This is a part answer (as this is outside of my field), but the article Self Regulation of the Heart: Natural Frequency and Damping of the
Heart Contractions (Bahramali et al. 2009) imply that 'normal' frequencies may not be possible to precisely define, due to:
ECG frequency analysis is complicated by the fact that ECG signals are non-stationary, that is, their activity
patterns change slowly or intermittently as a result of variations in a number of physiological and physical influences.
The article extensively reviews the general frequency analysis, divided into P, Q, R, S and T components. However, they define 'optimal conditions' (i.e. 'normal and healthy) as:
optimal conditions of the heart contraction as those under
which the dynamics of contraction shows the best
agreement with underdamped harmonic oscillation
An example of this is visualised in the diagram below:
this example is provided on the University of Pennsylvania tutorial site ECG Tutorial
In the article Comparison of Power Spectral Density (PSD) of Normal
and Abnormal ECGs (Das and Chakraborty, 2011), note comprisons for the QRS component, as
QRS complexes are chosen because
it shows distinct differences for different heart diseases
In other words, distinguishing healthy from not-so-healthy.