Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711

E-mail: Jack.Smith@cliftonlaboratories.com
 

Revision History
05 March 2009: Original

The Internet is a wonderful repository of information—both useful and less than useful—but it has gaps. A few days ago, I looked for information on the typical output (in millivolts) of a dynamic microphone as used in amateur radio.  Microphone manufacturers, from what I can see, use a two-step specification process. The first  step defines how the microphone responds to a defined sound level (mechanical) input. The second step is an estimate of the typical sound level produced by an average talker. Combined, these two factors result in the required answer, X millivolts.

Given the difference in microphone specifications and the squishiness in the talker output level, the best answer I came up with was the output is "a few millivolts."

Accordingly, I decided to measure the output of the microphone section of my old Heil "Proset" boom microphone. It's equipped with Heil's HC-5 "full range" element.

I make no pretence of claiming my measured data is representative of what others might measure. There's too much variation in voice, lip to microphone distance and the like for that. And, of course, different microphones will produce quite different outputs for the same speech loudness.

The oscilloscope image below is taken when I spoke as loudly as would be comfortable for a short duration. The oscilloscope is, of course, a high impedance device as are microphone input amplifiers in typical amateur radio SSB transmitters. The peak-to-peak voltage is 140 mV. In fact, this level overstates the expected voltage because I would not speak this loudly for more than a minute or two.
 

The figure below suggests that nearly all the voice sample (for normal speaking level) are within the range ±5 mV, with lesser probability excursions to ±10 mV, although samples are clearly visible at ±80 mV or more.

The figure below is a histogram plot of the microphone's output. The voltage levels are the instantaneous voltages and the plot is read with the vertical axis being the number of instantaneous measurements at the voltage level indicated on the horizontal axis. It's easier to understand the histogram plot than the dot plot presented above.

Note that the central value is slightly negative and the histogram is not symmetrical about the peak. This is because the human voice is non-symmetrical. This may be seen in the oscilloscope images, but the histogram is a more powerful visualization tool. Although the signals are distributed more on the the negative voltage side, the distribution tail is actually slightly heavier on the positive side. This means that there are more high value positive peaks than negative peaks.

Let's look at the 50% percent point, which is the center (horizontal) grid line. It crosses the curve at about -0.5 mV, which is similar to the histogram plot. (The histogram and CDF are not of the same speech sample.) In other words, half the instantaneous voltage samples are less than -0.5 mV and half are greater than -0.5 mV.

For design purposes, the 10%-90% or 5%-95% or 2%-98% values are more useful. I've placed the marker at the positive 95% percent point, which is 7.40 mV. The corresponding 5% point is -8.0 mV. Hence, we can say that 90% of the instantaneous microphone voltage values will be between -8.0 and +7.4 mV.

The sample I used to make this plot is way too small to be widely extended, but it suggests the magnitude of expected microphone voltages.