To achieve high resolution in a short acquisition time, a trade-off needs to be reached based on three parameters: modulation frequency, modulation amplitude, and lock-in filter bandwidth.
First, the spectral resolution depends on the signal-to-noise ratio (SNR) and on the spectral distortion, which are both influenced by the amplitude of the magnetic field modulation. Large modulation amplitudes increase the SNR following an increase in signal intensity. At large amplitudes, however, the detected EPR signal broadens and becomes distorted, hence decreasing the resolution because close lines cannot be resolved. A similar distortion effect applies when high modulation frequencies are used and the spin relaxation is too slow to follow fast changes in the magnetic field.
Moreover, the SNR and the spectral resolution directly depend on the modulation frequency. This is a consequence of lock-in detection, as explained in detail in the Principles of Lock-in Detection White Paper. A high modulation frequency results in high SNR - but it will also cause spectral distortion, as described above.
Finally, the filter bandwidth used for lock-in detection also impacts on the SNR and on the acquisition time. A low filter bandwidth causes a high SNR but leads to a slow acquisition time at each step of the magnetic field sweep, because low filter bandwidths need slow settling times. Another way to achieve high SNR is by averaging - remembering that the SNR is proportional to the averaging time - and using a large filter bandwidth with fast settling times and fast acquisition times. In a stable laboratory environment and with a stable spectrometer, signal averaging with a large filter bandwidth and few averages with a low filter bandwidth are equivalent. In reality, signal drifts must be taken into account. This requires finding suitable trade-offs between the filter bandwidths and the averaging time.