Noisemeter project advice

[StellarWind7]

Hello all.  I'm new to Arduino development.  I have some programming experience but mostly in high level languages. 
I am trying to build a 24/7/365 noisemeter.  I live in an area that has a severe noise problem but the citizens cannot fight back against the offender without empirical data.  They are claiming their noise does not exceed 55 dB but my iPad and cellphone noisemeter software consistently show 85 dB + noise levels throughout the day.  Unfortunately, I don't have a 24/7 surveillance of noise levels nor is it practical to keep an iPad or cell phone on the roof. 
So, I need advice on how to create a noisemeter with the following requirements:

• Sound measurement accuracy comparable to Type 1 noise meter (NIOSH developed an iPhone/iPad application using the native sound hardware and achieved performance near Type 1 accuracy:  https://www.cdc.gov/niosh/topics/noise/app.html  )  Ideally the measured dB level would be within 1-2% of what this software is reporting.
• Runs 24/7/365.  Should run of AC power via AC-DC converter but must not induce unacceptable levels of noise.
• Logs dB data to either:
•    o   A computer via Ethernet, WiFi, direct USB.
•    o   A memory card (SD, microSD).
• Saves in simple comma delimited text format or similar that can be easily parsed.  Data logged would be nothing more complex than:   date, time hhmmss, ADC_value, dB_level
• Samples at 1 Hz.
• Should be able to do up to 130 dB (like cheap $20 noisemeters).  Is a 24-bit analog-to-digital converter necessary to capture this level?  I'm not a noise expert by any means but I did read in the Audacity manual that low bit depths cannot capture higher dB levels.  I would think that even an 8-bit ADC just outputs 2^8 = 256 levels of voltage measurement, and this would somehow be correlated to a dB level?  So would it be fair to say that even an 8-bit ADC could measure within 0.5dB of accuracy if we desire to measure up to 128dB?  (256/128=2 levels of resolution per dB). 
• Would need some degree of resistance to weather and elements.  Breezes seem to generate high dB readings on my phone and iPad.  I would assume that covering the microphone from breezes/wind will reduce actual dB measurements.  Would it be best to enclose it in a coffee can or something like that?

Thanks for all your advice!

[DVDdoug]

Anything that's home-built isn't acceptable for legal purposes.    It gets rather complicated to make a calibrated A-weighed SPL meter, and of course it has to be professionally calibrated by a certified lab.  Most labs won't calibrate homemade equipment....  You can only make something that's approximately accurate if you can calibrate it to a particular "kind" of noise.

The kind of timed & logged measurements you want to do require some expensive equipment.  You might be able to rent it...  I don't know...

Should be able to do up to 130 dB (like cheap $20 noisemeters).  Is a 24-bit analog-to-digital converter necessary to capture this level? 

Typically there is a range control (sometimes auto-ranging) that adjusts the sensitivity or the gain of the amplifier.    You also don't need 130dB of dynamic range to read up-to 130dB SPL because you're not reading down to zero dB SPL. 

[MarkT]

Find a reputable calibrated noise meter that has serial output and connect it to a datalogger.  Find an expert
witness who can witness the use of the system, and the actual readings being observed, and is willing to
answer questions in court about the measurements and their relevance to the legal case - anyone capable of
claiming 85dB is 55dB is also capable of pulling a few sneaky tricks so be ready.  They certainly would trash
any claims based on home-brew equipment. 

I suspect the best course of action is to employ a professional company whose business and reputation is
with noise management - they might see an opportunity to get work from the offender in fixing the problem
once the spectre of a successful prosecution is raised.

[StellarWind7]

Anything that's home-built isn't acceptable for legal purposes.    It gets rather complicated to make a calibrated A-weighed SPL meter, and of course it has to be professionally calibrated by a certified lab.  Most labs won't calibrate homemade equipment....  You can only make something that's approximately accurate if you can calibrate it to a particular "kind" of noise.

The kind of timed & logged measurements you want to do require some expensive equipment.  You might be able to rent it...  I don't know...
Typically there is a range control (sometimes auto-ranging) that adjusts the sensitivity or the gain of the amplifier.    You also don't need 130dB of dynamic range to read up-to 130dB SPL because you're not reading down to zero dB SPL. 

I understand what you're saying about the dynamic range.  I think that's something I could process from the ADC output. 

As far as legal purposes, I'm not looking for something I can use in a legal battle.  I need a smoke detector, not a smoke analyzer.  I want to get approximate dB levels while logging the times they occur at over 365/24/7.  If I can show there's sufficient 'annoyance', I would have the regulatory agency place sophisticated equipment at the monitoring site to make the legal case.  Right now they stand by their "noise contour" analysis which has very simplistic and best case assumptions.

[StellarWind7]

Find a reputable calibrated noise meter that has serial output and connect it to a datalogger.  Find an expert
witness who can witness the use of the system, and the actual readings being observed, and is willing to
answer questions in court about the measurements and their relevance to the legal case - anyone capable of
claiming 85dB is 55dB is also capable of pulling a few sneaky tricks so be ready.  They certainly would trash
any claims based on home-brew equipment. 

I suspect the best course of action is to employ a professional company whose business and reputation is
with noise management - they might see an opportunity to get work from the offender in fixing the problem
once the spectre of a successful prosecution is raised.

In my case, I just want to show there's an "annoyance" based on anecdotal/semi-empircal data.  Simple citizen-science will be aimed at having an actual regulatory study.  Unfortunately right now we can't make much of a case with people only complaining about noise.

Do you happen to have a recommendation on noisemeters with a serial output?  This is precisely what I wanted originally but what I found at Amazon was $800+.  I was looking for < $50. 

[DVDdoug]

I understand what you're saying about the dynamic range.  I think that's something I could process from the ADC output.

No.  The ADC goes from 0-1023 and that's all the "digital information" you get.  Any range/scaling has to be handled on the analog side.    You can adjust that with amplification or by adjusting the ADC reference.

The 10-bit ADC gives you about 60dB of dynamic range.    So for example, you could read from 40-100dB SPL (if it's properly calibrated to max-out the ADC at 100dB).   That's not as much range as you wanted but it's enough should be enough for your purposes.

but what I found at Amazon was $800+.  I was looking for < $50.  

I don't know the exact price but like I said, this equipment is expensive.  You can rent one for less than $100 per day.

I live in an area that has a severe noise problem but the citizens cannot fight back against the offender without empirical data.

If your city has noise regulations, then the city ("code enforcement" department?) might have a way to measure the noise.   Otherwise, how can they enforce the regulations?

Of if the offender has "deep pockets" try contacting a lawyer.  If there's money to be made the lawyer might help you collect evidence.