Over the past 10 years, more and more high-end manufacturers have begun using
Integrated Circuits (ICs) in their analog electronics. While ICs offer advantages in terms
of cost and size compared to discrete circuits, even the best ICs fall short when it comes
to all-out sonic performance. This is why Ayre only uses discrete FET circuitry for all
analog stages (and associated power supply regulators), and mechanical switching elements
for level controls and input selectors.
Four Design Elements
Today, there are four different gain elements available to the audio designer: vacuum
tubes, bipolar transistors, integrated circuits, and field-effect transistors (FETs). Let
us briefly examine each of these choices.
Vacuum Tubes
The oldest type of electronic gain element still offers many strengths for audio
design, including excellent linearity, extended high-frequency performance, and tolerance
to short-term overload. However, these characteristics must be balanced against their
inevitable sonic degradation over time, the need for high-voltages and filament supplies
(with attendant reliability issues), and the difficulty of interfacing with low impedance
loads.
Nonetheless, allowing for their limitations, excellent performance is available from
tube designs. In fact, the characteristics of tubes are such that it is difficult to
design a poor sounding product when using tubes exclusively in the audio circuitry and
power supply.
Bipolar Transistors
These solid-state device offer virtually the opposite benefits and drawback of tubes.
They offer virtually no degradation over time, operate with low voltages and without
filaments, and are naturally suited to low-impedance loads.
Regrettably, they are not the perfect substitute for vacuum tubes. Their exponential
transfer function yields higher-order distortion products, while high-frequency
performance is constrained due to minority-carrier effects, and short-term overload
results in catastrophic failure.
Notwithstanding these limitations, a skilled designer can extract excellent performance
from bipolar transistors, as attested to by the many successful designs utilizing this
technology.
Integrated Circuits
Modern photo lithographic techniques allow for a complete circuit to be fabricated with
nearly the same ease and cost as a single transistor. Representative ICs used for audio
applications are op-amps, voltage regulators, and digitally-controlled level attenuators.
These devices typically include several dozen bipolar transistors, although FETs are
sometimes utilized in the input stages.
The chief advantages of ICs are low costs, potential for improved thermal tracking, and
the possibility of miniaturized products. Note however, that these benefits are not
generally applicable to high-performance audio designs. The disadvantages of ICs include
those associated with bipolar transistors in general, along with other constraints that
result from the prepackaging of complex circuitry.
One limitation with ICs is the fact that virtually all of the resistors are fabricated
from the same silicon substrate as the transistors. Silicon resistors inherently cannot
achieve the same linearity and performance that is available from high-quality discrete
resistors, due to the limitations of the resistive material. In addition, a nonlinear
parasitic capacitance exists between the silicon resistor and the substrate, which varies
with the applied voltage.
Another difficulty arises due to the need for massive amounts of negative feedback with
IC op-amps and voltage regulators. Critical listening tests by many observers have shown a
preference for little or no negative feedback. However, this is not an option for circuits
using ICs.
Perhaps the most pervasive shortcoming of ICs is the cookie-cutter approach to circuit
design that they engender. Although their easy application allows the inexperienced
designer to achieve reasonably good results from today's high performance ICs, gone is the
opportunity for an experienced designer to extract the ultimate performance level from any
given circuit.
Obviously, the overall circuit topology of an IC is determined by the manufacturer and
cannot be altered. Perhaps what is not so obvious is the loss of ability to fine-tune any
given circuit. The crucial details that separate a good sounding design from the truly
excellent are not available for optimization. Bias current levels, open-loop gain,
ultrasonic response characteristics, and other important parameters that affect the sonic
result remain inflexibly predetermined.
Field-Effect Transistors
The concept of the FET predates the bipolar transistor by several decades, although
manufacturing constraints gave them a late start in commercial electronics. This is
unfortunate, as the FET combines the strengths of vacuum tubes and bipolar transistors,
bypassing their respective weaknesses.
As solid-state devices, they exhibit no gradual deterioration. Their transfer function
obeys the square-order law, dramatically lowering high-order distortion products. Only
majority carriers are present, allowing for improved high-frequency performance. FETs are
also free from secondary breakdown effects that can destroy bipolar devices during
overload conditions.
Disadvantages of FETs include higher cost, device-to-device performance spread that
necessitates matching for optimal results, and a lack of readily available design
information.
Conclusion
Among the four basic building blocks of audio circuits, ICs offer the fewest advantages
for high-performance audio. Excellent results are achievable from both vacuum tubes and
bipolar transistors when a skilled designer optimizes the circuit to the application.
However at Ayre, we have chosen to use FETs and mechanical switching elements exclusively
as the best avenue to attaining the ultimate sonic performance combined with decades of
reliable operation.
|
