Recorders and dataloggers are the Mason jars of the instrument world, preserving carefully prepared data for future use. At least that’s how they have often been used in the past. Today, there are many test and measurement products that address data acquisition and archiving, and it can be difficult to make the best choice.
For example, why would you use a stand-alone recorder or datalogger when a PC-based data acquisition system can do the job so much better? If you regularly use more traditional instruments you may ask for what applications are PC-based systems “so much better?” Why shouldn’t you use a recorder or datalogger if it can deliver the performance you require at an affordable price?
There are several reasons to use PC-based systems: They analyze as well as acquire data, run report-generation software, interface to your company’s LAN, and consequently, save you time by providing answers in the form you need. Many young PC-weaned engineers wouldn’t consider using anything else. Have they overestimated the universal applicability of the ubiquitous PC? What are they missing?
Traditional Instrumentation
The traditional recorder/datalogger is a stand-alone, portable instrument with an operator interface including features such as buttons, displays, and a recording medium. Classic examples are strip-chart and circular-chart recorders and sampling voltmeters. Traditional strongholds for stand-alone recorders and dataloggers will continue to be those applications where cost is not the primary issue but ease-of-use and reliability are important.
According to Mike Bayda, a data acquisition marketing engineer at Keithley Instruments, “In these applications, operators are interested in quick setup and operation but can’t afford the time or real estate needed to deal with a PC, monitor, keyboard, mouse, or other peripherals that a computer imposes. Economics also may favor the situation where an operator can simply throw a switch to start datalogging.
“Medical applications are good examples,” he continued. “In a typical hospital, we find brain-wave monitors, EKG recorders, heartbeat dataloggers, and prenatal monitors all based on the strip-chart concept. Also, it is easier to certify a self-contained instrument for medical use rather than a plug-in data acquisition board, separate computer, and related equipment.”
Environmental conditions may influence your choice of instrument type. In a well-controlled laboratory or office, PC-based data acquisition is a good selection. Space probably is not a problem; power is readily available; and shock, vibration, and weatherproofing are not relevant.
“Stand-alone recorders and dataloggers are best suited for applications that occur outside the office or lab environment,” said Tom DeSantis, president of IOtech. “Because of their hard drives and display mechanisms, PC-based systems are limited to test environments that are not extreme or harsh.”
Making Up Your Mind
A good place to start when selecting a recorder or datalogger is signal conditioning. Voltage, temperature, pressure, acceleration, high voltage, flow, velocity, current, strain, frequency, force, displacement, phase, and vibration are examples of common signal types. Unless you have the expertise and time required to develop your electronics, you may need to reconsider the experiment you are planning so that the measured parameter corresponds to a readily available signal conditioner.
In the case of a stand-alone instrument, this means that you are limited to signal-conditioning modules made by the instrument manufacturer. Also, you may be subject to implied mechanical constraints.
For example, some manufacturers have reduced the range of specialized functionality available in newer models of recorders and dataloggers. The signal-conditioning function you need still is available, but because it is an older model, the function doesn’t fit into the new instrument. If you have a portable application, your system just grew an awkward bulge.
Many recorders and dataloggers offer so-called universal inputs, but the term really means that you don’t need additional signal conditioning as long as you have voltage, temperature, current, or resistance inputs. Anything else, such as charge or torque, and you will require a signal conditioner.
Of course, you also may have a problem finding a PC board that handles odd mixes of I/O channels, analog and digital signals, and specific signal processing. On the other hand, the range of suppliers is very large. The boards only need to cater for your signal-conditioning needs and meet the ISA or PCI specifications.
A second consideration is data use. Why are you recording/logging the data, and what is ultimately going to happen to it? Who is going to use it, how often, in what formats, and where? The answers to these questions help determine whether you need hard-copy output, a color display, a large display, or no display. Also, the answers affect the need for internal formatting capability as well as signal processing such as fast Fourier transform (FFT).
The comments about strip-chart recorders from Keithley Instruments’ Mr. Bayda reflect the need for an immediate, permanent copy of a patient’s EKG at the bedside. It also is necessary to record the signals electronically and archive them to a central hospital database. EKGs and other routine test results can be interpreted automatically by specialized medical software. Generally, the patient’s doctor and his cardiologist will review the program’s suggested prognosis, and they may each do so independently by accessing the database from their own offices.
As in this case, if the answer to the data-use question involves remote locations, Ethernet and Internet connectivity may be required, sometimes including the capability for instrument control. Although more complex technically, many products now offer full remote control and data access.
Performance Limitations
Not too many years ago, direct-writing recorders were judged on horsepower numbers such as chart speed and pen-tip amplitude response. Today, many applications have adopted thermal-array recording as well as high-speed memory systems. At speeds above a few hundred millimeters/second in traditional instruments, the chart transport mechanism soon overtakes the rest of the recorder in terms of power and size.
So although some recorders do offer speeds as high as 100 or 200 mm/s, many don’t, especially smaller, portable units. Instead, data recorded to high-speed memory is replayed to paper at a much lower transport speed. The equivalent chart speed can be thousands of millimeters/second, but the paper never moves faster than 10 or 20 mm/s. So-called memory recorders accomplish both higher frequency data acquisition and paper-speed down-conversion.
Direct-writing pen recorders are limited to a 50- to 100-Hz bandwidth. Paper speed restricts the bandwidth of thermal-array direct-writing recorders. Their bandwidths are higher, but you only see a solid band unless the paper moves very fast. Only the bandwidth of their input amplifiers and the analog-to-digital converter (ADC) sampling speed limit memory recorder bandwidth.
However, huge amounts of memory can be involved because applications may require recording for several minutes. A typical 100-kS/s sample rate with 16 channels translates into a 3.2-MB/s data rate, assuming 2 bytes per sample. That’s 180 MB/min. It’s not surprising, then, that most memory recorders offering long recording times use one or more hard disk drives.
As an extreme example, the StreamStor™ PCI-816 Disk Recorder from Boulder Instruments has an array of up to 16 hard disk drives to provide a 100-MB/s sustained data rate and 1,200 GB of storage capacity. Most other products use fewer drives, such as Nicolet Instrument Technologies’ 32-channel Odyssey with four drives and 36 GB of storage. The entry-level models of these recorders have only one or two disk drives.
It does not always follow that a high sample rate combined with a long recording time results in large memory requirements. Memory recorders often use their high-speed capabilities to support transient capture for a relatively short period of time while allowing continuous recording at a much lower paper speed. This class of instrument represents a lower cost solution for applications that don’t require high-speed sampling all the time.
Features Cross Instrument-Type Boundaries
The recorder has become an entirely electronic product to the extent that manufacturers offer hard-copy output as an option. Dataloggers also have become wholly electronic, but they never did have such a distinctively mechanical nature as recorders. The main difference between the two breeds remains storage of measurement results in a datalogger rather than raw data in a recorder. But the distinction is disappearing.
Memory recorders are available with on-the-fly post-acquisition mathematical capabilities. Sometimes this is termed computed traces or channel mathematics. It is the capability to perform computations involving one or more channels of acquired data in such a short time that the result can be plotted along with the raw data. Any small time skew encountered can be compensated by delaying the raw data one sample period, although generally calculations are completed within a single sample quantization period.
Single-channel examples include rms voltage or current, filtering, and averaging. Rather than use a special-purpose hardware signal conditioner, why not just compute the result on a sample-by-sample basis? Instantaneous power is a good example of a two-channel function. Of course, you may need to sum and scale several channels, each with separate weighting and filtering time constants. Some instruments offer this level of custom functionality via menu control similar to the wide range of options available in Excel. Others simply present a list of predefined choices.
A separate class of instrument is the oxymoronic paperless recorder. The name may sound like a euphemism for a data acquisition system, and it can be, but paperless recorders usually have a video display associated with them. Often used to replace circular paper charts in process-control industries, paperless recorders are memory recorders that emphasize connectivity and control simplicity. Acquired data can be made available immediately on the manufacturing LAN and used as required. The local display is useful for an operator but not needed in some applications.
Where the display is used, it may present a graphical user interface (GUI). This method of instrument control is another change from traditional knobs and buttons. Whether or not a GUI is objectionable tends to be a function of the type of industry using the device and another variable you need to keep in mind.
What about data-analysis software? Many data analysis programs are available, but only a few deal with very large data records. This distinction becomes important for anything over a megabyte and essential for truly large files. Things to look for include search capabilities, file appending and concatenation, and a generous assortment of data formats. An example is the DataViewer playback and analysis software available from Gould Instrument Systems. The company claims that the software allows you to “view megabytes of data in seconds.”
When paper records were popular, recorder users became adept at scanning many feet of paper to discover anomalies in their traces. Finding the exceptions within a long data file is too difficult to do manually, given the very small viewing window one PC-screen’s worth represents out of hundreds or thousands of screens. But, it is an ideal job for the PC to do automatically. The problem is one of defining what to look for and having sufficiently fast data search routines and hardware so the process doesn’t become an excuse for a long coffee break.
Recorders and dataloggers have indeed come a long way. Today, many more products are available than ever before to solve a particular application problem. On the other hand, you also have more complex choices to make.
The best advice is to understand your problem thoroughly before you worry about its solution. Signal conditioning is key to acquiring good data regardless of bits, bytes, numbers of hard disks, and type of display. If the analog signals that you started with have become corrupted or distorted, nothing will restore them.
However, once you cross the great analog/digital divide, the sky’s the limit. Recorders and dataloggers have borrowed from and adapted computer technology to the extent that the names of instrument types really have lost most of their meaning.
Data acquisition systems tend not to emphasize display and seldom have integral hard-copy output. But, many have PC software that supports trace display and further signal processing.
Recorders traditionally supplied hard-copy output, but many don’t today. They are encroaching into data acquisition markets with very long data-file handling capabilities. Dataloggers acquire the results of measurements, but so can data acquisition systems and recorders. Typically, dataloggers are less complex and, as a result, lower cost. If you understand your application thoroughly, you’re almost certain to find a good solution regardless of its name.
Published by EE-Evaluation Engineering
All contents © 2000 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.
October 2000