1) Nowadays preamps do not differ much, if at all ; sometimes a passive “preamp”  is the best choice if you don’t intend to buy a turntable. Passive preamplifiers may sound (sometimes) harsh or “dry” because they form a high-pass filter in combination with the system.

2) Multistrand cables (with silver-plated copper strands) less than 2,5 meters long (gauge 14 or 12) or equivalent flatwire, with the simplest possible structure of non PVC insulation, will do perfectly well. You may already own them, so leave them in peace.

3) Amplifiers are very important and they do make a difference concerning speed, bass power, dynamic headroom, upper mid band, THD (total harmonic distortion), TIM (intermodulation distortion) and soundstage. But before you make a new buy, try power conditioning or cascaded earthing or both ; the result is audible. Even a block of granite under your lightweight amp will give you a pleasant surprise. Although some top amplifiers may give you a pleasant (or unpleasant) surprise, amps are not the issue here : sometimes good amplifiers don’t combine well with loudspeakers.

4) Loudspeakers. It is possible that comparison of some amps’ circuitry design or component quality by audition will not give you any audible clue. This is never true about loudspeakers. Loudspeakers are -by far- the main responsibles for your system’s sound : they create it. They receive electric signals, they split them and they transform their energy in to acoustic (pressure waves) energy.  They should match electrically your amplifier and they should behave linearly or at least predictably ; unfortunately, they do not.  Loudspeakers are highly non-linear (even chaotic) systems for signals powerful enough. Transducers’ type, construction and quality, electroacoustic  design and combination of materials in a loudspeaker, will make a huge difference. In fact, you will never find a loudspeaker  sounding the same as any other  model of the same or other brand !

CONCLUSION : The main responsible for your high-end system’s sound on a scale from 1 to 100 is :

loudspeakers 50%  –> amplifiers 20% –> sources 15% –> power conditioners / massive bases 10% –> preamplifiers 3% –> cables 2%.

Of course this is a subjective scaling !  Some people would pay a fortune for a tiny extra 1% that upgrades their system to a champion… One more thing : 1% is audible !!

Image via Wikipedia

High end audio speaker cables can be very expensive. Byers feel that bi-wiring will improve the performance of their equipment. Is this true ? Is this extravaganza justified ?  A simple but effective engineering model for calculating your cables’ performance is the lossy transmission line model (see http://www.ece.uci.edu/docs/hspice/hspice_2001_2-269.html).

Image via Wikipedia

Simple wires present resistance & inductance in series, as well as capacitance to “earth”. All those quantities are  proportional to the wires’ length. Pairs of wires interact, so they present mutual capacitance C, combined inductance L, resistance R and dielectric/insulation impedance G (mainly resistive) between them. Speaker cables are complex-structure paired wires, including compound insulation, twisted groups of strands etc. In spite of their complexity they can be modeled as equivalent wires, presenting  resistance (ohm/m) and  inductance (mH/m) in series, as well as shunt capacitance (pF/m) and insulation conductance. Whatever the manufacturers say, only the abovementioned measured (if measured) values count for calculating signal attenuation, phase shift, time rise and bandwith ; and, yes, now you know that a speaker cable pair is a 2nd order filter.

Bi-wiring will cutout  half the resistance and the inductance of the connection, but it will double the capacitance.

To simplify things take as a rule of thumb the following :

1) With speaker cables (14 gauge ) less than 2.5 m long you will not have any audible effect either bi-wiring or not. In this case, there is no point, either, in bying a 12 gauge cable.

2) The more complex, bigger and heavier cables, the less resistance and inductance, the more capacitance. You don’t need to and you should not bi-wire using extra big, super complex, ultra expensive cables  ; complex cables usually present very low resistance and inductance and impresivelly high capacitance. Beware of the capacitance, too.

3) Keep away from the floor.

4) If the cable’s impedance  from 10 to 20.000 Hz is less than 5% of your loudspeakers’ nominal impedance, probably you won’t need bi-wiring for improving your audio system. For bandwidth, phaseshift vs frequency, Q,  resonances and standing waves, ask an engineer (and read more of these posts).

Image via Wikipedia

To bi-amp or not to bi-amp ? This is the question you must ask the vendor of your Loudspeakers twice, because if he says “YES” the next question you are going to ask is : “What is the order of the crossover for each way (tweeter /mid /bass sections) ?” If the answer is not “1RST ORDER for every way” then you are sure that this Loudspeaker is designed either for poly-amping only or  for operating with the binding posts paralleled only despite the beautiful set of  multiple pairs of terminals at the loudspeakers’ back ; a complex X-over cannot do both jobs equally well.

The reason for this, is the fact that for a Loudspeaker with paralleled (short-circuited) pairs of binding posts, each section (branch or way) of the X-over “sees” not only the amplifier’s output but it also “sees”  the other branches electrically, and what it sees is a parallel complex impedance Z(f) varying non-linearly with frequency  ; this highly variable – in magnitude & phase vs frequency- parallel load, seriously affects performance of the section. Unless the crossover is designed to be a multibranch one only (most designers don’t do this), the multiple pairs of binding posts must be amped independently.

Image via Wikipedia

Now if the argument of  the vendor  is that Zobel linearisers do their job very well and every “way” of the X-over is linear and resistive, so the whole X-over is linear and purely resistive, I would say that this miracle is impossible for human technology. One reason is that transducers (drivers) are higly non-linear electric motors depending on air-resistance, cone mass, etc. ; this is why Zobel linearisers work effectively within a small portion of the frequency band, and they also do interact with the rest of the circuitry in ways that superposition-linear-models do not like. In fact, anything parallel to a driver changes the order of the X-over. The other reason is that the underlying  ”principle” of  linear superposition of parallel-branch drivers’ impedance phase and magnitude is not valid for X-overs of order greater than 1, because of Algebra .

Moral : Only 1rst order crossovers without linearising networks have a chance to do the job equally well, either poly-amping or not. But go on poly-amping, most probably your loudspeaker will play better : designers think linear without even knowing it !

Image via Wikipedia

Design principles of high end audio equipment have evolved over the years. Concerning stages and circuit components the principle is : the lesser, the better. Concerning global feedback the principle is : zero negative global feedback is the best. Other design principles are : Class A triode operation is the pinnacle of the high end audio amplifier operation modes. Push pull topology is technically (in theory) better than any single ended configuration provided that matching of currents is regulated and transformers are perfectly made. Single ended triode amps mask effectively within certain limits huge harmonic distortions. Regulated fixed grid bias can be noisy. Cathode bias leads to lower power output. Triodes are better than pentodes distortion-wise in any kind of topology. Pentodes are noisy unless we operate them as triodes. Cascodes are better than pentodes as well. Handwired turret/tagboards can be better performers than PCBs. Heaters must always be supplied with dc current in order to avoid 50Hz buzzing. Separation of channels must be extreme. Direct stage coupling is better than capacitor or transformer coupling provided that the circuit remains stable. Cathode followers are almost ideal for stage coupling. Topology of μ followers in small signal stages provides the best engineering solution for high input/ low output impedance plus extreme HT noise immunity. A concertina inverter is better and simpler than any cathode coupled or floating paraphase inverter ; isodyne inverters are the best of them all, but produce less than half the signal input. Zeners produce noise ; replace them with leds,  you get light as a bonus! Solid state rectifiers in power supplies are supposed to give more dynamic sound, while thermionic diodes (valve rectifiers) produce a brighter and generally a more pleasant sound quality. E88CC/6922 is most probably the best overall small signal triode. Carbonized file 6SN7 presents the lowest harmonic distortion among all small signal triodes. Construction rigidity and high/low mechanical frequency damping of chassis is important. Electronic components quality, matching and individual testing is a sine qva no in High End Audio manufacturing.

The abovementioned “principles” are not holy rules. They are a mixture of empirical and scientific pieces of advice. The sound quality of the resulting system is everything that matters.