Design Criteria

Since distortion inherent in raw drive units compromises the sound quality of a speaker system the following specialized tests are performed to evaluate a potential driver.

1. Variations in frequency response dips and peaks in the amplitude vs. frequency response are caused by cone flex, energy storage in the rubber surround and spider, imbalances in the mechanical design of the moving assembly, and inadequate motor structure (voice coil and magnet).

Cone flex: ordinary cones made of polypropylene or treated paper bend and flex while vibrating, especially at high volume levels or at frequencies outside a narrow range of linearity. If the cone is allowed to flex parts of the cone will be going in the opposite direction that the voice coil is moving. This has the effect of allowing certain frequencies to be either canceled or boosted, depending on the wavelength of the frequency, the cone diameter and material. In order to cancel this effect an attempt is made to fabricate a cone that is very light and stiff. The ratio of weight to rigidity is called the Young's Modulus, and the higher the better. However, there is a penalty to be paid in that incorrect design will result in a cone that will "ring" at certain frequencies due to the lack of internal damping in the cone.

Cone flex can be measured by several means. In the middle 1980's, Celestion published laser interferometry photos of a soft tweeter dome undergoing distortion when attempting to reproduce a 10kHz tone. As laser interferometry measurement setups cost more than most manufacturers can afford, these important tests are often deleted from the measurement protocols and the resulting distortions ignored. In fact, most system designers will attempt to compensate for the mechanical errors in the driver by using equalization in the crossover, or ignoring it completely, hoping that the coloration are not perceived.

At Von Schweikert Research we have attempted to invent or recycle technology forgotten in the rush "to go digital." For instance, we are observing and measuring cone flex with a very simple apparatus consisting of a strobe light coupled to a signal generator. When the strobe light is set to blink at the same rate of cycles per second (CPS to you old timers out there), the actual in and out motion of the cone is "stopped" by the strobe in a darkened room, allowing the cone flex to be observed. In this manner, different types of cone materials can be tested quickly and results catalogued without the added problems inherent with sonic subjective evaluation. (another test regimen in itself, being highly dependent on the recordings used.)

Testing by visual strobe motion has enabled us to design drivers (or modify existing ones) with accurate mechanical translation of the electrical signal into sound waves. This naturally results in systems having more clarity, transparency, and realism.

For instance, the carbon fiber midrange driver used in the VR-4 has far less flexing distortion than any polypropylene or treated paper midrange driver we have measured. In addition, it was found by empirical test procedures that a woven cone surface breaks up standing waves, the energy being dissipated by the "traps" formed by the weave itself.

2. Problems with Frequency Response Measurements

Frequency response measurements are made by many manufacturers in an environment that does not correspond to an actual living room. Anechoic chambers or test instruments are generally used that will isolate the effects of the room from that of the speaker. However, it is a mistake to assume that a speaker that measures well in an anechoic chamber will sound good in an average living room.

It is our policy to ensure that the design of the finished system takes into account the room effects which include bass boost and reflected energy from walls, ceiling and floor that degrade image specificity.

The design of the VR-4 takes into consideration these effects and was "'voiced" or balanced with an average of several types of listening rooms and conditions. This is not to mean that the VR-4's will only work in certain rooms, but that they are not limited by the room conditions.

Most manufacturers will concern themselves solely with axial frequency response since they believe that most listeners will be content to sit in a narrow sweet spot and that reflections from room surfaces are not as critical to the ear as the direct response. They therefore allow huge variations in the off axis frequency response with 20dB roll-offs being quite common at 90 degrees off axis. In addition, "'normal" drivers and crossover configurations do not have consistent dispersion due to lobing effects caused by driver offset and wide overlap at the crossover points. It is quite common to find 15-18dB dips in the off axis response due to these effects, with the manufacturers being unconcerned since they are not axial effects. However, studies undertaken over a ten year period by Dr. Floyd Toule of the NRC proved that off axis response is very important to the perceived sound quality, since the reflected sound from the side walls comprises over 60% of the total sound field.

At VR, we use measurement equipment modified to overlay 20 pictures of the sound field at polar angles to the direct axial wavefront, in both the horizontal and vertical axis, out to 360 degrees. Rather than use the common but meaningless linear frequency response plot, we use polar response plots which show visually how the sound field emanates from the speaker. Only in this fashion can we determine if we are replicating the original sound field picked up by the recording mic. To our knowledge, no other manufacturer advertises that they use this method.

3. Transient Response Considerations

Dynamic, or "box" speakers have not been taken seriously in the past by some audiophiles who value the transparency of planar and electrostatic devices, since undamped enclosures, poor drivers and "textbook" crossovers which "ring" are the source of these colorations. Special tests were implemented 20 years ago by the Von Schweikert Research group to identify and eliminate these colorations. Energy storage (called ringing) added by undamped devices in the system are traced by pulses of fast, wide-band noise. All drivers in the VR-4 were designed to eliminate energy storage and sound as fast as any planar speaker. Cabinet resonances were eliminated with a high-tech method of constrained layer damping utilizing a very lossy plasticine compound with the molecular weight and specific gravity necessary to absorb vibrations in the frequency range desired (20Hz-600Hz). This material is compressed between twin sheets of 0.75 inch MDF fiberboard, resulting in a cabinet so dead that most first time listeners to the VR-4 believe that they are listening to a very fast electrostatic or ribbon that has 20Hz bass, tremendous dynamic range and "punch". Waterfall plots showing the decay of stored energy are used to verify the design goals and the success of damping techniques used in the VR-4.

4. Impedance Curves and Thiele Small Parameters

We use automated equipment by Bruel and Kajer, ATI, DRA Labs and others to measure technical specifications to insure both quality and repeatability. Impedance curves of the completed system are as important as frequency response measurements, especially in tracking down abnormal problems or variants in the quality control process.

5. Live vs. Speaker System

A final test at the design stage is the reproduction of live instruments and voice. Several calibrated mics are used to pick up the sound of an acoustic guitar, saxophone, cymbal, drum, and several human voices, both male and female. The speaker output is then acoustically compared to the original sound, with timbre and harmonic accuracy judged in real time under the same acoustic conditions.

Normal listening tests using recordings made by others (under dissimilar conditions with unknown mics) do not result in accurate conclusions, since no one knows how the recordings are supposed to sound except for the original recording engineers! This live vs. speaker system test is very humbling and has caused many of our designs to go back to the drawing board! However, it is the acid test of sonic truth; we know of no other test which will indicate whether the speaker is truly accurate, or merely pleasant.

NOTE: None of these tests were designed or pioneered by Von Schweikert Research; however, very few, if any, other manufacturers use this many tests. The process of cone design using strobe light was handed down to me by Dr. Oscar Heil, inventor of the Heil Air Motion Transformer (a folded ribbon tweeter with the finest measurements ever recorded) over 20 years ago. The live vs speaker test was first made known to me in a series of articles appearing in AUDIO magazine in 1954 by Edgar Vilkur of Acoustic Research. My experiments with cabinet wall damping hearken back to 1957, when my dad and I built a Jensen speaker kit which was supplied with a 3/8" plywood cabinet with no bracing or stuffing! although I am a firm believer in measurements, having worked with dr. Richard Heyser (the father of phase and inventor of time delay spectrometry) in the late 1970s, I also believe in empirical, scientific experiments, and use subjective listening tests based on a minimum of six auditioners (listeners) as the final arbiter.

Albert Von Schweikert