exerpts
EXERPT 1: The following excerpt from Working in the Front Line, (Stereophile, January 1991) now available as a download from HIFICRITIC.COM concerns some fundamental issues for audio engineering. While it has been controversial to describe sound quality differences between high quality amplifiers of close to identical technical performance, descriptions of sound quality for passive components has resulted in still greater outcry:

The sound quality of passive electronic components: capacitors, resistors, inductors, cables: Very small differences in subjective sound quality can be identified. For example, listening tests have revealed audible differences between groups of metal-film and other types of resistor used in audio equipment. In these tests, the listeners had no interest or foreknowledge of the resistor types, and would not have known how to identify them even had they felt like trying. These auditioning results have been given strong practical confirmation in practical amplifier designs.

Similar subjective tests involving capacitors have resulted in a number of improved-sounding components employed in loudspeakers and amplifiers. In one double-blind listening sequence, a group of electrolytic power-supply capacitors was assessed for their contribution to the sound of a complete high-grade stereo amplifier. All of the capacitors tested were used well within their ratings. Their internal design, foils, and electrolyte chemistry were different, however. The capacitors were properly formed, then uniformly disguised and soldered directly into circuit by an independent operator in a location remote from the listeners. There were no other variables in the experiment. The listeners were asked both to assign merit scores to each presentation and describe the sound quality.

The results showed good consistency for the limited number of repeats employed; the engineers involved were astonished to find that the capacitor differences were highly significant, determining between 20% and 30% of the overall performance of the amplifier. Each type showed complex differences in virtually all of the normal subjective audio characterizations, including bass damping, stereo focus and depth, timbre and treble distortion, and/or treble brightness. No measurable differences were observed for the complete amplifier using any of these capacitors. ‘

continued



EXERPT 2: High Performance Loudspeakers
(Excerpt from Ch 6 High Performance Loudspeakers 6th ed., illustrations omitted)

Bi-wiring and Multi-wiring

The term bi-wiring arose when extensive experiments showed that with many two-way systems a small but worthwhile improvement in sound quality, independent of timbre or frequency response, could be obtained by electrically separating the crossovers for the two frequency ranges and providing separate two-wire cables back to the amplifier terminal for each frequency section (Figure 6.47). This technique is proving increasingly popular with high-quality designs and can provide the sort of gain achieved by replacing a budget amplifier with one of superior quality.

Polyamplification

As a simple extension to multi-wiring for the separated crossover input ports of a loudspeaker system, where finances permit, similar additional improvements in sound quality have been observed by using separate power amplifiers to drive each frequency channel. Under the general term ‘polyamplification’, for a given stereo channel, power amplifiers are provided for each passive crossover port. All the amplifier inputs for that channel are connected in parallel, without the need for low-level pre-filtering. Improvements in clarity, stereo focus, precision and bass transients are evident. One explanation is that while each amplifier section is fed the whole frequency range current is only drawn in the frequency range appropriate to the passive crossover section to which it is connected.

Amplifier distortions are in general dependent on the frequency range, signal complexity and power delivered. If the power delivered to the load is filtered, intermodulation is reduced, whether or not the filter is before or after the amplifier. Polyamplification offers a simple route to multi-amplified ‘active’ systems without the need for custom-designed electronic active crossover filters, although the full sound quality benefit of an active system is not attained.

Speaker Cable Practice

For maximum benefit from bi-wiring or other aspects of high-quality sound reproduction some simple rules have been established for speaker cabling.
  1. Cable runs should be as short as possible.

  2. Lengths for the left- and right-hand channels should be equal, any spare cable should not be wound into a coil; it should be folded non-inductively (zigzag) and secured neatly.

  3. Virtually any kind of conductive wire will work but subtle sound quality differences are present between cables. These are significant in critical applications. Some care and expense on cable is justified and, in critical market areas, up to 20% of the loudspeaker system cost may be allocated to cable. For the industry as a whole the cable allocation is set at typically 1% of the price of the speakers.

  4. Speaker cables may be viewed as a profit point and not as a legitimate component in the audio chain.
Cable Design and Sound

It is easy to establish that cables can and do sound different simply on the basis of their conductivity and the effect of their series impedance (mainly resistance and a small series inductance) upon the generally non-uniform load impedance presented by most loudspeakers. With zero amplifier source resistance and zero cable resistance the frequency response of the loudspeaker will be as defined by the designer. Introduce some series resistance, say 1 ohm, and a nominal 8 Ohm speaker will lose a just audible, 1 dB of sensitivity. In addition, if it has the usual complex impedance varying over the frequency range, then this will now be reflected as errors in the axial frequency response.

In the case of the example shown in Figure 6.48, the effect is to increase the relative output at the upper bass resonance, decrease the 200 Hz ‘power band’ present in the low mid-range and elevate the treble. Some sectors of the hi-fi fraternity have experimented with single-strand speaker cable down to 0.4 mm diameter. In lengths above a few metres such thin grades add significant loop resistance, impair bass dynamics and impart response changes that outweigh any benefits. Thus, a sensibly low cable resistance is desirable in high-quality applications; a loop resistance of less than 5% of the nominal impedance is desirable. If the expense is justified considering the quality of the loudspeaker, 2% is a worthwhile goal. Low inductance is also important since 10 m of typical parallel ‘twin’ cable adds 6.5 µH, a further 0.8 Ohm at 20 kHz.

Effect of Damping Factor on Speaker Low-frequency Alignment

Low-frequency alignments are generally calculated for zero source resistance including amplifier and cable.

With a loudspeaker Re at a typical 6 Ohms and Qt at 0.707, maximally flat, consider an amplifier and cable combination with a damping factor of 16 for an 8 Ohm rating, i.e. a total source resistance of 0.5 Ohm. Qt must rise by around 11% to 0.77 and results, by itself, in little change in local frequency response. For example, the tiny peak due to under-damping is only 0.12 dB high, with a slightly larger variation at lower frequencies of perhaps 0.6 dB. On the other hand, the finite series resistance will attenuate the rest of the frequency range, with a small audible difference on an A/B test. Some small changes in frequency response will also be present, reflecting the usual variation of a speaker’s load impedance with frequency. Interestingly, in tests with high-quality audio systems using a range of matched cables, and where the significant variable was limited to loop resistance only, where the amplifier had a low source impedance and the speaker had a powerful, well-damped and extended bass response, simulated cable resistance showed a greater effect than was anticipated. Unexpectedly, it proved possible to readily characterize sound quality differences even in the range 0.2 Ohm down to 0.05 Ohm, with the most solid and articulate low-frequency sound related to the inserted lowest resistance.

The physical weight and rigidity of a cable is also proportional to lowered resistance, so it is possible that some mass loaded electromechanical factor is also at work here. Accelerometer analysis has shown that at high currents speaker cable conductors do move relative to each other if they are not sufficiently restrained by the construction.

Loudspeaker cable performance does matter in critical applications and many refined types have become available, often designed on an empirical basis and, in consequence, marketed with a considerable degree of conflicting pseudo-technological claims. These do the designers little credit. However, careful subjective analysis indicates that significant sound quality differences between cables do exist which are also independent of loop resistance and impedance. From a review of a database of 100 cable types, it can be shown that conductor type and purity, its state of annealing and any surface plating are relevant. In addition single or multiple stranding, including the use of differential strand diameters or separately insulated strands and/or more complex strand winding methods are influential. Geometry also plays a part, for example flat twin, twisted pair, coaxial, planar ribbon and tube. Finally the dielectric quality of the insulating medium matters together with mechanical properties such as stiffness and damping. For one cable design, a wrap of lead tape was added to increase the mass and damping to a high-power speaker cable, with a clearly audible and positive result.

Cost-effective results can be obtained using 0.8–1 mm high-purity, single-strand copper conductor with polyethylene insulation, arranged as a tight twisted pair. Such a cable offers low dielectric loss, a mechanically rigid construction, low inductance and moderate resistance, together with a well-defined geometry and conductive path.

Electromagnetic Screening Loudspeaker and Cable, EMC Effects

At the upper limit of reproduced sound quality, the final attainment will depend on an accumulation or summation of small improvements. One of these is the electromagnetic screening. As our environment becomes increasingly radio-emission rich, the impact on the performance of a replay system becomes apparent. This includes numerous transmitters present both locally and in the home itself, portable and mobile telephones, radio linked remote control and data transmission systems, and the computer, radio and video equipment.

Radiation does not mysteriously affect a loudspeaker; rather, the loudspeaker’s internal circuit, crossover, wiring loom and the cable connecting to the amplifier constitute a receiving aerial feeding EMC into the amplifier. Amplifiers vary considerably in their susceptibility and a simple reporting of measured interference levels or susceptibility will only go part way in determining whether sound quality will be affected.

A significant engineering step is to provide for common grounding of the main metal components of the loudspeaker, possible shielding of the crossover network (taking care to avoid coupling to the internal components), shielding of the speaker cable via a third wire/braid, and provision to take this radio signal shield connection either to the amplifier chassis, or to the main signal ground, whichever is preferable. (Some power amplifiers have a balanced output requiring a chassis ground connection for this.)

I have observed in blind A/B testing that this grounding technique is moderately beneficial with several types of high-quality power amplifiers significantly, if modestly, improving the sound quality score for loudspeaker/cable combinations that are so equipped.

The Sound of Metal Conductors

A simple proof of the audibility of sound quality differences between silver and copper conductors was devised. A small high-performance two-way speaker, which in manufacture was built to high precision and close tolerance using normal copper conductors, was replicated with the electrically conducting parts made in silver (99.99% purity or better). During its development individual components were built and auditioned, revealing interesting differences, but it proved impossible to anticipate the effect of executing the whole signal path in silver, this including the speaker cable.

Several aspects of sound quality were altered, with general agreement that there were not merely differences but genuine improvements from the substitution of silver for copper. Most interesting was the apparent reduction in mid-range colouration. A degree of previously cone-related colouration and upper mid ‘hardness’ showed improvement, indicating that a component of this unwanted sound was in fact due to the conductor. With this came an improvement in mid-range clarity and transparency. Transients were more articulate, while the bass sounded more precise and better focused. In the treble, the tweeter appeared to have a purer, sweeter sound, with reduced ‘grain’ and related subjective distortion.


continued


HIFICRITIC.COM Stereophile Magazine Hi Fi News COLLOMS.COM