Complete Sensor-to-Bits Solution Simplifies Industrial Data-Acquisition System Design ±20 mA ranges. When long cables with substantial electromagnetic interference (EMI) are encountered, current loops are often used due to their inherently high noise immunity. Analog output modules typically control actuators, such as relays, solenoids, and valves, to complete the automated-control system. They typically provide output voltages with 5 V, 10 V, ±5 V, and ±10 V full-scale ranges and 4 mA to 20 mA current-loop outputs. By Maithil Pachchigar Introduction Typical analog I/O modules include 2, 4, 8, or 16 channels. To meet stringent industry standards, these modules require protection against overvoltage, overcurrent, and EMI surges. Most PLCs include digital isolation between the ADC and the CPU and between the CPU and the DAC. High-end PLCs may also incorporate channel-to-channel isolation, as specified by the International Electrotechnical Commission (IEC) standards. Many I/O modules include per-channel software programmable single-ended or differential input ranges, bandwidth, and throughput rate. Thus, designers of industrial equipment and critical infrastructure encounter significant challenges in meeting their customers’ stringent accuracy, noise, drift, speed, and safety requirements. Using a PLC as an example, this article shows how the versatile, low cost, highly integrated ADAS3022 reduces complexity, solving many challenges encountered in the design of multichannel dataacquisition systems by replacing the analog front-end (AFE) stages. Ideally suited for precision industrial, instrumentation, power line, and medical data-acquisition cards with multiple input ranges, this high-performance device reduces cost and time to market while offering a small, easy-to-use footprint and true 16-bit precision at 1 MSPS. In modern PLCs, the CPU performs numerous control tasks in an automated manner, employing real-time access to information to make intelligent decisions. The CPU may embody advanced software and algorithms, and web connectivity for diagnostic error checking and fault detection. Commonly used communication interfaces include RS-232, RS-485, industrial Ethernet, SPI, and UART. At the heart of many industrial automation and process-control systems, programmable logic controllers (PLCs) monitor and control complex system variables. Employing multiple sensors and actuators, PLC-based systems measure and control analog process variables such as pressure, temperature, and flow. PLCs are found in diverse applications—such as factories, oil refineries, medical equipment, and aerospace systems—that require high accuracy and robust, long-term operation. In addition, the competitive marketplace demands lower cost and shorter design times. Discrete Implementation of Data-Acquisition System Industrial designers can build analog modules for PLCs or similar data-acquisition systems with discrete high-performance components, as shown in Figure 2. Key design considerations include input signal configuration and overall system speed, accuracy, and precision. The signal chain presented here utilizes the ADG1208/ADG1209 low-leakage multiplexer, AD8251 fastsettling programmable-gain instrumentation amplifier (PGIA), AD8475 high-speed funnel amplifier, AD7982 differential-input 18-bit PulSAR ® ADC, and ADR4550 ultralow-noise voltage reference. This solution provides four different gain ranges, but with maximum input signals of ±10 V, designers will have to worry about the multiplexer’s switching and settling times, as well as other analog signal conditioning challenges. In addition, achieving true 16-bit performance at 1 MSPS can be a major challenge, even when using these high-performance components. PLC Application Example Figure 1 shows a simplified signal chain for a PLC used in industrial automation and process-control systems. The PLC typically comprises analog and digital input/output (I/O) modules, a central processing unit (CPU), and power-management circuitry. In industrial applications, analog input modules acquire and monitor signals from remote sensors located in harsh environments characterized by extreme temperature and humidity, vibration, and explosive chemicals. Typical signals include single-ended or differential voltages with 5 V, 10 V, ±5 V, and ±10 V full-scale ranges, or current loops with 0 mA to 20 mA, 4 mA to 20 mA, and POWER MANAGEMENT COMMON INDUSTRIAL V/I INPUTS 0mA TO 20mA 4mA TO 20mA ±20mA DIGITAL ISOLATOR IN-AMP/ PGA MUX ADC BUFFER ANALOG INPUT MODULE DIGITAL ISOLATOR CPU REF OP AMP DAC BUFFER PROTECTION CURRENT LOOPS TO ALL COMPONENTS BUFFER PROTECTION 0V TO 5V 0V TO 10V ±5V ±10V TO INDUSTRIAL ACTUATORS V/I OUTPUTS 0V TO 5V, 0V TO 10V, ±5V, ±10V 0mA TO 20mA, 4mA TO 20mA REF ANALOG OUTPUT MODULE Figure 1. Typical PLC signal chain. Analog Dialogue 47-04, April (2013) www.analog.com/analogdialogue 1 The AD7982 specifies a 290-ns transient response from a fullscale step. Thus, to guarantee the specified performance while converting at 1 MSPS, the PGIA and funnel amp must settle in less than 710 ns. The AD8251 specifies 785-ns settling time to 16 bits (0.001%) for a 10-V step, however, so the maximum throughput that can be guaranteed for this signal chain will be less than 1 MSPS. +2.5V +0.5V GAIN = 0.16, 0.2, 0.4, 0.8, 1.6 INL MAX = 0.649 INL MIN = –0.592 1.5 1.0 0.5 INL (LSB) +4.5V 4V 2.0 0 –0.5 ±10V RANGE ±5V RANGE ±2.5V RANGE ±1.25V RANGE MUX IN AMP ADG1208/ ADG1209 ±10V FUNNEL TRUE-DIFFERENTIAL ADC AMP AD8251 AD8475 G = +1 G = +2 G = +4 G = +8 G = +0.4 –1.0 AD7982 LPF +4.5V 4V +2.5V –1.5 REF ADR455x +0.5V MUX CONTROL –2.0 0 8192 16384 24576 32768 40960 49152 57344 65536 Figure 2. Analog input signal chain using discrete components. CODE 0 Integrated Solution Simplifies Data-Acquisition System Design iCMOS ®, VDDH DIFF PAIR IN0/IN1 DIFF COM AVDD DVDD VIO RESET PD ADAS3022 CNV LOGIC/ INTERFACE IN0 IN4 PulSAR ADC PGIA MUX IN5 IN6/IN7 COM –100 –120 SCK DIN REFIN BUF REF AUX– VSSH AGND –160 –180 0 100 200 300 400 500 FREQUENCY (kHz) TEMP SENSOR AUX+ DGND REFx Figure 3. Functional block diagram of ADAS3022. This complete sensor-to-bits solution utilizes only one-third of the board space of discrete implementations, helping engineers to simplify their designs while reducing the size, time to market, and cost of advanced industrial data-acquisition systems. Eliminating the necessity to buffer, level shift, amplify, attenuate, or otherwise condition the input signal, and the concerns regarding commonmode rejection, noise, and settling time, it alleviates many of the challenges associated with designing a precision 16-bit, 1-MSPS data-acquisition system. It delivers the best-in-class 16-bit accuracy (±0.6-LSB typical INL), low offset voltage, low drift overtemperature, and optimized noise performance at 1 MSPS (91-dB typical SNR), as shown in Figure 4. The device is specified over the –40°C to +85°C industrial temperature range. 2 –80 –140 SDO IN6 IN7 –60 CS IN2 IN3 IN4/IN5 –40 BUSY IN1 IN2/IN3 GAIN = 0.8 fS = 1000kSPS fIN = 10.1kHz SNR = 90.7dB SINAD = 90.6dB THD = –107dB SFDR = 106dB –20 AMPLITUDE (dBFS) Manufactured in a proprietary, high-voltage industrial process technology, the 16-bit, 1-MSPS ADAS3022 dataacquisition IC integrates an 8-channel, low-leakage multiplexer; a high-impedance PGIA with high common-mode rejection; a precision, low-drift 4.096-V reference and buffer; and a 16-bit successive-approximation ADC, as shown in Figure 3. Figure 4. INL and FFT performance of the ADAS3022. The PGIA has a large common-mode input range, true highimpedance inputs (>500 MΩ), and a wide dynamic range, allowing it to accommodate 4-mA to 20-mA current loops, accurately measure small sensor signals, and reject interference from ac power lines, electric motors, and other sources (90-dB minimum CMR). An auxiliary differential input channel can accommodate ±4.096 V input signals. It bypasses the multiplexer and PGIA stages, allowing direct interface to the 16-bit SAR ADC. An on-chip temperature sensor can monitor the local temperature. This high level of integration saves board space and lowers the overall parts’ cost, making the ADAS3022 ideal for space-constrained applications, such as automatic test equipment, power-line monitoring, industrial automation, process control, patient monitoring, and other industrial and instrumentation systems that operate with ±10-V industrial signal levels. Analog Dialogue 47-04, April (2013) D2 L2 47 H COUT3 4.7 F + C2 1 F + 1.78 RFILT L1 47 H +5V + D1 VIN = +5V CIN + 1 F C1 1 F RB0 1 COUT1 + 1 F L3 1 F +15V + COUT2 2.2 F VDDH REN ENABLE 50k ADP1613 CC1 + 12nF + COMP CC2 10pF RC1 100k FREQ EN VIN GND SW CSS + 1 F CNV MUX PulSAR ADC PGIA BUSY SCK DIN SDO BUF TEMP SENSOR REFIN REF AUX– Z1 DNI PD CS COM AUX+ CV5 + RS2 1 F DNI RESET LOGIC/ INTERFACE IN0/IN1 SS FB RF2 4.22k IN0 IN1 IN2 IN2/IN3 IN3 IN4 IN4/IN5 IN5 IN6 IN6/IN7 IN7 VIO ADAS3022 DIFF DIFF PAIR COM RS1 0 AVDD DVDD VSSH AGND DGND REFx –15V RF1B 47.5k Figure 5. Complete 5-V, single-supply, 8-channel data-acquisition solution with integrated PGA. Figure 5 shows a complete 8-channel data acquisition system (DAS). The ADAS3022 operates with ±15-V and +5-V analog and digital supplies, and a 1.8-V to 5-V logic I/O supply. The ADP1613 high-efficiency, low-ripple dc-to-dc boost converter allows the DAS to operate with a single 5-V supply. Configured as a single-ended, primary inductance Ćuk (SEPIC) topology using the ADIsimPower ™ design tool, the ADP1613 furnishes the ±15-V bipolar supplies required for the multiplexer and PGIA without compromising performance. The noise performance of the ADAS3022 and the discrete signal chain are compared in Table 1, which uses the input signal amplitude, gain, equivalent noise bandwidth (ENBW), and input-referred (RTI) noise of each component to calculate the total noise of the complete signal chain. Table 1. Noise Performance for the ADAS3022 and the Discrete Signal Chain ADG1209 AD8251 AD8475 AD7982 Total Noise ADAS3022 RTI RTI RTI RTI SNR RTI Total SNR SNR (μV rms) (μV rms) (μV rms) (μV rms) (dB) (μV rms) (dB) (dB) (V rms) Gain = 1 (±10 V) 6.56 124 77.5 148 95.5 208 90.6 91.5 7.07 Gain = 2 (±5 V) 6.56 83.7 38.8 74.2 95.5 119 89.5 91.0 3.54 Gain = 4 (±2.5 V) 6.56 68.2 19.4 37.1 95.5 80.3 86.8 89.7 1.77 Gain = 8 (±1.25 V) 6.56 55.8 9.69 18.5 95.5 60.0 83.4 86.8 0.88 Input Signal Noise Analog Dialogue 47-04, April (2013) 3 The single-pole low-pass filter (LPF) between the AD8475 and AD7982 (Figure 2) attenuates the kick coming from the switched-capacitor input of the AD7982 and limits the amount of high-frequency noise. The –3-dB bandwidth (f –3dB) of the LPF is 6.1 MHz (R = 20 Ω, C = 1.3 nF), allowing fast settling of the input signals while converting at 1 MSPS. The ENBW of the LPF can be calculated as ENBW = π/2 × f–3dB = 9.6 MHz. Note that this calculation ignores the noise from the voltage reference and LPF as it does not significantly affect the total noise, which is dominated by the PGIA. Consider an example using the ±5-V input range. In this case, the AD8251 is set for a gain of 2. The funnel amplifier is set to a fixed gain of 0.4 for all four input ranges, so a 0.5-V to 4.5-V differential signal (4 V p-p) will be applied to the AD7982. The RTI noise of the ADG1208 is derived from the Johnson/Nyquist noise equation (en2 = 4K BTRON, where K B = 1.38 × 10 –23 J/K, T = 300K, and RON = 270 Ω). The RTI noise of the AD8251 is derived from its 27-nV/√Hz noise density as specified in the data sheet for a gain of 2. Similarly, the RTI noise of the AD8475 is derived from its 10-nV/√Hz noise density using a gain of 0.8 (2 × 0.4). In each calculation, ENBW = 9.6 MHz. The RTI noise of the AD7982 is calculated from its 95.5-dB SNR as specified in the data sheet using a gain of 0.8. The total RTI noise of the entire signal chain is calculated based on the root-sum-square (rss) of the RTI noise from the discrete components. The total SNR of 89.5 dB can be computed from the equation SNR = 20 log(VINrms/RTITotal ). Although the theoretical noise estimate (SNR) and the overall performance of the discrete signal chain is comparable to that of the ADAS3022, especially at lower gains (G = 1 and G = 2) and lower throughput rates (much less than 1 MSPS), it’s not an ideal solution. The ADAS3022 can reduce cost by about 50% and board space by about 67%, as compared to the discrete solution, and it can also accept three additional input ranges (±0.64 V, ±20.48 V, and ±24.576 V) that the discrete solution cannot offer. 4 Conclusion The next generation of industrial PLC modules will demand high accuracy, reliable operation, and functional flexibility, all in a small, low-cost form factor. The ADAS3022, with industry leading integration and performance, supports a wide range of voltage and current inputs to handle a variety of sensors in industrial automation and process control. An ideal fit for PLC analog input modules and other data-acquisition cards, the ADAS3022 will enable industrial manufacturers to differentiate their systems while meeting stringent user requirements. References Kessler, Matthew. Synchronous Inverse SEPIC Topology Provides High Efficiency for Noninverting Buck/Boost Voltage Converters, Analog Dialogue, Vol. 44, No. 2, 2010. Slattery, Colm, Derrick Hartmann, and Li Ke. PLC Evaluation Board Simplifies Design of Industrial Process Control Systems, Analog Dialogue, Vol. 43, No. 2, 2009. Circuit Note CN0201. Complete 5 V, Single-Supply, 8-Channel Multiplexed Data Acquisition System with PGIA for Industrial Signal Levels. MT-048 Tutorial. Op Amp Noise Relationships; 1/f Noise, RMS Noise, and Equivalent Noise Bandwidth. Author Maithil Pachchigar [[email protected]] is an applications engineer in ADI’s Precision Converters business unit in Wilmington, MA. He joined ADI in 2010 and supports the precision ADC product portfolio and customers in the industrial, instrumentation, medical, and energy segments. Having worked in the semiconductor industry since 2005, Maithil has published several technical articles. He received an MSEE from San Jose State University in 2006 and an MBA from Silicon Valley University in 2010. Analog Dialogue 47-04, April (2013)