Background: In electrical brain signals such as Local Field Potential (LFP) and Electroencephalogram (EEG), oscillations emerge as a result of neural network activity. The oscillations extend over several frequency bands. Between their dominant components, various couplings can be observed. Of these, Phase-Amplitude Coupling (PAC) is intensively studied in relation to brain function. In the time-frequency domain, however, PAC measurement faces a dilemma in the choice of filter bandwidth. For a frequency m modulating a frequency n, filters narrowly tuned around the latter frequency will miss the modulatory components at frequencies n+m and n-m; wide band tuning will pass increasing levels of noise. New Method: Our CFC measurement uses three identical narrow band filters with center frequencies located on n-m, n, and n+m. The method therefore is free from the bandwidth dilemma. Comparison with Existing Method(s): The method was tested on diagnostic artificial signals modeled on local field potentials and compared with four established PAC detection algorithms. While the proposed method detected the simulated PAC in high frequency resolution, the other methods detected with poor frequency resolution, or completely missed the PAC. Conclusion: Using the proposed triplet-filter banks instead of wideband filtering allows for high resolution detection of PAC. Moreover, the method successfully detected PAC in wide range of modulation frequency. Finally, bandwidth is not chosen subjectively in our new method which makes the comparison of PAC more convenient among different studies.