Measuring S/N ratio

[jeffbreen]

Hello,

I'm quite new to audio measurement, and therefore Smaart. I thought I'd start learning by measuring every piece of gear I have lying around to get a feel for things. I'm starting off by trying to measure the noise floor of a DSP and an analog mixer.

I'm trying to figure out how noisy this setup will be: A typical DSP with analog ins and outs configured to pass a mic level input signal to the output with no gain and no processing w/ the output feeding the preamp of a low priced analog mixer. Obviously, all the gain is applied at the mixer input, after the signal has passed through the DSP. I expect this will be bad, but how do I measure it and put a number on it?

I've set it up so I have a pink noise generator feeding the DSP analog input. Configured the DSP to wye this input to two outputs. Out 1 feeds the mixer input, out 2 is my reference signal. Main out of mixer is my test signal. Can I figure this out comparing relative levels?

With pink noise ON I set the mixer's gain to the point where the channel peak indicator started to flash. In Smaart the RTA shows a flat freq. response with an average of -15dB. With pink noise off the average is -95dB. Do I have a S/N ratio of 80dB?

The reference signal is -120dB with PN off, -48 with it on. When I drive the signal to the point the DSP output peak indicators start to flash it registers as -29 in smaart. So, is the DSP's S/N ratio 91 dB? The published dynamic range for this DSP is 108dB. I'm sure they didn't determine that by looking at an RTA, so how SHOULD I be measuring this?

[Jamie]

Hey Jeff,

This is not as straight-forward a measurement (or question) as it would seem. First, you have a few devices involved here including your Smaart measurement rig. For your best measurement, it would be best to take them individually - of course you need the Smaart rig involved when measuring the DSP or console, but it would be good to at least start by evaluating your rig by itself, then measure the DSP unit and mixer seperately (using your Smaart rig).

Second, you seem to be conflating two different (but directly related) measures: Signal to Noise Ratio (SNR) and Dynamic Range. SNR is the ratio between signal power levels, generally a reference signal (e.g. 1 kHz sine wave at +4 dBu (1.228 VRMS)) compared to the power level of the noise (no signal present). The reference signal used for this measurement is normally NOT the loudest, undistorted signal level the device can handle, but rather an agreed upon "standard" signal level like noted above.

The easier measurement to do with Smaart would be characterizing the dynamic range. Here you are simply measuring the difference between the loudest undistorted signal level (input a sine wave and increase level till you see clip harmonics in the spectrum) - generally this will be right at FS (0dB) - and the level of the noise floor.

SNR will be significantly less than dynamic range. 108 dB makes sense for a dynamic range for a 24 bit device.

Regards,
-j