Stand Loudspeaker Reviews

Silicon Valley–based Velodyne was founded in 1983 to develop a range of subwoofers that used servo-control to reduce non-linear distortion to vanishingly small levels. They succeeded in this goal to the extent that Velodyne is now perhaps the best-known subwoofer company in the US, currently employing 65 people. At the 1994 Winter CES, Velodyne launched the subject of this review: the DF-661 ($1800–$2600/pair), their first full-range loudspeaker (the "DF" stands for "Distortion-Free").

The three-way DF-661 was designed from the ground up to continue the Velodyne tradition of ultra-low distortion. "We had developed the technology and resources to attack distortion elsewhere in the audio chain," wrote company President David Hall, "and started with the premise that, by definition...distortion in loudspeakers is wrong." (His italics.) "We went to the laboratory for a solution, with the living room as the ultimate goal." Velodyne calls this attention to technological detail "The Silicon Valley approach to sound."

To achieve their goal of less than 0.1% THD at any frequency in the new speaker's passband, Hall and his team developed radically new midrange- and low-frequency units for the DF-661. Velodyne manufactures both drive-units in-house, which has meant considerable capital investment in the necessary machines, tools, and jigs.

The DF-661 woofer has a roll surround and features a spun-aluminum cone, intended to act as a perfect piston. The cone profile was carefully shaped to reduce the amplitude of higher-frequency breakup modes, which will contribute to the measured distortion. The cone is attached to a long voice-coil former, this filled with a lightweight damping material to keep it from flexing radially. The woofer's diecast chassis was designed to minimize reflections from the rear of the cone: the metal arms joining the front of the chassis to the motor are carefully profiled and extend straight back so that the only immediate acoustic obstruction is the spider. The chassis is also decoupled from the magnet structure with rubber bushings, so that cabinet excitation is kept to a minimum.

All of this reduces the level of distortion components. But the largest reduction in non-linearity is due to the use of a stationary counter-coil between the magnet polepiece and the voice-coil. Electrically in series with the moving coil, the counter-coil renders the woofer's impedance purely resistive. More importantly, it cancels distortion components arising from magnetic nonlinearities, the modulation of the fixed field by the varying field generated by the current passing through the voice-coil, for example. The price to be paid for this drastic improvement in linearity, however, is that the driver's efficiency is halved. Not only does the magnetic gap have to be wider, but only half of the total coil winding generates sound.

The midrange unit has a similar chassis and motor to the woofer, but features a laminated cone with two layers of aluminum separated by a visco-elastic damping layer. Its surround is also different, being very non-compliant. Only the tweeter is not made by Velodyne; it's a version of the well-known SEAS 1" aluminum-dome unit. Because this will not meet the Velodyne 0.1% THD at 1W goal in its lower octave, it is crossed over to the midrange unit at 5.5kHz.

Physically, the DF-661 looks elegant, and is taller than it is wide. The midrange unit is mounted at the top of the baffle with the tweeter and woofer beneath it. The woofer is reflex-loaded with a rear-panel port, 1.85" in diameter and just under 3" deep. The Velodyne grilles—black cloth stretched over bulky wooden frames—were left off for the tests.

The DF-61's crossover is constructed on a printed circuit board mounted to the rear panel behind the terminal posts on standoff pillars. Air-cored coils are used to avoid core saturation, and an additional series capacitor is used in the woofer feed to give an overall fifth-order rollout.

Distortion
Understandably, the first thing I examined about the DF was its non-linear distortion performance. Velodyne specifies less than 0.1% THD, –60dB, at a 1W level, though no frequency range is mentioned. I therefore drove the speaker with 50Hz, 100Hz, 200Hz, 500Hz, 1kHz, 2kHz, and 5kHz sinewaves at voltage levels of 2.83V and 8.95V. Measured at the speaker terminals with a Fluke 87 true-RMS multimeter, these represent nominal 1W and 10W levels. With each signal, I captured the speaker's output waveform with a B&K 4006 microphone with its capsule positioned 12" from the DF-661's tweeter (footnote 1). An EAR mike preamp fed the magazine's Audio Precision System One DSP analyzer, which calculated the sound's spectrum, hence the levels of all the harmonic components, using the Fast Fourier Transform.

Table 1: Velodyne DF-661
Distortion Harmonic Levels in dB ref. Fundamental Level

Frequency (Impedance)	Drive Power	2nd	3rd	4th	5th
50Hz (13 ohms)	2.83V 0.62W	-34.75	-39.6	-50.3	-46.3
	8.95V 6.16W	-32.9	-28.1	-42.4	-41.5
100Hz (8.3 ohms)	2.83V 0.96W	-53.3	-54.8	-	-
	8.95V 9.65W	-42.6	-52.3	-72.4	-68.3
200Hz (3.4 ohms)	2.83V 2.35W	-55.9	-64.9	-69.6	-65.1
	8.95V 23.5W	-47.2	-51.5	-60.7	-65.1
500Hz (5.4 ohms)	2.83V 1.48W	-62.4	-65.8	-83.1	-72.75
	8.95V 14.8W	-51.0	-56.1	-75.3	-70.1
1kHz (5.1 ohms)	2.83V 1.57W	-60.6	-66.6	-88.1	-67.7
	8.95V 15.7W	-57.2	-71.0	-82.9	-74.2
2kHz (3.65 ohms)	2.83V 2.19W	-57.7	-67.9	-78.9	-71.6
	8.95V 21.9W	-51.1	-61.5	-70.5	-77.0
5kHz (5.6 ohms)	2.83V 1.43W	-49.1	-62.9	-	-

Table 2: B&W Silver Signature
Distortion Harmonic Levels in dB ref. Fundamental Level

Frequency (Impedance)	Drive Power	2nd	3rd	4th	5th
100Hz (11.9 ohms)	2.83V 0.67W	-40.5	-53.3	-75.7	-68.6
	8.95V 6.7W	-31.0	-40.0	-57.6	-62.1
1kHz (31.2 ohms)	2.83V 0.26W	-55.8	-45.0	-78.0	-59.6
	8.95V 2.58W	-45.3	-35.2	-59.5	-67.0

The levels of the DF-661's individual harmonics are given in Table 1, while Table 2 shows the performance of the B&W Silver Signature loudspeaker measured under the same conditions. (Note that I've listed each speaker's impedance magnitude at each of the test frequencies. This is because the 2.83V and 8.95V voltage levels I used will only be equivalent to 1W and 10W power levels if the impedance is exactly 8 ohms.)