Proven Condition Monitoring Solutions

Accelerometer Overload

Temperature environment, frequency range and maximum input are all parameters which are considered in the design of an accelerometer to meet a particular vibration monitoring requirement. There are situations when these criteria can be inadvertently exceeded, which will lead to the processing and subsequent presentation of an inherently bad signal.

Consider for a moment that the accelerometer and its environment is similar to a bungee jumper leaping from a bridge, and bounces up and down on the end of the rubber cords – ignore friction damping for now – if his motion with respect to time were plotted it would be a sine wave. If this were analysed using an FFT it would result in a single spectral line with a certain ‘bounce’ amplitude and at a particular frequency. If however the choice of jump height and bungee strength was not chosen correctly there is a distinct possibility of hitting both the ground at the bottom of the plunge and the bridge on the rebound, which may cause not only some distress to our bungee jumper but also clip the peaks off the time domain signal and creating a non periodic wave. If this signal were analysed using an FFT we would no longer see the clear single spectral line in the frequency domain but a mass of lines at the lower end of the spectrum which resembles a ‘Ski slope’. The more severe the distortion the greater the ‘ski slope’ effect.

The problem is that within the ‘skirt’ of the ‘ski slope’ may be a particular frequency which we are extracting for other diagnostic purposes, for example dynamic balance. In the example it can be seen that the 1R is within this range of influence, indeed other frequencies that can be identified will have their amplitudes modified, resulting in incorrect interpretation.

A typical example of an FFT produced by a saturated accelerometer

The accelerometer is subject to an almost infinite range of frequencies from low to very high emanating from the machine to which it is mounted. Some may be way out of the frequency range we are interested in but nevertheless have an influence on the signal. All accelerometers and their mounts have a natural resonant frequency; it may be that the machine is generating other high amplitude, high frequency vibration which is exciting the accelerometer into saturation.

The non-periodic signal may also be generated by electrical contact faults, associated with broken wires or loose connectors and mountings, or poor power supply causing the accelerometer output to drift. If this happens whilst data is being collected the same ‘ski-slope’ will be seen in the frequency domain.

If a cold accelerometer is mounted to a hot, heat soaked, machine the thermal shock data can not only damage the accelerometer but modify its output until equilibrium is attained, if data collection is initiated under these conditions it too can result in the ‘ski slope’ effect.

In conclusion always look critically at the data presented, if a balance chart such as that shown above is generated with a test condition value markedly out of kilter take an FFT if a ‘ski slope’ results there is a problem.

In real life Helitune create vibration signature ranges at the optimum frequency range and number of lines on the FFT to produce the best possible resolution for the task and the shortest possible data collection time.