THE ENGINEERING BEHIND SPECTRON SOUND

Spectron does not buy its class D modules from a third party! Our designs are our own, on the cutting edge of the technology. Because we have complete control over our designs, we are constantly improving them. Moreover, Spectron's president and chief designer, John Ulrick, introduced the first commercially available class D amplifier into the audio world at CES 1974. At that time John was president and co-founder of Infinity Systems, famous for its pioneering development of the combination of the servo woofer with the electrostatic speaker. John has devoted the last 25 years to extending a "control system" approach to class D amplifiers, and the Musician III MK2 is his and the Spectron design team's latest achievement

Here's how class-D works. The input audio is converted to a pulse width signal called a modulated carrier, not unlike the AM or FM modulation associated with radio. Like radio, the audio signal is contained in the modulation of the carrier. This carrier frequency is 500kHz, several times greater than the highest audio frequency. This carrier is a square wave with a modulation index that varies from 0 to 1 or PWM (Pulse Width Modulation). The modulator drives a power section that converts this modulation index into a high voltage level which can drive a speaker. This is done using two switches which are high speed transistors switching ON and OFF at the 500kHz carrier. These two switches raise the voltage level up to the required level to drive the speaker. The amount of time each of the two switches are ON or OFF is controlled by the modulation index. At this point, we have a high voltage (+125V) PWM signal to drive the speaker. We now need to pass this signal through a low pass filter in order to stop the carrier from passing onto the speaker but allowing only the audio to pass on to the speaker. This describes the typical open loop class-D amplifier which would be useable as is, but….. with feedback we can make substantial improvement in terms of both measured and sonic performance.

How Feedback is implemented is the Key to Outstanding Sound in a Class-D Amplifier: A power amplifier is a voltage control system with sufficient current delivery capability to maintain the output voltage no matter what the speaker load. So let's look at it from that point of view. Power amplifiers simply take an input voltage and amplify it by a factor of 20 to make the output voltage high enough to drive the speaker. When you load the amp with a speaker, the current drawn by the speaker causes the output voltage to fall. So the entire key to low distortion and flat frequency response, in other words the accuracy of the amplifier, is control of the output voltage. This is where feedback comes in. Control systems are implemented by feedback loops.

What is a feedback loop? A feedback loop is a circuit that compares the output to the input and drives the errors in the output towards zero. In general there are three important stages that should be included in the feedback loop. They are the modulator, power section and the output filter. Most class-D amplifiers will have at least some of the stages included in the feedback loop. The Spectron Musician III has all of the stages included in the feedback loop. The stage most often not included in the feedback loop is the output filter. It is by far the most difficult to implement.

Why did Spectron go to the trouble of including the output filter in the feedback loop? The output filter passes the audio signal to the speaker and blocks the high frequency carrier. All filters have group delay errors. Group delay occurs when the various frequencies of an instrument arrive at the listener's ears out of alignment compared with the original recorded sound. Including the output filter in the feedback loop greatly minimizes these group delay errors. All of the harmonics of the music therefore appear at the output of the amplifier with the same time alignment in which they were recorded. Consequently, the four most important advantages of including the output filter in the feedback loop are 1) proper time alignment, 2) flatter frequency response, 3) lower distortion and 4) lower output impedance, which improves speaker damping.

Another very important aspect of a control system is called transit time, the amount of time it takes from the time an error is detected at the input until corrections are made at the output. For example, a typical transistor power amplifier has a three primary sections: a low noise high gain differential input stage, feeding a differential to single ended conversion driven a by a high current output stage. Each of these three stages is designed to have low distortion and noise. These attributes come at the sacrifice of speed. Typical transit time of linear amplifiers will have about 2000 - 3000 nanoseconds which is too slow for effective implementation of global feedback and error correction. Consequently, feedback has gotten a bad rap in audiophile literature and from magazine writers.

In contrast, the Spectron amplifiers don't use low distortion circuits, they use much faster digital logic circuits. The Musician III transit time is 200 nanoseconds; such a short transit time allows the amplifier to correct for many small errors and the control loop can follow the input much more accurately. This result is a more detailed, transparent sound with less noise and louder music

Group Delay: It is important that all of the frequency components of the original recording retain their time alignment. These are the position cues to the listener. Group delay is a measurement of time alignment. If a drum strike is being reproduced and the low frequency has one position in time and the high frequency has another, the spatial position of the drum will be confusing to the listener. Spectron's fast feedback loop, however, retains accurate group delay and preserves each instrument's position in space.

Distortion: Distortion (or non-linearity) results when the output of the amplifier has signals added to the output that were not part of the input music signal. This distortion is mostly the harmonics of input signal components, that is multiples of the input signal 2, 3, 4, 5 etc. The published measurements of many class D amplifiers reveal that while their THD is primarily second-harmonic in nature, there are also some higher-order harmonics present. This pattern of distortion is very similar to the pattern of distortion in poorly designed tube amplifiers. Thus, sonically, this euphonic coloration is confused by some with the warmth of tube amplifiers. In reality this deviation acts as both an annoyance and murky veil. In exchange for euphonic, overly rich harmonic texture (of absolutely artificial origin) the listener gets no transparency and a lower level of true detail. In class D amplifiers, the output impedance at low frequencies, as a rule, is low. This results in a commanding, dynamic bass. However, as frequency increases, the level of this distortion as well as output impedance, often increases too, making the amplifier sound bright. The distortion at high frequencies of the Musician III is exceptionally low, resulting in an amplifier that is not only not bright or muted, but also has a three-dimensional, transparent, and natural musical sound.

Output impedance: Effective feedback lowers the output impedance of the amplifier and it is the lower output impedance that minimizes the amplifier's interaction with the speaker, which manifests itself when an amplifier sounds better with one speaker than with another. For example, in TAS review issue 166 Nov 2006, one of the class D amplifiers under review sounded wonderful in some reviewer's systems and awful in others. Another example is the Tripath amplifier chip. It uses feedback but it doesn't measure the feedback signal from the output, but rather measures the feedback prior to the low pass filter. Therefore, errors caused by the low pass filter are not corrected. The Tripath chip also can't use overall feedback because the feedback signal would be too far out of time alignment to be useful. As a result, typical Tripath based amplifiers, as well as other class D amplifiers with similar high level of distortion and output impedance cannot drive speakers with difficult loads such as electrostatic speakers (e.g. see reports in Stereophile). Most of class D amp manufacturers publish measurements of THD and output impedance obtained at 1 kHz level. While this is customary for solid state and tube amplifiers (the 1kHz is chosen because it well represents the harmonics found in real music), its not enough for class D amplifiers. Because the weakest area of these amplifiers is in the high frequency the relevant specs should also be obtained at 20kHz or above and one can more or less predict many aspects of sonic behavior of the amplifier in question based on these specs.The Musician III relevant specs are presented and its low impedance creates sounds virtually the same regardless of which speaker is used

Stability: The Musician III is stable to loads down to .1ohm. It will actually deliver 300 watts into .1ohm. This means it will easily drive ribbon, electrostatic or any complex impedance speakers. There are many speakers which will cause currently available amplifiers, including some class D amplifiers to go unstable because of poor control loop stability. No matter what type of speaker, the Spectron will yield a nearly identical sonic signature.

Frequency response: The audio components should sound neutral and uniform across the listening range. A frequency response that deviates from flat overemphasizes some frequencies at the expense of others. For example, recent discussion in TAS (Nov 2006) indicated that many class D amplifiers are bass heavy, indicating a rise in frequency response in the bass region. The measured frequency response of a Musician III in the audio band is
+/- 0.1db, which is extremely flat.

Bandwidth: Wider bandwidth translates into faster transients and greater detail, better preserving phase information. This results in a bigger soundstage, sharper image and more holographic representation. The potential drawback in wide bandwidth amplification is an increased level of distortion if the output filter is less than perfect. Spectron avoids this by a sophisticated design of the output filter and feedback loop, which are mathematically optimized for maximum phase margin. Spectron's bandwidth is an exceptional 100 kHz, resulting in optimum square wave response.

Headroom: Spectron's Musician III Mk2 has a huge amount of headroom. Headroom is a measure of the amplifier's ability to reproduce large transients. These large transients can be in the form of a loud passage of music, e.g., the cannons in the 1812 Overture, or more often in piano music, or the rim strike of a snare drum often found in jazz. It's true that power is the product of the voltage times the current. However, in reality an amplifier's voltage headroom is limited by the power supply voltage, which limits the peak voltage. Spectron's amplifiers are the highest voltage we know of in the audiophile world, using a plus and minus supply of 120V. By comparison most amplifiers use power supply voltages on the order of plus and minus 65V or even lower. An engineer evaluating an amplifier to determine the required headroom would look at the output voltage of the amplifier with an oscilloscope while playing the loudest music he would reasonably anticipate would be listened to. At Spectron we have done that. What we see is that it is common in high quality recordings to see voltages near 100V peak, with medium efficiency speakers. This means that when listening to the same music with most other amplifiers, the signal would be "clipped". These amplifiers would be unable to deliver the transient voltage to the speaker. The effect is that the music loses some of its lifelike qualities.

Let's turn our attention to current, the other important aspect of headroom. High current is required to deliver power into a low impedance speaker, or a medium impedance speaker that has a dip in its impedance curve. It is important to note that most speakers have some dip in their impedance curve. In general, high current demands are associated with the lower frequencies in music. Speakers have a rated impedance, for example 4 or 6 ohms. However, few speakers stay within 25% of this value throughout the frequency range. There are many well regarded speakers whose impedance dips down, some even lower than 1 ohm. When a musical note is played at these frequencies where the impedance dips, the current demands skyrocket. When this happens with amplifiers that do not have large output current capability, they "current clip". These transients will be both attenuated and quite distorted. Spectron amplifiers deliver peak currents of 65 amps which allows the amplifier to deliver the full transient (burst of music) without "current clipping". Spectron not only delivers very high peak current, but also will hold that current up long enough to play loud passages. Specifically, Spectron will deliver this high current, 65 amps peak (with a staggering peak power of 3500 watts per channel) for 500 msec. On the other hand, most other amplifiers only deliver their rated peak current for sometimes a fraction of this time. This means that the Spectron will not only deliver very high current, but will play very loud music passages long enough to complete the passage.

In summary, voltage headroom means the peak voltage the amplifier will deliver, and current headroom means the peak current the amplifier will deliver. The Spectron MIII Signature was designed with more than adequate voltage and current capability, including duration, to ensure that no matter what the speaker, the amplifier will deliver the full dynamics of the original recording. This means that the Musician IIL Signature can drive everything from conventional speakers to complex loads such as ribbon speakers or electrostatics and represent effortlessly the fury of a symphonic crescendo, unlike any other amplifier!

Noise and Hum
Noise in audio is very important. We often think in terms of the hiss we hear out of the speaker with no signal. This is of secondary importance. When music is playing, the noise (hiss) and hum (audible noise coming from the speakers and related to the AC line) increase dramatically. Comments about amplifiers with the least rise in noise are: "I hear more detail" or "I hear a voice I never heard before". This is because the noise is not coming up and masking the music. Comments about amplifiers with low rise in hum are: "The bass is tight and well defined" or "Voices don't have that muffled sound", again lack of hum masking. Some Class-D amps tend to have less rise in noise when music is playing, resulting in less masking. This is because the logic circuits in class-D contribute little noise. Also, class-D draws significantly less power from the power supply, yielding less ripple on the power supply, thereby reducing hum. The measured noise in Spectron amplifiers is less than 83db, below one watt, which is extremely low.

Mechanical humming from the mains transformer is caused by a varying DC component on the mains grid's AC voltage. The DC component comes from unbalanced loads on the mains grid, such as hair dryers and is present in most households.The amplifier main transformer's vibrations degrade performance of the power supply and when audible can be annoyance as well. Spectron solved this problem by building appropriate filtering circuit.below one watt, which is extremely low.

Efficiency: Spectron amplifiers are class D, which use about 1/3 of the current of class-AB amplifiers. Class D is the most environmentally friendly topology possible, and of course the amplifier does not heat up the room.

Consistency: Spectron class D amplifiers get rid of the problem of the amp degrading over time. Logic gates and MOSFETS don't change much with aging. Also, they are not dependent on matching of components. You should expect the amplifier to sound about the same in 20 years.

Quantizer: Nyquist's rule tells us that we must sample a signal at twice the highest signal frequency. Since our hearing goes to approximately 20 kHz, the minimum for music reproduction is 40K samples per second. Conventional CDs are sampled at 44 kHz and the most advanced recording systems record at up to 192 kHz. Spectron's analog input class D amplifiers sample at 500 kHz samples per second, about 11 times oversampled. This high over sampling rate minimizes alias distortion.

Energy Storage Capacitors: In the original Musician III as well as the Premiere amplifiers there are two 20,000uf energy storage capacitors, one for the +125V rail and a second 20,000uf capacitor for the -125V rail. The Musician III Mk2 has a very large power supply, with an amount of filter caps used in few, if any other, class-D amplifier. This provides for stable, low ripple power to the amp module. In the Musician III Mk2, the two large electrolytics have been replaced by four banks of 25 330uf capacitors, one for the +125V right channel and one set of 25 capacitors for its -125V rail and a second set for the left channel. This lowers the power supply impedance by a factor of more than 10. Therefore the supplies for the left and right channel are separated, resulting in much improved midrange and high frequency fidelity.

Isolated low level power supply: Spectron powers the low level circuits using a high frequency isolation transformer. This gives about 500 times more isolation than deriving the low level power from the conventional line transformer. This virtually eliminates the effect of noise and transients from the AC power line on the low level circuitry. In addition, the Musician III Signature Edition has two of these independent supplies, one for the right channel and a second for the left channel. Having independent supplies minimizes interaction between the two channels as well as cross talk. These two +15V power supplies have input noise rejection filters and 2% regulators.

Current limits: The Musician III has a full fold back current limit. This means that when the output is shorted or draws more than 65 amps, the current limiter reduces the maximum current to 12 amps. When the short is removed, the amplifier resumes normal operation automatically. Spectron provides modules to professional amplifier manufacturers where the amplifiers must be especially bullet proof. The Musician III meets this requirement.

Speaker protection: To protect your speakers, the Musician III has Spectron's protection circuitry that detects excessive DC or high frequency signals not associated with normal music. If excessive DC or a high frequency signal is detected at the output of either channel (which could damage your speakers), the protection circuit will latch the amplifier off. This protects the speaker from damage. The power is simply latched off and a red LED is illuminated on the rear panel. If the reset button, also on the back panel, is pushed and the fault still exists, it will just reset again until the fault is removed. Spectron knows of no other amplifier with this protection.

Optional Remote Sense Cables: The Musician III has rear panel connectors and the required circuitry to accept Spectron's optional Remote Sense Speaker Cables. The Remote Sense Cables put the speaker cable in the feedback loop of the amplifier, thus using the amplifier to correct for distortion of the cables. This means that the amplifier controls the voltage at the speaker, not at the output of the amplifier. It's the signal at the speaker that is important, not at the amplifier's output. So the Remote Sense Cables dramatically reduce the distortion inherent in the cables.

Conclusion: We hear time and time again that the numbers on the data sheet do not predict an amplifier's sound quality. While largely true in the past, today more and more parameters, measured correctly and even more importantly - truthfully reported - do reflect well some very important amplifier sonic characteristics. When you are considering buying any amplifier, the first question must be if it can drive your speakers at all. If it can drive your speakers, then can it drive them well. Can it drive them, at the very least, without euphonic coloration, muted treble and veil etc? The right approach is to check the amplifier's distortion level and output impedance at high frequencies and full output. Also look for peak current and its duration, flatness of frequency response, noise level, bandwidth etc. Of course, Spectron also considers the speed of the feedback loop and inclusion of the output filter in the feedback loop to be of great importance, even if an audiophile cannot really check on this. Each and every one of these measurements contributes to a gorgeous musical experience, and each deficiency will be readily apparent to the critical listener.

THE SQUARE WAVE

The above graph is of a 1 kHz 16V p-p square wave with a 4 Ohm load using Musician III Mk2 Amplifier. The 1kHz is chosen because it well represents the harmonics found in real music. The blue band represent residual 500 kHz carrier frequency which is way out of listening range (20-50kHz) and thus inconsequential. The noise, measured by spectrum -analyzer, is 0.0025% (or -89 dB), and it is remarkably low. Note sharpness of the square waves, specifically, the tremendous steepness of the rise and fall portions of the square wave, reflecting lightingly fast amplifier's responses, and the total absence of ringing noise.

In brief, this signal transfer graph demonstrates the "iron fist" control the device exerts over the signal, and this is exactly what we want from an amplifier!