2

This is sort of a cross between a data processing and Physics question, but I would be grateful if anyone here could help me.

My objective is to calculate the sound pressure level (in dB) of a sound generated 1 meter away. I would like to use my condenser microphone to calculate this. The microphone I have is the Blue Yeti, which has a sensitivity of 4.5mV/Pa (reference value 20 microPascal).

So my question is is there a way that the digital output of the microphone can be directly monitored and recorded in units of Volts. Hence, I would use these values to calculate Pascals. Can anyone point me to a (preferably free) program that I can use to do this direct monitoring?

LPC16
  • 41
  • 1
  • 1
    You will need to know the pre-amp gain and the reference level of your DAC.

    I am not aware of such program, but programs aimed at acoustic analysis are most likely your best bet.

    Or just invest into a SPL meter... your phone might even have one (not so sure about the accuracy though).

     – Dole Jun 29 '16 at 04:34

1 Answers1

1

So my question is is there a way that the digital output of the microphone can be directly monitored and recorded in units of Volts. Hence, I would use these values to calculate Pascals. Can anyone point me to a (preferably free) program that I can use to do this direct monitoring?

So, in fact, your sound card is a Analog-to-Digital Converter, which does exactly that: convert the voltage coming from a microphone into a stream of numbers. Congratulations, you already have your data acquisition output :)

That stream of numbers has no physical unit (it is "best-case proportional to input Voltage", but it is a number, not a physical entity anymore)– you will need to calibrate it with a couple of known signals.

To make matters more complicated, the amplitude response of a soundcard can't, technically, be flat. Due to the [Shannon-Nyquist theorem], the microphone voltage signal must first go through a low-pass filter that limits the maximum frequency contained in the signal (you're a physicist: apply Fourier transform to time signal $s(t)$, giving you $S(f)$, find $f_{max}$ for which $S(f>f_{max}) \approx 0$).

Since perfect analog filters don't exist, the suppression "edge" can't be infinitely steep, so you'll see changing effect of voltage on digital numbers with changing frequency.

If you work with only one (or few) frequency(ies), and you have a multiple reference pressure wave source for this(those) at different powers, then go ahead and use the sound card, calibrate using the known powers, and calculate correction factors.

If you need something that actually promises linearity, sound cards with their analog filters optimized for audio perception might not be the tool of choice. Verify your approach with an oscilloscope, then use a suitable data acquisition unit – some digital oscilloscope can record signals for you, there's even digital oscilloscopes that have no display and are only meant for usage with a PC, other people I know use Software Defined Radio (SDR) devices to sample audio/sonar frequency signals instead of RF, yet others have just built their own microcontroller-based data acquisition devices based on 10€ evaluation boards.

Marcus Müller
  • 30,525
  • 4
  • 34
  • 58
  • 1
    Just moving the microphone by a few cm will give you larger variation than anything a typical ADC and filter would do, so I wouldn't worry about that unless you need a high precision measurement – Hilmar Jun 30 '16 at 12:30
  • Can you please elaborate further on the topic of 10€ based data acquisition devices? Any offers for good analog signal quality ? – Fat32 Dec 27 '16 at 18:40
  • 1
    @Fat32 well, my first DAQ thingie was actually some TI eval board. Still selling such; the TivaC has some ARM Cortex M4F, which means you get >66MHz 32bit processor with FPU, and also, at max 500kS/s ADC – at 12bits, I think, with a wildly guessed ENOB=10. "Good analog sig quality" definitely not, but you get the full flexibility over your sampling rates, and the boards are sufficiently simple that you can be relatively sure that you can rule out interference at relevant frequencies. – Marcus Müller Dec 28 '16 at 17:19
  • But, yeah, I've been looking for boards that are little more than an SPI ADC, USB-capable microcontroller and reliable clocking as well as a set of reasonably good analog AA filters that would allow quick deployment. I'll probably let the world know if I one day get fed up and do a low-speed, multichannel USB DAQ myself. – Marcus Müller Dec 28 '16 at 17:22
  • Microcontrollers and PCs have evolved to amazing potentials, but yet the most critical component (from a dSP point of view) of capturing high resolution and accurate samples of physical signals is still the weakest part! Consider a USB 3.1 port with a theoretical speed of 10Gbps! Get a USB 3 controller and a Gpbs sampling ADC (at a moderate price of course) and create your DIY dream capture card? But I cannot see any such implementations yet? I'm afraid of missing available & affordable technologies in this overcrowded world of industrial, commercial and scientific electronics... – Fat32 Dec 28 '16 at 23:49
  • @Fat32 disclaimer: I'm affiliated with those guys, but: http://ettus.com 's USRP X310: two ADCs + two DACs, each dual-channel (IQ), 200MS/s, 10GE <-> host. You could use the LFRX/TX analog front end daughterboards (32MHz Anti-aliasing/reconstruction filters), or BasicRX/TX (~250MHz impedance matching transformer), or one of the multiple mixer- and amplifier-containing daughterboards, or design one of your own. If you hit up myfirstname.mueller@ettus.com, we could discuss alternatives. – Marcus Müller Dec 29 '16 at 15:27
  • @MarcusMüller that's the thing I was looking for. Indeed I was searching web for an RTL-SDR that's extremely affordable through chinese suppliers (as a personal purchase). Now I would like to get in touch but I must be quite clear that my bugget for this (hobby-like) endeavour is exremely tight... some of those products are thousands of $s range! :)) Not suitable for my non-profit personal work. Certainly good for R&D firms though. – Fat32 Dec 29 '16 at 18:45
  • I think you might actually make an interesting DSP.SE question out of this, if you'd put down your requirements and described a draft of what you think you'd use to build such a daq and asked for things to look out for; you should also incorporate devices that you've considered but which didn't meet your demands and/or budget. That way it would not inherently be a shopping question, and it would work both on DSP. as well as on electronics.SE if it turns out to be more on the electrical engineering side of things and less on the signal processing side – Marcus Müller Dec 30 '16 at 10:19