Why do you not manufacture a fully digital amplifier?


It has only been possible to design a high quality, high power 'digital' amplifier in the last 5 years or so, because before this the major components: - output devices, Magnetic materials and high frequency (greater than 500kHz) pulse width modulators, were not available.

Prior Art Class D designs were around, but the sound quality was just not acceptable. (Please see below for an explanation of Prior Art Class D Amplification.)

The output stage, and in this we include the power supply, is arguably the most critical stage in all amplifiers whether digital or analogue. A long time is needed to fully evaluate the output stage under the various load and programme material combinations, and to reach a conclusion as to whether a change of component or circuit does indeed improve the sound quality. All changes must be done one at a time in an ordered manor under different conditions, auditioned by many people. That's the only way we can make real progress.

So, the answer to the question is simply this: -

We are now relatively happy that our output stage and the Tripath topology are achieving the best sound quality at this time. The next generation of amplifiers may well include a fully digital chain, but we will not release the amplifier if we cannot demonstrate a further Improvement in sound quality.

Prior Art Class D

'Digital' Amplifiers have been around since the middle 1950s. The class of amplification was then described as Class D. The "D" did not, as some are led to believe, refer to Digital but was simply the next letter after Classes A, AB, B, and C.

Amplifiers at that time were limited to very low powers in the order of 2 to 6 Watts. The Input signal was compared with a high frequency (approximately 100kHz) triangular waveform, which produced a pulse width modulated output signal with two discreet levels (on or off), which were then passed to the output stage. As these signals are either on or off the output stage dissipation is greatly reduced providing high efficiencies due to the much lower losses. The instantaneous sampling of the input signal was represented by the width of a pulse. To reconstruct the analogue of the digital signal an output filter was necessary. It was very difficult if not Impossible to Include feedback around this arrangement because to filter the 100kHz switching waveform to acceptable levels would necessitate the filter rolling off well into the audio band which would limit high frequency response, and also produce wide phase shifts with increasing frequency. Without feedback, distortion was destined to be very high. The early Class D amplifiers were only really suitable for low frequency, bass or sub-bass amplification at the higher powers (20 to 50 Watts), or car radio amplifiers at the lower powers (2 to 6 Watts).

It is interesting (for us anyway) to note that we did in fact make several Class D amplifiers in the mid 1960s for Mr Radford's car radio to replace his Motorola Valve Amplifier, from circuits published in Wireless World, with great success.