[an error occurred while processing this directive]

The Mix Speaker Tests, February 1998

Feb 1, 1998 3:26 PM, By Mike Klasco and Jack Hidley


Note: Starting this month, Mix is enhancing its loudspeaker “Field Tests” by adding a range of acoustic analysis results. The tests were performed by Menlo Scientific, an independent acoustical test facility in Berkeley, Calif. To help readers interpret the results of other monitors in the future, we examined characteristics of three near-field speakers that many readers would be familiar with: Genelec 1031s, Meyer HD-1s and Yamaha NS-10Ms. The 1031s and HD-1s retail in the $4,000-per-pair range. The NS-10Ms—priced about a tenth that amount—were selected for their ability to translate to a “typical” home stereo bookshelf speaker and because every studio seems to have a pair.—Eds.

On-axis (upper trace) and 30-degree off-axis (lower trace) frequency response of a Meyer HD-1. Note the smooth response and sharp roll-off above 15 kHz.

Studio monitors are subjective tools, so why bother measuring speakers at all? Too often, manufacturers state speaker specifications that are questionable, either as a frequency response plot with a 90dB vertical axis reproduced with a thick line that obscures a system’s true performance or response figures that are meaningless without a range specifier (such as ±2 dB, ±3 dB, etc.). In such cases, an 8-inch ceiling speaker could spec out as 20 to 20k Hz. (The ±60dB part of the spec is somehow omitted.)

Is a perceived “midrange clarity” of some studio monitors merely a rise in the midrange frequency response or because the speaker exhibits low distortion? Is “punchy bass” the result of a mid-bass bump in the frequency response, outstanding transient response with a fast settling time, low power compression or some other factor? For these and dozens of similar loudspeaker personality characteristics, acoustic test instrumentation provides a quantitative analysis.

The acoustic analyzer used for the tests is called SYSid, short for SYStem IDentification. Developed by Bell Labs, SYSid is a PC-based system combining signal generation and acquisition, and DSP using an Ariel DSP-16 board. The measurement mic and preamp are Earthworks models M30 and Lab 1.

On-axis (upper trace) and 30-degree off-axis (lower trace) frequency response of a Yamaha NS-10M. Large peaks and dips in the crossover ranger like these are usually from crossover interactions.

The Tests
The frequency response plot is taken from data collected both in the near-field (1/4-inch) and at 1 meter. The close proximity of the mic in the near-field measurements loads the speaker diaphragm and creates reflections. By comparing the near-field and 1-meter data, room effects in the 1-meter data and mic reflections in the near-field test can be isolated and identified. We also measured each speaker 30 degrees off-center (shown offset in the graphs by 5 dB for clarity) as off-axis response is important, particularly when reaching for controls or outboard devices located out of the “sweet spot.” All frequency response measurements are made with a 40dB window, with each vertical division being 5 dB.

Transient response graphs reveal both a speaker’s ability to reproduce the fast rise of an impulse and how quickly the speaker settles. If the speaker is still ringing from a previous waveform, then it will mask the following waveform. The typical muddy sound of bandpass subwoofers, for example, is due to the long settling time of the enclosure design.

Impulse response measurement of Meyer HD-1, a powered monitor with internal signal processing that delays the tweeter's signal to align with phase of woofer output. Impulse response after initial pulse is quite good. The unusual pre-ringing is the result of delay circuit in signal processor.

Harmonic distortion graphs show second- and third-harmonic distortion and total harmonic distortion, plus noise. There can be dramatic differences in distortion levels among speakers. Most “hi-fi” speakers typically exhibit about 1-percent distortion above bass frequencies. By contrast, some high-quality audiophile and studio monitors deliver as little as 0.1 percent, and, at low excursion, distortion levels can drop to 0.01 percent over a narrow frequency band. Distortion rises dramatically at higher and lower frequencies; however, the ear is most sensitive in the mid-band.

Low distortion is hard to achieve. Refinements that can reduce distortion include features such as a copper cap on the pole piece, flat (as opposed to “cupped”) outside edge on the spider, improved tweeter dome and woofer cone materials, fancy passive crossover network components or, even better, an active crossover/bi-amp configuration. But sometimes, a speaker with less-than-exceptional data does excel. This can be due to smooth frequency and dispersion transitions between drivers at their crossover, the right octave-to-octave balance and optimum radiation pattern control in the midrange (to minimize reflections off the console).

There are many ways to measure distortion, the most common being second, third and total harmonic distortion (THD). These specs are important, yet drivers can exhibit substantial second (or other even-order) distortion products and still sound acceptable. Higher harmonic distortions, particularly odd-order (fifth, seventh, ninth, etc.), are more objectionable.

All of the above measurements quantify distortion components that are in some way related to the original input signal. What about distortion products that have nothing to do with the input signal frequencies? This nonlinear distortion is reflected in intermodulation (IM) distortion levels. However, IM distortion tests usually use only two input frequencies, whereas music and speech contain far more than two tones. Multitone IM distortion tests will reveal nonlinearities that generate distortion products at frequencies where no signal is present.

Impulse response of a Yamaha NS-10. Woofer output (wide spike) lags tweeter output (narrow spike). This type of impulse response is typical in many speakers.

Such multitone test measurements produce a graphic analysis of the cross-modulation products (“self-noise”) generated by a speaker system that is excited by a multifrequency signal. The spectral contamination test (a term coined by Deane Jensen of Jensen Transformers) offers a unique approach to measuring and quantifying speaker clarity and definition in a subjectively meaningful way. A speaker with low spectral contamination may enable the recording engineer to discern more subtle aspects of a recording such as phase cancellation effects between mics.

With this analysis, we hope to provide the reader with a comprehensive set of data that will correlate—but not replace—subjective evaluations. And perhaps over time, we will all learn to better see what we are hearing.

Modern Recording and Mixing

This 2-DVD set will show you how the best in the music industry set up a studio to make world-class records. Regardless of what gear you are using, the information you'll find here will allow you to take advantage of decades of expert knowledge. Order now $39.95

Mastering Cubase 4

Electronic Musician magazine and Thomson Course Technology PTR have joined forces again to create the second volume in their Personal Studio Series, Mastering Steinberg's Cubase(tm). Edited and produced by the staff of Electronic Musician, this special issue is not only a must-read for users of Cubase(tm) software, but it also delivers essential information for anyone recording/producing music in a personal-studio. Order now $12.95

[an error occurred while processing this directive]