Dynamic Range Versus Ambient Noise
A practical solution involving metal-cone loudspeakers and high power amplifiers.
By George Izzard O'Veering
Note: Although this article is not meant to be taken seriously, it still manages to raise salient points and is remarkably accurate in many areas - the article is over 30 years old. See editor's notes at the end of the page.
The essential requirements for a high quality sound reproduction system are adequate power and adequate bandwidth. Since loudspeakers are inefficient, and the attainment of wide bandwidth systems is generally incompatible with high efficiency, the achievement of the desired acoustic spectrum from the subsonic to the ultrasonic makes heavy demands on amplifier output. Moreover, it will be apparent on reflection that many of the musical and other instruments, the acoustic output of which it is desired to reproduce, are themselves both powerful and developed to a high degree of acoustic efficiency. It is clearly laughable to suppose that the majestic splendour of a full orchestral fortissimo or the lung power of a Wagnerian tenor in full cry can be represented adequately on an acoustic budget of a few hundred milliwatts.
Inconvenient though it may be, there can be no doubt that to recreate the true dynamic range of much recorded sound over the required sonic spectrum makes demands on the output power of the audio amplifier/reproducer system which are well beyond the capabilities of most, if not all, of the equipment at present on the market.
The quietest sound which can be heard in a given environment depends entirely upon the background noise level of that environment. Unfortunately, most people five in close proximity to traffic, neighbours with television sets, dogs, and noisy children, and these things, together with the normal background sounds of the home, combine to give an ambient noise level of about 50dB. The minimum sound level which can be distinguished clearly above this background level is therefore 53dB. The dynamic range of orchestral music can be as much as 70dB, therefore in order to be able to hear the pianissimo as well as the fortissimo passages, a peak level of 123dB is required.
The acoustic power in watts required to produce a sound intensity level of 53dB is about 6 uW for an average-size living room. Since a 10-dB increase in power output requires a tenfold increase in power, the 123-dB peak-power level will therefore require a maximum acoustic output of some 50W. If the loudspeaker efficiency is 5% (and this is significantly better than is obtained from most commercially available loudspeaker systems) a peak-power output of 1000W per stereo channel is obviously required if the total dynamic range of a symphony orchestra is to be heard in comfort.
It was clear from discussions both with manufacturers and distributors that no serious attempt had been made to meet the requirement for drive units capable of handling as little as 250W. Initial trials made with some of the more likely units were generally unsatisfactory. In particular there was a tendency for the cone and speech coil to become detached, and for fraying of the surround. In addition, the failure was often made more serious by partial combustion of the flammable materials within or in proximity to the speech-coil assembly.
When more substantial reproducer units had been evolved, this only brought to light the flimsy nature of the housings which had been supplied, and considerable annoyance was caused by an injury sustained when one of the cabinets burst during an orchestral transient and the room was filled with flying splinters. At this stage it was accepted that the cabinets used would require to be of comparable strength to the reproducers, and the assistance of the specialist who constructed the metal cone loudspeaker assemblies was sought to manufacture four sheet-steel column loaded units, of a suitable type to take the 23" x 14" elliptical wide-band speakers. These are situated at the four corners of the listening room, and the opposite units are connected in parallel but antiphase. This has the effect of increasing the apparent dimensions of the listening room, in addition to reducing the I²R losses in the speaker wires.
Each unit is rated at 500 Watts, with a nominal 20 ohm impedance. The required output from the amplifier is therefore 10A at 100 Volts RMS (282 Volts peak-peak) per channel.
The use of a solid-state, transformerless amplifier to provide an output of 1kW into a 10 ohm load imposes certain limitations on the designer. In particular, the normal complementary or quasi-complementary output stage configurations are no longer practicable since the only useful and relatively cheap high-voltage transistors which are available are all of the n-p-n construction.
The basic output stage configuration employed, to provide a fully symmetrical push-pull class B output stage using only n-p-n transistors, is shown in Fig. 1. As shown, this would be satisfactory for power outputs up to about 50W. In this circuit arrangement, Tr2/Tr3 and Tr4/Tr5 are Darlington pairs with Tr2 and Tr4 being normal small-power driver transistors. Tr1 in combination with R1 and R2 provides the necessary signal level and amplitude transformation for the lower half of the output stage, and ZDI effectively stabilises the voltage level at the power output point. This is chosen so that the largest symmetrical voltage swing is obtainable. The symmetry of this stage is maintained up to a frequency determined by the resistance of R1 and R2, and the input shunt capacitance of Tr1. This will normally be well above the audible spectrum.
The final circuit employed is shown in Fig. 2. Although for simplicity only four parallel-connected output transistors are shown in each half of the output stage, this is only adequate for intermittent use at 1kW output. In practice six Parallel connected transistors are required in each half of the output stage.
The paraphase input is obtained from two medium-power high-voltage transistors, Tr3 and Tr4, the h.t. supply for which is obtained from a separately smoothed 400 V line, because bootstrapping is not practicable with this type of driver stage.
The input is derived from a long-tailed pair of p-n-p transistors, of a type chosen for high voltage linearity, and freedom from avalanche or collector leakage (Early effect) distortion. Although 150V is applied to the end of the 'tail', the maximum collector-emitter voltage is limited to about 52V, because the base of Tr2 is retuned to the 50V tap on the zener diode chain. A variable resistor is included in the tail to set the current through Tr1 and Tr2. This controls the current through Tr3 and Tr4, and, since the output dc level is determined by ZD1, thereby controls the quiescent current of the output stage.
This should be set to about 200mA. Because of the absence of coupling or bootstrapping capacitors the gain of the circuit from the base of Tr1 to the output of the power transistors is constant from the h.f. roll-off point down to d.c. The l.f. roll-off point is therefore determined solely by the 2uF input capacitor and the output time constant.
The input impedance is 2k ohms in series with 2uF. The h.f. roll-off point and the phase stability margin is determined by C1 (the input-lag capacitor) C2 and R3, and C3 and R5. The loop gain is determined by resistors R1 and R2 and is approximately 100. The full output is given by an input of IV r.m.s., which can be obtained from any suitable high-quality pre-amplifier capable of operating into a 2 k ohm load.
The construction of the power amplifier unit follows conventional lines, and no unusual precautions are required apart from the need for generous heat sinks. Very satisfactory results were given in the prototype by the use of a pair of old cast-iron radiators, such as can be found second-hand for a few pounds in a builders yard, to which the transistors can be individually attached by small bridges made from a suitably substantial gauge of copper sheet.
The bottom and sides of an old copper preserving pan would be ideal. Care should, of course, be taken in drilling the attachment holes to make sure that the radiator shell is still capable of retaining water without leakage. If such radiators cannot be found, a copper hot-water storage cylinder would serve admirably, but it would probably be more difficult to introduce such an item inconspicuously into the listening room. The siting of the output transistors should combine shortness of signal leads with the required thermal separation of the power transistors one from another.
It should also be borne in mind that the circulating currents at full power are of the order of 30A. The leads to the loudspeaker terminal bosses - for which old car battery connectors are suggested - to the collector and emitter rails of the output transistors, and to the h.t. and earthy ends of the h.t. decoupling capacitor block must be substantial. A 3/8" x 1/4" bore copper pipe is preferable, but as an alternative, lengths of 12 s.w.g. copper wire may be plaited together. After assembly, it is recommended that the amplifier units be bench-tested on a dummy load before attachment to the speaker units, since quite trifling faults can lead to a surprising amount of energy being released. For example, in preliminary listening trials with the prototype, an intermittent open circuit in the earth braiding on an input to the pre-amp, led to the necessity for the listening room ceiling to be substantially restored and replastered.
Although the results obtained with good quality gramophone recordings have been most astonishing, and have brought home to the author in the most vivid way the qualities of stamina and emotional detachment required of an instrumental player situated, as the fortunate listener, in the midst of a large orchestra, it is clear that there are a large number of residual problems in the life-like reproduction of disc recordings, of which the major one is the avoidance of acoustic feedback. As with many other of these problems, it is suspected that the manufacturers of the equipment have not really got down to serious thought on this matter, and the solution which the author feels most people must adopt, that of housing the record player unit is some detached building, such as a small garden shed, is inconvenient and prevents the listener from hearing the beginning of the recorded piece. Moreover, if in one's hurry to return to the audition room, the pickup cartridge is let fall too rapidly upon the record, extensive damage can be caused to windows and other glazing.
The performance of the equipment as installed is entirely satisfactory, and a wide variety of sound sources have been explored during the assessment of the scope of this system, and many sounds have been recaptured with a degree of realism not previously encountered. However, the development of this apparatus has not been without difficulty, scepticism and expense, and it has been suspected at times that unnecessary difficulties have been placed in the author's way. For these reasons, it is thought unlikely to appeal to those for whom high-fidelity reproduction is merely a passing interest. On the other hand, it has proved possible to purchase several of the adjoining properties at a very advantageous price, and this has undoubtedly offset a large part of the constructional costs.
Although the above article is intended as humour (note the month of publication), there is actually much truth in the details of dynamic range, and the power levels needed to reproduce the maximum and minimum sound levels experienced in a live concert. It is worth noting however that the 5% loudspeaker efficiency quoted equates to a sensitivity of about 99 dB/m/w - a figure that is extremely hard to achieve in reality for a wide range loudspeaker.
I must say that this article is typical of Wireless World (and its modern equivalent Electronics World), that even in humour, their articles will still attempt to be factual (if a little fatuous). I have been a reader for many years, and cannot recommend the publication(s) highly enough. The detail below is the least I can do for breaching their copyright on this article - I hope it appeases the publishers such that my sentence may be reduced to only a brief incarceration.
Sutton, Surrey SM2 5A5
It was explained in a later issue that the power amplifier described would indeed work, although the descriptions of the heatsinking and cabling required may have been a little over the top. Considering that this was written over 30 years ago (and given the "state of the art" at that time), there is no doubt that George Izzard O'Veering (say it quickly if you haven't got the pun yet) was something of a prophet! Note that the comments about NPN transistors being the only generally available high voltage and power types was quite true 30-odd years ago. Today, PNP devices are available which match all but the most ambitious NPN offerings, although very high voltage PNP devices are still uncommon.
Although the 'requirement' for 2000 Watts was considered outrageous at the time this was written, the power levels used now exceed George's system in some home installations, and as for professional use, I'm sure that if he were alive today, he'd be turning in his grave!