Can You Trust Your Sensor Output?

Follow these tips to obtain accurate data and minimize maintenance costs.

By Victor Viola, IMI Sensors, Industrial Division

In the past decade, trending vibration parameters has become the most widely used technique for monitoring machinery health, and industrial accelerometers have become the workhorse in the predictive-maintenance market. Systems continue to evolve to accommodate modern manufacturing practices and streamline recordkeeping, but sensor accuracy remains critical in protecting valuable assets. 

Despite the risks associated with inaccurate output, the performance specifications of sensors used for machine-health monitoring are often taken at face value. Reduce the likelihood that bad data leads to equipment damage and other costly errors with these tips for careful sensor selection and consistent performance monitoring. 

Check the spec (and read between the lines!)  

When you’re new in the market for an accelerometer, it pays to compare the quality standards of various sensor manufacturers. After all, an advertised specification is only as good as its method of verification. Several national accrediting bodies can help eliminate the guesswork. They include:

NIST: National Institute of Standards and Technology (nist.gov)

NVLAP: National Voluntary Lab Accreditation Program (nist.gov)

A2LA: American Association for Lab Accreditation (a2la.org)

In the U.S., NIST is the authority on calibration equipment standards, while NVLAP and A2LA are two of the most popular accreditation bodies when it comes to calibration processes. A lab that uses NIST traceable calibration equipment is not necessarily accredited to ISO standards for calibration practices. However, an accreditation by NVLAP or A2LA (or another certifying body) guarantees NIST traceability of equipment and lab accreditation to ISO standards. Note that these are not the only valid accreditation bodies to ISO practices. They, and others, serve as shorthand for a thorough background check of your manufacturer’s calibration standards and provide assurance that your valuable assets are in good hands. 

When credibility is established, the specification data can be used to determine whether the sensor was fully tested after manufacturing and offer clues to the testing methodology. Physical attributes, such as size, housing material, hermeticity, and cabling, are good starting points to determine if a sensor is a fit for its intended environment. They are included on all specification sheets. You may find that performance attributes vary by manufacturer.

A few key specifications should always be included in sensor data, regardless of manufacturer, to guarantee its operation within an intended application and performance range. These include measurement range, frequency range, resonant frequency, output impedance, output bias voltage, and discharge time constant. When reviewing the specs, be sure to ask:

• Are any of these performance specs omitted?

• Was the last calibration single point/limited range?

• Is historic trend data available and reviewed?

Lab calibration report shows how sensitivity data from a single point calibration provided an incomplete profile of the sensor’s performance.

Reading between the lines for missing data can give you a sense of how a calibration was performed. Consider the results (Figure 1) of an accelerometer that was tested directly from the factory, with an advertised sensitivity of 50.39 pC/g at 120 Hz. While the sensor tested was of very high quality, its performance was measured within a very specific range at 120 Hz, 5 g peak, and 23 C (73 F).  

In this example, the sensor failed to stay within +/-5% range of sensitivity as the frequency range decreased. Sensitivity data from a single point calibration provided an incomplete profile of the sensor’s performance, especially when applied to turbine testing (this product’s intended use), where the sensor will need to operate in extreme temperatures, be hermetically sealed against environmental debris, and live with high electric fields. 

Consider your own application to determine if any missing data points might become an issue down the line. A manufacturer may be able to provide further performance data upon request. The more information you can get on a sensor’s proven performance over time, the better. 

Take sensor calibration into your own hands.

Among the must-have attributes, there’s one line on any spec sheet that you’d be better off ignoring: “Additional calibrations are not necessary.” Don’t let this statement inspire false confidence in a sensor’s performance over time.A few scenarios illustrate this point:

• shock events (dropping a sensor, exceeding shock limit), which can result in a broken crystal (sensing element)

• temperature shifts that occur when a sensor is exposed to temperatures beyond its operating range, which can cause a crystal to depolarize and lose its charge output

• damaged or de-bonded crystal during manufacturing 

• natural decay in crystal output over time.

In a straightforward cost-versus-risk analysis, calibration can be costly in terms of downtime to machinery and calibration lab fees. On the other hand, the cost of inaccurate data could be catastrophic. Or, it could be a minor setback, as in a repeatable test. In the example of turbine monitoring, the severity of failure is always major. 

In his whitepaper, Managing Machine Assets Using Predictive Maintenance, David A. Corelli, IMI Sensors’ Director of Application Engineering writes: “When it comes to plant operations, it seems there is always time and money available to put out a fire but there is often no time or money available to prevent one. In the end, even seemingly small failures can lead to big consequences. So, it boils down to this question: How much risk are you willing to take?”

Combat costs with careful planning.

What regular calibration looks like can vary from company to company according to convenience, cost, likelihood of failure, and associated risks. If this seems difficult to quantify, it helps to remember that the most common cause of failure is damage to a sensor’s crystal, which usually occurs with use outside of specifications. Accelerometers used in portable applications are more likely to incur damage from being dropped or mishandled. When used within specification in a low-risk setting, natural decay in output over time can be estimated based on the sensing material.  

Piezotronic sensors are typically made of either quartz or a ceramic composite. Quartz is a naturally piezoelectric mineral that is considered the most stable of all such materials. For quartz sensors, the calibration value is guaranteed to remain stable for a minimum of five years when used within the published operating guidelines for these sensors. 

Polycrystalline ceramics are manmade materials that have been polarized to exhibit piezoelectric properties. It is recommended that sensors with ceramic sensing elements be calibrated at least once a year, as well as after any suspected physical damage due to excessive mechanical shock, extreme thermal transients, excessive temperatures, or other extreme environmental influences.

With a schedule in mind, options for calibration can be broken into three basic categories: on-site/portable calibration, lab calibration, and field calibration. Each comes with its own benefits and downfalls, which should be considered against the needs of your facility. 

Portable vibration-calibration units include handheld shakers or more sophisticated units with built-in data systems. Generally, benefits include reduced downtime and improved personnel efficiency, as nuisance vibration trips can be checked on the spot. However, the initial investment may be expensive, and many are not capable of generating calibration reports for quality purposes. 

Sending your sensor to the manufacturer or outside calibration lab can be a good option for less-frequent calibration. Benefits include calibration certificates by ISO-certified labs and the peace of mind that comes from leaving it to the experts. A major drawback to this method is the sensor’s time away from the facility, so planning for downtime or having a backup monitoring plan is recommended.  

Field calibration may be one of the most expensive options, as travel is often a factor in the cost. But it is one of the best for minimizing downtime, with added benefits of testing on-site, such as verification and calibration of vibration transducers and related test systems, verification of connector and cabling integrity, and confirmation that machine vibration alarm trip points are set properly to ensure end-to-end functionality of vibration-monitoring systems.

With an understanding of how manufacturing data holds up against your application, a sensor calibration plan that makes sense for your facility, and common sense safe-handling practices, you can be sure that your predictive-maintenance system is functioning as intended.  EP

Victor Viola  is Director, Industrial Division, at PCB-IMI Sensors, Depew, NY (pcb.com), where he is responsible for global sales and marketing of all IMI products in the following areas: Predictive Maintenance, Process Monitoring, and Protection/Control Products. He has spent nine years in the industrial-sensor market with a focus on Condition Monitoring.