The power-line hum is caused by a 60 Hz tone. You choose a bandstop filter when you want to remove frequencies over a given band.Ĭonsider an audio signal that has a power-line hum and white noise. You can also choose a bandpass filter to remove low and high frequency content while leaving an intermediate band of frequencies intact. You choose a lowpass filter when you want to remove high frequency content, or a highpass filter when you want to remove low frequency content. You can choose from a variety of filters to do this. Removing Unwanted Spectral Content from a Signalįilters are commonly used to remove unwanted spectral content from a signal. Filters that introduce frequency-dependent delay are non-linear phase filters. Zero-phase filtering is a great tool if your application allows for the non-causal forward/backward filtering operations, and for the change of the filter response to the square of the original response.įilters that introduce constant delay are linear phase filters. Design a 7th-order lowpass IIR elliptic filter with a cutoff frequency of 75 Hz. Filter this signal with and without delay compensation.

Other effects are that you get a filter transfer function which equals the squared magnitude of the original filter transfer function, and a filter order that is double the order of the original filter.Ĭonsider the ECG signal defined in the previous section. The main effect is that you obtain zero-phase distortion, i.e., you filter data with an equivalent filter that has a constant delay of 0 samples. filtfilt performs zero-phase filtering by processing the input data in both the forward and reverse directions. If your application allows off-line processing, you can remove the frequency-dependent delay by implementing zero-phase filtering using the filtfilt function. Compensating for this type of delay is not as trivial as for the constant delay case. You want to apply an FIR lowpass filter and compensate for the filter delay so that the noisy and filtered signals are aligned correctly and can be plotted on top of each other for comparison.Ĭompensating for Frequency-Dependent Delayįrequency-dependent delay causes phase distortion in the signal. You can use the grpdelay function to measure the filter delay, D, and compensate for this delay by appending D zeros to the input signal and shifting the output signal in time by D samples.Ĭonsider a noisy electrocardiogram signal that you want to filter to remove high frequency noise above 75 Hz. IIR filters introduce frequency-dependent delay.Īs mentioned before, you can measure the group of delay of the filter to verify that it is a constant function of frequency. Compensating for frequency-dependent delay is not as trivial as for the constant delay case. On the other hand, delay that varies with frequency causes phase distortion and can alter a signal waveform significantly. Looking at the output of this function allows you to identify if the delay of the filter is constant or if it varies with frequency (in other words, if it is frequency-dependent).įilter delay that is constant over all frequencies can be easily compensated for by shifting the signal in time. The grpdelay function allows you to look at the filter delay as a function of frequency. The type of delay determines the actions you have to take to compensate for it. Depending on the filter characteristics, the delay can be constant over all frequencies, or it can vary with frequency. Compensating for Delay Introduced by Filteringĭigital filters introduce delay in your signal.

Matlab filter designer integrator