CUBES, I SEE CUBES EVERYWHERE!
Save me from the invasion of cubes. They are flooding everywhere with their weird colors, everywhere like I was on dope. Ok, stop it … a few more serious words about the electronics we are developing for low signal audio. This line, dubbed cubes, is based on jFET's and sports a compact and customizable chassis.
jFET's can provide outstanding benefits to audio amplifiers in terms of lower noise, increased bandwidth and flexibility of the design. They are nearly ideal three-terminals devices and can sound as good as tubes but, clearly, cannot deliver the power of the triodes nor can they withstand high voltages. The superior efficiency is an added advantage that allows compact size designs and calls for negligible requirements for ventilation and dissipation.
Toshiba managed to compress noise of 2sk170 down to a figure of 0.95nV/√Hz. A low noise audio opamp such as OPA161x by Texas Instruments stops the clock at 1.1nV/√Hz, a couple other opamps by AD are specified to be slightly under 1nV/√Hz. Vacuum tubes can do great as well: the ECC88 is around 2 nV/√Hz and other low impedance/high transconductance triodes can do a bit better. In high gain application such as a phono stage, low noise is utmost important and tubes will suffer due to the higher value resistors they need to be properly biased as they contribute to increasing the total resulting noise. jFET's will win. Opamp's will have lower noise figures as well but according to a wide audience they do not sound as good as tubes and jFET's.
But things are more complex in reality and using the best available component is no guarantee for success as its characteristics drift with temperature impairing the performance and noise figure. Even though it is possible to design the circuit to operate close to the zero tempco (a point where the influence of drift is lower) to get stable performance, a different approach would offer higher immunity and accuracy. The differential topology uses two symmetrical amplifiers working together and each deals with a mirrored image of the incoming signal. The consequence is that noise which is common to the two halves is cancelled and the differential amplifier will have higher rejection of power supply noise and incoming common mode noise. For the very same reason the differential amplifier will exhibit odd order distortion only but in a much lower value than the typical second order distortion of a single amplifier. Perfectly matched halves will make distortion and noise astonishingly low and this is possible using dual monolithic jFET's: two devices are built on the same die sharing the same substrate and have nearly identical performance parameters and drift coefficients. Dual triodes cannot be so easily matched with the same precision.
The 2SK389 is an ultra-low noise monolithic n-channel JFET pair based on the 2SK170 and it is the ideal component for application where accuracy and low noise of differential amplifiers are mandatory. Unfortunately Toshiba discontinued the production many years ago but Linear Systems reintroduced on the market a large offer of single and dual jFET's again making possible the production of discrete differential amplifiers. Cubes use LSk389, the equivalent from LS for the old 2SK389, to bring back the benefits of jFET's in audio.
The phono cube employs a 2SK389 cascaded differential amplifier at the input to reduce capacitance and intermodulation distortion. The second stage is a balanced buffer, again based on the 2SK389, and it drives the discrete spit RIAA network with a <0.1dB deviation. The third stage is a 2SK389 differential amplifier buffered with two more 2SK389’s for a direct coupled low impedance output stage. Gain is selectable in 42 or 48 dB, output impedance is 80+80 ohm and bandwidth much wider than what is typically needed, input impedance 47k+47k with user selectable capacitance. It is the best companion for MM cartridges and coupled with SUT N° 3 can offer a complete high performance equalization system for vinyls.
The line cube is a 9dB preamplifier that uses the output stage of the phono, so it's a differential pair followed by a balanced buffer, and capable of delivering 9.5Vrms of signal before approaching the clipping. THD is excellent everywhere and reaches 0.1% at the maximum output voltage.
A balanced DAC with jFET's buffer is in the early development phase and more news will be published in future.