The Basics of Anti-Aliasing: Using Switched-Capacitor Filters - AN3494

Aliasing due to unwanted, spurious out-of-band signals is a problem in many applications that use A/ D converters. These signals, if not filtered properly, can seriously impact the performance of a data-conversion system. This application note discusses the principles of anti-aliasing and its effect on system performance. It also offers advice, via an elegant and cost-effective, complete-system example, on using an integrated switched-capacitor device for this important function. It covers in detail all the important performance parameters and practical issues associated with designing a high-performance system. What causes aliasing? The effect is explained in the Nyquist criterion. This criterion says that a time-continuous signal needs to be sampled more than twice per period if converted into a timediscrete signal. If the signal is not sampled often enough, information is irretrievably lost. This problem can be seen in Figure 1. If a signal is sampled once per period, a DC signal (with arbitrary amplitude) is generated (Figure 1a); if it is sampled exactly twice per period, a square wave is generated (Figure 1b). It is important to note that sampling at exactly twice the period of the input signal is a very special case and should always be avoided. Figure 1c shows a 190kHz signal sampled at 200kHz. The result is a perfect sine wave, but with a completely wrong frequency. This frequency change is caused by aliasing. Figure 1a. A sine wave sampled only once per period results in a DC signal with arbritary amplitude.