TECHNOLOGY

It took over 10 years of dedicated research in which time we explored the world of quality amplification. Our goal was to reach the absolute limits of excellence. In order to achieve this, one has to evaluate and experience every available technology in amplification circuits and components. Our experience to true live music gave us the opportunity to have a reference. Combined with our background in electronics and sound engineering we were able to explore what is possible in high quality reproduction.

There are two schools of how an audio signal can be amplified depending on the active components used: Solid State and Vacuum Tubes.

We came to the conclusion that the real dilemma is not Solid State versus Vacuum Tubes but Single-ended amplification versus Push-Pull amplification.

PUSH PULL

In Push-Pull class A or class AB two (or bank of parallel) active components are used, where one sinks current and the other sources current. In class AB operation the problem is that crossover distortion produces a cold and harsh sound. In class A operation most of the times the two halves are not the same components (PNP with NPN transistor or P-channel with N-channel mosfet). In quasi-complementary topology where two same components are used in the two halves, the problem arises from the different topology e.g. One half NPN transistor common emitter, the other half common collector. To minimize distortion various topologies have been used with different types of feedback. e.g. voltage feedback, current feedback, nested feedback/error correction, leading down a one way root to lifeless music. Even when there are two same halves in a Class A pentode or a triode P-P amplifier the sound is not as convincingly natural. This happens because one half acts somewhat as an "active" current source to the other half and thus loading each other producing a mechanical sound. An additional problem is the phase-splliter stage. There is no way it can be done in a consistent manner with active devices. The conclusion is that P-P is not the way forward for reaching the best.

SINGLE-ENDED

In Single-Ended amplification only one (or bank of parallel) active component is used. This demands operation in Class A, where current flows independently of the audio signal. Generally single-ended amplifiers are low wattage tube amplifiers. They provide musical involvement when realized properly. Most commonly used big output tubes are 211,845 and 833. The drawback is that in order to achieve maximum available power they have to be driven in class A2 (Grid starts to draw current from the previous stage). The result is a difficult and awkward load for the driver stage that starts loosing its consistency. E.g. 211 in pure class A delivers about 12 watts, after this and up to 25-30 watts starts to draw up to 30-50mA. The load that the driver stage sees is not constant during the full sinewave. Paralleling multiple tubes, also, is not a solution. Each tube loads the others in a strange way due to differences between each other. This causes a harsh and edgy sound. Also the measured distortion contains more odd harmonics (3th,5th,7th). Another issue that needs to be considered is the output impedance. Without feedback this is normally more than 1,5-2 ohms. The amplifier will alter its frequency response in loudspeakers with big dips and peaks in their impedance curve changing the tonal accuracy of the loudspeaker. A loudspeaker with a very even impedance curve should be used with such amplifiers.

THE SOLUTION

Single-ended amplification provides something that no P-P could ever provide. It is closer to the "real thing", music flows in a way that happens only in live unamplified performances. By incorporating a unique single-ended mosfet output stage on the SET-100 we achieved on having all the virtues of a big single-ended triode output stage without having its drawbacks. We manage to have more power and drive with transparency, musical involvement and above all with music flowing naturally. With only two gain stages, tube input with tube rectification and mosfet output, without using overall feedback we achieved on having output power more than 100 watts, enough gain, and sufficient output impedance. In the SET-100VS amplifier which is a 2X 40Watt rms stereo amplifier we chose the GM-70 output tube. It is a linear ragged direct heat triode able to deliver high power for a S-E design, with a sound that is very musical and powerful at the same time. By driving the tube with a interstage transformer coupled driver stage, with matched characteristics to the output stage, the distortion is kept low for a no-feedback design. The outcome is an all valve S-E-T amplifier that provides a tonal palette of immense width and produces music in its true natural scale, but above all it brings you as close as possible to the music event, to feel the music rather than hear it, to be touched and overwhelmed by the deeper feeling, with music emerging and not only sounding in a clear undistorted way.

DIGITAL

In the first CD players presented to the market in the 1980's, analogue filters were used at 22 kHz to reject out of band images of the audio signal. These filters cause a big phase shift in the audio spectrum and a slow response to transient signals. They are also very expensive to implement, so the stop band attenuation was applied in the digital domain with so called digital filters. In digital to analogue conversion these are interpolation filters, whereas in analogue to digital conversion filtering is done with decimation filters. With interpolation filters, data is added mathematically and calculated from the originally retrieved data of 16/44 kHz. This process is called "oversampling". The result is that the sampling frequency of 44khz is increased to 96khz or 192khz and cheaper and more effective analogue filters can be used to reject the out of band noise. Even when this process is performed by powerful DSP (Digital Signal Processing) devices the end result is never like the originally retrieved data. Tremendous accuracy is required to retrieve clock (master, bit) data in order to keep jitter levels low. With interpolation filters, music sounds more processed and clinical. To avoid the problem of the high accuracy requirement of retrieved master clock data, a technique was used called "upsampling". The data in the DAC input are interpolated and re-clocked by a local clock generator, thus achieving low jitter since the clock generator itself is within a short distance of the DAC. Then data are sent to the digital filter and the upsampling is done again with interpolation. By not using interpolation filters, the sound is more natural, the image depth and dimensions are open and better defined. Also, analogue filters colour the sound and affect dynamics in a negative way. Of course, by not using analogue filters, the DAC's measurements include the out of band noise. But, it sounds more open, with bigger scale and a more analogue-like presentation. Remember that the basic idea behind DSD technology was to get rid of the digital filters used in PCM. Unfortunately the industry did not embrace it but instead, kept interpolation in one way or another. With the Ypsilon CDT-100 and DAC-100, neither oversampling nor upsampling are used. A very linear and accurate chipset are implemented. I/V conversion is accomplished by a specially designed transformer, designed and built in-house. The analogue stage of the DAC1-00 is a single ended class triode transformer coupled at the output. The power supply uses a valve rectifier and choke regulation. All signal and power supply transformers are designed and manufactured by YPSILON ELECTRONICS.

By using only the best materials available in the DAC-100 and combining it with CDT-100 the sound can only be compared with the best analogue sources. You will be astonished!!