Seven tips for making better accelerometer measurements

One of the most common transducers used in automotive tests is the accelerometer; it measures forces as well as the amount of vibration experienced by components and systems. Like all things analog, making accurate accelerometer measurements means paying attention to how you select and connect the sensor and then properly analyzing the data you collect.

To help you make better accelerometer measurements I turned to the experts for some tips:

1. Find the right accelerometer. Think about whether you need an accelerometer with good AC or DC frequency response. According to Gary Rendla, director, MEMS & accelerometers for the IC Sensors Division of Measurement Specialties (Fairfield, NJ; http://www.msiusa.com), you need good DC response when measuring tilt and when making long constant measurements, such as measurements of velocity (single integration) and displacement (double integration). Piezo-resistive, capacitive, force balance and thermal-mass accelerometers are the types that provide the best DC response. You would use these accelerometers in navigation and robotic applications. One disadvantage to this type of accelerometer is that it generally has poor high-frequency response.

For making higher frequency measurements, such as high-frequency vibration and high-level shock measurements, use accelerometers with good AC response. Most AC accelerometers use piezoelectric ceramic or quartz sensing elements and generate a charge output, and many of them now have built-in electronics to provide a voltage output. These accelerometers are often used in harsh environment applications.

2. Mount the sensor securely to the structure under test. According to Dan Wolf, an application engineer for Kistler Instrument Corp. (Amherst, NY; http://www.kistler.com), how you mount an accelerometer will affect the measurements you make. While it may be convenient to mount a sensor with a cyanoacrylate adhesive, such as a super glue, a magnet, or even double-sided tape, you will obtain the most accurate measurements if you screw the sensor to a stud mounted on the test structure.

Screwing an accelerometer to the test structure yields the best results, but other methods can be used as long as you take into account that the frequency response will be degraded somewhat.

3. Power it properly. When configuring a system to make accelerometer measurements, determine how you are going to power the accelerometer and make sure you have the proper power supply. Armando Valim, sound and vibration product manager for National Instruments (Austin, TX; http://www.ni.com), says you need to know what type of accelerometer you will be using so you can supply the appropriate power source.

Valim says there are three main types of accelerometers: integrated electronic piezoelectric (IEPE), charge-mode, and MEMS accelerometers. IEPE accelerometers are the most common and require a constant current (usually 4 mA) to power a small preamplifier embedded in the transducer. The data-acquisition system or measurement device may supply this current.

Charge-mode accelerometers require charge amplifiers, which are electronic amplifiers sensitive to changes in the charge on the device. When connecting an accelerometer to an external charge amplifier, you should use low-capacitance cables and mount them carefully to minimize vibration pickup by the cables. Because a cable has a given capacitance, vibrating it may dynamically change the capacitance (and hence, its charge), making it an undesired part of the vibration transducer. MEMS accelerometers, like IEPE sensors, have internal amplifiers. They have output voltage levels, however, and require an external power supply.

4. Use the right data-acquisition system. Make sure your data acquisition system has enough dynamic range to make your measurements. To do this, say Joel Roop, senior applications engineer at Keithley Instruments (Cleveland, OH; http://www.keithley.com), you determine both the highest value and the smallest difference that you will measure. Then, choose a data-acquisition board or measurement device that has the right full-scale value and has the proper number of bits of resolution.

Roop also advises that you estimate the highest frequency component in the signal you are going to measure. This will determine the sampling rate required of your data acquisition. Roop notes that to accurately recreate a waveform with a frequency of Ω, you need to sample at a rate of at least 5 Ω to 7 Ω, preferably 10 Ω to prevent aliasing errors.

5. Calibrate your accelerometers regularly. Jim Pierson, VP of applications engineering for Entran Devices (Fairfield, NJ; http://www.entran.com), recommends that you initially calibrate accelerometers every six months. He notes that you should keep a calibration history for each sensor in order to determine how your sensors react to use in different environments. This will give you the confidence needed to extend the calibration interval to a year or more.

Of course, in severe environments, sensors require more frequent calibration. Accelerometers used in crash tests, for example, must be calibrated prior to every barrier impact test or at most every couple of days. One reason for calibrating so often is that some tests are very expensive to conduct and repeat testing is simply not practical or affordable. This easily justifies pre-test calibration of sensors. In other cases, such as in brake-system testing, government regulations require that you perform pre-test or post-test calibrations.

Pierson also points out that accelerometers should be calibrated even if they are new but have sat on a shelf for a long time. Sensors often use epoxies and other organic compounds whose mechanical and thermal characteristics change over time. These changes will affect the calibration of the sensor by changing the stresses applied to components, such as span resistors or thermal compensation resistors. The metals used in accelerometers also age, affecting measurements. To negate these effects, calibrate sensors before putting them into service.

6. Don't overload the test structure. Adding a mass, such as an accelerometer, to a vibrating structure can alter the frequency of vibration. This phenomenon is called mass loading. Kistler's Wolf says the mass of the sensor and its mounting accessories should not be more than 10% of the mass of the test structure.

7. Use smart sensors to avoid setup errors. In some cases, it may be wise to use sensors with an IEEE 1451 (Ref. 2) interface. The 1451 standard defines a digital point-to-point interface that connects a smart transducer to a microprocessor-based network adapter. It also defines a transducer electronic data sheet (TEDS) that gives the measurement system information about the transducer, including the type of transducer and calibration parameters.

By using these sensors, and a measurement system that can make use of the information from the sensor, you can avoid setup mistakes that could render test results invalid. The disadvantages are that the sensors cost more and that the measurement systems that support them are more complex than systems for "dumb" sensors. Even so, the added expense and complexity may be warranted if you run expensive tests.