- Mon Aug 21, 2006 11:23 pm
#17632
Bear in mind that virtually all sound level meters or "audio level meters" use RMS (Root Mean Square) signal detection of signal amplitude. The reason is that for complex waveforms with harmonically and non-harmonically related components and random noise, RMS detection indicates the actual energy content of the signal, independent of the "crest factor" (ratio between peak voltage and RMS voltage) and independent of the phase relationship between various harmonics (which affects the crest factor). Crest factor can vary from 1.4 for sine waves to five or ten or even more for music with strong impulsive components like drums.
Peak detection can be wildly off the true energy content of the signal based upon the magnitude of the peaks of the signal. The peak detected voltage is related to the energy content of the signal ONLY for DC or for a sine wave. For any other waveform, the energy content (which is, along with other parameters, directly related to the subjective loudness), cannot be predicted from a peak detected signal unless you know the exact waveform and have calculated or measured the relationship, and this is, of course, impossible in continusouly varying music.
In the olden days, it was difficult to get true RMS detection without elaborate circuits and devices. Bruel & Kjaer for many decades used a biased resistor-capacitor network in their sound level meters, calling it a "quasi-RMS detector," and it worked quite well for the limited range of the analog meter it was driving (perhaps 20 dB).
Fortunately, modern ICs make very accurate, wide dynamic range true RMS detection easy and cheap. See Analog Devices' AD 536 (I think) and its offshoots. Google "true RMS detection," to see the cheap and very good ICs available.
If you don't use RMS detection in an analog meter, your'e just kidding yourself.
awright
Acoustical Consultant