PSD (Power Spectral Density Function) is a spectral function expressed as a power value per unit frequency range (1 Hz width) which is independent of the frequency resolution Δ f computed with the FFT, and is frequently used as an unsynchronized signal forming a continuous spectrum, in other words, for evaluation of irregular (random) signals. When the signal x (t) physical unit is an EU, this unit becomes EU2/Hz. In practice, if the vertical axis is voltage (V), it becomes V2/Hz or its square root V/√Hz, and is frequently used as a value for evaluation of noise in amps, etc.
In the field of random vibration tests, the vertical axis is acceleration vibration (in m/s2), and is referred to as ASD (Acceleration Spectral Density), and is expressed in (m/s2)2/Hz or m2/s3. See the following for details of the method of computation.
ESD (Energy Spectral Density Function) is the function expressing the energy frequency distribution of the time signal x (t), and is used for evaluation of transient signal (e.g. impulse signals) spectra. It is computed without the averaging operation in the spectrum-related FAQ [What is Signal Power?], and is therefore obtained by multiplying the PSD value by the FFT time window length T.
Table 1 Comparison of Three Spectrum Types
| Spectrum type | Physical meaning | Relevant signal |
|---|---|---|
| Power spectrum | Power distribution for each frequency band | Periodic signal |
| PDS | Power distribution for each unit frequency | Continuous random signal |
| ESD | Energy distribution for each unit frequency | Transient signal |
Note: PSD and ESD are used primarily in vibration analysis rather than in noise analysis.
Revised:2009/11/16
