In Figure 2 and Figure 3, the size of the spectrum is expressed in decibel milliwatts (dBm), which is a common measurement unit for spectrum analyzers. One decibel milliwatt is the power ratio measured in decibels relative to one milliwatt. For the spectrum analyzer in this example, the decibel milliwatt measurement also assumes in advance that the input impedance is 50 ohms. For most spectrum analyzers, this is also the case when the input impedance is chosen to be 1M ohms. Figure 4 shows the derivation of the formula used to convert the decibel milliwatts into voltage rms. In Fig. 5, this formula is used to calculate the measurement results listed in Fig. 2 – 3 – the R voltage of the –10 dBm signal.
From Figures 5.13 – 5.14, we can see that when the resolution bandwidth decreases, the intrinsic noise increases from –87 dBm to –80 dBm. On the other hand, when the resolution bandwidth changes, the signal amplitude at 67 kHz and 72 kHz does not change. The inherent noise is affected by the resolution bandwidth because it is thermal noise. Therefore, the increase of the bandwidth also increases the total amount of thermal noise. In addition, since the signal waveform is a sine wave curve and the amplitude inside the band-pass filter remains constant regardless of the bandwidth, the signal amplitude at 67 kHz and 72 kHz is not affected by the resolution bandwidth. Because we must understand that discrete signals should not be included in spectral density calculations, the characteristics related to noise analysis should cause us enough attention. For example, when measuring the noise spectral density of an op amp, you will find a discrete signal that occurs at 60 Hz (power rising line). Because this 60 Hz signal is not a spectral density but a discrete signal, it is not included in the power noise spectral density curve.