PSW: To obtain free field measurements which are as close to perfect as humanly possible, you'd need an environment completely free of reflective surfaces and background noise. In a practical sense, what steps does Community take toward reaching this ideal?

Dave: We place the test system as far away as possible from reflective surfaces. Then we choose TEF parameters that don't allow for any possible reflections to interfere with our measurements. We also utilize the noise immunity advantage of TEF TDS measurements to reduce the interference of background noise.

In translating all of those concepts in real world terms, the first step we took in reducing the effects of background noise was to measure its level as it relates to time. We used TEF NLA software for this test (see Figure 2).

Figure 2.

Since testing takes place outside above our parking lot, and not far from a major highway, several noise events were measured. They included a truck idling below the test microphone, a truck accelerating out of the parking lot, and vehicles passing along the highway.

The testing scenario.

Typically, the noise levels varied from 60 to 85 dB SPL. Since noise varies significantly with time, tests to determine frequency content were averaged over time to obtain an accurate picture of frequency content. Measurements we collected using TEF NC software (see Figure 3) showed how noise levels vary by frequency.

Figure 3.

These tests proved that noise levels are significantly greater at low frequencies as compared to high frequencies. To conduct testing in an environment like this without attenuating for noise would require a sound pressure level at the microphone of 125 dB at the low end, and 85 dB at the extreme high end, 40 dB above the noise floor. Since that's not practical, we call upon the advantages of TDS measurements to help us out.

It's fairly simple to measure background noise with TEF TDS; you simply disconnect the loudspeaker under test and perform the TDS analysis. What you get is a measurement of background noise reduced by the TDS tracking filter. As before, a measurement of this type shows that background noise is more of a problem at lower frequencies than high frequencies. Knowing this, we're able to measure high and mid frequency devices at a lower level than full-range or subwoofer systems while maintaining an on-axis SPL 40 dB above the noise floor.

So, in simpler terms, what does all this mean? Through the use of TDS measurement techniques, an environment free of reflective surfaces close to the loudspeaker under test, and long measurement distances, we are able to collect polar, SPL, and harmonic distortion data in an accurate and dependable manner.

PSW: Speaking of noises, what types of signals do you typically use for actual testing?

Dave: For all tests-except sensitivity-we use swept sine waves that lie within the operating range of whatever loudspeaker component or system we are testing.

PSW: What would that sound like to the human ear?

Dave: Well, it sort of whistles from high to low. A BAH-WEEEEEEP kind of thing. It could start out low, rumbling the legs of your pants. Or, if we're testing a high frequency device, it could start out at a point where the neighborhood dogs are going to be covering their ears with their paws. These swept tones are typically between 5 and 12 seconds long. Longer sweep times yield more detailed data, but at the expense of valuable engineering time. We use pink noise to measure the sensitivity of speakers, and generally run it at between 50 and 100 percent of the device's rated output. That way, we obtain a real world figure. In the real world, you're typically not going to be playing a loudspeaker at one watt, so why test it that way? One of Community's big testing concerns has always been to provide numbers that you can really use.

PSW: The test and measurement arena in the pro sound world has traditionally been a hotbed of controversy. Where do you think Community fits in?

Dave: I think one of the most important things to note about Community is that all of our published polar data is actually five degree, one-third octave data. Some companies have been known to lay a loudspeaker on its side and measure it horizontally, stand it up and measure it vertically, then extrapolate everything in between. None of our measurements are faked-this is the real deal, across the board.

Community also provides measurement of Axial Q, an important but often overlooked spec. This one is for the WET218 loudspeaker.

PSW: How long did it take you to re-measure the entire Community catalog? Dave: Fourteen months, and I was at it pretty much every day. At the time, I was just doing the measurements, too.

My predecessor, Chuck McGregor, was doing the post-processing required. Now I do everything.I've got my hands in the process right from the first moments of product development to proofing actual printed spec sheets and PDF files for the Web. Right now, I have four to six loudspeakers I'm collecting full data on. Our goal it to always have the measurements out before the products begin shipping. This is a wide-ranging job, and like just about all others, it's never done.

(Editor’s Note: In case you missed the earlier link, Bruce Howze offers additional commentary on Community test and measurement philosophy. Click here to view it.)