I'm searching for a sound/audio sensor with ultra-low power consumption to detect the presence of a really loud noise (100dB+). The controller runs on battery and there is no option for a different power supply.

For me as a non-expert there seem to be two major types of sensors:

  • Electret microphone (with potentiometer): Here I had a battery life of not more than a week. But the device works as intended due to the right sensitivity.

  • MEMS microphone: These do look like the real power savers but the byproduct seems to be some way too high sensitivity. I read about that Vesper VM1010 (https://vespermems.com/products/vm1010/) which looks like a perfect fit (if the noise threshold can be set high enough) with about 10us current but as a hobbyist I'm totally unable to do anything with PCB and I can not find any (cheap!) breakout boards.

The common use case for sound sensor seems to be the detection of (very) low noises and I have been searching for a while now to find my product. And yes I even considered to design my own PCB circuit but I have a lack of electronic knowledge and I just don't want to believe that there isn't any smart solution for simple people.

I really appreciate any help or suggestion you can give. Thanks a lot!

  • \$\begingroup\$ What exactly is the power budget? \$\endgroup\$
    – Jon Nordby
    Jun 8, 2023 at 19:47

4 Answers 4


Try using an ordinary loud speaker as a microphone. And it doesn't require any applied power.

The moving coil and magnet generate a voltage.

Loud speakers come in a variety of low impedances.

  • \$\begingroup\$ Wow! Thank you so much for that great idea! Can you provide an example of a speaker that may be appropriate? (Or simply some tags I could go for) \$\endgroup\$
    – MaxPower
    Mar 31, 2019 at 16:24
  • 1
    \$\begingroup\$ Search here on EE stack exchange for "speaker as microphone". \$\endgroup\$
    – Marla
    Mar 31, 2019 at 16:28
  • \$\begingroup\$ A large speaker is a cool way to harvest energy from sound. \$\endgroup\$
    – tobalt
    Jun 9, 2023 at 17:35

A piezo transducer might suffice to provide a voltage needing no amplifier. However, these are high-impedance devices, and to preserve the large output voltage, it must see a load resistance that is large.
As a test, a piezo element was measured with a 10Mohm oscilloscope. A "hand-clap" was the source of loud noise, about six inches away. When changed to a 1Mohm load, results were similar. It produced the following voltage, whose peak-to-peak maximum amplitude reached about six volts:
hand clap into piezo: volts vs. time This particular piezo transducer was surrounded by a plastic shell, which provides a large resonant peak at about 3.4 kHz. That results in a sinusoidal wave that swings above and below zero volts. There are about seven cycles within each (dotted) division of the oscilloscope scale (0.002 seconds per division).
You could quite possibly hose this transducer up to a microcontroller's analog-to-digital converter directly - no amplifier needed. A few large-value bias resistors might be necessary to establish a known DC voltage at the input pin.

This might even work hosed up to a CMOS or HCMOS logic gate input.

  • \$\begingroup\$ I have a remote control for my house fan, and this is how it works. Actual end-user product. You push the switch, and it goes "snap". No batteries. \$\endgroup\$
    – gbarry
    Mar 31, 2019 at 19:03
  • \$\begingroup\$ Piezo may produce enough high voltage, for long enough time, to overheat the protection diodes on the MCU's ADC input. Place a 100Kohm resistor in series. \$\endgroup\$ Mar 31, 2019 at 21:56

The VesperMems product seems to be ideal for your requirements; amplitude detection, variable threshold and ultra-low standby power.

All you need is about 80 dB of attenuation, across the whole audio frequency which is harder than it sounds ( pun intended ) as lots of materials behave more like a low pass filter, so you can still hear the bass.

This may require some further research on your part to identify the spectral sensitivity of 100dB which to humans becomes an almost flat response. enter image description here Courtesy WIki

When you use a speaker or microphone, you can imagine how loud it is when you tap the diaphragm. So protection from these near-field vibrations must be isolated or prevented.

Acoustic Pressure is logarithmic (dB SPL) meaning it spans many decades of power levels and choosing a calibrated 100dB can be done, but this is another task for you to define, as well as defining the minimum event duration and stretching the event to stay ON or to some interface device.

So there is more to your design specs than simply choosing the sensor.


For gunshot detection in a battery powered sensor, there are two critical parameters for the microphone: Acoustic Overload Point (AOP) and power consumption.

Standard analog MEMS microphone

There are many MEMS microphones with excellent performance in both areas. For example:

  • ISC-40638. AOP 138 dB SPL, 170 µA
  • IM73A135. AOP 128 dB SPL, 170/70 µA
  • Vesper VM2020. AOP 152dB SPL, 248 µA

For level trigger one might need a rectifier and preamplifier, which would bring power consumption of 100-300 uA, depending on the chosen microphone.

MEMS microphones with built-in triggers

An alternative to a standard analog MEMS microphone, would be to use one that includes trigger.

The VM1010 mentioned in the post has the highest trigger point at 89 dB SPL.

There was also the VM3011 (now discontinued), which has the trigger level configurable over I2C instead of analog. The highest dB level it can trigger at is 91.5 dB SPL. Power consumption in idle is 10 uA. However, note that these soundlevels are measured in the area of 2-800 Hz. So one needs to check if that fits the intended application.

If a trigger point higher than this is needed, one could consider acoustic dampening to achieve that. For very loud noises, like explosions and gunshots, one may also want that to protect the microphone.

For a complete sensor, one may also need a microcontroller, as well as a data transmission system. The power budget for data transmission would depend a lot on the range needed, choice of networking technology.

  • \$\begingroup\$ 1 mA is about 2 orders of magnitude more than necessary for noise detection, the other answers should easily do less than 10 µA. \$\endgroup\$
    – pipe
    Jun 8, 2023 at 22:21
  • \$\begingroup\$ Sure. But as the power budget was not specified, so I thought I would give an answer of what is achievable with a standard MEMS mic, as it can be an option for some battery powered scenarios. But let me add one more, which can do the trigger detection on chip, at 10 uA \$\endgroup\$
    – Jon Nordby
    Jun 9, 2023 at 7:35
  • \$\begingroup\$ The VM3011 is no longer an option I'm afraid (EOL, no replacement), and the company doesn't want to talk about it. The founder jumped ship and moved to another company... (Maybe a reason that product recommendations aren't allowed as questions!) \$\endgroup\$
    – pipe
    Jun 9, 2023 at 14:21
  • \$\begingroup\$ Wow, I was not aware. I was considering it for a product of my own... Thanks for the heads up. Answer updated with that. \$\endgroup\$
    – Jon Nordby
    Jun 9, 2023 at 15:42
  • \$\begingroup\$ It was an awesome product, I have no idea why they stopped it. Afaik the nearest competitor (TDK) does 20 µA - twice as much. I've spent some time evaluating it for a future product too but nope... \$\endgroup\$
    – pipe
    Jun 10, 2023 at 16:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.