Editor’s Note: Be sure to see the related article by Jamie entitled “Anatomy Of A System Measurement Rig: Probes, Preamps & Processors”.
While it may be counter-intuitive, for 99.52367 percent (roughly speaking) of common applications, the signal quality that our system measurement/analyzer rigs require to produce good measurements is, by pro audio standards, not really that high—particularly when compared to the signal quality demanded for studio recording or even simple listening.
What we require from the signal transmission path in our measurement rigs is really quite humble.
Frequency Response: Flat (+/-.25 dB) throughout the measurement bandwidth, a spec that should be easily achievable from 20 Hz to 20 kHz by all levels of audio gear.
The biggest offenders here are unexpected filters—HPF caused by wiring issues or faulty DC blocking and phantom isolation circuits, LPF from long transmission lines and poorly implemented anti-aliasing filters, and of course, the random unintentionally inserted EQ filter (a frustratingly common occurrence when grabbing measurement signal feeds off of extra board and processor outputs.)
Generally, this can be checked by simply examining the heavily averaged spectrum (RTA) of a known spectrally flat pink noise source. Any major FR deviations will show themselves quickly.
Channel Consistency (in FR and Latency): The requirement that the measurement signal channels have virtually the same response and latency (which is by far the norm) so that dual-channel system response measurements—made through the comparison of two signal channels—are not biased by any transfer function (FR) or latency differences between the signal channels of the measurement rig.
A quick way to check this is by applying the exact same signal to both input channels (or all channels in the case of a multi-channel rig) and then performing transfer function (FR) and delay (IR) measurements between channels. The FR should be flat in magnitude and phase (signifying no latency issues), and the delay between the channels should be zero.
It is possible that your delay measurement could show zero time offset, and yet the phase response shows some deflection from flat in the VHF—this occurs when the two channels are off by a factor of 1/2 sample due to uncorrected interlacing in the ADC (the ADC samples at 96 kHz and alternates between channels to produce two 48 kHz signals – the driver should correct for this 1/2 sample latency offset)
Signal-to-Noise and THD: Hold onto your hats—for most most applications in standard measurement rigs, we only truly need an signal-to-noise ratio of >70 dB and THD performance
<1 percent! Of course we expect performance much higher in the electrical path of our measurement rigs, but for the basic requirements of the spectrum, FR and IR measurements we most commonly make, this level of performance will not significantly impact our data.
Think of it this way: it’s not uncommon to have the SNR of the acoustic environment for our measurement be far below 70 dB (the old Spectrum Arena in Philadelphia comes to mind). The key here is to be aware of your noise floors (acoustic and electrical), measure well above them, and use all the tools at your disposal to help protect and improve your data quality (Amplitude and Coherence Thresholding, and liberal application of data averaging.)
Note that this is also why few measurement systems and situations see any appreciable difference between the use of 16-bit and 24-bit A/D conversion.
Channel Isolation: Verify that there is no significant cross-talk between your measurement channels (>70 dB isolation @ 1 kHz). There are a couple easy ways to do this. The simplest is to input a sine wave on one channel, and then view the spectrum of the other.
Another way is to measure a piece of electronic gear with a bit of delay/latency. If there is significant (in terms of our measurement world) cross-talk, a look at the IR measurement will show an impulse at zero time as well as the correct delay through the device.
The measurement signal path quality requirements detailed above would probably be horrifying to a recording engineer, audiophile nut, or even your standard pro-audio system engineer. But for the purposes of acquiring valid, stable and useful measurements with our rigs, that is where the bar is set. And this is good news to us, because our (99.52367 percent of us) measurement rigs don’t need to be comprised of high-end, esoteric, built-from-unobtainium gear.
Standard professional quality gear in functioning order will do most of us just fine. If you are involved in the (0.48633 percent), well, most professional quality gear will work for you too.
There are of course exceptions created by the demands of situations like measuring very low noise levels (low NC measurements)—and those are the applications where you need spend the extra $$ on your signal chain.
Jamie Anderson is a founding member of Rational Acoustics, which provides training courses, hardware products/packages, and professional consulting for sound system measurement, analysis, and alignment. He has been teaching and working in the field of sound system engineering, measurement and alignment for almost 20 years. During his career, Jamie has worked as a technical support manager and SIM instructor for Meyer Sound Laboratories, as a system engineer on tour for A-1 Audio (kd Lang) and UltraSound (Dave Matthews Band).