The NSM Model 50, despite its conventional appearance, is not just another conventional column loudspeaker. The goal of its designers was to produce a "compact acoustic suspension system with the bass response of a subwoofer, and the imaging and soundstaging of a minimonitor."

To achieve those goals, they located its single 6/2-inch bass driver in a separate sealed subenclosure at the bottom of the cabinet, with heavy bracing to ensure rigidity and damped with open-cell foam and Dacron. At the top of the cabinet they placed another 6l/2 inch cone driver to handle the midbass and midrange frequencies. NSM does not indicate the effective crossover frequency between these drivers, or to the single 1-inch soft-dome tweeter located midway between the two cone speakers.

NSM stresses the Model 50's imaging characteristics, which are said to be due in part to its minimum-phase crossover network and careful matching of the drivers and crossover component values in each pair of speakers. The Model 50 is sold only in pairs, bearing identical numbers and distinguished by their "A" and "B" designations. The specific component values used in each Pair of speakers are recorded on the production schedule, so that the speakers can readily be restored to their original matched performance in the event of damage. Although NSM states that it "breaks in" drivers and matches them before assembly, the owner's manual recommends that the speakers be "broken in" for 200 hours at low to moderate volume to insure the best possible performance.

The upper two-thirds of the front panel is covered by a removable black cloth grille. Near the bottom of the rear panel are two pairs of recessed, gold-plated binding posts, normally joined by gold-plated straps. With the straps removed, the system can be biwired or biamplified.

The Model 50's frequency response is rated as a "room response" of the pair of speakers under specific conditions that do not include the room dimensions; the rating cannot be correlated with any measurements we might be able to make. The sensitivity of each speaker is given as relatively low 85 dB sound-pressure level (SPL) on the tweeter axis at 1 meter with a 1 watt input (presumably the standard 2.83 volts). Unconventionally, the manufacturer gives three impedance ratings: minimum 4 ohms, rated 8 ohms, and "nominal" 16 ohms.

NSM recommends that the speakers be fitted with the supplied spiked feet and, if possible, biwired. Neither was practical for our measurements and listening tests. We did

place them about 3 feet from the walls and 8 feet apart, however, essentially as recommended.

The Model 50's measured sensitivity was 85 dB, exactly as rated. The room response above a few hundred Hertz was very uniform and smooth, with a variation of only +2.5 dB from 300 Hz to 20 kHz. When we measured the quasi-anechoic responses of the two cone drivers, with close microphone spacing, we found that the woofer (lower driver) output reached its maximum at 100 Hz, falling at 12 dB per octave below about 70 Hz and at 6 dB per octave from 100 Hz to 300 Hz. The upper driver (midrange) had a similar response below 100 Hz (about 3 dB lower in level than the lower one), but did not drop off as rapidly at higher frequencies and delivered several times as much power above 500 Hz as did the lower driver. Both drivers appeared to cut off above 3 kHz.

Splicing the woofer curves to the room response to create a composite response was difficult. A reasonable match existed in the range of 300 to 700 Hz, resulting in a composite curve flat within +2.5 dB from 250 Hz to 20 kHz, rising about 6 dB from 250 to 100 Hz, and returning to its midrange level at 45 Hz. Even at 32 Hz, the output was only 6 or 7 dB below the average midrange level, and (as listening confirmed) it was perfectly audible and usable at that frequency.

Quasi-anechoic MLS frequency-response measurements at several different distances from the speaker shared a number of common characteristics. All showed a maximum output at 1.5 to 2 kHz, dropping to a minimum at about 5 kHz, and returning to the 300Hz level (the low-frequency limit of this measurement) at several frequencies between 7 and 20 kHz. Despite these minor variations, typical of most speakers in this sort of measurement,
the overall response varied a mere +3 dB from 300 Hz to 20 kHz, confirming the room-response measurements.

The system's horizontal directivity was typical of a l-inch dome tweeter. At 45 degrees off-axis, the output at 10 kHz fell 4 or 5 dB relative to the on-axis response, and the difference increased to 12 dB at 15 kHz and 18 dB at 20 kHz. The Model 50's group delay (an indicator of its phase linearity) was among the lowest and most uniform we have measured to date, with a peak-to-peak variation of less than 0.2 millisecond from a few hundred Hertz to 20 kHz.

The Model 50's impedance curve was rather unusual (as suggested by its unconventional manner of specification). Over most of the audio range, from 1 to 20 kHz, the impedance ranged between 20 and 28 ohms. At lower frequencies it dropped

steeply, to a minimum of 5 ohms at 120 Hz, climbing to a bass resonance of 11 ohms at 65 Hz and dropping to 4.8 ohms at 29 Hz.

We measured the woofer distortion with a 5-volt input (corresponding to our 90-dB reference level for this measurement). From 60 to 220 Hz, it was between 0.55 and 1 percent, rising at lower frequencies to 7.5 percent at 40 Hz and 13 percent at 30 Hz. In the upper part of the woofer's range, the distortion rose to 1.5 percent at 400 Hz and somewhat more at higher frequencies. But it was the upper cone driver, whose output was comparable to that of the lower driver up to about 300 Hz, that dominated the system output above 400 Hz, with a roughly constant distortion level of 1 percent from 300 Hz to 1 kHz and less than 2 percent up to 2.5 kHz, where our measurement stopped.

 Because of its high average impedance, the Model 50 can probably handle the full output of most amplifiers likely to be driving it. Our pulse power tests resulted in the amplifier clipping (at about 330 watts) at 1 and 10 kHz before the speaker emitted any signs of distress. Only at 100 Hz were we able to generate the rasping sound of a cone reaching its excursion limits, which also required an input of well over 300 watts (and left the speaker unscathed).

All of this suggests that the NSM Model 50 is a very good speaker, which had been apparent from our first listening. How did it sound? For one thing, it was unobtrusive, in the sense that we were never particularly aware that the sound was emanating from those black boxes, even when quite close to one of the speakers. Its stereo imaging was excellent, as demonstrated by the way it dealt with the imaging tests of the Chesky JD37 CD.

It is difficult, however, to describe the tonal characteristics of a loudspeaker in terms that convey one person's reaction to another person who has not heard it. I compared the NSM Model 50's to some other speakers of similar stature that happened to be on hand, whose sound I had admired. I could get used to either one (and like its sound very much), yet when switching between them I found myself hearing properties of each that I preferred at one moment or another, depending on the program material.

I suspect that the NSM speakers were the more "accurate" of the two (whatever that may mean), and that its spatial performance was more realistic. I like sonically-unobtrusive speakers that do not call attention to themselves yet give a believable impression of a concert-hall performance. By those criteria, the NSM Model 50 is a great success. Hear it for yourself if you can.