Metal Marshmallow Pro's Subsonic Frequency Response

Introduction

A few people have asked recently about Metal Marshmallow Pro's ability to record very low frequencies below the range of human hearing. This is difficult to measure for several reasons. The initial challenge is finding something that can drive the microphone at just a few Hz. Second, standard audio hardware typically attenuates very low frequencies, and the details are not usually published. Consequently it is difficult to obtain an experimental setup with known properties. Third, while playing sound into the microphone, it is difficult to know how much force is being applied. Finally, as I have pointed out before, the frequency response of a contact mic is not constant, but changes depending on what you attach it to.

Setup

As a partial solution to some of these issues, I made the setup shown in Figure 1.

This setup consists of a long flat piece of spring steel 'fish tape' clapmed to a Metal Marshmallow Pro contact mic. A second clamp holds the apparatus in a convenient orientation. I adjusted the length of the spring steel until it vibrated at approximately 5 Hz.

Experiment

I then recorded some sound through the mic with the spring dampened and plotted the spectrum between 1 and 100 Hz. This establishes the noise floor of the microphone. Then I set the spring vibrating and again made a short recording and plotted the spectrum. Both spectra are plotted together in Figure 2.

This graph shows that the spring steel was vibrating closer to 6.5 Hz. It also shows 80 dB of separation between the signal and the noise at this frequency. Note that this should not be taken as THE signal-to-noise ratio of the mic at this frequency. This is because I set the spring vibrating at a moderate amplitude which quickly dampened during the measurement. Consequently, this is not the loudest signal that can be recorded undistorted by the mic. Instead it is just a moderate-ampitude signal in the middle of the mic's amplitude range. So 80 dB represents a lower bound on the SNR at this frequency; the actual SNR is higher.

The mic has internal protection circuitry that will kick-in in response to very high amplitude input. This will result in a garbled waveform that is nonetheless correct in its fundamental frequency. Springs exhibit harmonic motion so the mic output should be sinusoidal. To check that the mic's output is correct (sinusoidal) and not some other garbled or otherwise incorrect waveform, I plotted an oscilloscope trace of the mic's output with the spring vibrating. The resulting waveform is shown in Figure 3.

This shows two superimposed sinusoids: the main one at about 6.5 Hz, and a second one about 19 times higher in frequency. The second higher frequency is the result of the complex physics of vibrating rods. In any event, the spring is actually vibrating at both frequencies simultaneously, and the second frequency is not due to the the mic's protection circuitry or anything else in the microphone. The main 6.5 Hz signal is nice and sunusoidal and appears otherwise undistorted. Separately, I was also able to produce a garbled waveform (not shown) by vibrating the spring at high amplidude, which is the expected behaviour.

Thoughts and Caveats

This shows that the mic has exceptional SNR for very low subsonic sounds below the range of human hearing. What is does not show is what the mic's sensitivity in this range is. The sensitivity would tell you how many Volts the microphone outputs for a certain amount of pressure applied to it. This would probably be more informative than a lower-bound on the SNR. Nonetheless, I have not measured the sensitivity due to the difficulties discussed in the introduction to this article. In short, I don't know how to measure the sensitivity, so I'm leaving it as future work.