How do you test audio gear analytically? Audio Precision has some useful white papers on its Web site, including "Fundamentals of Modern Audio Measurement" (http://ap.com/download/Fundamentals_Modern_Audio_Meas.pdf), which first appeared in the Journal of the Audio Engineering Society. If you're more interested in mixed-signal audio measurements, check out Cirrus Logic's "Personal Computer Audio Quality Measurements" (www.cirrus.com/en/pubs/whitePaper/meas100.pdf).
In the way of test equipment, designers need analog and digital signal generators and analyzers, along with something like a PC to control them and crunch the data. Naturally, the test equipment should have precision that's at least an order of magnitude better than the devices being tested.
That may not seem like too tall an order, since the frequencies involved are only measured in the tens of kilohertz. But distortion specs on components like those in National Instruments' custom sound room are down around 0.00003 dB while dynamic range covers roughly 120 dB, almost from thermal noise to the threshold of pain. A further equipment challenge is how the hardware is partitioned. There are several approaches, with accompanying price/performance tradeoffs.
That's the test-equipment hardware challenge. Equally or more important is knowing how to make the measurements. The plusses and minuses of simply measuring audio amplitude take up a page and a half in that Audio Precision white paper. When it comes to the potential pitfalls of fast-Fourier-transform (FFT) measurements, one passage from the white paper is illustrative of the challenges:
"Since all hypothetical sine and cosine frequencies in the FFT are multiples of the reciprocal of the waveform length, the analysis is inherently equal \[to\] resolution in the frequency domain. This analysis also presupposes that the signal components are at exact multiples of the reciprocal of the waveform length; serious problems occur when this is violated. Stated differently, the FFT assumes that the waveform being analyzed is periodic with a period equal to the length of the data record being analyzed...
"Consequently, if the beginning and end of the record do not meet with the same value and slope when looped back on themselves the discontinuity will result in artifacts in the spectrum. The usual way to deal with this is to 'window' the data and drive its value to zero at the end points. This turns the waveform into a 'shaped burst,' whose spectrum is the convolution of the window spectrum and the signal spectrum."
This leads to a detailed description of different types of windows, and Audio Precision's alternative approach, described in the paper. As if that weren't enough, there are equally detailed considerations of the advantages and disadvantages of frequency, dynamic range, noise, and intermodulation measurements.