The MEMS (micro-electro-mechanical systems) accelerometer has been around now for at least 50 years. They have both electronic circuits and mechanical devices on a silicon chip. They are composed of small moving mechanical elements that generally range from 0.001 to 0.1 mm in size. Because these types of sensors are used in all kinds of products (game controllers, smartphones, cars, etc), they are inexpensive and very small.
More and more we are seeing the use of these MEMS accelerometers in vibration monitoring equipment. And there are advantages to this type of sensor – for example, because they are small and relatively inexpensive, it becomes more viable to deploy a number of them across a site to give a more comprehensive view of indicative vibration levels being generated. Because of this feature, we refer to them as indicative monitor types. In other words, they can give an indication of exactly where on your site vibration levels are high and where they are low. But once you have determined where levels are high, you will need greater accuracy than these MEMS sensors can produce.
Why is this so?
All standards relating to vibration limits are expressed in velocity terms (mm/s or in the US in/s). Accelerometers give readings in acceleration terms (mm/s2). To go from acceleration to velocity, the data must be integrated. This is not a major problem except when the signals are of low frequency – less than about 5 Hz. Low frequencies mean that the change in velocity over time (i.e. acceleration) is very small ( e.g. a 10 mm/s signal at 2 Hz has an acceleration of less than 0.01 g), so integrated results will often be so low that they blend into background noise. To make this issue worse, powered sensors (such as accelerometers) have higher background noise levels than un-powered sensors (such as geophones).
Low-pass filtering can help but may introduce lag or miss high-frequency dynamics. High-pass filtering can remove drift but may remove real low-frequency dynamics.
A further complication with the MEMS accelerometers is the cross talk they generate – for example the feedthrough from radial to transverse elements.
The issue that makes this point about integration confusing for many is the claim that accelerometers can record all the way down to 0 Hz – so where is the problem? It is perfectly true – accelerometers can record levels down to 0 Hz – they can also record levels much higher than geophones too. But to meet your compliance standards, you must report the results in velocity, not acceleration. It is this need to integrate the acceleration numbers to velocity numbers at low frequencies that causes the inaccuracies. To further clarify this point: if all you are looking for at low frequencies are peak levels, then this integration process may be adequate. But if you need frequency/phase information, it will almost certainly be inadequate.
The problem is that it is precisely these low frequency vibration levels that cause the most concern – especially in the construction and infrastructure industries. If a dispute arises, it is the accuracy of the monitoring results at low frequencies that will be critical. If it can be shown that these results are likely to be inaccurate, then this data could not be accepted.
While Texcel will always recommend the sensor most appropriate for each situation, we do recommend the use of geophones where low frequencies are, or could be, important – for us data integrity is paramount.