Roger Sanders of Sanders Sound Systems
Laurence Borden: Roger, welcome to Dagogo. Please begin by telling us how you first got interested in audio.
Roger Sanders: My parents introduced me to music when I was a child. I grew up playing in school bands and later in symphony orchestras and big band jazz orchestras. I played many instruments and learned to love music.
My father built his own speaker system back in the 1950s before commercial high fidelity systems were available. He also built Heathkit electronics to drive them. I helped him and was fascinated by what he created and by the amazing sound it produced.
LB: Did you have formal training in audio electronics, or are you self-taught?
RS: Both. I had formal training in music recording when I was in college in the late 1960s. This training resulted in the university eventually hiring me to do much of their concert hall recordings using the finest equipment of the era. While these recordings were always made on half-track, open-reel tape, I often had them produced into LPs for general distribution and sale.
After graduation, I bought my own recording equipment and made both live and studio recordings, a passion I have pursued all my life. In the 1970’s I had a small business doing on-location recordings and making LPs of concerts that the musicians could later sell. As part of my recording training, I also worked in the university’s electronics lab doing audio equipment calibration, measurements, maintenance, and repairs. This experience gave me the tools needed for me to design, build, test, and measure equipment of my own design in the future.
In addition, I was majoring in mechanical engineering and physics at the university. But the Vietnam War forced me to change to medicine before graduation. So although I had a lot of education in engineering, I never received a degree in it.
In any case, there were no specific engineering classes that involved electrostatic loudspeakers (ESLs). I had to learn most of this on my own. The two major sources of information on these were Peter Walker’s articles in 1955 in Wireless World Magazine and David Hermeyer’s articles in 1972 in The Audio Amateur.
I worked with David when he was at the California Institute of Technology. We discussed improvements to his ESL designs that I eventually incorporated into my early ESLs. He also was key in the development of my direct-coupled ESL amplifier design.
LB: Your name has become synonymous with electrostatic speakers. What sparked this fascination?
RS: During my university years, I worked with many speaker systems. But I was always frustrated with the muddled, distorted, and strained sound produced by the cone, dome, and horn speakers of the day. They could not reproduce my recordings in a realistic manner.
One afternoon I visited the home of a friend and discovered that he had the most unusual speakers I had ever seen — speakers that were in the form of a thin panel rather than a box. They were KLH Model 9 speakers and were my first contact with electrostatic speakers.
I was astonished by the clarity and detail of sound they produced. They sounded vastly clearer, more natural, and less distorted than any other speaker system. They simply reproduced the most realistic sound I had ever heard. I was hooked, and from that moment on I was determined to have electrostatic speakers in my home in the near future.
LB: When did you first start building speakers? Did you begin with ‘stats, or with more conventional designs? Did you dabble with magneto-planars?
RS: As a college student, I didn’t have much money so could not afford to buy KLH 9 speakers. I decided to build my own ESLs because in theory, they are relatively cheap and easy to build. Because I would do it myself, I figured that I would not have to make the compromises that KLH did, so I could actually build a better speaker than KLH.
Ah, the idealism of youth! Things are not as easy as they appear. Had I known everything that would eventually be involved, I might not have ever started down this road. But I charged forward and eventually succeeded. I did not build magnetic speakers as some of my friends were doing because conventional magnetic cone and dome speakers are relatively massive. Due to my training in physics, I knew that due to their high mass, magnetic speakers simply cannot match the performance of an essentially massless ESL.
Also nobody really builds their own magnetic cone/dome speakers. Such drivers are quite difficult to manufacture and are beyond the capabilities of amateurs to do.
So it is not accurate to say that one builds his own magnetic speakers. What amateur speaker builders really do is build boxes, not speakers. They then obtain drivers from manufacturers to put in those boxes.
I do not consider building boxes to house commercial drivers to be actually designing and building speakers. Doing so is more like rearranging the chairs on the Titanic when what you really need to do is build a better ship. So I started by building actual electrostatic transducers. I built my first electrostatic speaker system in 1972. It was a full-range, crossoverless design that was floor-to-ceiling in height and 2 feet in width.
I also built a special amplifier to drive them. This amplifier was a direct-coupled, high-voltage, Class A, transformerless design using 8068 radio transmitting tubes operating at 2,200 volts. It was based on David Hermeyer’s larger amplifier design and he was instrumental in helping me design it. I published several articles on ESL design, including my direct-coupled amplifier over the years. These are still available and can be accessed through my website on the “Audio Related Articles” page. Here is the direct link for your readers who may want to see these:
Many years later, I also started working with magnetic cone woofers due to the need to produce deep powerful bass, which is impossible to produce with a full-range ESL. But I started with ESLs and only later got into magnetic woofers out of necessity. Later, I designed and built true ribbon drivers. I had a strong interest in these because they are the only magnetic driver that can have mass that is as low as an ESL.
But ribbon drivers cannot reproduce sufficient output at midrange frequencies (much less in the bass), so they can only be used as tweeters. Due to this limitation, I abandoned them.
I have also dabbled in planar-magnetic speakers, but their mass is excessive and like all dipole radiators, they cannot produce deep powerful bass. I will not compromise, so I use ESLs for my best speakers and only use planar magnetic speakers for noncritical applications.
LB: Your book, “The Electrostatic Loudspeaker Design Cookbook,” was published in 1993. Have your thoughts on ‘stats changed dramatically since then?
RS: No. There is no magic. An ESL is an ESL. The only things that have changed are the construction methods and materials used to solve the extremely difficult engineering problems of ESLs. The only thing in my book that I would change is to stress the importance of stator insulation, and reveal better ways of construction due to improvements in materials.
LB: When and why did you switch from full-range ‘stats to a hybid design in which the ‘stat is mated to a traditional magnetic woofer?
RS: Having been young and idealistic at the time I built my first speaker system, I decided that I would not compromise. I would build the perfect speaker — no matter what it would take. So I went straight for the Holy Grail of speakers — the full-range, crossoverless ESL.
One of the wonderful things about ESLs is that they have essentially no moving mass. Therefore, they can reproduce the entire audio bandwidth with a single driver. They also are the only type of speaker that is driven uniformly over their entire surface. So they act as a piston at all frequencies (no cone breakup), so they have vanishingly low distortion — usually lower than the electronics that are used to drive them.
So my first ESL was a full-range, crossoverless design. However, it quickly became apparent that this approach was unsatisfactory. While the sound was very clean, detailed, and distortionless, the bass was a disaster. The problem is that a dipole radiator suffers from massive amounts of phase cancellation and radiation resistance losses when you get into the deep bass. While there are many ways of compensating for this, none are practical enough to solve the problem. These problems are compounded by the fact that electrostatic drive forces are quite weak, which makes it especially difficult to produce sufficient output at low frequencies.
While the problem of insufficient deep bass and poor output in full-range ESLs is widely recognized, what is not commonly understood is the other insoluble problem with electrostatic bass. I am referring to its high Q behavior.
Q is the engineering term for “quality” and it refers to the damping or control of a device. A tuning fork, bell, or drum is an example of a high Q device. Such devices are poorly damped. If you strike them, they will exhibit large amounts of overshoot and ringing that will cause them to continue to make sound long after the exciting force is removed. A block of foam rubber is a low Q device. If you strike it, nothing happens. It simply stops instantly, makes no sound of its own, and has no overshoot and ringing.
The low mass of an ESL makes it inherently a low Q device. It cannot ring or make any sound of its own because its mass is swamped by the mass of the air around it. This low Q behavior is one fundamental reasons that ESLs sound so clear and detailed. However, an ESL is built like a drum. It has a diaphragm that is under high tension. So at some relatively low frequency, the spring rate of the diaphragm will cause the air to which the ESL’s diaphragm is coupled to resonate.
Resonance is the very definition of high Q behavior. So an ESL will exhibit high Q behavior at its fundamental resonance. This causes the awful, flabby, wimpy, poorly controlled sound that is so characteristic of electrostatic bass. The resonance also causes gross errors in the bass frequency response.
Because of all the above factors, electrostatic bass is truly abysmal in both quality and quantity. I worked with every technique imaginable to solve these problems over the years, but the laws of physics cannot be circumvented. Electrostatic bass is simply unacceptable if you want realistic sound. One of the essential requirements of a SOTA (State Of The Art) speaker is that it be able to reproduce all the bass frequencies and do so at levels like those heard at a concert. Since electrostatic bass cannot do that, I had no choice but to solve the problem by making hybrid ESLs.
This is unfortunate, because many audiophiles falsely believe that full-range, crossoverless ESLs are the best speakers, when it is obvious that they have tremendous flaws that prevent them from ever being great speakers. Manufacturing full range ESLs is the cheap and easy way to meet these audiophiles demands. By comparison, a good hybrid speaker is far more difficult and expensive to build — but it is the only way to achieve SOTA performance. I do not compromise, therefore I build only hybrid speakers.
LB: The problem with many hybrid designs is that the ‘stat and the woofer don’t blend well. Your speakers are, in my opinion, a notable exception. Possible explanations for this phenomenon — which are not necessarily mutually exclusive — include improper crossover design, poorer transient response of the woofer, and differences in dispersion. What are your thoughts on this?
RS: I find it interesting that I am known as an electrostatic speaker designer when actually I have spent far more time and effort developing woofer systems to mate with them. ESLs are quite straightforward to design because their massless nature makes it relatively easy to predict their behavior. Well-behaved woofers are far more difficult to design.
The topic of making an ideal woofer is fascinating and too extensive to discuss here. But suffice it to say that the challenge is to eliminate the overshoot and ringing and general high Q behavior that plague conventional woofer designs. The difference in Q between massive woofers and massless electrostats is the main factor that prevents them from being well integrated. Only if you can produce low Q behavior from a woofer can it match an ESL.
You mention transient response. You are correct that this is a critical factor. It is very difficult to get a massive woofer to accelerate and decelerate as quickly as a massless electrostat. Any difference in acceleration will be heard as poor integration. But it is quite easy to get a woofer’s rise time to be fast enough to match the music signal at bass frequencies. All you need is a really powerful amplifier.
The main problem therefore is not starting the woofer, but stopping it. Most woofer systems have large amounts of overshoot and ringing (high Q behavior) because they won’t stop quickly. Obtaining low Q behavior from a woofer is impossible if the typical infinite baffle (closed box) or reflex (vented) enclosure is used. This topic is extensive so I cannot discuss it here. Your readers can contact me for a technical analysis if they wish.
I have found that only horns and transmission lines can produce low Q bass. Bass horns must be gigantic to reproduce deep bass properly, so they cannot be used in most homes. This leaves only transmission lines as the enclosure of choice. Crossovers are also extremely important to integration, and I will discuss them shortly. However, dispersion does not affect integration to a significant degree.
LB: As I alluded to above, your speakers have a reputation for having the best panel/woofer coherence of any hybrid. Without giving away trade secrets, how did you accomplish this?
RS: The following factors were key to producing the excellent integration of the drivers:
· The woofer needed to have a low Q behavior.
· It had to be small, but with a long length.
· It had to have no resonances.
· It had to be extremely rigid.
· It had to stop quickly enough to match an ESL — at least at bass frequencies
In addition, it had to have an aesthetically pleasing external shape.
LB: Was the choice of woofer driver critical?
RS: Yes. As I mentioned above, the Thiele/Small parameters of the woofer had to have extremely low Q. Additionally, the moving mass of the woofer had to be as low as possible. This required a relatively small/light voice coil, light suspension system, and paper cone.
To achieve high rigidity in the cone combined with light weight, it was made of woven paper fibers. The cone needed to be shaped to have excellent breakup characteristics. It needed to be impregnated with a thin film of Baxtrene to produce superb damping characteristics. Many audiophiles think that modern, exotic materials like light metals, Kevlar, or carbon fiber make better cones. But my research showed that paper remains the best material for woofer cones used to mate with ESLs, especially if the paper is made to be very stiff, light, and thin.
While the Q of the woofer’s cone can be controlled with a transmission line, the voice coil cannot because there is significant flexing between the cone and the voice coil. To control the voice coil, I used a magnetic damping system to control the woofer voice coil’s mechanical Q. An extremely large magnet structure was used as this is crucial to lowering the Qes and Qts. Finally, it is necessary to use a powerful amplifier with a very high damping factor. The combination of all these factors combined with a TL enclosure will finally produce satisfactory woofer transient performance.
There are surprisingly high forces in a woofer. So a very strong and rigid cast frame was required to prevent flexing of the structure that would have prevented the cone from following the electrical signal with precision. Of course, the frequency response had to be very linear. I had SEAS of Norway make a special driver for my application.
Passive crossovers have dreadful performance. This is another extensive topic that I cannot address adequately in this interview. So just allow me to summarize by saying that passive crossovers have the well-known problems of phase shift, hysteresis losses, group delay, inadequately steep crossover slopes, and distortion. Less appreciated is the fact that they insert inductors, capacitors, and resistors between the amplifier and the speaker’s drivers. This isolates the woofer from the amplifier and prevents the amplifier from having tight control of the driver. Or to put it another way, the damping factor of the amplifier is degraded by passive crossovers.
Since it is essential to use the amplifier to control the woofer, the amplifier must be connected directly to its driver without any intervening crossover components. Therefore electronic/active crossovers are essential to obtaining good integration. The amplifier must have a high damping factor. This excludes the use of tube amplifiers for driving woofers because their output impedance is too high. Powerful, well-designed solid state amplifiers have incredibly low output impedance (typically less than 0.1 Ohm), and massive current flow capacity, so they have the high damping factor and power needed to really control a woofer.
It is necessary to use electronic/active crossovers to achieve the best from any speaker system. I continue to be amazed by all the speakers on the market with passive crossovers that claim to be “reference” quality or SOTA systems. The truth is that electronic/active crossovers and multiamplifier systems are superior to any passive crossover system. All speakers will be improved with the proper use of electronic/active crossovers. So any speaker that uses passive crossovers simply cannot be considered the finest available. I am unwilling to compromise and so do not use passive crossovers in any of my current speakers.
There are many diverse types of electronic/active crossovers with many different features. Also, both analog and digital electronic crossovers now are available. I consider digital crossovers to be one of the greatest advances in modern audio. They offer features that I have wanted crossovers to have for years, but which were unavailable.
I am referring to the ability of digital crossovers to produce extremely steep crossover slopes (up to 96 dB/octave), the ability to eliminate phase shift, and the ability to insert delay into the signal. The delay can be used to time-align the speaker without having to physically position the drivers. Generally, aligning the drivers physically is cosmetically impossible at worst, and ugly as sin at best. It also isn’t all that accurate. A digital crossover will allow you to arrange for the sound from all drivers to arrive at your ears simultaneously, regardless of their location. This is a wonderful development.
A digital crossover has far greater precision than other crossovers because its filters are generated by computer calculations rather than by physical components that always have significantly flawed precision. The tremendous precision of digital crossovers makes steep crossover slopes possible. Steep crossover slopes are one of the most important factors in the woofer/panel integration in my speaker systems. There are several reasons for this. First, most audiophiles tend to think that the crossover point is where a driver stops. This is false.
The crossover point is the point where the driver just starts to roll off. Until the output from the driver has dropped by at least 48 dB, the driver will still be heard and will still be affecting the sound quality.
Since most passive crossovers operate at 12 dB/octave, this means that most woofers will still be producing significant output at midrange frequencies. No massive woofer can produce midrange frequencies well much less anywhere nearly as well as a massless ESL. So passive crossovers guarantee poor midrange performance due to the presence of woofer energy in the critical midrange.
Digital crossovers typically operate using 48 dB/octave slopes. By combining this steep crossover slope with a low crossover point (172 Hz in my larger speakers), the output from the woofer is gone by around 350 Hz. As a result, the woofer has no effect on the quality of the midrange. The midrange is produced completely by the electrostatic panel and therefore my hybrid speaker sounds every bit as clean and detailed as a fullrange ESL.
Another factor that is affected by steep crossover slopes is the distortion produced around the crossover point. Understand that no two drivers will ever act exactly the same way, especially when the drivers are as different as a massive woofer and a massless ESL. So when both drivers are reproducing the same frequencies around the crossover point, their different characteristics as they mix will cause distortion and phase anomalies that degrade the sound. By using very steep crossover slopes, I minimize the range of frequencies where this mixing and distortion occur.
While our ears are keenly sensitive to phase errors and distortion at midrange frequencies, they are quite insensitive to these flaws in the bass. So by using a very low crossover point and steep slopes, I am able to eliminate all sonic errors due to crossover issues.
LB: Let’s talk now a bit about the panels. In the early days, electrostats were plagued by arcing, sensitivity to humidity, and inability to play loudly. How were these shortcomings overcome?
RS: Electrostatics still are plagued by these problems. Even today, most ESLs cannot play loudly and are not very reliable.
ESLs operate on several thousand volts. This high voltage is very difficult to control and if it causes arcing, the panel will be damaged. The Achilles Heel of ESLs is their diaphragm coating. Panel failures are caused by a combination of diaphragm coating failure and inadequate stator insulation.
I have solved both these problems. The most difficult challenge was figuring out a manufacturing process that would achieve perfect stator insulation to over 20,000 volts. After many years of trying every technique imaginable, I finally succeeded by completely encapsulating the conductors in my stators inside a layer of glass. This makes it impossible for arcing to occur, even in high humidity conditions. My panels will play at ear-bleeding levels using multi-thousand watt amplifiers without arcing or any need for protective circuitry. They are also immune to dust, insects, and small foreign objects. No dust covers are needed. The diaphragms are designed to repel dust, so do not need cleaning.
My diaphragm coatings are utterly reliable, cannot be wiped off the diaphragm, are not damaged by arcing, and are flexible so they can’t develop cracks. My panels are very expensive and difficult to produce. But they are more rugged than conventional magnetic drivers and are so reliable that I can and do sell them with a lifetime warranty.
LB: Unless one listens with headphones, room acoustics contribute to the overall sound, often significantly. There are a number of studies suggesting that an ideal speaker would have an even power spectrum, whereby the off-axis frequency response is similar to that on-axis. Because of their width, panel speakers are at a disadvantage to conventional designs in terms of “beaming.” You are credited with being the first to implement curved ‘stat panels – which would lessen this effect – yet you ultimately rejected that design. Please share with us your thoughts on this topic.
RS: There is a lot more involved in the issue of off-axis performance than just frequency response. It also involves transient response, phase response, and imaging. I cannot discuss this huge topic here. I would encourage your readers to study my White Paper on the topic that can be found on my website. Here is the direct link:
But to try to summarize this topic, let me point out that no matter how good a speaker inherently is, the reflected, delayed sounds bouncing off room surfaces tends to ruin the performance of loudspeakers. This became obvious after I invented the curved electrostatic panel and then did research on its performance. It quickly became apparent that the common belief that a loudspeaker should have wide dispersion is a mistake. It is easy to prove that the best performance is produced by a speaker that minimizes the room’s sound and its interaction with the speaker. So I abandoned the curved panel design in favor of a planar panel that minimized room reflections and interactions.
I therefore will have to take exception to your comment that panel speakers are at a disadvantage to conventional designs. The fact is that the “beaming” to which you refer is actually a huge benefit and offers a great advantage over conventional, wide-dispersion speakers. This is because it makes it possible to focus a speaker’s energy on the listening location instead of spraying the sound all over the room, thereby producing delayed reflections that ruin the transient response, frequency response, and imaging of the speaker.
Let me also point out that many audiophiles who haven’t experienced a high-performance, narrow-dispersion speaker assume that it has poor sound when listening off-axis. This is not true. The nature of a speaker’s dispersion primarily involves the issue of imaging at the sweet spot. No speaker can produce a good image when you are not at the sweet spot because the sound from the two speakers will arrive at your ears at different times (out of phase). Also, the sound will be dominated by reflected/delayed room reflections.
Without accurate phase information, your brain cannot recreate an accurate, 3-dimensional image anywhere but at the sweet spot. This is why everybody always listens to their speakers at the sweet spot for serious listening. When casually listening to speakers out of the sweet spot, the image is very unrealistic. It will have a diffuse, directionless, and ethereal quality without any localization of the instruments. This doesn’t mean that the sound is bad — it may be excellent. But there will simply not be a realistic image associated with it.
When listening to narrow dispersion speakers out of the sweet spot, you will hear reflected sounds from the room like you do with conventional, wide dispersion speakers. So the sound will be similar to wide dispersion speakers out of the sweet spot. Where the real difference becomes apparent is when you go to the sweet spot where you will discover that the imaging and transient response is dramatically better with narrow-dispersion speakers than with wide-dispersion ones.
Of course, I must struggle with these common misconceptions and it is an endless task to try to educate audiophiles about the importance of eliminating room acoustics from the sound. But I simply won’t compromise the performance of my speakers by using curved panels. So I have an uphill battle to fight.
The good news is that once audiophiles here my speakers, the battle is won. Narrow dispersion speakers are so obviously superior that once an audiophile hears them, I don’t have to put forth any more effort to convince them. So I encourage your readers to keep an open mind about this issue so that they do not allow false assumptions to deprive themselves of truly great sound.
LB: My experiences of hearing your speakers under less-than-optimal “show conditions” (i.e., small, untreated rooms) certainly supports your claims.
Let’s switch gears now from speakers to electronics. Though you are best known for your speakers, your amplifiers (of which there are two distinct models) have received accolades in the audio community. Are they suitable only for ‘stats, or can they be used with conventional speakers as well? How do they differ from other amplifiers, and from one another?
RS: The quick answer to your question is that my amplifiers are suitable for all speaker types. It is a misconception that my ESL amplifier can only be used with electrostatic speakers.
The reason for this is that when I first started selling ESLs, I quickly discovered that most amplifiers could not drive them well. This was not because manufacturers made bad amplifiers, but rather that they did not make their amplifiers to drive the difficult and unusual load that an ESL presents. Specifically, an electrostatic speaker is a capacitor, which presents a very different load to an amplifier from that of a conventional magnetic speaker, which is primarily a resistor.
This situation forced me to develop an amplifier that could drive the extremely difficult load of an electrostatic speaker — and do so at high voltage and power without the need for any protective circuitry. I called the resulting amplifier “The Electrostatic Amplifier” because it was designed to drive electrostatic speakers.
Your technically-inclined readers may wish to have more detail of this amplifier design. I have written a White Paper on the topic and posted it on the “Technical White Papers” page of my website. The direct link to the ESL amplifier is:
The ESL amp’s ability to drive difficult loads makes it very easy for it to drive relatively easy loads like magnetic speakers. At the same time, it is very powerful (300 W/channel @ 8 ohms), which is the kind of power needed to make most speakers perform well.
As an aside, I continue to be amazed by the fact that most audiophiles are not aware that they are using underpowered amplifiers that are clipping and distorting most of the time. It is easy to show that most speakers need around 400-500 W/channel to play dynamic music at the loud levels audiophiles enjoy. If you doubt this, just connect an oscilloscope to your speakers and watch it. You will see the trace run into an invisible “brick wall” when the amplifier runs out of power and clips. In most systems, you will see the amplifier clipping constantly on musical peaks.
A clipping amplifier has very high distortion and awful performance. But it does not sound obviously distorted because only the music peaks are clipped, which are of very short duration. The average levels are about a tenth the level of the peaks. The average levels usually will not be clipped (when they are, you will hear obvious and gross distortion). Since the average level is much longer than the peaks, the sound will not seem to be distorted and listeners will think that the amp has adequate power.
But it doesn’t. A clipping amp will be behaving horribly. It compresses the dynamics, it will sound strained, muddy, and harsh — particularly if protective circuitry is triggered, which is usually the case. In fact, most of the sonic flaws heard in amplifiers are due simply to clipping.
Audiophiles often make false assumptions about these flaws in that they assume they are caused by various amplifier features (like class of operation, feedback, type of output device, etc.), when in fact the flaws are produced by clipping. In other words, amplifier power is the most basic and fundamental requirement in any audio system. An underpowered amplifier will always sound bad. So what is the point of using a low power amp that simply guarantees poor performance? I do not compromise, so I only manufacture very powerful amps.
My latest ESL amp produces 400 W/channel, so it will not only drive electrostatic speakers well, it will also drive magnetic speakers better than most other amplifiers due to its high power output. It did not take long for magnetic speaker owners to discover that the ESL amp would drive their speakers superbly. Magnepan owners were especially excited to discover that the ESL amp with its high power and lack of protective circuitry drove their power-hungry speakers very well. It became very popular with them. But eventually, I started to get requests to build an even better amplifier designed specifically for magnetic speakers.
I don’t produce products unless they offer a real improvement. So after studying the needs of magnetic speakers, it became clear that the key issue in the performance of magnetic speaker amplifiers is the need for a regulated power supply. This is because magnetic speakers make tremendous current demands on an amplifier’s power supply. This causes the voltage to fluctuate by around 30% in most amplifiers.
Such voltage fluctuations mean that the music will modulate the amplifier’s distortion and bias because these can only be optimized for one voltage. Obviously this is an undesirable condition. A regulated power supply will maintain a constant voltage regardless of load and therefore will improve the performance of an amplifier significantly. But the greatest advantage of a regulated power supply is amplifier power. Since an amplifier’s power varies by the square of its voltage, being able to maintain full voltage means that such an amplifier will produce much more power than the same amplifier without a voltage regulated power supply.
In short, amplifiers designed to drive magnetic speakers really need a regulated power supply. You wouldn’t even consider buying a preamp or CD player that didn’t have a regulated power supply, so why would you buy an amplifier without one? This is especially important because an amplifier needs a regulator even more than a low level component whose voltages are not modulated by the music. There are almost no amplifiers available that have regulated power supplies. The vast majority of amps on the market have free-floating, unregulated, power supplies.
Why? Because conventional regulators convert a large percentage of their power supply power into waste heat. This is not a problem in low power devices like preamps where only a few watts are wasted, but in a powerful amplifier the heat produced by conventional regulators is intolerable. I eventually figured out a way to make a power supply regulator that produced no waste heat. It ran cold. Nobody has invented anything like it. It currently has a patent pending on it.
Your technically-inclined readers can see the details on how this regulator works on my website at the following link:
When I added this regulator circuit to the ESL amplifier, I ended up with an amplifier that would drive any type of load and it would do so with a stable power supply voltage. The resulting amplifier eliminates the problem of the music modulating the bias and distortion and the power is increased to 500 W/channel. I named the new amp the “Magtech.” The Magtech is ideal for driving magnetic loudspeakers and it will do so even better than the ESL amp. But the Magtech costs $1,000 more than the ESL amp, so for those on limited budgets, the ESL amp remains a great deal. Where else can you get a high quality, high power (400 W/channel) linear amp for only $4,000?
Electrostatic speakers do not use power in the usual sense. So they do not modulate the voltages inside unregulated power supply amplifiers. Therefore the regulated power supply in the Magtech is not as important when driving ESLs as it is when driving magnetic speakers. But even though ESLs don’t modulate an amplifier’s voltages much, your power line voltages are not stable. So the Magtech still offers advantages over a unregulated amplifier and therefore is slightly better than the ESL amp even when driving electrostatic speakers.
In short, both my ESL and Magtech amplifiers can be used to drive all types of speakers. The Magtech is substantially better than the ESL amp when driving magnetic speakers, and slightly better when driving ESLs. As with most things, it all comes down to money. Take your pick.
LB: Do your speakers require special wires?
RS: The key word here is “require.” And the answer is no, my speakers do not “require” special speaker cables. Most speaker cable designs will work reasonably well on electrostatic speakers of all types. That said, it is also true that ESLs have unique cable needs that are not met by most cable designs, which are made for magnetic speakers. So I designed, recommend, and manufacture speaker cables that are optimized for electrostatic speakers.
Audiophiles are not required to use ESL cables, but they do work better in most cases. Again, I cannot discuss this extensive topic in detail in this interview, so I recommend that interested audiophiles read my Cable White Paper to learn more about electrostatic speaker cables. The direct link is:
LB: A hotly debated topic in audio is the role of measurements. How much importance do measurements play in the design of your gear? Have there been instances when the measurements flew in the face of subjective listening (and vice versa)?
RS: I find it astonishing that audiophiles question the value of objective measurements. The simple fact is that designing audio equipment is a scientific and engineering exercise. There is no magic involved, so scientists and engineers design and manufacture audio equipment — not magicians.
One can no more design and build a quality amplifier without using measurement equipment like volt meters and spectrum analyzers than one can build a quality house without using measuring tapes and levels. Listening tests have their place, but instruments are vastly more sensitive and accurate than human hearing and will reveal far more about the performance of electronics than what we can hear.
For example, adjusting the bias on one of my amplifiers involves measuring its distortion while adjusting the bias to achieve the ideal balance between distortion and waste heat (I want my amplifiers to run as cool and efficiently as possible). So I use a spectrum analyzer to observe the amplifier’s 5th harmonic distortion product and adjust the bias to set it to 115 dB compared to the reference level. This harmonic product is less than 0.0001% of the signal and is so quiet that no human could hear it. In fact, humans cannot hear less than 1% distortion. So even if the 5th harmonic was loud enough to be heard, one would not hear a distorted signal as the test tone would sound utterly pure.
In other words, it is simply impossible to adjust the bias by listening to the amplifier. It must be done using instruments. That said, it is also true that there are some limits to measurements. We cannot measure everything in audio.
For example, we cannot measure the “imaging” of a loudspeaker. This is because the image is produced by our brains and there is no way to measure it in the physical world. We can get clues to the imaging properties of loudspeakers by measuring their phasing and impulse response, but only our brains can directly form the virtual image produced by a pair of stereo loudspeakers.
On the other hand, when discussing electronics, modern test instruments (particularly the spectrum analyzer) have revolutionized our measuring techniques and can reveal everything there is to know about the sound that will be reproduced. It is now easy to predict with proven accuracy what you will hear through electronics by measuring them. Of course, many audiophiles will adamantly reject what I just said. They will insist that they have heard differences between amplifiers that measured well. They are right — and I agree with them. It is this type of difference in experience between what is measured and what is heard that has resulted in the unfortunate tendency for most audiophiles to reject objective measurements.
So what gives? Why do amplifiers often sound very different from what their measurements would suggest? I have studied this phenomena in detail and can explain the reason. It is quite simple once you understand what is happening.
Amplifiers are only measured when they are operating properly within their design parameters. A clipping amplifier measures horribly, so there is no point in measuring one that is clipping. But audiophiles usually are clipping their amplifiers when they are doing listening tests, as I previously described. You can safely assume that amplifiers that produce less than several hundred Watts/channel will be clipping musical peaks most of the time. I discuss amplifier performance in detail in my White Paper on the topic. The direct link is:
So the reason that measurements don’t correlate to most amplifier listening tests is that we measure amplifiers under one set of conditions while measuring another. You simply can’t compare a clipping amp to one that isn’t clipping and draw any valid conclusions about the inherent performance of the amplifiers because they will sound quite different. Please note very carefully that the measurements were correct — and so were the listeners. But because the test conditions were different, they could not be compared.
Another good example of this phenomenon is loudspeaker frequency response. A speaker manufacture may show that their speaker has reasonably linear frequency response when measured under anechoic conditions. But when that same speaker is placed in a typical reverberant room, its frequency response will be altered by the room’s acoustics and the speaker’s interaction with them. So the speaker probably will not sound like its published frequency response would have you believe.
Once again, we are measuring one set of conditions and listening to another. So the speaker will not sound like its measurements suggest, even though its measurements were accurate.
There is yet another problem when comparing measurements to listening tests: Most audiophile listening tests are seriously flawed because they contain multiple variables. By this I mean that there is more than one factor in the test that can is affecting the sound. When there is more than one variable present, it is impossible to know what is causing the differences in the sound you hear.
Now everybody recognizes the importance of having only one variable in a test. It is the most fundamental rule of testing. But audiophiles fail to realize that they break this rule all the time in their listening tests. For example, if you want to test amplifiers, you cannot change speakers at the same time you change amplifiers. If you do, you won’t know whether it is the speaker or the amplifier (or both) that is causing the change in the sound you hear. Without knowing the cause, the test is meaningless.
Cause is the thing we are trying to determine in our tests. So it is absolutely essential that we only have one variable in our tests. Most audiophiles think that if they test amplifiers by listening to one amp for a while and then changing to a different one that the only variable in the test is the amplifiers. But this is not true. There are actually many uncontrolled variables in the typical audiophile test.
One of the most obvious variables (in addition to the amplifiers under test) is that the output levels will be different between the amps since amplifiers have different gains and power capabilities. Although our hearing is acutely sensitive to level differences, we will not hear small differences in level as “louder” because the dynamic nature of music makes it difficult. Instead, we will perceive slightly louder music as sounding “better.” This is why we prefer to listen to our music at loud levels instead of quietly. So unless the levels of the amplifiers are matched exactly, we will hear the louder one as “better” even if it actually sounds identical to the quieter one. Obviously such a test is invalid because we would draw a false cause/effect conclusion.
So to do a useful test where we can really draw accurate cause/effect relationships, we must eliminate the loudness variable (as well as many others). And note that it is essential to use a volt meter to measure the levels because they must be matched to within 1/10th dB and our hearing simply is not sufficiently accurate to do so. The topic of testing is extensive, and I have just touched on the problem in this interview. I strongly recommend that readers check out my “Testing White Paper”, which discusses this topic in the detail it deserves. It will then be clear what has to be done to do meaningful and valid listening tests. Here is the direct link:
Accurate listening tests are possible, but they are difficult to do properly. They can reveal the basic cause of what we hear, but they are very limited when it comes to identifying problems and correcting them. For that, instruments are far more useful. For example, a skilled listener can easily hear that a certain speaker has a “suck out” say in the “lower midrange.” That is easy to fix using a DSP (Digital Signal Processor) — but only if you know the exact frequencies involved.
But the problem is that the subjective “lower midrange” involves too broad a frequency range to adjust the DSP by ear. Just what frequencies are causing the problem? It is nearly impossible to fix such frequency response problems by just listening to the speaker and fiddling with the controls on the DSP. So to use the DSP, you must measure the speaker’s frequency response using something like an RTA (Real Time Analyzer), which normally comes as part of a DSP. It is then easy to see the frequencies that are deficient and compensate for them as required. This just isn’t practical to do by ear. So once again, measurements are absolutely essential.
As an aside, let me point out that the RTA would reveal what the frequency response of the speaker would sound like in that particular room. You don’t have to listen to the speaker to know that it will have a sucked out lower midrange since the RTA will measure this accurately and predict the sound.
In summary, measurements are not optional. They are essential when designing, manufacturing, testing, and evaluating audio equipment. Measurements are easy and fast to do — and they are extremely accurate. They are far more sensitive and will tell you much more about audio equipment than listening tests. Measurements can be of tremendous value to audiophiles. The only exception to this is some aspects of loudspeaker performance that cannot be measured with instruments. But all aspects of electronics can be measured with greater sensitivity and accuracy than what a human can hear.
Listening tests are also useful and accurate — if they are done well. But they are much more difficult to do than measurements and are very limited in their usefulness. Listening tests and objective measurements correlate perfectly as long as the same conditions are being tested and if the listening tests are done properly. Whenever the two don’t agree, you may safely assume that you have overlooked something in your testing that is causing the results to differ.
Instead of rejecting scientific testing (both instrument and listening tests), audiophiles should embrace it. The lack of valid testing of all types is destroying the audio industry. This is because inaccurate and invalid testing deceives audiophiles and allows the snake oil salesmen, charlatans, and unscrupulous manufacturers to ply their trade and relieve gullible audiophiles of their cash. If the truth were known, there would be good agreement about what equipment sounded better and what didn’t. We wouldn’t have all the conflicting and confusing opinions and just plain lies that plague the audiophile industry today.
The truth is available, but the industry in general doesn’t want audiophiles to know it. It is essential that confusion abound so that all sorts of false claims can be made to sell audio equipment and nobody will believe measurements or valid listening tests that prove that these claims are bogus. I find it utterly astonishing that audiophiles are so gullible and willing to believe all the voodoo science out there. They should demand proof of claims before dropping tens of thousands of dollars on equipment. This situation is an extremely sad and frustrating commentary on our industry.
LB: High-end audio has been undergoing numerous changes in the past decade, and the recent economic downturn has certainly not helped. Sanders Sound Systems currently sells direct to the consumer. Do you see this as the wave of the future? Will, as some have suggested, audio shows take the place of “Brick and Mortar” audio shops?
RS: In today’s economy, a manufacture needs to reach out to make sales wherever possible. The traditional business model of selling only through audio dealers is very limiting and makes it difficult to produce adequate sales volume. Selling through dealers is very expensive and limited. In particular audio dealers are too few and far between to serve all my customers, especially because I sell globally. In fact, 83% of my sales are international and when selling in places like India, China, Brazil, and Pakistan dealers are essentially nonexistent. So factory-direct sales is the they way I can serve customers in many countries.
But this does not mean that I do not sell through brick and mortar dealers. I have several such dealers, but I just don’t limit myself to them. I also work through “in home” dealers who will demo my gear for audiophiles in their area to make good commissions to supplement their incomes from other jobs. In short, I work through many channels to serve customers everywhere.
I do see this as the future of audio. It appears to me that there is no other way to do it if you want to be successful — and we are. We are having to work hard to keep up with demand. We work 7 days/week now. We have no debt, own our factory, and have plenty of cash reserves in the bank. My business model works.
I do not think that audio shows will replace dealers. Audio shows are a good way for customers to hear my products, but audio shows do not provide the service that customers expect. Also, the demos done at audio shows are very limited due to time constraints and the need to run many listeners through the demo room in a very short time. So I consider shows to be a form of advertising, not service. The only thing that takes the place of dealers is direct sales using my 30-day, risk-free, in-home trial program. Customers can then have plenty of time to listen to my equipment and they can contact us for support and answers to their questions.
LB: Your products are all very reasonably priced, given their quality and performance. I suspect I know the answer to this question but let me ask it anyway: What are your thoughts on the “sky’s the limit” pricing that seems to dominate high-end audio?
RS: Overpriced equipment is the other factor that is killing the audio industry (bad testing and deceit are the others). Young people today simply don’t have the disposable income available to throw money away on absurdly expensive products. And youth is the future of our industry as the older audiophiles are fading away.
Oh sure, some wealthy individuals are still deceived and think that a more expensive product must sound better than one that is reasonably priced. But once you reach a price point that is high enough that compromises don’t have to be made, that simply is not true. For example, my amplifiers and speakers perform as well or better as anything on the market. I’ll stack them up to any product at any price. And I can prove by using valid testing (both instruments and listening) that they match or exceed the performance of all comers. For less than $20K I can supply a SOTA system. You do not have to spend hundreds of thousands of dollars on an audio system to get state of the art sound. I’ll be happy to prove it to anybody who cares to contact me about this.
LB: Are there any new products on the horizon, or will you be sticking with the current lineup for the foreseeable future?
RS: I don’t make new models for styling or marketing reasons like the car industry and some high-end manufacturers. I only introduce new models or upgrade equipment when I genuinely have something new and better to offer. I don’t like to play such games. I prefer to make superb performing equipment at reasonable prices and sell it to those who appreciate what I am doing.
This marketing plan doesn’t appeal to everyone as many are duped by the latest gimmick. But I have plenty of sales, so I don’t need to make cosmetic changes to basic equipment in an attempt to attract new sales. Instead, I will continue to do what is ethical and honest. I will tell customers the truth, educate them, sell the best performing equipment at reasonable cost, provide the best service and warranty in the industry, and introduce new products when I have something genuinely better to offer.
Currently I build two different sizes of hybrid ESLs. I build the ESL and Magtech amplifiers in both stereo and monoblock versions. I have recently consolidated my preamps from a separate line stage and phono stage into a single unit at half the price. I did this to give better value to my customers. I make custom cables for both ESLs and magnetic speakers. I sell digital crossovers. My plate is full, but I am working on new products to my line in the near future.
I hope your readers find this information helpful and entertaining. They may feel free to contact me with questions. Thank you for the interview.
LB: Roger, on behalf of Dagogo and our readership, I’d like to thank you for taking the time to share with us your considerable knowledge and experience. We wish you continued success.
- (Page 1 of 1)