Data acquisition quality relies upon how closely the representation of reality matches what’s being measured. Drivers who are high insurance risks may explain that the only measurement that matters is the instant when a police radar gun measures their speed. Reality, however, may differ significantly depending on the point of measurement and how variable their driving process.

A sampling rate that’s too infrequent (see graphic) may not represent the process accurately enough for sound decision-making. Elliott Middleton, Wonderware Historian product manager, says many processes can change significantly between sample times. For a more accurate representation [of automobile speed], “You’d probably need to sample every 5 seconds or so. However, if we were, instead, recording your speed on I-10 traveling from Palm Dessert to Blythe, sampling every 2 minutes might be adequate—the difference is the variability in the data and the level of accuracy (and number of samples) you need.”

According to Brian Betts, group manager for data acquisition marketing at National Instruments, the five elements of a basic data acquisition (DAQ) system are transducers and sensors; signals; signal conditioning; DAQ hardware; and driver and application software.

DAQ can be do-it-yourself, rely on purchased or integrated systems, or be contracted through a laboratory or service. Rockwell Automation, for instance, is among vendors with a testing and certification lab that performs DAQ testing and documents results. Rockwell Automation says that type of high end system, calibrated by standards traceable to National Institute of Standards and Technology (NIST), also has been inspected and approved by UL and CSA for test requirements.

If the graph measured a car's speed on surface streets every two minutes, an inaccurate representation would result, no matter how good the sensors, network, and software delivering the analysis.

Measure the right thing

While chapters and books provide DAQ advice, doing it right involves ensuring 1) that what’s being measured represents the process, 2) signal integrity, and 3) accurate analysis.

Most common DAQ errors, according to John Lehman, engineering manager, Dataforth Corp., are capacitive coupling, magnetic coupling, ground loops, over-voltage and transients, electromagnetic interference/radio frequency interference (EMI/RFI); and aliasing. Sources of signal interference vary.

“In today’s dynamic industrial environments,” Lehman says, “signal and power wiring, electronic devices, and other electrical plant/process equipment often interact to create noise that can degrade critical measurement and control signals.”

Signal conditioning technologies can help. When signals are too faint or dangerous to measure directly, signal conditioning can amplify, attenuate, isolate, excite the sensor, multiplex the signal, and offer other functions, such as bridge completion and simultaneous sampling, says Betts of NI.

After (or as) information is acquired, various software programs can analyze and keep records, often for regulatory purposes, adds Keith Jones, Wonderware marketing program manager for HMI, supervisory, SCADA and platforms. How much and how fast? Systems vary in bandwidth and speed by need.

Recent DAQ software from Yokogowa, for instance, records from as many as 32 recorders, data loggers, controllers, measuring instruments, or other devices, enabling acquisition of data on up to 1,600 channels at intervals as short as one second, says Todd Stanier, network solutions technical marketing, Yokogawa Corporation of America. Others can record more than a thousand samples per second. Acquired data can then be used to make a decision, alert an operator, or actuate a change in process, as desired.

ONLINE

Control Engineering provide more about data acquisition.

Have a DAQ question? Post it to the “Ask Charlie” blog.

Do you keep data forever? Here's insight about "Data deletion, or letting go is hard to do."