Square waves contain rich harmonics, including any odd harmonics of the fundamental frequency of square waves. Theoretically, filtering out any harmonics will change the shape of the square wave and distort it.

The higher the harmonic frequency, the lower the amplitude. Assume that the fundamental frequency of a square wave is 100Hz, the frequency of its 99th harmonic is about 10kHz, and the amplitude is1100 of the fundamental wave. If the frequency of noise is close to 10kHz, or even less than 10kHz, and the amplitude is less than the fundamental frequency of square wave1100, then if noise is to be filtered, the effective harmonic loss of square wave will be greater than the noise loss.

When the frequency of clutter is much higher than the fundamental frequency (at least 100 times), the influence of clutter on square wave can be smaller. In this case, a low-pass elliptic filter or Chebyshev filter with the fastest amplitude-frequency curve decline can be used.

Because the square wave is easy to produce, it is suggested to use the second-order passive RC low-pass filter with the simplest circuit form to filter out the square wave and all kinds of high-frequency clutter. Then a hysteresis comparator is connected to regenerate a relatively pure square wave with the same frequency.