• Achieve maximum reliability. (Remember, the THD performance of a non-functional audio power amplifier is 100%.)
   
• Minimize all distortion components to the lowest possible levels, without sacrificing any other critical performance characteristics.
   
• Incorporate effective noise reduction techniques for a minimum of internal noise generation.
   
• Provide an extraordinarily wide power bandwidth for the accurate processing of high-level transients and ultrasonic components. (Wide bandwidth is necessary to eliminate the undesirable effect of slew-distorting ultrasonic components into audible coloration of the original program.)
   
• Dramatically reduce dissonant clipping distortion artifacts, for unimpeded processing of modern wide dynamic range program material.
   
• Install bulletproof protection circuitry, so the electronics will be protected in the events of short-circuit loading, thermal overloading, and operator error. Also provide protection circuitry for external devices (i.e. DC speaker protection and on-delay muting), and meet full compliance to internationally-recognized operator/equipment safety standards.
   
• Incorporate extensive self-correction circuitry, to ensure long-term stability and performance excellence in situations of time-related component drift or user abuse.

      In addition to the aforementioned highlights of the development goals, the project goals also included ensuring the compatibility of the design in terms of modern audio standards, such as input sensitivity, input impedance, absolute phase response, and conventional speaker load impedances.

     The remainder of this writing will detail the methodology and reasoning that were involved in the development of the ZUS (OPTI-MOS) topology.

     The fundamental architecture of the ZUS (OPTI-MOS) topology is the conventional Lin 3-stage topology, consisting of an input stage, voltage amplifier stage, and output stage. This decision to base the design on the Lin topology was made without much effort, since the overall linearity and stability benefits of this architecture are unparalleled, and it has been well-proven in at least 99% of all solid-state amplifiers throughout the last 50 years.

     In considering the basic input stage design, there are only two practical high-performance foundations; a single differential input design or a dual-complementary differential input design (i.e. a fully-complementary, or mirror-image, design). A well-designed single differential stage, incorporating a constant-current tail source and active loading, is capable of very low distortion and suitably high gain factors. However, it suffers from the disadvantage of asymmetrical clipping characteristics. In contrast, a well-designed dual-differential stage will produce slightly higher distortion, but symmetrical clipping is automatically ensured, and the cancellation effects of the complementary action allow for lower-noise operation and less sensitivity to component tolerances. Considering these weighty advantages, the dual-complementary input stage design was chosen for the ZUS (OPTI-MOS) topology. The small increase in distortion is at least several magnitudes below human perception levels and consists almost entirely of non-dissonant, even-order artifacts. Thus, it is appropriately considered insignificant.

     To enhance the desirable attributes of the input stage operation, dual constant current sources were incorporated to provide tail current for the dual differential stages, and four cascode stages were incorporated to provide active loading and increased gain for the differential stage collectors. In addition to the benefits of improved gain and linearity, the aforementioned enhancements provide the highly advantageous characteristic of virtually isolating the dual differential input stage from the power supply rails. Such isolation reduces the input stage's susceptibility to noise, ripple, and voltage variations from the power supply rails, as well as improving the amplifier's PSRR (power supply rejection ratio), which is a term used to define how well the amplifier rejects unwanted interference from other amplifiers utilizing a common power supply for operational power. Thus, any suspicions of crosstalk problems from adjacent audio power amplifier channels can be summarily dismissed.

     The decision to incorporate a fully-complementary, dual-differential input stage forces the decision to utilize a class-A, push-pull type of voltage amplifier stage. The cascode stages included in the input stage configuration provide for very good quiescent current regulation and temperature stability in the voltage amplifier stage operation. Dual two-pole compensation networks were used in the voltage amplifier design, since these provide for lower distortion and superior stability in comparison to the more conventional dominant-pole compensation techniques. Two pseudo-cascode stages were included in the voltage amplifier design, and these stages provide three highly-desirable characteristics to the voltage amplifier stage operation. First, these pseudo-cascode stages provide active loading for the push-pull gain stages, which enhances linearity. Secondly, they provide a moderately-high voltage amplifier stage output impedance, which is preferable for driving the Lateral MOSFET output devices, since this is the primary means of eliminating gate oscillation problems. Thirdly, they are called pseudo-cascode stages because, unlike true cascode stages, they do not include a stable reference source. Rather, they are referenced to the dynamic difference potential between the output rail of the amplifier and the power supply rails. This design strategy results in a non-linear loss of transconductance whenever the amplifier's output rail approaches the potential of either power supply rail (i.e. at the onset of clipping). The result is a soft clipping response, with the added enhancement that the clipping action is forced to be symmetrical due to the fully-complementary topology.

     Much thought and research went into determining the best overall design for the output stage. In considering fundamental output stage topologies, there are four practical options that can be taken. These are (1) complementary-feedback, (2) emitter or source follower, (3) quasi-complementary, and (4) a combination of complementary-feedback and emitter (source) follower techniques. Complementary-feedback output stage designs have been virtually abandoned by all commercial manufacturers, due to their inherent susceptibility to instability and cross-conduction problems. They are capable of providing the lowest distortion performance of any major output stage topology, but their self-destructive nature when driven to higher frequencies have deemed then unacceptable for high-reliability audio power amplifiers. Emitter follower designs (or source follower, when incorporating MOSFET output devices) provide an excellent compromise between the characteristics of distortion, reliability, and stability concerns. The output devices are in their ideal environment for providing high levels of power gain, without any sensitivity to frequency concerns or tendencies to break into parasitic oscillations. Many years ago, it was discovered that quasi-complementary output stages were more reliable than fully complementary stages within bipolar designs, due to the physics involved in the manufacture and operation of bipolar NPN power devices. Consequently, many modern high-power professional audio power amplifiers still use quasi-complementary output stage designs, since low-distortion is not of paramount importance in most stage/professional environments, and the NPN devices are relatively cheap. Complementary feedback and emitter-follower stages can be merged into a single output stage topology. This technique has a moderating influence on the inherent stability problems of complementary feedback designs, while providing a marginal improvement on the distortion characteristics of emitter follower designs. Unfortunately, such combinational output stage designs suffer from excessive cross-conduction problems at higher frequencies, which results in reliability concerns as well as detrimental high-frequency sonic effects.

     After fully weighing all of the advantages and disadvantages of the design options, it was decided to incorporate an emitter (source) follower output stage architecture for the ZUS (OPTI-MOS) amplifier line. This was one of those situations wherein the examination of the options went full circle but returned to the majority convention.

     Regarding the choice of output stage devices to utilize, the best choice was obvious and rather non-conventional. Lateral MOSFETs were originally developed exclusively for high-power audio applications (one of the few devices in history to hold such a classification), and their advantages of speed, reliability, and insensitivity to thermal extremes deems them the clear favorite. Unfortunately, Lateral MOSFETs have been rejected by the vast majority of competitive commercial audio power amplifier manufacturers due to the much higher cost, and the abundance of conventional bipolar designs already available. While it is both necessary and unavoidable to make some performance compromises in the development of practical audio power amplifiers, it is not mandatory to make cost compromises. It was therefore decided to choose Lateral MOSFETs for the critical high-power output devices incorporated into the ZUS (OPTI-MOS) amplifier line. At least 95% of all failures that occur within audio power amplifiers can be traced back to a failure of one, or more, of the output devices. Consequently, the decision to utilize Lateral MOSFETs for output stage devices is the single most critical reliability issue to consider in the development of any audio power amplifier.

     Internal audio power amplifier protection is provided by a multi-slope overload protection circuit, merged into the output stage design of the ZUS (OPTI-MOS) line. This is a non-conventional approach, since the inherent ruggedness of Lateral MOSFETs has prompted most designers to simply neglect the use of any type of overload and short-circuit protection, allowing the power supply rail fuses to provide the brute-force protection for the output devices, as well as the external devices. While this technique is effective, since Lateral MOSFETs will withstand extraordinarily high current surges for the duration of the fuse blowing process, it is not the safest and most practical approach.

     The end results of bringing all of the individual stage designs and component choices together into a direct-coupled amplifier topology are an anticipated excellence in sonic performance as well as non-temperamental reliability. The physics of operation are balanced to the aesthetic requirements. For example, the sonic benefits of wide bandwidth are not sacrificed in the effort to play a promotional numbers game with distortion, and reliability is not compromised in the process of providing the best dollars-per-watt bargain.

     The ZUS (OPTI-MOS) design was originally introduced in an engineering textbook, and it has been field-tested for more than two years by hundreds of audiophiles internationally. It was anticipated that the users would be completely satisfied with the sonic and reliability attributes, and this has been the case unilaterally. However, a surprising result has been the perception of improved sonic performance by uninvolved listeners, such as family members associated with the audiophile using a ZUS (OPTI-MOS) system. It is even more significant that the majority of these evaluations were made in comparison to other high-end audio power amplification systems. Consequently, we who have been involved in the development of the ZUS (OPTI-MOS) program must conclude, from both subjective and empirical evidences, that the correct design choices coupled with a no-compromise materials quality, will result in a superior level of sonic realism that can even be perceived by the untrained ear.