By K.C. Dahl, Engineer, DLI Engineering
A common diagnosis issued by the Azima AI-powered diagnostic system is accelerometer overload, which is when a significant ski slope is present in the spectral vibration data. A ski slope is an elevated noise floor that decreases with increasing frequency, often engulfing the peaks at discrete frequencies that are present in normal measurements and preventing analysis of the data.
When the system indicates accelerometer overload, further investigation is usually required to determine the real cause of the bad data.
True Accelerometer Overload vs. Other Causes
Collecting time waveform data or increasing the analysis frequency range may help point to the problem causing accelerometer overload. Intermittent glitches in the time record may be caused by physically driving the accelerometer into saturation or by temporarily interrupting the power supply to the sensor.
True accelerometer overload is caused by exceeding the vibration limit of the transducer, which saturates the internal amplifier. When this occurs, the electrical response of the transducer becomes distorted or temporarily unstable. Like all physical structures, accelerometers and their mounting system are subject to natural frequencies that, when excited, can result in massive increases in measured amplitude and therefore exceed the accelerometer vibration limit. This excitation of natural frequencies can be the root cause of accelerometer overload without any clear indication to the less experienced user.
An accelerometer overload reading may also result from intermittent electrical faults such as broken wires or loose connections. If the power supply to the sensor is temporarily interrupted, then the signal will drift over a brief settling period. If this occurs while a measurement takes place, then glitches will appear in the time domain, as will a ski slope in the frequency domain data.
Temperature gradients, which may occur when a cold accelerometer is attached to a hot machine, can also cause a ski slope in the frequency domain. This can be avoided by ensuring that the machine and sensor reach temperature equilibrium before initiating a measurement
Mechanical and Installation Factors
Random or non-periodic motion within the measurement limit of the transducer will also result in a ski slope. This type of motion usually comes from a source outside the machine or by the operating conditions changing during the measurement period. An example of this might be a nearby machine starting up midway through data collection, which can produce large amplitude transient pulses that may overload the accelerometer. A loosely mounted sensor may also produce this type of signal. To avoid a ski slope, always ensure that the machine remains in a steady-state condition while taking vibration readings and keep external sources of motion at a minimum.
Make sure to eliminate bad cables, faulty connections, and improper sensor mounting as a cause of the ski slope before investigating true accelerometer overload. Cable problems should always be suspected when a ski slope is seen in data from multiple machines collected with the same transducer/data collector, particularly if the issues are always present on the same channel. For accelerometers that are not ground-isolated, a ski slope may be seen on all channels when the ground connection to the accelerometer block is broken. When you suspect a faulty cable or connector, it’s time for repair or replacement.
When Ski Slopes Appear Consistently
It is common for one machine bearing location to consistently indicate accelerometer overload no matter how many times you measure even if the machine is running at steady state. This may not necessarily be an issue with the accelerometer mounting. Rather, the high level of vibration might be inherently normal for that machine and therefore does not necessarily indicate a problem.
However, it’s also important to consider whether the consistent ski slope readings are a recent trend or whether they have been consistently present. If historical measurements from the same bearing location were considered valid and normal, and you have also eliminated the above mentioned causes for overload, then the presence of accelerometer overload may indicate a developing mechanical fault. This is sometimes the case when ceramic bearings deteriorate, which can generate significant high-frequency vibration.
Final Considerations
Machine locations that consistently produce indications of accelerometer overload but don’t appear to have other problems might require technicians to take additional measurements with a different type of sensor. The accelerometer data may also need to be analysed with a technique other than frequency analysis.
It is also important to realize that measurements are generally not valid if physical or electrical sensor overload occurs. Technicians can rarely deduce information about the state of the machine from this data.