# Why is the dB plot of a signal usually non-positive?

Almost every decibel plot of a signal is usually below 0. Why is this the case? I recently just plotted a transfer function with a very high decibel plot above 0.

## 3 Answers

In power spectra of signals, 0 dB is some agreed power level. It can for example represent the level of a full-scale (FS) sinusoid. In that case it would be rare to see values above 0 dBFS (sinusoid), as it would require peak cancellation as with $1.125\sin(x) + 0.125\sin(3x)$ to stay within a numerical amplitude range of $[-1, 1]$. When a power spectrum is given in dB, the 0 dB reference should be specified. Otherwise one can't be sure what it means.

You saw the above 0 dB values not with a power spectrum of a signal but with a transfer function, or equivalently with a magnitude frequency response. In a magnitude frequency response, 0 dB means a magnitude or gain of 1, that there is neither attenuation or amplification. If you followed this standard convention and saw values above 0 dB, it means that the corresponding frequencies are amplified.

$0$ dB is the limit at which the power of the signal and the power of the noise (or a standardized level) are equal: their ratio is equal to $1$, and the $\log$ is zero. Looking at a spectrum, areas with higher dBs than other are frequency bands where the presence of a useful signal is more expected. Areas with very negative dBs are suspect of having very low signal content. However, they can be a target for the detection of very weak signals of interest.

The dB is a ratio, thus a reference is needed. A negative dB value means the signal level is less than the reference value. Note that the signal-to-noise (SNR) always uses the dB format, but the dB is not only for the SNR.