requirements to amplification:

  • adjustment of the levels of currents and voltages in amplitude and phase
  • matching of input- and output-impedance values


The current and voltage levels of the audio signal at the beginning of the signal chain can be tiny. At the end of the signal processing chain large amounts of power, resp. large values of voltage and current are required.

Besides the pure signal amplification to the required levels additional signal processing is optional.

For example are viynl LPs and Tapes processed with a preequalizing process because of mechanical and electrical reasons. For playback a mirror equalizer is required. Equalizers allow to adapt for human hearing curves, to raise or lower highs and bass, or come as filters with specialised functions for special demands.

Even if You restrict on a linear amplification process the number of possibilities, circuit topoliogies and useable parts is huge.

Basically it is technically useful to restrict the amplification factors to as low as possible values and to keep the bandwidth small.

The more important question for DIY is usually what kind of amplifying parts, active parts to use. Should one use transistors or tubes or integrated circuits, ICs? The latter sometimes contains hundreds of parts, their internal connection and circuit topology remains mostly unknown and unaccessible and hence out of our range of our influence. Their behaviour is to a great extent predetermined. Besides power supply pins only a few pins are connected to the outside world to feed signals in and out.

Datasheets and electronic simulator models allow for easy and quite precise evaluation of the circuit behaviour before breadboarding.

The ICs reduce the effort for research and development and the time-to-market alot, because one needs to concentrate mainly on the peripheral parts of the circuit.

The small, sometimes tiny casings allow for very compact layouts. Regarding measurement values modern ICs are hard to beat in many applications and are often much more cost effective.

Regardless of the advances in modern IC-technology (OP-amplifiers) do well designed circuits using discrete parts like transistors or tubes remain sonically ahead.

The differences are not so much related to typical HiFi-prosa like dynamics, tonality, THD or detail, but rather a impression of a lifelikelyness and authenticity. The soberly diction of OP-amplifier circuits never letsYour brain forget to listen to artificial technical playback. The illusion of beeing there, life, will never emerge. The foot-tapping factor remains low.

A reason might be the inherent need of global feedback (NFB). "Might" because the verification is hardly possible. There exists no possibility to switch from feedback to no-feedback mode by simply toggling a switch to allow for direct A/B comparison.

Global feedback (negative global feedback to be precise) is a way to correct for the strongly non-linear behaviour of the OP-amplifier by external parts. In theory, if the open loop is closed by the NFB-network the circuit adopts a behaviour which is in theory solely dictated by the external parts. In practise we can come close to the ideal within certain limits. It seems though that our hearing sense works right at those limits or beyond, so that a discrepancy between technically measured performance and listening impression occurs.

In any case is the sonic character of a circuit influenced at most by the circuit topology. Regardles of all differing internal details, can OP-amps be divided into two main groups, the VFAs, Voltage-Feedback-Amplifiers and the CFAs, the Current Feedback Amplifiers.

VFAs almost always are build after a 3-stage concept, consisting of input-, driver- and output-buffer-stage. Quite renowned as audio OP-amp is the AD797 of ADI, which features a different topology. Its circuit structure is a 2-stage concept featuring a superhigh-gain complementary-folded-cascode-differential amplifier stage and a output buffer stage.

The most prominent difference in a CFA is that the inverting input is a low-impedance node, typically the emitters of the input stage. This leads to a different behaviour that also requires a different design of the feedback parts. One finds CFAs often amongst high-speed video and fast buffer amps.

If one wants to put away with the limits of standard 3-stage concepts, one is almost inevitably forced to switch to concepts using disctrete parts.

Since the requirements for audio circuitry are mostly easy, it is no problem to design ´discrete´ here. Typical gain factors are low, apart from phono-stages. Bandwidth may be limited to a small range around the human hearing range and distortion figures (THD) are of rather low interest, since classically measured figures don´t correlate with hearing impression.