Data Acquisition Addresses Multifaceted Applications

Nobody buys test instrumentation because it’s trendy—cars and clothing, yes, but not data acquisition systems. These products are bought to solve problems. They help you understand how equipment under test (EUT) is operating by presenting graphical and numerical signal descriptions. When that information is combined with knowledge and experience, insight into the causes of performance issues quickly follows.

As test engineers will confirm, satisfying basic signal conditioning and acquisition requirements can be the easy part of the job. A test system’s size, weight, EMC performance, isolation, data transfer speed, and ruggedness often are at least as important. It all depends on the particular application being addressed.

And, it’s apparent from the features being offered in today’s equipment that a very wide range of applications is being considered. Some of the new capabilities result from improvements in FPGAs, which now are large enough and reasonably priced so all the logic needed for a data acquisition system is provided by one device. In addition, extra FPGA capacity is available to support customization.

The functional density gained by using an FPGA may result in reduced size or more channels in the same space. An FPGA also makes possible signal processing that otherwise wouldn’t be practical. Especially for multichannel high-speed and high-resolution data acquisition, on-board data reduction through filtering or a more specialized algorithm significantly offloads the host processor, facilitating greater throughput.

Improved ADCs also have contributed to new data acquisition system capabilities. For example, 24-b devices have such a wide dynamic range that less gain switching is needed. In addition to saving money, a simpler system can improve measurement accuracy by eliminating range-to-range uncertainty.

FPGA and ADC devices are only two of many factors influencing data acquisition system performance. Because a large number of capabilities are needed in some situations, their relationships to one another can best be shown by presenting a number of actual application examples.

Applications

Nuclear Fusion Monitoring
If controlled nuclear fusion can be successfully developed, a sustainable source of virtually unlimited power will be available. Uncontrolled fusion occurs continuously in stars and has been accomplished by man in the hydrogen bomb, so there’s no question that huge amounts of energy can be produced. The problem is in the control and containment of the fusion reaction.

Very expensive, long-term research is getting closer to achieving that goal. Typically, a Tokomak-type reactor is used to heat circulating plasma to approximately 100,000,000°C and contain it in a strong magnetic field. The ITER experiment is the latest large project under way and expected to produce much more energy than it consumes. Until now, the best experimental results generated as much energy as was used.

Originally, ITER was an acronym for International Thermonuclear Experimental Reactor. This interpretation has been discontinued. Today ITER is associated with its Latin meaning “the way.” The term ITER refers to both the organization running the project and the Tokomak device itself. An artist’s impression of the proposed device is shown in Figure 1.

Figure 1. Proposed ITER Tokomak Fusion ReactorCourtesy of The ITER Organization

The electron temperature and density in the plasma can be accurately measured by detecting the effect of Thomson scattering on a high-intensity laser beam. Thomson scattering refers to the scattering of electromagnetic radiation from an accelerating particle. It’s a weak effect that requires a strong laser source.

As Klaas Vogel, market development manager for Agilent Technologies’ Acqiris Operation, described it, “Two features of Thomson scattering make it a particularly attractive measurement technique. First, for all practical purposes, it does not perturb the plasma, requiring only the access of laser beams to it. Second, Thomson scattering offers the potential to determine detailed information about the distribution function of electrons and ions in the plasma. These advantages outweigh the fact that the measurements are technically very difficult to perform.”

In a monitoring solution deployed at the University of Wisconsin, photomultiplier tubes positioned around the school’s Tokomak torus detect photons, and the tube outputs are digitized by up to 160 channels of Agilent’s high-speed Model U1063-002 Digitizers. At least a 250-MHz bandwidth and a 1-GS/s sampling rate are required along with tight synchronization of all channels. This type of instrumentation also should be appropriate for the much larger ITER Tokomak when it is completed.

In addition to basic speed, considerations include the need for sensitive inputs, high channel density and relatively compact size, low power consumption, and EMI immunity. The last factor is especially important for instrumentation operating near the intense Tokomak electric and magnetic fields. The University of Wisconsin solution uses Agilent’s Model U1091 AK08 Fiber-Optic Interface with 30 meters of fiber-optic cable to ensure isolation.

Testing Lithium-Ion Batteries
It’s easy to test one battery cell at a time, but the job becomes more challenging when there are several in series. A manufacturer of portable industrial tools needed to accurately measure 10 separate cell voltages as well as temperatures at 36 points in a battery pack. With a little external signal conditioning help, Data Translation’s 48-channel TEMPpoint™ instrument was used.

The 36 thermocouple inputs were no problem. The instrument accepts a mix of most kinds of thermocouples, and each channel has a 24-b ADC and a dedicated cold-junction compensation circuit.

However, when used to measure voltage, the input range is limited to ±1.25 V. The approximate 4-V output from each of the cells in the battery pack had to be reduced by a 4:1 resistive divider. The TEMPpoint instrument’s typical 5-M? input impedance allowed relatively high-resistance dividers to be used, drawing little battery current.

Another consideration was the common-mode voltage that increases as you progress up the battery stack. It’s fine that after voltage division each cell presents only a 1-V signal differentially, but at the upper-most cell, the common-mode voltage is about 36.5 V. This proved to be well within the TEMPpoint’s ±500-V common-mode voltage rating.

A related specification, channel-to-channel crosstalk, concerns the effect of one channel on another. In this case, what effect would the 36.5-V input have on the channel measuring the cell at the bottom of the stack or on the channel physically closest inside the TEMPpoint instrument?

This specification is not listed although the channel-to-channel isolation is given as ±500 V. This means that any two channels could have input voltages that were different by up to 500 V without damaging the instrument.

Nevertheless, although the crosstalk no doubt is small, you have to assume there is none. Typically, in a low-bandwidth instrument, crosstalk is very good to the point of being difficult to measure. It’s usually at high signal frequencies that it increases and degrades accuracy.

The instrument’s ±7.5-ppm gain accuracy and ±10-µV zero offset error ensure meaningful measurements. In this case, a 1-V signal could measure between 0.9999825 and 1.0000175, resulting in a 4-V cell voltage uncertainty of about ±70 µV assuming the voltage divider ratio is exactly 4:1.

Channel gain has a temperature dependence of ±5 ppm/°C, so the measurement uncertainty could be affected if the battery-pack temperature changed significantly during a load discharge test, for example. Nevertheless, the measurement error caused by the TEMPpoint instrument is likely to be less than the IR drops in interconnecting wiring or the tolerance of the 4:1 dividers.

Once the measurements have been made, the data must be transferred to a PC. Here, the unit’s 500-V isolation to the host computer ensures that the PC isn’t going to be affected by high common-mode input voltages or possible ground loops.

A Driving School Teaching Aid
Most of us can drive a car, but seldom do we deliberately and in a controlled manner operate one close to its limits. Yet, training law enforcement officials, bodyguards, and racecar drivers to do precisely this is the job of Advanced Driving & Security in Rhode Island. To measure the advancements made by students, the company needed some way to record and assess performance that also could provide immediate driver feedback (Figure 2).

Figure 2. Driving School Multiple-Information DisplayCourtesy of Dewetron and ADSI

Dewetron provided a complete turnkey solution that combined the 16-channel DEWE-201 Dynamic Datalogger with the GPS-based DEWE-VGPS-200C Speed and Displacement Sensor; a video camera; an accelerometer for each of the three axes; and a MEMS-based gyro to determine roll, pitch, and yaw. The DEWE-201 unit also has two optional high-speed CAN interfaces that allow direct acquisition of vehicle parameters such as wheel speeds, pedal positions, rpm, and cooling temperatures.

The key factor to making sense of all this data is the capability to record it synchronously. Grant Smith, the company president, said, “Synchronous recording of analog, digital, video, GPS, CAN, serial, PCM, and Ethernet data to a single file and displaying all these disparate data types together are central to our data acquisition systems. It’s not good enough to simply record analog sensor data. Today’s customers demand that many different streams of digital data also be recorded in sync with the analog data or sometimes entirely in place of analog data.”

You may be wondering what special attributes qualify a data acquisition system to be described as a dynamic datalogger. One characteristic is lots of data storage. The DEWE-201 has a 4-GB flash disk memory as standard and an optional 30-GB hard drive.

Also, Dewetron distinguishes between faster, dynamic signal types and much slower quasistatic things like temperature. Because of the system architecture, you can mix the kinds of interfaces needed depending on the types of signal sources you have. The company also supplies cameras, accelerometers, and solid-state gyros used in this application.

Other considerations that make the DEWE-201 a good solution include its 11.2″ x 8.6″ x 3.5″ size, low 8.4-lb weight, and 6 to 24-V DC power requirement. Also particularly important for in-vehicle data recording, the unit has been tested to MIL-STD-810F for both shock and vibration. Finally, because the integral microprocessor system runs Windows XP Professional and supports USB and Ethernet interfaces, it’s straightforward to export the data file as a movie that can be played on another computer.

Rain Forest Monitoring
Rain forests play a major role in CO₂ absorption and oxygen production, affecting both local and global climate. This activity has been known for many years but still is not well understood. A research project that will measure the exchange of CO₂ and other materials between the forest floor and the atmosphere hopes to change that situation.

The work is being done at La Selva Biological Station in the Costa Rican rain forest, an area that averages 13 ft of rainfall per year. To acquire the necessary temperature, CO², humidity, 3-D wind movement, heat flux, solar radiation, and photosynthesis active radiation information, a wireless measurement network was developed based on National Instruments’ (NI) LabVIEW and CompactRIO hardware.

A robotic sensor system is suspended on a high horizontal cable and traverses through the forest. Initial experiments involved taking measurements for 30 s at each 1-m increment along a 60-m distance (Figure 3). Similar wireless sensors were deployed on the forest floor. NI’s Compact FieldPoint Network Interface was used for distributed wireless measurements and the company’s WAP-3701 Wireless Access Point transferring data between the sensors and the forest floor.

In this application, LabVIEW allows flexible measurement configuration, channel selection, and scaling. In addition, it provides advanced analysis tools for real-time embedded processing. With LabVIEW’s Web capabilities, data access can be extended to researchers worldwide via a Web browser.

Lightning Detection
Lightning detection involves an array of eight antennas that detect the RF burst caused by a lightning strike. The antennas are positioned around the perimeter of a 10-km circle, and it’s the job of a data acquisition system to record the detected burst activity with sufficient timing accuracy to determine the location of the strike: A 10-ns error corresponds to about a 3-m position uncertainty.

The digitizer resolution must be 12-b or 14-b because lightning can have a large dynamic range. A 100-MS/s sampling rate meets the 10-ns error criterion. Although a strike can be acquired within about 100 µs, another could occur with little delay. As a result, the system needs to run continuously. Temporary on-board storage could not be used as often is done to transfer bursts of data to the host PC.

Instead, GaGe implemented a special peak detection algorithm in firmware on the digitizer’s FPGA. Four Model CS14200 CompuScopes were used in master/slave mode for timing synchronization, and each one ran the algorithm in a customized version of the system FPGA.

Five parameters are determined: the strike peak positive and negative values and times of occurrence as well as the trigger event identification. The data is grouped as a set of information and requires only 216 B. This is in comparison to the raw data volume of 160 kB. Obviously, the data transfer requirement has been reduced by a factor of about 800, allowing the basic PCI bus to be used. The result is near continuous operation, with rearming within 1 µs after each event.

Trends

FPGAs
The lightning example highlights the growing importance of on-board signal processing. Dr. Andrew Dawson, GaGe’s business development manager, elaborated, “The latest innovations incorporated into the GaGe CompuScope Digitizers relate to expanded use of the on-board FPGAs. Although we have offered on-board data reduction for a few years, we’ve now added an FFT option.

“The user acquires waveform data as usual. But once captured, the data is processed by the FFT algorithm in the FPGA,” he explained. “The result is provided as integer numbers proportional to the real and imaginary frequency components. This approach significantly reduces the host PC processing load.”

Microstar Laboratories also uses FPGAs in its products. Neil Fenichel, the company’s president, said, “Advances in FPGAs, serial ADCs, digital isolators, and power supply ICs have made possible a new generation of isolated test and measurement components with improved capabilities at relatively low cost. The logic designs of the isolated boards contain a common core of VHDL code for a Xilinx FPGA,” he continued. “This code is adjusted for the unique input/output structure of each product and then stored in an on-board flash memory. The VHDL code can be customized for unique applications.”

Continuing the FPGA theme, United Electronic Industries’ Director of Marketing Bob Judd commented, “We’re putting more and more functionality into FPGAs. They have become powerful and simple enough to program that we’re putting increasing amounts of real-time math into them. For example, our new DNx-AI-254 RVDT/LVDT Interface and the DNx-AI-255 Synchro/Resolver Interface both have the actual decoding performed real-time in on-board FPGAs. This greatly reduces the burden on the host CPU, allowing it to be used to run expanded application code.”

Isolation
As some of the application examples highlighted and Microstar’s Mr. Fenichel reinforced, isolation can be an important factor. “Data acquisition technology is diffusing from its core of high-tech users to new users that require robust, compact packaging with protection from harsh environments. One of our customers makes small measurement and control systems that are used in the field and subject to transient voltages, sometimes induced by lightning. Isolation is a major concern that our new series of isolated boards addresses,” he explained.

Galvanic isolation, often shortened to just isolation, eliminates a direct ohmic electrical path between two circuits. A transformer or opto-isolator can be used to couple a signal between two circuits without having a direct ohmic connection. When a sensor and the acquisition system to which it’s connected are referenced to different ground voltages, isolation is necessary.

Typically, isolation also minimizes the amount of electrical noise that will be coupled into the measuring instrument by breaking any possible ground current path. For low-frequency applications, isolation usually is accompanied by low crosstalk between channels. At higher frequencies, low crosstalk is much more difficult to achieve.

At the output of a data acquisition system, galvanic isolation allows a computer bus to be connected without corrupting the data or damaging either the computer or the acquisition system. If the two devices are a large distance apart with the possibility of different ground potentials, isolation breaks that direct ohmic connection. Digital data amplitude variations caused by crosstalk usually aren’t a problem, but timing jitter caused by one signal interfering with another can be.

Isolation is familiar to industrial users of data acquisition systems. However, as Mr. Fenichel commented, data acquisition systems are being used for a wide variety of applications. New users understand the kinds of tests they need to run but may not fully appreciate the need for isolation.

Flexible Configuration
Also commenting on the changing nature of data acquisition systems, Sheri DeTomasi, an Agilent Technologies product manager, said, “Customers need to perform more distributed data acquisition with the computer controlling multiple instruments in multiple locations. For example, they may be monitoring temperature in many locations in a building or monitoring data on a railway system for collision avoidance. LXI instruments support widely distributed test systems and often can be placed close to the signal source, allowing shorter connecting wires and typically less noise,” she explained.

Flexibility also is a major benefit of KineticSystems’ data acquisition systems. According to Steve Krebs, director of engineering, “The DAQ Director™ software the company developed to operate its data acquisition hardware was of particular interest to Boeing for a wind-tunnel test application. Because the software supported an unlimited number of user-defined configurations and required no programming, it helped Boeing facilitate test setup while reducing the need for specialized skills.”

Post-acquisition processing also is a facility that needs to be wide-ranging and easy to use, and this was discussed by Ms. DeTomasi. “Agilent’s Model 34832A BenchLink Data Logger Pro Software helps users create application software with limit checking and decision making without writing code. For example, you can set rules so that only every nth measurement is saved or that only out-of-limit data is saved.”

Summary

Many data acquisition systems will handle most applications. To the degree that you may need something unusual such as very high speed, hundreds of channels, or a wireless system qualified to extreme levels of shock and vibration, the selection process becomes easier because there are fewer products to choose among.

However, as the separate application discussions have demonstrated, even seemingly straightforward jobs may be hiding the need for isolation or a high common-mode voltage rating. The key to choosing the best solution is to thoroughly understand the full implications of the signal types being measured and the tests to be performed.

This is a process that can and should be undertaken rigorously. Start by splitting the task into input, output, data analysis, and physical environment requirements.

Although only four requirements have been suggested, the number you prepare and their contents will depend on your experience and emphasis. So, the final part of the job is to design sets of tests that will both acquire the data you need as well as verify the data acquisition system’s performance.

October 2008