Components for hi-res continuous data logging sensor (beginner)

For Adafruit customers who seek help with microcontrollers

Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.
Locked
User avatar
gjoy
 
Posts: 2
Joined: Fri Apr 14, 2023 2:38 am

Components for hi-res continuous data logging sensor (beginner)

Post by gjoy »

Hi all, I'm really hoping for a little guidance for 1st project after trying so long to deciphering the specs, protocols, boards...

I want to build an accurate sound level meter which will log data to local web page (Windows laptop via WiFi), cloud/app or Google sheets (if all else fails 🙃). I am interested to use ESP since it seems so useful for future IoT projects, but I'm open to just getting started with something.

I'm mostly worried about sufficient resolution (bit depth, sampling rate, memory, logging/transfer speed) so that I can get high quality data. I am not sure what is necessary to reliably record transient, sudden impulse sounds (like a door slamming down the hall) and good sensitivity. I don't need frequency/FFT analysis, just accurate/consistent level (mV/dBfs). I'm also unsure about the ADC quality and power source affecting data resolution/quality. I would like to get the right products so the accuracy isn't limited there, and later I can add more complex signal analysis. Hopefully this much data won't cause trouble on the chip/WiFi/laptop, running continuously, but I'm thinking I can play with some filtering if this happens.

I know this is basic stuff, please forgive my ignorance... Ill learn quickly hands on, but I've gotten stuck endlessly reading specs & can't even begin the fun part!
Any suggestions would be really great :)

User avatar
blnkjns
 
Posts: 963
Joined: Fri Oct 02, 2020 3:33 am

Re: Components for hi-res continuous data logging sensor (beginner)

Post by blnkjns »

I have no experience with dB meters, but it sounds like you need a good dB sensor. It will probably set you back a bit, but put your money there. Read the spec sheets and look for accuracy, measure intervals, possibilities to grab spikes. The logging should not be the issue. It is always wise to do some data selection at the measuring already, to keep your data searchable and prevent clogging. Yet the bitrate of ESP Wifi is plenty to do continuous measurements. Memory is no issue, as the cloud side probably takes care of that.

User avatar
adafruit_support_mike
 
Posts: 67454
Joined: Thu Feb 11, 2010 2:51 pm

Re: Components for hi-res continuous data logging sensor (beginner)

Post by adafruit_support_mike »

The desire for more accuracy, more resolution, etc is a bottomless pit. Diving into it can quickly get difficult and expensive for no good reason.

The questions that push back the other way are, "what do you intend to do with the data once you have it?" and "what do you actually need to get the results you want?"

Every physical system is subject to limits that I call "engineer's zero" and "engineer's infinity":

Engineer's Zero: any number so small that, if you made it smaller, I wouldn't care about the difference.

Engineer's Infinity: any number so large that, if you made it bigger, I wouldn't care about the difference.

For audio systems, things like "the variation in speed of sound due to temperature and humidity" are usually below the Engineer's Zero threshold.

Every calculation involving physical data contains some amount of approximation. Asking, "what effect does this much change have on my calculation?", "what would ignoring that difference cost me?" are critically important design questions.

As one of the engineers for the Perseverance Mars Rover said, "a mathematician's job is to make things perfect. An engineer's job is to make things just perfect enough."

User avatar
gjoy
 
Posts: 2
Joined: Fri Apr 14, 2023 2:38 am

Re: Components for hi-res continuous data logging sensor (beginner)

Post by gjoy »

Thanks, this is really helpful. I was worried where the 'bottleneck' could be and I think you're right. I've looked at a ton of specs and some simple sound level meters projects, and the low end mic boards and ESP's internal ADC (10- bit) are clearly not sufficient. I think I'm going to try a MEMS mic, or even a 16-bit ADC with external mic. It seems that I2S output will be important to avoid picking up noise on an analog input signal, especially during WiFi transmit.
It sounds like the rest of the system will be fine. So once I get it running I'm happy to play with the data filtering from there. This is just enough to get started!

Locked
Please be positive and constructive with your questions and comments.

Return to “Microcontrollers”