[SoundStage!]SoundStage! Interviews
August 1998

Richard Vandersteen of Vandersteen Audio: Part One

by Doug Blackburn (db@soundstage.com)

What’s the best way to decompress after a five-day high-pressure event like HI-FI ’98? Take a couple of days off then go on another audio safari involving six hours of driving! In this case, a factory tour of Vandersteen Audio, the first ever done, and an interview with the company founder and energizing force, Richard Vandersteen, were my goals. The factory tour will be coming up, but the interview was just too interesting (and too long!) to wait that long. So we’ll be giving you the interview in digestible pieces to help you get through it all.

We don’t often get to hear the deep-down rationale behind the design of high-end products, but in this interview, Richard Vandersteen makes some interesting observations about his speakers, other types of speakers, the criticisms Vandersteen loudspeakers have taken for 20 years from some quarters of the audiophile brotherhood, and other interesting facts both small and large.

I started the "Vander-Day" from my brother’s home in the western suburbs of San Francisco. Vandersteen Audio is located in Hanford, CA almost midway between San Francisco and Los Angeles. This meant a three-hour drive down the simultaneously fascinating and boring Interstate 5. It's fascinating because much of I-5 follows the western edge of the San Joaquin Valley where some unbelievably large portion of all the food grown in the US comes from. The fields are truly amazing. But the road is flat, straight and featureless. Where there is so much growing vegetation, there are bugs and the windshield was nearly covered with smashes within 60 to 90 minutes. This necessitated cleaning the windshield two times at gas stations during the three-hour drive.

Once at the Hanford exit, there was another 30-mile drive east toward the Sierra Nevada mountains which remained snow covered along the top 20% of their peaks. Arriving in Hanford, you are struck by the feeling of California the way it used to be -- before LA boomed, before Silicon Valley, before the crazy real estate markets. Vandersteen Audio occupies a well-kept industrial building just off the main east-west highway through Hanford. People and mail use the door on 4th Street. Shipping and receiving percolate through the dock on 5th Street. To get away from the factory noise and telephone interruptions, Richard and I talked at the Vandersteen homestead, less than a 15-minute drive from the factory.

We start our interview with a question about first-order crossovers, one of the hallmarks of Vandersteen loudspeaker designs. You can’t help thinking that the additional effort needed to build successful first-order crossover loudspeakers wouldn’t be expended if there weren’t some awfully good reasons. Richard Vandersteen thinks he has some awfully good reasons for making speakers the way he does.

A Brief Explanation of First-, Second-, Third-, and Fourth-Order Crossover Slopes

In the figures below, the black lines represent the operating range of a woofer. The blue lines represent the operating range of the tweeter. You will notice as the crossover slope changes, the amount of overlap in the range of frequencies covered by the two drivers decreases. In the fourth-order crossover, there is a very limited amount of overlap between drivers. Normally, you would like to have a small amount of overlap between the two drivers; however, the electronic components required to create second-, third-, and fourth-order crossover slopes increase time delays and phase shifts that are applied unequally to the signals going to each driver.

NOTE: The graphic above is intentionally simplified. It is used only to illustrate the concept of crossover slopes. The graphic does not accurately depict frequency response of a loudspeaker.

The flat parts of these simulated response graphs indicate the main operating range of the woofer and tweeter. The sloped portions show how quickly the output of the woofer and tweeter drop off. The point where the woofer and tweeter response touch is supposed to be the -3dB point, meaning this is the point where the woofer and the tweeter are 3dB down in response from flat. Below the -3dB point is the zone where sounds are reproduced by both the woofer and tweeter simultaneously, the overlap zone.

Doug Blackburn: Tell us about how you came to decide that you were going to build your loudspeakers with first-order crossovers.

Richard Vandersteen: In the ‘70s we were beginning to play with loudspeaker designs and the realities of what external driver diffraction due to the enclosure surfaces and internal diffractions inside the loudspeaker enclosure were doing sonically. We started to look at things that would remove these secondary distortions that come from secondary radiation. Secondary radiation is most easily defined as sound which exists because of the reproduction method/device. Secondary radiation is 100% distortion because none of it exists in the original input signal. The more secondary radiation you can remove or prevent, the better the speaker will sound.

To see what was possible sonically without the secondary radiation products, we mounted drivers naked in space, no enclosure or baffle at all. In the ‘70s, second- and third-order crossovers were by far the most common types in use and that is what I was working with. Experimentally, but almost by accident, we discovered that a tweeter in this free-air fixture took on a completely different sonic character depending on whether it was driven by a first-, second-, or third-order filter. It’s wasn’t just different because the 3dB-down points were different for each crossover or because of the slower roll-off in the lower-order crossovers. The tweeter took on a noticeable pinched, twangy sound with a second-order crossover, which was even more noticeable with a third-order crossover. Only with the first-order crossover did the tweeter sound natural.

In this case, "natural" refers to how accurately the driver or drivers under test reproduced sounds and music we recorded. Besides musical instruments and voice, we would record things like a shovel scraping on concrete or shaking car keys. We’d then use the same shovel on the same bit of concrete in the test area, live, to compare how drivers (and crossovers) sounded playing the recordings of the same sounds. It was really helpful to be able to duplicate these recorded sounds live right in the same space where the drivers and crossovers were being tested. We tested woofers, midranges and tweeters separately and as sets to see how individual and ensemble performance of the drivers worked in these controlled conditions. In fact, one phase of testing involved putting the driver setup that was under test behind a curtain and bringing in a panel of listeners who could not see what they were hearing and did not know what they were listening to. We found from this testing that our own observations and the observations of the blind panel were in strong agreement: The setup with first-order crossovers always sounded far more natural than the same setup with well-designed second- or third-order crossovers. At the time, we didn’t know why this was true. We were just getting started and first-order crossovers weren’t something that had been written much about in audio reference books of the day. But a little research and we found what still seems to be the important thing today -- first-order crossovers are the only type which do not introduce time delays or phase distortion.

dB: Why is time and phase distortion so noticeable once you make comparisons to sound produced with and without time and phase distortion present?

RV: The audio-products industry is fixated on amplitude-related performance of loudspeakers and electronic components. They all look for amplitude-based performance parameters like frequency-response curves. Consumers have been lead down this "amplitude response is everything path" forever by the magazines and manufacturers. So it’s hard to get people to understand that there are other factors that are even more important than amplitude response.

Early on, it became obvious to me from the testing that we were doing that the ear-brain mechanism could easily forgive minor, up to even modest narrow-band amplitude imperfections. But in the time/phase domain, even small amounts of time/phase distortion would greatly affect the live versus recorded test results. In the 20 to 30 years that have passed since our early experiments, I’ve become even more convinced that time and phase performance are far more important than the high-end industry has recognized. All of this knowledge we accumulated about the importance of very low levels of time/phase distortion came about as a result of those first experiments which were really looking for answers to other questions -- how to reduce and eliminate secondary distortions in loudspeakers.

dB: What are we missing or what don’t most people know or understand about the relationship between the audibility of amplitude differences versus time/phase differences?

RV: Here’s an example: If you set up a test so that you can vary frequency response over a controlled frequency band, say half an octave. And you set the test up very carefully so that when you increase or decrease the sound level in this half octave, that there is no change at all in the time and phase domains, you find that 3dB differences are just barely audible. But if you make a 0.5dB change and shift the phase at the same time, the difference is immediately noticeable. Only when phase or time delay were changed along with amplitude did the 0.5dB level changes become obvious. The digital test equipment that has been available for a while is really useful for this kind of testing. You can select any level change you want and get it with zero time/phase changes. You can also select time delays and phase shifts with or without level changes. It was harder to do this kind of testing in the pre-digital days, but we’d done some things just well enough to know we we’re on the right road. Today’s digital test equipment just helps prove the point even more convincingly. Time/phase distortion is simply more audible to us than most of us realize or recognize. The magazines and most manufacturers don’t get this yet. [grins] After all, it has only been 20+ years since time delay and phase shift were looked at seriously by us and some other companies. [still grinning]

dB: What can you say to people at home reading this who say, "3dB? Hell, I can hear MUCH smaller changes than 3dB. The stepped attenuator on my preamp has 0.5 (or 1 or 1.5dB) steps and I can hear each and every step completely and obviously."

RV: Well there you’re talking about a full-range change from 20Hz to 20kHz. And I agree with that, 3dB is easy to hear if it is full range. The ear-brain discrimination mechanism has more trouble when you limit the range of frequencies altered to half an octave or so. Our panel tests proved that this discrimination threshold varies from individual to individual a little bit. I believe that given enough time and familiarity, most of our panelists would eventually get quite a bit better at discrimination thresholds, but it’s something you have to give people a means to do accurately then give them a lot of time to work with it in a system and room that they are very familiar with. Test conditions mask about half of people’s perceptive abilities.

dB: When doing the tests you were doing with amplitude and time/phase audibility, you varied the sound level 0.5dB over half an octave. Did you also introduce the time/phase shift only over the same half octave?

RV: Yes. And going back to the issue of the audibility of a 3dB overall sound level shift, if you did that using a continuously variable pot you’d get a lot different results than with a stepped attenuator. Turning up a continuous pot very slowly is not immediately obvious and most people won’t notice a level increase until you get to 3dB or so -- but only if you adjust the level slowly.

dB: What makes loudspeakers with first-order crossover slopes and with no time or phase distortion harder to make than speakers which ignore time and phase distortion.

RV: The disadvantage first-order crossover designs have is the large area of overlap between drivers. Unless the drivers remain in a completely pistonic mode through that entire large operating range you have to manipulate the phase and amplitude response within the crossover, within the driver and within the loudspeaker enclosure to get the drivers to produce the flat phase, time and amplitude response pattern you want at the listening position. For the drivers to remain completely pistonic over their entire operating range means there can be no cone breakup or resonances over that entire operating range. This isn’t an easy thing to achieve, and even the best drivers will need some help from the crossover, enclosure and special attention to design and materials selection to be good enough to work well in loudspeakers with first-order crossovers. An interesting side effect of driver development for first-order crossovers is that as you eliminate more and more of the shortcomings of the driver in the problem areas, the performance of the driver improves noticeably in non-problem areas too.

You can’t just take two or three or four drivers and put first-order crossovers on them and have a time and phase correct loudspeaker. In a two-way speaker, first-order crossovers mean a single capacitor in series with the tweeter to roll off the bottom end of the tweeter and a single inductor in series with the woofer to roll off the top end response of the woofer. If you put normal drivers (off-the-shelf) in this two-way speaker, you’d more than likely have to wire the woofer and tweeter with opposite electrical polarity to get anything approaching linear amplitude (frequency) response. But it’s my opinion that putting drivers out of polarity in a loudspeaker is something you do in a cheap speaker, not in high-end speakers. Imagine a crossover at 3kHz and a sound being reproduced at that frequency. Half the energy (in our two-way example) will come from the woofer, the other half of the energy will come from the tweeter. One driver will be moving forward while the other is moving backward. You get a full or partial cancellation of that sound because of the polarity reversal. As you move away from the crossover point, the amount of cancellation decreases. To get those two drivers working in the same polarity requires a crossover with a lot of additional components and probably some significant mechanical or material design changes to the driver itself. But that’s what high-end design is all about -- doing the design the right way to end up with a product with the best possible performance in every possible way.

If you are using second- or third-order crossovers and one of the better computer programs for speaker design, you can plug in stock drivers with the right specifications and assemble a crossover that that works well with the drivers and end up with something pretty close to a finished product. First-order crossover designs aren’t that easy. It takes a lot more work and incredible demands on the drivers themselves. It isn’t that doing the work on a good first-order speaker design is impossible or anything, but it is expensive and takes a lot more time than designing loudspeakers with second-, third-, or fourth-order crossovers. I look at this as another thing that separates high-end designs from other products: True high-end designs, to be worthy of being called that, should be paying attention to the difficult details.

dB: There are some loudspeakers being made which have physically staggered drivers, the old "line up the voice coils" look to them. They either slope the front baffle or make a series of progressively smaller sub-enclosures which are staggered in space. But they are not time and phase correct loudspeakers because they do not employ first-order crossovers. Just how far would drivers have to be staggered to eliminate the time delays in second-, third- or fourth-order crossovers?

RV: Well it certainly isn’t practical, but I think I understand the point of your question. If the crossover was fourth order, the distance between drivers would be huge. Some manufacturers confuse the "time and phase correct" phrase for the consumer, making it harder to understand what their speakers are really doing. They will claim that their speakers are time and phase correct if they get all the drivers working in the same electrical polarity. Most of the speakers made today have every other driver connected with reversed electrical polarity to get the frequency response to look reasonable. With that kind of speaker (mixture of driver polarities) you could never get the speaker into absolute phase because you could only get every other driver into absolute phase at any one time. So in the last ten years or so, some loudspeaker designers have confused the issue by building, for example, a fourth-order design where they have worked hard to get two or three or four drivers connected with the same electrical polarity. They then refer to their design as time and phase correct.

To my way of thinking this is really misleading to the consumer since these fourth-order designs still contain large amounts of phase shift and time delays between drivers in spite of the fact that all the drivers are in the same electrical polarity. To have a truly time and phase correct loudspeaker design, you should be able to input a broadband square wave into the loudspeaker and get a triangular wave out of the speaker (because loudspeakers can’t do DC to light, so the edges of the square wave will be rounded or angled off by the speaker). The time and phase correct fourth-order loudspeaker will not reproduce this broadband square wave pulse with anything approaching the coherency of a true time and phase correct loudspeaker with first-order crossovers.

If you look at the impulse response graph in Stereophile magazine, this is exactly the kind of response I’m talking about. The leading edge is very steep, just like a square wave with the tweeter response starting right at the top of this vertical line. In a first-order loudspeaker with correct time and phase response, the midrange driver’s response connects to the tweeter’s response and gradually decreases in level on an angle (a slope). The woofer’s contribution is next and it further continues the slope of the falloff. If you send the same impulse to any speaker which does not have correct time and phase response, you’ll see one of two things: separate responses for each driver but all in the same polarity or a positive tweeter response followed by a negative midrange response because the phase response is radically different, 360 degrees or more phase shift, followed by a positive woofer response. The latter examples are not reproducing the impulse accurately, so those loudspeakers cannot reproduce music accurately. The flatter the sloped line is in the impulse test graph, the flatter the frequency response of the loudspeaker being tested. A lot of people don’t understand the importance of the impulse test and the resulting graph of impulse response as it relates to audible performance of the loudspeaker.

dB: Looking at various Vandersteen loudspeakers, you can see that there isn’t more than 2" to 4" of stagger in the positioning of the drivers to achieve the correct time-aligned position of the drivers. Explain two things: What are you really lining up by staggering the drivers and how big the stagger between drivers would have to be in speakers with something other than first-order crossovers.

RV: Well, the stagger distance would be very large for second-, third-, or fourth-order loudspeakers. With narrow and deep loudspeakers being fashionable these days, there might be enough depth in some of them to physically stagger the drivers enough without making the speaker any deeper. But that much stagger introduces such severe diffraction problems that it’s really impossible to make a commercial product that way. As far as what we are lining up, it isn’t the voice coils as consumers have been taught by the magazines. We’re really lining up the acoustic center of the drivers. This is an imaginary point where the sound can essentially be said to emanate or originate from. This point is different for different drivers, so you have to figure out where this acoustic center is for each driver in each different loudspeaker model you use it in. The design of the crossover is a significant element of determining just where that acoustic center is located. So you cannot just make a blanket statement that you "line up the voice coils" to get the drivers staggered correctly. If fact, it’s unlikely that lining up the voice coils would give you the right amount of driver stagger.

Our stagger is part of our minimum baffle design. The baffle is the flat surface the driver is mounted on. We try to make that surface as small as possible because our research proved and keeps proving as we revisit those tests that the smaller the baffle, the less surface there is around the driver, the better the driver sounds. Using foam or felt around the driver to attempt to reduce the effect of the baffle is better than nothing, but no baffle at all is much better. This is a direct result of learning about diffraction and how detrimental it is to the sound of loudspeakers by introducing large amounts of the secondary distortions we set out to eliminate in our speakers way back in the ‘70s. Small or no baffle produces no diffraction distortions. We put each driver in a separate enclosure so we can minimize the size of the baffle around the driver.

Internal diffractions or reflections are another thing we try to control. We use long damped transmission lines in an attempt to reduce the amount of energy that can bounce back against the rear surface of the driver and cause it to move in the absence of an electrical signal. That reflected energy, whether from a baffle or internal reflection, is not a part of the original input signal, so loudspeaker designers should be doing all that they can to keep those diffractions and reflections from contributing to the sound heard at the listening position. Those sounds are in fact 100% distortion. They weren’t in the original input signal; they originated at the loudspeaker as artifacts of a large baffle or poorly controlled internal reflections. Another trick we employ and have patented is reducing the diameter of the magnet structure and shaping the magnet structure of the driver. In conventional drivers, pressure waves coming off the back of the driver’s cone do hit the magnet and reflect right back onto the driver cone and cause the cone to make a sound that should never be there, another secondary distortion. We have been eliminating more and more of this reflection as we refine our drivers over time.

dB: [laughing] Richard, this could be something that will go right over people’s heads. I mean, it’s obvious to me what you are saying is true, but you said it in such a simple way that the impact, the importance of the facts, isn’t going to be fully appreciated by a lot of people reading this interview. I hope they do get it though. It’s a significant fact of life that is not being addressed very well out there in speaker-land.

There are people, smart people, in various aspects of high-end audio from publishing to manufacturing to research who are convinced that time differences smaller than 4 or 5 milliseconds (.004 - .005 seconds) are inaudible. Most of the time delay artifacts from loudspeaker crossovers and lack of stagger of drivers fall within this 4 to 5 millisecond window so people think their designs are as good as they can get as long as they are within the window. Let me do the math quickly -- using 750 mph as a rough equivalent for the speed of sound translates to approximately 13" of physical distance for each millisecond of time delay. Five milliseconds would translate to just about 5 1/2'. To me it seems incredible that the time delay between two sources separated by 1' to 5' to my ear would be inaudible.

RV: There are two things going on here. Number one is that there are people who can’t hear differences between amplifiers or interconnects or speakers cables or different CD players doing the research that leads to bogus rules of thumb like this 4 - 5ms window being the threshold of audibility. That doesn’t mean an audible difference doesn’t exist; it just means the people doing the research honestly did the best they knew how and ended up with a result that isn’t accurate. How could you expect a different outcome? They certainly would never have made the startling discovery that 1ms or 0.5ms is the real threshold of audibility.

Number two has to do with what goes on with many of these tests that some sharp people have accepted as being true. These are, after all, tests. When you put human beings under test and they know they are under test, results change. Human beings under test do not react to those things our right brain tells us are important under relaxed, non-test conditions. We respond to things very differently when we are in our own listening rooms listening to our own equipment and our own favorite music. In that environment, our right brain becomes an integral part of the listening experience, and our discrimination faculties are doubled or squared.

It’s very difficult to take a person who is known to be a good and experienced listener and put him in an environment where he knows he is being tested and get anything remotely resembling the results he gets in his own familiar listening environment.

Fundamentally, it comes down to the fact that the tests that give us all of these magic numbers are flawed. None of them have done anything to get the test subjects into the proper state of mind. By that I mean, as conducted, the tests exercised only the analytical side of the human mind. The thing that we need to make the test accurate is unavailable to us in the test environment because of the way human minds work. Put us under test and the right brain takes a hike and leaves all the work to the analytical left brain. So far, in these test environments it has been impossible to get people into the same relaxed state of mind that they are in when they are out enjoying live music or when they are listening to their own systems for pleasure.

Many audiophiles suffer from this too. They can’t stop analyzing their system and enjoy music. They have to constantly compare new components or wires. They tweak incessantly. They clean connections. They are constantly fiddling. These people can’t ever let their right brain loose long enough to just sit down and enjoy the music. Their left brain is in overdrive all the time.

dB: Perhaps this is why a certain herb is so popular among some audiophiles [grins] -- it beats the left brain into submission and lets the right brain come out to play. Engineers, scientists, and other technical types seem to benefit the most while artists, musicians, and fiction writers wonder what the big deal is.

RV: OK. But I can’t endorse that kind of thing though, you know. [grins] If you ever find yourself in "test conditions" and unable to hear differences you know you should be able to hear, try this:

Stop trying to quantitatively analyze the bass, the midrange, the dynamics, the transparency and all that and concentrate on your emotional response to the music being played. Remember your feelings about the music and you’ll find that all of a sudden, differences exist where they did not exist a few minutes ago. If you keep trying to be objective about the sound quality, everything just runs together and you find yourself unable to form an opinion.

[SoundStage!]All Contents
Copyright 1998 SoundStage!
All Rights Reserved