Using PCs for machine condition monitoring: Part 3

Questions to ask when selecting your data acquisition system

By Plant Engineering Staff September 10, 2004

Armando Valim, Chris DeFilippo, and Todd Dobberstein, National Instruments, Austin, TX

(Editor’s note: This is the third article in a series on PC-based machine monitoring. The next article in the series will appear in the next issue of Plant Engineering HotWire.)

Surprise! The analog-to-digital (A/D) convertervendors provide a variety of tools for creating control and monitoring systems for plant applications. The good news is that you can use many different combinations of hardware and software, depending on the requirements of your system. The bad news is you have to choose from a gigantic list.

This article lists 10 questions that you should ask in order to select the best hardware and software for your plant measurement system.

1. Which data acquisition platform — PC, PLC, and/or PAC — fits your requirements?

In the last decade, industry experts have predicted that PC-based control would end the PLC’s reign in industrial control. They predicted that features, such as floating-point processors, RAM, powerful software tools, and graphical interfaces would make the PC the ultimate industrial automation platform. They were partially right. An increasing number of engineers are investigating the use of PCs for advanced functionality, such as analog control and simulation, database connectivity, web-based applications, and communication with third party devices. However, the PC faces difficulty competing with the PLC for control-based applications.

Standard PCs, and even some industrial computers, cannot provide the reliability demanded by industrial automation control applications. PCs with a standard operating system and off-the-shelf hardware are too fragile and temperamental to deliver the reliability demanded in embedded industrial control.

Traditionally, engineers have had to choose either a PLC for industrial ruggedness and reliability, or a PC for analog measurement and communications. As a result, many engineers sacrifice advanced control functionality that they cannot easily accomplish with a PLC, or they cobble together a system that includes a PLC for discrete control and a PC for more advanced functionality. This is why in many plants, PLCs are being used in conjunction with PCs for logging data, connecting to bar code scanners, inserting information into databases, and publishing data to the web. This configuration presents major problems because these systems often are difficult to construct, troubleshoot, and maintain. Therefore, system engineers struggle to incorporate hardware and software from multiple vendors that was not designed to work together.

Today, there is a third option. Engineers now can use products that provide a hybrid of the PC and the PLC. Industry analyst group ARC has termed this new class of hybrid controllers “programmable automation controllers” (PACs). PACs combine the best features of the PC, including the processor, RAM, and powerful software, with the reliability, ruggedness, and distributed nature of the PLC (Fig. 1).

Fig. 1. PACs combine the packaging and ruggedness of a PLC with the software flexibility and functionality of a PC. These new platforms are ideal for sophisticated control and logging in rugged environments.

One example of the PAC platform is based on the PCI extensions for instrumentation (PXI) industry standard.plications requiring tens of channels. For higher channel-count applications, PXI chasses have up to 18 slots. Multiple chasses can even be daisy-chained for higher channel counts.

Timing and synchronization are critical, but often overlooked, aspects of a measurement platform. If your application requires tight timing relationships among signals or measurement synchronization tasks, PXI is an ideal platform. With PXI, you have access to eight bussed trigger lines that link all PXI slots so that measurement modules can interact, trigger, and control each other. The PXI backplane includes a star trigger line for very accurate timing with minimal skew, and a common 10-MHz clock to synchronize multiple modules. These technologies can dramatically increase test and manufacturing throughput. As an emerging measurement and automation platform, PXI is closing the gap — there are more than 500 different modules available for PXI and hundreds of companies developing CompactPCI peripherals compatible with PXI.

2. Which sensors are you using?

The types of sensors you choose depends on which signals you are measuring. For machine monitoring applications, common signals are acceleration, displacement, and sound pressure level (SPL). Common sensors to measure these parameters are accelerometers, proximity probes, and microphones, respectively. Temperature and pressure transducers are also very common in plant applications.

You should first know the type of sensor you intend to use, because this largely governs your choice of hardware. Some of the questions that your choice of sensor places on the hardware are as follows:

  • What kind of signal conditioning is required?

  • What is the frequency range of the sensor?

  • How much dynamic range does the sensor require?

    • For more information on sensor choices, refer to Part 2 of this series. Click here .

      3. What signal conditioning is needed?

      Before it is digitized by the data acquisition hardware, the signal from a sensor nearly always requires some type of conditioning, such as amplification, filtering, sensor excitation, and input configuration. Many vendors have hardware products that have built-in excitation for voltage mode or integrated electronic piezoelectric (IEPE) microphones and accelerometers. Another consideration is transducer electronic data sheet (TEDS) sensors. For more information on TEDS, refer to Part 2 of this series. Click here .

      4. What is your frequency range of interest?

      All sensors have a frequency range over which they are designed to operate. Your sensor should have a frequency range large enough to cover the frequency range of interest. Likewise, your digitizing hardware should have a large enough frequency range to cover these signals of interest. To prevent aliasing, look for products that come with antialiasing filters, which cut the maximum frequency range of the device to a little less than one-half the maximum sampling rate, as prescribed by the Nyquist sampling theorem.

      5. What is the required dynamic range?

      Dynamic range is a measure of how small you can measure a signal relative to the maximum input signal the device can measure. Expressed in decibels (dB), the dynamic range is:

      20 log (Vmax/Vmin)

      More than ever before, plant engineers can leverage commercial technology in their applications. In fact, the historical price of A/D converters continues to decrease, while new technologies surface as prime for plant applications. A 16-bit A/D is now priced lower than a 12-bit A/D of the early 1990s. As a result of trends like these, the industry has been able to leverage PAC devices to provide better DAQ performance and more I/O channels for an equal or better price as the years go by. A similar trend of increasing processor performance and decreasing price has occurred in the PC industry as well, and is known as Moore’s Law. This trend also benefits the new PAC platform by providing increased software performance. The lower cost A/D converter trend opens a new front for digitizers with up to 24 bits of resolution that deliver nearly 120-dB dynamic range, making machinery protection and fault prediction more accurate than ever before.

      Advanced filtering techniques and built-in signal conditioning are other hardware improvements making modular, PAC-based instruments the modern choice for machine condition monitoring. For example, a device that has an input range of

      6. How many analog input and output channels do you need?

      Plant applications require various numbers of analog inputs and outputs depending on the application. This is one of the key factors when selecting your system platform. Using PXI, you can expand your system to the number of channels of your choice by simply adding more devices.

      For example, using an 8-input channel device, you can fill an 18-slot PXI chassis with seven modules and achieve 112 analog input channels. In addition, you can synchronize multiple-chassis up to 5,000 channels. Some systems allow you to network hundreds of channels using low cost communications such as Ethernet.

      7. Do you need phase information (simultaneous sampling)?

      Simultaneous sampling is needed in applications where phase information between measurements of two separate channels is required. Phase information is used in applications that require a frequency response of an object or applications that use cross-channel measurements, such as axial correlations, orbital plots, or structural analysis. Simultaneous sampling is not required on very slow measurement types such as level or temperature.

      Multiple device synchronization for phase information across multiple channels is possible with several products. PXI enables synchronization across multiple modules by providing the PXI star trigger bus — a special trigger bus where all the trigger paths have the same physical length. This guarantees a phase mismatch of & 0.5 deg across the entire frequency range of the product between any two channels in the same chassis. Using the RTSI bus — a PCI synchronization scheme, you can synchronize the devices for PCI. In addition, some devices have built-in simultaneous sampling between channels on the same device, but cannot be synchronized between devices. The golden rule is to understand the need for synchronization — if synchronization is needed, ask and look for it; if it is not needed, don’t worry about it.

      8. How much processing and disk power do you need?

      Real-time operation or online operation for a measurement system is typically defined as the capability to perform all desired analysis, display, and storage tasks at least as quickly as the data are acquired. Some devices use an onboard digital signal processing (DSP) chip to guarantee real-time analysis under most conditions. Other measurement hardware devices rely on the host computer for processing tasks. The processing capabilities of modern desktop and PXI computers exceed that of most DSP chips.

      Gap-free processing means that all acquired data undergoes online analysis. In some applications gap-free analysis may not be required. For example, it is common for high-channel machine monitoring applications to log all data to disk, but to perform only rudimentary online analysis on a small subset of the data. Further analysis may be performed offline on the stored data as required.

      It is important to include enough computing and hard drive power in your measurement system to meet the processing and storage requirements of the application. A few pertinent suggestions are provided in the following list. Keep in mind that these examples assume no other requirements (for example, disk storage examples assume no online FFTs or other processing), and that they include only a minimal safety tolerance.

    • A 3.2-GHz desktop machine with a typical IDE drive can continuously stream to disk 40 channels at 102.4 kS/s (as raw binary data)

    • A 3.2-GHz desktop machine can continuously compute FFT power spectra on 32 channels running at 102.4 kS/s

    • A desktop server machine using the Intel E7505 chipset with a redundant array of inexpensive drives (RAID) system and multiple IDE drives can continuously stream to disk 72 channels at 102.4 kS/s

    • A dual processor machine will yield an improvement of about 80% for octave analysis and digital filtering operations as compared to a single processor running at the same clock rate.

      • However, dual processors do not offer a significant improvement for FFT analysis. In most cases, PLCs are not optimized for streaming or analysis applications

        9. Which analysis do you need?

        Your application may require analyses such as power spectrum, octave, or order analysis. Some data acquisition systems offer graphical programming environments that provide frequency measurements and analysis tools for various applications. Some add on software tools provide application-specific analysis.

        10. Can other parts of my system be integrated?

        You benefit from more advanced embedded control, data logging, and data sharing, as well as software that you can use in other applications throughout your system. One of the strong trends in the market is the integration among data acquisition, vision, motion, PLC, and SCADA.

        Armando Valim is a group manager responsible for sound and vibration products at National Instruments, Austin, TX. He holds a bachelors in engineering as well as graduate degrees in engineering, marketing, and finance. For more information about using PC-based machine monitoring, Armando can be reached at 512-683-5861 or armado.valim@ni.com .

        Chris DeFilippo is a sound and vibration product manager at National Instruments. His current projects include coordinating web development for sound and vibration products, implementing product launches, and developing partnerships. Chris has worked as an instrumentation engineer at Penn State University’s astronomy department and as an applications engineer at National Instruments.

        Todd Dobberstein works as an industrial data acquisition and control product marketing engineer for National Instruments. His current projects include Reconfigurable I/O (RIO) hardware such as CompactRIO and plug-in R-Series devices. He began at NI in March 2002 and holds a BSEE from Kansas State University.