of view, the delta-sigma modulator uses the oversampling technique to push a portion of the quantization noise power to frequencies outside of the desired frequency band. While oversampling reduces the noise power in the frequency band of interest, it evenly distributes the noise power over the frequency range from DC to half the sampling frequency.
To further improve the noise performance, a delta-sigma modulator offers a noise shaping feature to further reduce the noise in the frequency band of interest. This is shown below.
Image used courtesy of Analog Devices
The digital filter following the delta-sigma modulator keeps the desired signal and suppresses the out-of-band noise.
Digital Filters in Delta-Sigma ADCs
There are two types of filters commonly employed in delta-sigma ADCs: sinc filters (the name stems from the sin(x)/x frequency response of the filter) and wide-bandwidth flat-passband filters. The frequency response of a wide-bandwidth filter and a third-order sinc filter (sinc3) is shown below.
.jpg)
Image used courtesy of Texas Instruments
A wideband filter has a flat response up to the Nyquist bandwidth of the output data rate (fDR/2) and exhibits a very sharp transition from the passband to the stopband. Therefore, the desired signal passes through the filter without being distorted while the out-of-band noise is significantly attenuated (about 120 dB stopband attenuation in the above figure).
A sinc3 filter doesn’t have a flat passband response. It has an attenuation of 3dB at 0.262✕fDR and exhibits a slow transition from the passband to the stopband. Moreover, the sinc3 filter exhibits a much smaller stopband attenuation. As a result, with a sinc-type filter, a larger portion of the out-of-band noise can fold back into the bandwidth of interest.
Although the wideband filter has a nice frequency response, its time-domain response is not appealing. The sharp transition band of a digital filter is achieved at the cost of increasing the filter order.
The disadvantage is that the higher the filter order, the longer it takes the filter to settle to a final value upon receiving a step input. A typical wideband filter might need 80 output samples to settle to its final value whereas a sinc3 filter can settle in the time duration of three samples—after a step is applied to the input.
Due to this low latency, sinc-type filters are useful when we need to multiplex between multiple sensor inputs relatively quickly.
The ADS127L11 Provides Both Filter Options
A simplified block diagram of the ADS127L11 is shown below.
Image used courtesy of Texas Instruments
The new device incorporates both wideband and low-latency filter types. Depending on the application, the user can choose the appropriate filter type and optimize either the frequency response characteristics (wideband filter mode) or the time domain characteristics (low-latency filter mode). The wideband mode offers data rates as high as 400 kSPS.
According to TI, the wideband mode can improve AC measurement with a 50% wider bandwidth and a 30% higher signal-to-noise ratio than competing data converters. The low-latency mode also offers 25% lower latency at up to 1,067 kSPS. The offset drift in this mode is 50 nV/°C which, according to TI, corresponds to an 83.3% improvement over similar data converters. The low-latency mode should improve DC measurement resolution, data throughput, and response times in data acquisition and condition-monitoring applications.
Other Salient Features
The new ADC comes in a compact 3 mm x 3 mm quad flat no-lead (WQFN) package. With the ADC footprint reduced by 50%, the converter is no longer the largest component in a typical data acquisition board.
The figure below shows a layout of a data acquisition system that uses TI’s fully differential amplifier (THS4551) and voltage reference (the REF6041)—along with the new ADC.
Image used courtesy of Texas Instruments
The new ADC is also suitable for simultaneous sampling systems. The ADS127L11 burns only 3.3 mW at a sampling rate as high as 50 kSPS. According to TI, this is half the power of other similar products and can extend the battery run time in portable electroencephalogram (EEG) machines and power-quality analyzers.
Designed for Digital Audio and Industrial Measurements
Delta-sigma ADCs are well-suited for low-frequency, high-accuracy measurements—for example, applications that require greater than 20 bits of noise-free resolution. Digital audio and precision industrial measurements are two common use cases of delta-sigma ADCs.
The following figure provides some typical numbers for the resolution and sample rate of sigma-delta ADCs as well as the SAR and pipeline architectures.
Typical number of bits and conversion rates for different types of A/D converters. Image used courtesy of Texas Instruments
Note that the figure only gives some typical values and doesn’t specify the performance limits of the ADCs. For example, 32-bit delta-sigma ADCs with a sample rate of 38-kSPS are available on the market, such as the ADS1263 from TI.
TI’s new ADC is designed for shock and vibration instruments, acoustics and dynamic strain gauges, electroencephalogram (EEG), and power-quality analyzer applications.
