Aliasing is an effect caused by the conversion of continuous data to a digital format. The result can be signals appearing in your data with frequencies which never actually occurred.
     It is as if you could hear a low frequency tone in some audio data; the tone is not actually there until the data is digitized, but there is no way to know if the tone was really there or not. A common example is the effect that causes a wheel to appear to move backwards when seen on a television.
The Nyquist Theorem states that the sample rate must be more than twice the highest frequency in order to accurately measure all the frequencies present. In fact, if the sample rate is too slow, you not only don't measure the highest frequency, but lower frequencies are created in the data that didn't exist in the signal.
     In order to be sure there are no frequencies greater than the sample rate, an anti-aliasing filter can remove them. This must be done before the data is digitized. Once data is captured in the PC there is no way to correct it, or even to determine if aliasing happened. An anti-aliasing filter is also called a low pass filter because it only passes low frequencies.
     Vibration measurement is one of the applications where aliasing is most common. Unexpected high frequency vibrations sometimes occur. Even if the high frequency signal is unimportant, its presence can cause aliasing that alters the data by creating a lower frequency vibration in the data, a signal that is not really present.
     An anti-aliasing filter is important to remove the aliasing. Its cut-off frequency should be set slightly higher than the highest vibration frequency of interest, and the sample rate should be about five to ten times the cut-off frequency for a 5-pole filter.
     Sound measurement is another application where aliasing is very common. There can be high frequency sounds that are inaudible to a human, but they can cause aliasing, making a lower frequency sound appear in the data. Fast strain and pressure measurements can also have aliasing problems.
On the other hand, in some applications aliasing happens rarely or not at all. Temperature and humidity, for example, do not usually change very rapidly and signals are generally not affected by aliasing, primarily because you do not measure frequency.
     Strawberry Tree offers several models of data acquisition devices with integrating converters that are very good at rejecting noise for measuring slowly changing low level signals, such as thermocouple signals. With these integrating converters, high frequency noise is rejected and anti-aliasing filters are unnecessary.
     Slow strain and slow pressure measurements usually do not have aliasing problems, if you are not measuring frequency. Be careful, however, that there are no unexpected high frequencies when you are looking at the frequency content of a signal (use an anti-aliasing filter or check the signal at a much higher sample rate). Aliasing problems can be avoided if any noise is removed.
     However, in general, when you are looking at the frequency content of a signal, anti-aliasing filtering is advisable.
     Strawberry Tree's DATAshuttle Express is a widely used device for connecting sensors to PCs, and users have a solution to the aliasing problem: software programmable low-pass filter "drawers" which slide into the DATAshuttle Express enclosure.
     The aliasing problem - and its solution - is just one of the topics covered by a new series of Application Notes now available from the Strawberry Tree section of our website, http://www.adeptscience.com/.
     There's a wealth of information in these Application Notes, to guide you through what is often a complicated specifying and setup process. They cover a range of topics, from How to Design a Data Acquisition System to Selecting Data Acquisition Software; from Signal Conditioning to Exporting Data and Creating Reports; from Computer System Recommendations to the Environmental Limitations of Data Acquisition Equipment.
     If you're into data acquisition, we think you'll find this a very useful resource.

Mike Montague