Looking for dBA weighted sound level sensor

[npelov]

Hi,

Is there a reasonably priced sound level sensor that outputs the dBA weighted sound levels in easily readable way (linear analog or digital - SPI/I2C/UART). I'm looking for a device that has all the math/filtering built in. I have to measure 30-35 dBA Accuracy above that doesn't matter.

[barry14]

Just google "sound level meter".  There are many manufacturers of such meters and they vary considerably in price based on performance.  You can find many used ones on Ebay from such manufacturers as General Radio, Simpson, and B&K. Most analog meters have an audio output which is a representation of the input signal.  Some have outputs proportional to the measured sound level. Many of them will not measure sound pressure levels as low as you want (30 dBA is a very low sound level; normal indoor background levels are around 50 dBA  Most meters have accuracies of around 2 dB. If you need much better than that, it is going to be very costly and probably will require access to an acoustic calibrator which can cost more than the meter.

[npelov]

I'm looking for a sensor only, rather than the whole meter. I know 30 dBA is pretty low, but there are cheap meters that claim to have that limit. Accuracy requirement is 1.5-2 dBA. The reason I want a sensor only is because meters with proper output are very costly. Logging metters cost a lot. What I want is:
1. logging
2. real time readings over wifi or other wireless method.

[Richard Crowley]

It seems rather unlikely that you will find an off-the-shelf ready-made module that delivers A-weighted SPL measurement.

OTOH, it isn't rocket surgery, either.  Small, very inexpensive electret condenser microphone capsules are readily available. The circuit to implement an analog filter to execute the A-Weighting is relatively simple. And full-wave rectification and Analog to Digital conversion are all common circuit modules widely documented on the internet. 

IMHO, the two major barriers to making your own are finding mic capsules with low-enough SNR (signal-to-noise ratio). Because 30-35 dB is a rather low SPL, and inexpensive electret capsules generally suffer from self-noise, but you can buy a bunch and select for lowest noise, or else buy more premium-price capsules with better performance specs.

And the other is simply calibration.  How accurate does it need to be?  Do you need standards-authority traceable calibration?  It isn't difficult to DIY if you don't need documented traceability.

Ref: http://sound.westhost.com/project17.htm

[npelov]

Well, I should have mentioned more about my situation. I'm having troubles with noise in two places - one is day time, the other night time. Neighbours complain. For the day time I'm guilty. I run noisy machines. I want to put sound insulation and test it by having remote sensor there and readings where the machines are. The daytime sound level must be 35 dBA.
The other place is at my home. Neighbors below are new and have very sensitive years. If I drop a pen on the floor they complain on the next morning. Of course dropping a pen wouldn't violate any sound level rules, because all rules are about equivalent sound levels. For what I understand these are spread over time. If the levels are withing spec (I don't need to be that accurate) - my neighbors have to do the sound insulation. If not - I'll have to do something with the floor. Maybe a little of both.

I don't need traceable accuracy. All I need is to see how bad/good the situation is. I can get a (relatively) cheap sound level meter and calibrate the one I build with it, by generating different frequency tones and measure it with my tool and the one I bought at the same time. Like I said, I don't need that high accuracy. If the A-weighted filters are not hard to implement, then maybe the noise is the only thing left to figure out.

[npelov]

This microphone has the highest SNR (68dB) of the ones I found with my quick search. However I'm not sure how sensivity relates to my task. What does -35dBA mean? Is it not very sensitive? Can it be amplified? Or it's better to have -42dBA sensivity and 58dB SNR (this one I found locally for less than $1)?

[Richard Crowley]

Those local inexpensive mic capsules are likely quite adequate for your project.
At that price, you can buy several and hand-select the ones that perform best.
(Lowest self-noise, highest sensitivity, etc.)

Yes, they are intended to be amplified.
To make an A-weighted microphone, I would take the first stage (U1A) of this circuit...

(from Elliot Sound Products  http://sound.westhost.com/project17.htm)

And replace the first stage (U1) of this circuit (which includes the A-weight filter)...

(from Elliot Sound Products  http://sound.westhost.com/project93.htm)

And then turn the audio into a DC voltage with this circuit...

(from National/TI LM3915 data sheet: http://www.ti.com/lit/ds/symlink/lm3915.pdf)

After you have a DC voltage, you can sample and log the data with any micro-controller (Arduino, et-al).

Of course, you can also buy audio SPL meters quite inexpensively, and maybe even some with built-in logging.

[npelov]

Thanks for the circuits and explanation. I found few logging amplifiers that will do the job, but they are usb and log to a PC. I don't have suitable fan-less PC that runs windows. Even if they have linux software it might be hard to get it working on ARM (Raspberry Pi). The ones that log into own memory are a bit expensive and still won't do a good job because I can't monitor them remotely.
I'll see what I can do with these circuits. Can you tell me if there are any critical parameters to watch for if I need replacement opamps. I can find TL072 17nV/Hz @1kHz, but maybe I can use OP07C which is 0.13nV/Hz @1kHz for first two opamps. I guess opamps in the rectifier are not a big deal - the frequency is not high and also the offset doesn't matter much. The signal should be high enough to doesn't care about temperature drifts. So almost any opamp will do the job. Maybe they have to be JFET because of the input current. Correct me if I'm wrong.

Edit: well OP07 is not JFET. So maybe it's not a good idea for such small signals.

[Richard Crowley]

None of the components are that critical for a simple noise-logging system.
But you are correct, using a conventional op-amp in place of a FET device may not be a great idea.
Chances are pretty good that environmental background noise will be greater than any self-noise of the mic capsules.

[Kalvin]

Typically, a simple sound level meter contain a microphone, a-weighted filter, and a RMS-converter (AD536A etc.). The RMS-converter is fed by the a-weighted filter which will compute required RMS power estimate, time-weighting (fast, slow) and provide a nice dB output for the analog or digital display.

If you feed the linear output of the RMS-converter (not the dB-ouput, that is) in to a Voltage-to-Frequency converter and count the output pulses, you can create an integrating sound level meter which will give you noise dose estimate over a period of time. This is how it was done before the fast microcontrollers with fast ADC were available.

If you have an arduino-thing available, you can use it to sample the RMS-converter outputs and do the maths. The Arduino's 10-bit ADC will give you at best 50dB - 60dB dynamic range.

Edit: Here is one example of how to implement a simple sound level meter using a microcontroller. It might give you some ideas how to do things using DSP-algorithms. The problem with this implementation is that it is not sampling continuously (it samples ADC into a buffer and the goes into computing the data, losing the sound signal during the computation period) so it is only good for measuring steady or very slowly varying noise level:
http://davidegironi.blogspot.fi/2014/02/a-simple-sound-pressure-level-meter-spl.html#.Vo0PlF7AN0w

[npelov]

I think I just found what I need. This meter looks like it has an output. As everything from china, there is nothing said about the outputs, but common sense would be that AC should be the amplified output with the A-weighting applied and DC should be voltage proportional to the SPL. I think I'll buy it just to see if it does something. At least I hope to be able to use it as a sound level meter.

[npelov]

  .... Dave recently talked about the evil negative center pin power connectors. It's good they had it marked!