Speakers at low SPLs
This was a quick investigation to see how speakers react at very low sound levels. There is some speculation online that speakers may "lose accuracy" as they start to operate very close to their resting point.
There are some difficulties posed by this investigation. As we try to measure what's happening at very low sound levels, lots of things begin to interfere with the results:
- Ambient acoustic noise
- Electrical noise of the microphone
- Electrical noise of the preamp
I'll discuss how I worked to minimise those issues, show the test setup, and present the results below.
Optimising the setup
By itself, REW can help with some of the problems mentioned above. By repeating the sweeps (up to 8x times), REW can look for correlated data (ie, data where the same thing happened in most/all sweeps, and discard any anomalies.
Other things that I did to help out were:
- Putting the microphone very close to the speaker. This serves two purposes. Firstly, it ensures the best possible acoustic signal-to-noise ratio by maximising the microphone's pickup of the desired source. Secondly, by ensuring the maximum level possible hits the microphone, the electrical signal-to-noise ratio is also improved.
- Using a directional microphone. This helps to exclude unwanted acoustic noise.
- Using a microphone with very low electrical noise. A comparison of the datasheets was in order:
- SE4400a - Large diaphragm condenser, 25mV/Pa, 16dBA noise
- Beyerdynamic MC930 - Small diaphragm condenser, 30mV/Pa, 16dBA noise
- AKG C3000 - Large diaphragm condenser, 20mV/Pa, 18dBA noise.
- Beyer MM1 - Omni measurement mic, 15mV/Pa, 26dBA noise
- A few dynamic mics, but with 1/10th the voltage output.
Here, then, the Beyerdynamic MC930 was the winner. It showed the joint lowest electrical noise, while having the highest output. That would mean the preamp would need the least gain to output adequate levels, reducing background noise from there. In theory, dynamic microphones have almost zero background noise, but because they provide relatively little voltage output, that would've meant even more gain would be needed at the next stage.
- Finally, using a preamp that was as low-noise as possible. In this case, the QSC TouchMix 30 Pro has proven itself very useful.
Here's a photo of the test setup, using a Bose MusicMonitor as a test speaker: