Today’s market is full of data acquisition components and accessories. This abundance is prompted by the huge variety of products needed to support the thousands of applications, signal types and ranges that are possible. With so many choices, configuring a data acquisition system (DAS) can be a real challenge.
How do you know what you really need for your specific requirements? How can you avoid adding extra cost to your system? Admittedly, you’re going to have to do some research. Also, you will really need to know exactly what your requirements are, and spend some time talking with the technical support personnel from the company whose equipment you choose.
To get you started, we have compiled a list of some of the most commonly asked technical questions about configuring a DAS. This should help you focus on what your system will require before you begin talking to vendors.
1. How do I get my captured signals into my PC?
This isn’t as simple as merely connecting wires from your sensors into your PC. Let’s assume that you’ll be using a data acquisition board installed in your PC. At a minimum, you will still need a termination panel to connect your sensor wires to and a ribbon cable to connect the panel to the data acquisition board.
Signal conditioning is an important part of many systems. Signal conditioning converts signals generated by your sensors or transducers into voltages that a data acquisition board can digitize.
Data acquisition boards typically accept voltages from ± 10 mV to ± 10 V. Getting signals into this range can require amplification, current-to-voltage or resistance-to-voltage conversion, and voltage division. Signal conditioning may also provide isolation, which protects your computer and data acquisition equipment from spurious and potentially damaging signals.
An extensive variety of industry-standard signal-conditioning modules converts virtually any signal of interest. They plug directly into termination panels designed for these modules (Figure 1). The types of signals you are measuring and the types of transducers you are using determine what signal conditioning you need.
2. How far can my PC be from my signal source?
As you extend the distance between your PC and your signal source, you increase the level of noise introduced to the signal you are measuring. The answer to this question, then, depends on many factors. Signal-source type, signal level, cable type, noise source type(s), noise intensity, distance between the cable and the noise source(s), noise frequency, signal-frequency range and required accuracy are just some of the variables to consider.
Noise is most likely to affect low-level voltage signals. It can also be very disruptive to high-speed measurements. As a rule, you can generally extend the distance between your PC and your signal source if you are measuring analog current source signals or periodic high-level analog voltage source signals, or if you’re using shielded or twisted-pair cable.
The distance between your sensor or transducer and your termination panel will depend on the sensor you are using and its output. For example, some sensors output a 4- to 20-mA current, allowing you to place it a considerable distance from your termination panel.
As a rule of thumb, the distance between your PC and your termination panel should not exceed 6 ft in a typical system. If you need to exceed this distance, you can probably do so without worry if you are measuring PC-level voltages at a slow sampling rate in a relatively non-noisy environment. Otherwise, you may want to consider using a two-wire transmitter/receiver configuration. In this setup, a two-wire transmitter at the signal-source end converts your signal to a current. Then you can transmit this current across a noisy environment without contaminating your data.
The signal is transmitted to a receiver located near your DAS, where it is converted back to a voltage. Remember that a DC power supply is required to transmit the signal, so the two-wire transmitter/receiver configuration will add cost to your system.
3. Should I use a 12- or 16-bit board? Is 16-bit resolution worth the cost?
The resolution of the analog-to-digital converter (ADC) on your data acquisition hardware determines how closely the digitized waveform your PC sees resembles the analog waveform that’s going into your PC. Generally, the most common resolution used on data acquisition boards is 12 bits, or one part in 4,096. This is more than sufficient for most applications.
For applications which require more information from the measured signal, 16-bit (one part in 65,536) boards usually provide the necessary resolution. Audio applications needing wide dynamic range measurements represent typical 16-bit applications.
An important factor when considering 16-bit resolution for your application is the degradation of your measured signal caused by your transducer or signal conditioner. Most industry-standard signal-conditioning modules are only linear to 10 bits, and many transducers are only accurate to 12 or fewer bits.
For example, most thermocouples are only rated to 0.1% accuracy, which is equivalent to about 10-bit accuracy. If your measured signal is only accurate to 12 or fewer bits, then using a 16-bit ADC is needless overdesign.
For low-speed applications, you can statistically increase the accuracy of your measurements by using software to average a number of measurements of the desired signal.
4. Should I use DMA for my application?
Direct memory access (DMA) is a popular technique used to enhance system throughput on an ISA or EISA computer. It allows you to store samples directly into memory without any intervention from the CPU, freeing the PC to perform other tasks. During the process, the data acquisition board communicates directly with the bus using a special set of hardware handshake signals.
DMA is ideal for a number of data acquisition tasks. Any time you capture a stream of data, either to PC memory or to disk, DMA will ensure a more accurate, gap-free transfer.
DMA is also necessary for providing accurate hardware timing when performing periodic sampling. This accurate time base is required for many digital signal processing applications. However, if your application only requires low-speed sampling and/or control, you probably don’t need DMA. Most environmental monitoring applications fall into this category.
Incorporating DMA can be tricky if you are writing your own software to control your application. If you are using a data acquisition software package that supports DMA, then implementing it is an easy and recommended technique.
5. Do I need isolation? Or is protection sufficient?
One important aspect to consider in almost any data acquisition application is protecting your PC from dangerous voltages which could come in contact with your signal path. Overvoltage protection up to ± 35 V is built into most data acquisition boards.
Figure 2 illustrates two examples of such overvoltage protection circuits. This type of solution is probably sufficient for systems that don’t involve power measurements or for situations where sensors are placed on machinery that could incur high-voltage transients. However, it won’t protect you from some of the higher process voltages that can exist in control rooms.
You can easily add isolation to your system in the form of isolated signal-conditioning modules. Analog modules allow you to safely connect common-mode voltages up to 1,500 Vrms and up to 240 V differential to your system. Digital modules protect your equipment up to 4,000 V input to output. As you would expect, though, isolation will add some cost to your system.
Isolation also lets you accurately measure signals when a floating ground exists. This relatively common condition exists when there is not a common ground between your PC and your signal source. For example, if you’re taking measurements from equipment that is powered off a different circuit, there is probably a substantial potential between the equipment and your PC.
6. How do I reduce noise on analog input signal measurements?
Any time you measure a signal, it contains unwanted noise. Whether this noise is troublesome depends on the signal-to-noise ratio and the specific application. If the transducer you are using acts like a current source, your system is inherently less sensitive to external noise sources than if you are using a voltage-driven device. Otherwise, the noise voltage signal is added directly to the source voltage signal without attenuation.
Most noise problems can be solved by applying proper grounding and shielding principles. A common mistake is to assume that your ground line and return line are the same. The ground line connects your equipment to earth and does not normally carry current. The return line is an active part of your circuit and should not have more than one connection to your ground line. The return path should also have as low an impedance as possible to reduce the effects of EMI.
Using appropriate interconnections is essential. Balanced cables are indispensable to rejecting ground difference potential. Shielded or twisted-pair wire is suggested when measuring low-level signals. With twisted-pair cable, EMI is induced equally into both wires, effectively canceling out any error in your measured signal.
You may also have to use filtering to remove noise from your signal. AC power lines and radio frequencies may cause noise in your signal that cannot be removed by grounding and shielding techniques.
Filtering can be implemented in two ways. A hardware filter consisting of a combination of resistors, capacitors, inductors and amplifiers can remove unwanted noise from your signal prior to A/D conversion. Figure 3 illustrates two examples of such filters.
Filtering can also be accomplished with software after the conversion, using digital signal processing to perform averaging. By averaging a series of incoming data points, the signal-to-noise ratio is effectively increased.
7. Will my DAS have an intrusive effect on the rest of my system? Is it OK to leave the DAS connected to my system when powered down?
By its very nature, a DAS will have an intrusive effect on the system you are monitoring or analyzing. To measure a signal, the DAS must add a load to the system. When the DAS is powered down, it still provides a path to ground. To determine if you can safely leave your DAS connected after you’ve powered it down, consider the input impedance and capacitive loading of your data acquisition board (or signal conditioning, if used) during both power-on and power-off conditions. The input impedance will be significantly lower on most data acquisition boards during power-off due to built-in overvoltage protection. As a result, your DAS is likely to have a more intrusive effect on your system when powered down than when you’re actually monitoring the system. Specific information is available from your vendor.
8. Why do I need to linearize my thermocouple signals? How should I do it?
Thermocouples are, by far, the most common devices used to measure temperature. Rugged and inexpensive, they are ideal for applications in both industry and science. Consisting of two wires of dissimilar metals joined together, the thermocouple produces a voltage dependent on the temperature of the metals (the Seebeck effect).
Different types of thermocouples are available for different temperature ranges. However, there is one problem with thermocouples: The voltage they produce is not linear with respect to temperature. This is why thermocouple measurements must be linearized.
Fortunately, this is easily done through either hardware signal conditioning or software. Signal-conditioning modules that perform linearization in addition to isolation and cold-junction compensation are readily available.
You can also use software to perform linearization. The method you choose is completely arbitrary. It is generally easier to accomplish it with hardware signal conditioning; however, software linearization is less expensive and more accurate.
9. What are the pros and cons of using a visual software package instead of writing my own software using a high-level language?
Turnkey application packages that perform a specific task, such as strip-chart recording or oscilloscope simulation, are widely available. Many software packages let you design your own applications using graphical tools.
Most of these packages make it easy to design your own run-time and user-interface screens, such as the one created by Visual Designer in Figure 4. These packages are much more flexible than turnkey packages, making them suitable for a larger variety of projects.
If you are an experienced programmer, there are some advantages to writing your own code to control your application. This approach will make your program smaller and more specialized, incorporating only the features you need. Program control will probably be faster, an important consideration if real-time data is critical.
If your primary concern is cost, investing a little extra for graphical software tools will probably be worth it in the long run. No matter what your level of programming expertise, using a visual software package will dramatically reduce your development time and virtually eliminate the sometimes lengthy debugging process. What’s more, these packages make it much easier to document, maintain and modify your system.
Many vendors will provide free demo disks so you can see if the software will help you; some vendors even provide free evaluation copies, so you can actually design your application before you purchase the software. If you have unique requirements, most vendors provide a toolkit so you can add your own functions to their software.
Conclusion
As you can see, there are many issues to consider when putting together your system. But don’t let that intimidate you. Most data acquisition vendors offer free technical support before and after you buy.
About the Author
Robert M. Winkler, who joined Intelligent Instrumentation in 1994, is the company’s Product Marketing Engineer. He is a graduate of Lehigh University with a B.S.E.E. degree. Intelligent Instrumentation, 6550 S. Bay Colony Dr., MS130, Tucson, AZ 85706, (520) 295-2047.
Robert M. Winkler, who joined Intelligent Instrumentation in 1994, is the company’s Product Marketing Engineer. He is a graduate of Lehigh University with a B.S.E.E. degree. Intelligent Instrumentation, 6550 S. Bay Colony Dr., MS130, Tucson, AZ 85706, (520) 295-2047.
Copyright 1996 Nelson Publishing Inc.
January 1996
|
|