Vibration sensors need to be matched to operating and environmental conditions. As Chris Hansford, managing director of UK manufacturer Hansford Sensors, puts it: “Selecting the right sensor for the job is key. It’s ‘garbage in, garbage out’… and if you’ve not understood the mechanical constraints and the IP ratings that are required then you might as well not bother.”
Some sensors are only intended to detect a certain level of vibration, and to send a simple on/off signal when this is reached – to prompt inspection or maintenance. Some are activated if a severe shock load is applied, to shut the system down. But most are used to monitor vibration levels either periodically or continuously. Sensors can be sited around the machine, with connectors for periodic reading using a data collection device. Or they can be permanently connected to a datalogger, a controller or a building management system (BMS) for continuous monitoring.
Vibration is described in terms of acceleration (see box, right), so vibration sensors are generally in fact accelerometers which assess the motion of an internal mass using some form of force measurement. The sensor (Hansford HS-150 pictured above) must be specified so that it measures the appropriate range of acceleration: if the measurement range of the sensor is too broad, precision may be low; but if it is too narrow, the sensor could be overloaded.
The other vital specification is frequency range – will the sensor respond to vibrations across the range of speeds it will encounter? This is not just converting the rotation speed to a frequency (for example, assuming that 6,000rpm equates to 100Hz): it should cover much higher frequencies, to cover issues like ‘squeal’ due to incorrect lubrication. Frequency response is usually quoted using a decibel range or within a tolerance.
And, of course, the sensor must be rugged enough for its environment (defined by IP ratings, and suitability for immersion in different liquids) and it must comply with safety regs such as the ATEX directives.
Most accelerometers work on the principle of the piezoelectric effect, whereby stress (and strain, or internal movement) in a material generates a small but proportional electric charge. Piezoelectric materials used to be typically crystals, but now most are more robust ‘piezoceramics’.
Piezoelectric accelerometers generally work in either ‘compression mode’ or ‘shear mode’. There is no need to know which is which – the specifications of the device are much more important – although shear mode accelerometers are a more recent development. Chris Hansford says that “Shear devices are a lot more stable, and able to be used in harsher conditions, such as where you have changes in temperature.”
The small signal from the piezoelectric material is amplified and ‘conditioned’, often by electronics built into the sensor itself.
The output is typically described as either AC or DC. A constant-current AC sensor produces a voltage proportional to the acceleration, with a sensitivity expressed as (for instance) 100mV/g. A DC accelerometer puts out the proportional 4-20mA current signal required by a BMS.
Piezoelectric sensors usually come with a calibration sheet so that vibration amplitude can be assessed accurately.
Either type can detect vibrations due to imbalance and misalignment, but Hansford says: “AC output can give you so much more detail” – allowing diagnosis of issues such as belt and gear problems, and cavitation in the oil film.
INTEGRATED EXAMPLE
MEMS (Micro Electro-Mechanical Systems) accelerometers can fit into a microchip-type package, and are often used for sensing purposes in mass-produced items such as mobile phones and cars. They are also available as vibration sensors for machinery, sometimes with sophisticated communication and signal processing functions built in. Their abilities are described by Analog Devices: “This type of programmable device can wake itself up periodically, capture time-domain vibration data, perform a fast Fourier transform (FFT) on the data record, apply user-configurable spectral analysis on the FFT result, offer simple pass/fail results over an efficient wireless transmission, provide access to data and results, and then go back to sleep.”
However, Hansford says: “MEMS has limitations in terms of environment and temperature… the worldwide market is growing, and MEMS is actually expanding the market, but is not impacting the traditional condition monitoring markets like process industries.”
Strain gauge accelerometers use a resistive element to give an AC output, but they are prone to variation with temperature and need careful calibration, so they are less common than piezoelectrics. Accelerometers often also incorporate a temperature sensor, which can be an equally important diagnostic tool.
Eddy current transducers, sometimes known as proximity sensors, work on the principle of inducing a current (and hence a magnetic field) in a metallic object such as a gear. The probe is connected via a shielded cable to an oscillator/demodulator ‘driver’ unit which outputs a voltage proportional to the distance of the probe from the surface. The whole assembly must be calibrated before use.
Proximity sensors have no lower frequency limit, so they can be used on slow-moving machines; however, they have an upper frequency limit of typically 10kHz. They have typically been used on journal-type bearings, where they provide a reading of the radial displacement of the shaft relative to the journal, and can also be used to measure axial vibration or end float in a shaft, or ‘rod drop’ in a worn reciprocating mechanism. But they are less common in conventional machine monitoring applications.
MOUNTING
Choosing the right sensor is one thing, but Chris Hansford says, “Where you mount it and how you mount it are very important.” To measure radial vibration, sensors are typically mounted on the plane of a bearing; mounting a sensor between two bearings runs the risk that two instances of bearing play might cancel each other out. Depending on the type of sensor, you might also need to fit more than one sensor per bearing, to detect motion in perpendicular axes. However, biaxial and triaxial sensors are available.
Hansford points out that bearings usually present a curved surface, but this “needs a flat on it to enable good coupling of the sensor”. Gluing the sensor on is one option, but surfaces can be greasy and doing this (or ideally drilling and tapping a mounting point) may invalidate the warranty.
Some modern electric motors incorporate optimised mounting points for sensors, but for those that don’t there are solutions such as grease adaptor mounts (pictured above), which combine a lubrication point with a stud or threaded hole to ensure a good interface between machine and sensor.
Sensors are usually pretty rugged, but problems are more likely to occur with the cables connected to them; most suppliers recommend using pre-made cable assemblies, and ensuring that connectors are always fitted with liquid- and dust-proof covers.
BOX: DEFINING VIBRATION
Vibration can be described as periodic motion around an equilibrium point; in rotating machinery, it takes the form of radial vibration (perpendicular to the axis of rotation), axial vibration (along the axis) or torsional vibration (a varying couple twisting a component around its axis).
Whatever the type, vibration is measured in terms of frequency (usually given in Hz) and amplitude; for radial and axial vibration, this is expressed as an acceleration, with units of metres per second squared (m/s2).
A vibrating component usually produces an acceleration graph that looks like a sine wave; but the peak-to-peak measurement of this wave is not the figure you usually want to take. Instead, amplitude is conventionally expressed as a root mean square (RMS) value, approximating the average amplitude throughout a cycle.
NOTE: standard BS/ISO 20816 covers this area.