I'm curious about a strange effect I observed.
I was designing a device, which sleeps most of the time and can be woken up by a loud sound signal, like an opening door, a clap, etc.
The relevant part of the schematic:
simulate this circuit – Schematic created using CircuitLab
The ambient noise is picked up by an electret mic, amplified, and sent to a digital input on the MCU. The trimpot serves for sensitivity adjustment, so the user can tune the sound pressure level that should trigger the device. All this is very crude, as the pin threshold is not strictly defined, and the pot is not in the usual "divider" configuration (the gain selection is not linear), but that is not the issue.
When I soldered this, I noticed it didn't work; it wasn't picking sounds unless they were very loud, so I increased the amplification to absurd levels (trimpot turned to the limit, think of 19990 ohms on the right side and 10 ohms on the left side). Even at that levels, clapping hands only registered as 300 mV pulses.
While walking around the room, pondering on what can be wrong, I noticed that the oscilloscope was picking enormous signals on the opamp output, up to 5 volts. Thinking that must be the noise of my walking, I experimented more, and discovered that even in a completely quiet room, just lifting my foot of the ground produced that signal. It wasn't the sound of it, just the movement. And I was 2-3 meters away from the circuit. In a sense, the slight act of lifting my foot (or just altering its position) got registered by the sensor; whereas flipping a light switch, making a big noise with hands, etc. did not.
In the end I found what the problem was. The sensor I was using is a ECM30B. It has the traditional two pins, along with a metal tab protruding from the casing, that also seemed that needs to be soldered. Usually these mics have the ground pin and the casing inherently connected. But this one wasn't. When I connected them, the circuit behaved as expected, and with the huge amplification it picked even minute sound noises.
In a sense, the circuit was open at the "X" point. Not really, since the ground pin of the mic was soldered to ground, but just the casing was floating. With this in mind, what could have been the effect that captured my foot moving meters away (even considering the enormous amplification), but ignored similar movements with my hands?
Best Answer
The case of your microphone is probably electrically connected to the flexible membrane inside the microphone, and is responsible for grounding it, while the "ground pin" is only used to provide the return path for the FET preamplifier. In normal operation, both need to be grounded so that there's a complete circuit to drive the gate of the FET.
With the case "floating", the microphone is no longer senstive to sounds, but it does become very sensitive to external electric fields of any sort. Walking around, or even just lifting your foot, can create large amounts of static electricity, particularly in a dry environment, and the field associated with your body was affecting the case of the microphone. Your body might have a potential of several thousand volts, and it only takes a few mV to drive the gate of the FET.