Within a digital world, the parameters of real-world information (temperature, voltage, current, speed, flow, pressure, distance, etc.) used for process communications and control are analog in nature. Processing this information is achieved primarily using digital signal processing techniques. The age of microelectronics has made it possible for digital processing power to be extended into the field where sensors are located; thus, much preprocessing can be accomplished outside the main computer or PC.

This graph uses MATLAB program functionality for Fast Fourier Transforms. Aliasing occurs since sample frequency (400 Hz) is less than twice the highest input frequency (213 Hz). The 213-Hz input frequency component is reconstructed as 187 Hz (400-213).
There are many issues that can create serious problems along the way, such as EMI noise, grounding, accuracy, resolution, aging, drift, isolation, and noisy power supplies. In addition, there is one very subtle problem that often goes unnoticed but can lurk in the background. Known as aliasing, this problem exists in isolated signal conditioning modules (SCMs) and anywhere an analog-to-digital converter is active.

The basic concept of aliasing is this: Converting analog data into digital data requires sampling the signal at a specific rate, known as the sampling frequency. The result of this conversion process is a new function, which is a sequence of digital samples. This new function has a frequency spectrum, which contains all the frequency components of the original signal. The Fourier transform mathematics of this process show that the frequency spectrum of the sequence of digital samples consists of the original signal’s frequency spectrum plus the spectrum shifted by all the harmonics of the sampling frequency. If the original analog signal is sampled in the conversion process at a minimum of twice the highest frequency component contained in the analog signal, and if the reconstruction process is limited to the highest frequency of the original signal, then the reconstructed signal accurately duplicates the original analog signal. It is this process that can give birth to aliasing.

To illustrate, consider a time function signal consisting of two sinusoids, one at 25 Hz, and the other expected to be at 180 Hz. This sampling process uses a sampling frequency of 400 Hz (which is greater than 2*180 Hz) and the reconstruction process has a bandwidth limit of 200 Hz. The only frequency components that will appear in the conversion are 25 Hz and 180 Hz. However, if the original signal’s highest frequency is not the expected 180 Hz but 213 Hz, a reconstructed signal consisting of 25 Hz and 187 Hz (400-213) will result. The 213-Hz signal component appears in the output as the 187-Hz component (187 Hz “alias” 213 Hz).

Since all signal conditioning modules that provide isolation typically use some form of analog sampling conversion process to move analog information across an isolation barrier, it is critical for system engineers to examine their application carefully to determine whether aliasing can occur.

This work was done by Dataforth Corporation. For more information, click here.