|
The amount of information which has been published about the electrostatic or capacitor loud speaker is almost negligible. Being such an unknown quantity it has not yet come into general use, although it has been on the way for over fifty years. This neglect is not altogether deserved, for in some respects, particularly reproduction of is the highest frequencies, it is unrivalled. Being fundamentally different from other types of loud speaker it is unfair simply to substitute it and expect the best results. The purpose of the author of this contribution will be to explain how the electrostatic loud speaker works and to give advice on its use.
If a sufficiently high voltage were applied between the plates of a variable capacitor, and the moving plates were sufficiently freely suspended, they would rotate into the maximum capacity position. In practice there is no great likelihoods of our tuning dials moving round in this way when an extra, powerful atmospheric arrives, because the friction at the bearings is so great that the voltage would spark acrossbbetween the vanes long-before being enough to shift them round. But the electrostatic voltmeter, which is simply a very tiny and lightly suspended variable capacitor fitted with a pointer, demonstrates the truth of this principle every time it is used.
The principle, of course, is that a difference of potential or voltage between any two conductors causes them to attract one another, and, where they are free to move, to approach one another.
If the potential continually varies, the degree of attraction varies and the motion is continuous. A back and forward motion repeated at anything between about 16 and 16,000 times a second causes sound. Therefore, a capacitor is, in principle at least, a loud speaker. Actually an ordinary paper fixed capacitor may sometimes be heard to emit a faint note on being connected across the AC mains or other source of high alternating voltage. But generally the manufacturer has taken good care that the two strips of foil that make up the capacitor are too tightly sandwiched between the waxed paper dielectric strips to vibrate at all.
Practical Considerations
The Ferranti electrostatic voltmeter illustrates the principle of electrostatic attraction.
If a capacitor is to be any good as a sound reproducer it should be exposed to the open air, and one at least of its two conducting elements should be reasonably tree to move to and from the other. This could be done by suspending it a little distance away, with a layer of air in between. There are several disadvantages in this. The layer of air acts as a cushion, tending to damp out the motion, just as a pneumatic stop prevents a door from being slammed, however hard it is pushed. Then the force set up by a given voltage falls off very rapidly as the plates are separated even slightly. And to keep them separated at a fixed distance without solid material in between it is necessary to impede their free motion still further, either by making the plates very thick and stiff or by stretching them tightly.
Component parts of the Vogt stretched diaphragm loud speaker.
The result is that the simple capacitor loud speaker tends to be very inefficient and insensitive, and the object of design is to remove these objections. An example of the stretched diaphragm type is that by Hans Vogt, of Ferrocart fame. He used an extremely thin moving diaphragm, stretched so tightly close to a perforated fixed plate, that the natural frequency was above the usual audio frequency range - about 15 kHz, in fact. The perforations were intended, of course, to remove the air damping.
Fig. 1. Section of the 'Kyle' electrostatic loud speaker.
The Kyle loud speaker of American origin has the effect of splitting up the total surface into a vast number of tiny diaphragms, by stretching the moving element directly on the fixed one, separated only by a thin layer of flexible insulation. Fig. 1 shows in section a small portion of this arrangement. The fixed plate is slotted and ribbed, and so shaped where it makes contact with the composite moving plate that an increase in voltage between them causes the latter to cling closer to the fixed plate and squeeze some of the air out through the slots. The reverse action takes place when the voltage relaxes.
Fig. 2.- Section of the 'Primustatic' loud speaker. The dotted line shows the position of the diaphragm under the force of attraction.
A still closer approach to unrestricted motion is obtained in the Primustatic speaker, which, as it is one readily obtainable in this country, will be principally considered. Fig. 2 shows an enlarged section of it, in which the fixed plate, of perforated aluminium, is slightly curved. Behind it is a tinfoil-coated sheet of waxed paper, folded so as to form tiny triangular sectioned air spaces behind the perforations at the 'dead' lines, in between the rows of perforations, where there is no motion, the foil paper is held in position by a special sort of hairy thread which makes very light, but adequate, contact everywhere. It is possible to use graphite-coated paper instead of metal, as a fairly high resistance does not interfere with the operation.
It will be clear that when a voltage is set up between the plates a sort of rolling action takes place, causing the line contact to spread out into a strip, as shown dotted, and air to be expelled through the perforations.
Possibilities of Distortion
Fig. 3. - Without a polarising voltage, frequency doubling would occur in an electrostatic loud speaker.
As the attractive force takes place whenever a difference of voltage exists between the plates, it is obvious that they will move together during both positive and negative halves of an alternating wave. Thus, in Fig. 3, if (a) represents two complete periods of a 50 Hz supply, (b) represents the corresponding attractive force set up, and it will be seen that there are four complete waves. So instead of a 50 Hz note we get a 100 Hz note.
Reproduction in which every frequency is double what it ought to be is not likely to be considered satisfactory. It is interesting to note that precisely the same result is obtained in a moving-iron loud speaker or headphones in which no permanent magnetism is provided, the only attractive force being due to the signal current. The remedy in the latter case gives us the clue to that for the electrostatic speaker defect the provision of a steady initial force considerably larger than any due to the signal.
Fig. 3 (c) shows the combination of the high steady voltage and the alternating signal voltage, and one important feature is that it never reverses - the initial voltage being relatively large keeps it on one side of the base line throughout. Consequently the frequency-doubling effect is absent, and the attractive force closely follows the outline of the voltage, Fig. 3 (d).
The Polarising Voltage
It can be shown mathematically that not only is the distortion reduced to an unimportant quantity, but also the sensitivity is considerably increased. In practice it is not advantageous to increase the initial polarising voltage indefinitely, even if it were convenient. There would be the danger of breaking down the insulation - air or solid according to the type of construction - between the plates. Also the freedom of motion would be impaired by an excessive displacement in one direction. So the actual voltage is something of a compromise, and one is not far wrong in imposing the voltage used at the anode of the output valve. The signal voltage at this point is, of course, always substantially less. If the latter is stepped up, however, it may be necessary to increase the polarising voltage also, to make sure that it is in the correct proportion.
A polarising voltage of 1,000 used to be necessary for the earlier types, which is sufficient to explain the disfavour with which they were regarded for general use, and an advantage of the Primustatic type is that about 250 Volts is usually enough. So, what used to be a serious drawback now seldom presents much difficulty. The loud speaker being in effect a capacitor draws no current from the polarising source.
A rather more serious criticism is that the amplitude of motion is not very great, and that there is therefore difficulty in obtaining strong reproduction of the lowest frequencies. This can be got over to some extent by increasing the area of the diaphragm, which can be done as much as one pleases by adding more units in parallel.
Lastly, on the debit side of the account there is a problem in efficiently coupling the speaker to the output valve. This matter will be gone into in detail later.
Back and front views of the 'Primustatic' loud speaker unit.
High-frequency Response
Now for the credit side. It has already been pointed out that the high note reproduction is a strong feature. There are several reasons for this. Instead of the relatively heavy and complex moving system of any of the magnetic types of speaker, which renders it difficult to get upper frequency reproduction except in the form of resonances, there is a light uniform moving diaphragm, with extremely low inertia. Moreover, this is actuated all over its surface instead of at one part, as in other types, thus avoiding the complicated modes of vibration with resulting resonances and irregularities that distinguish the latter. Further, the area of the diaphragm can be made as large as one pleases, and the focusing and interference effects of a small cone are avoided. It is a better sound radiator, in other words. There is practically no upper frequency limit, and therefore it is in advance of present day microphones and. transmitting systems, and advantage can be taken of any improvements in the latter. But even with existing standards of transmission, so long as local interference does not impose a severe limit, it is possible to appreciate a very greatly increased clarity and faithfulness of reproduction. It is significant that for special experiments in America, where substantially uniform output of sound was required up to 14 kHz, a capacitor reproducer was employed.
The advantage is particularly marked in speech, string tone, and sounds involving transients, such as clapping, tapping, paper rustling, and cymbals.
The construction is much simpler than that of any other type. There are no coils, magnets, or field excitation. For the same reason it is extraordinarily light, and as regards compactness it can be made into panels of very little depth - an inch or so, for example, or even less if necessary.
An electrostatic, loud speaker, used alone, is far superior in naturalness and clarity of speech to a moving coil speaker, particularly of the cheap sort now almost universal. But unless very large there is not enough depth to music to please most listeners, although even this deficiency can be made considerably more tolerable by the delightful distinctiveness of the instruments in the upper registers, which is variously described as brilliance, 'life', and crispness. It is a great relief to get away from the thumpy whoofy reproduction that is so common, or the still commoner apology for 'brilliant' tone produced by a fierce high note resonance.
The best overall reproduction is therefore given by a combination of moving coil and electrostatic speakers. To obtain a satisfactory distribution of labour between the two, or even to run an electrostatic on its own, we must consider how it behaves as a load in the valve circuit. Unless attention is given to this the results can be very bad indeed. So it will form the subject of the next part of this article.
Matching Loud Speaker and Output Valve
We have seen the fundamental difference between the electrostatic loud speaker and all other types. Before it can bbe effectively used it is necessary to consider it as a load in a valve circuit.
A moving-iron speaker can be considered approximately as an inductance. A moving-coil speaker is conveniently (but not very accurately) assumed to be a resistance. An electrostatic speaker may be represented as a capacity, perhaps not quite so pure (i.e., free from resistance) as the very best capacitors, but one does not go far wrong in neglecting the impurity.
Fig. 4 - Variation of capacity with polarising voltage in the 'Primustatic'18 in × 20 in loud speaker
It has already been explained that the effect of a voltage between the two plates is to draw them closer together. Therefore, one would expect the capacity to be increased thereby, and this is exactly what happens. Fig. 4 shows the measured capacity of a 'Primustatic' loud speaker with an 18 in. by 20 in. diaphragm. The capacity averages about 0.008 μF per square foot, or a reactance of about 20 megohms divided by the frequency in Hertz.
The Load Diagram
In drawing load curves on a valve diagram the loud speaker is usually represented by a straight line, which means a resistance, constant at all frequencies. No loud speaker ever does act just like that, but a moving-coil type is near enough to it over the middle range of frequencies for one to get at least a hazy idea of how to match it to the valve. Our capacitor speaker has this advantage at least, that it is very closely a capacity load, but, unfortunately, it is therefore not a straight line at all, but an ellipse, and one of a different size at every frequency. So it is rather an exasperating business trying to fit it comfortably into the valve diagram. To start with, it would be worse than manufacturing 1 'Mickey Mouse' film to draw a diagram for every frequency, so let us select three only - 80, 800, and 8,000 Hz. The respective reactances of a 2½ square foot speaker are 100,000, 10,000, and 1,000Ω.
There are two ways in which these figures can be altered to suit the valve: first, by selecting a diaphragm of different area, and, secondly. by using a step-up or step-down transformer, which has the effect, looked at from the primary side, of multiplying the reactance by the square of the transformer ratio. A step-up is equivalent to an increase in capacity. But it is inevitable that with any one arrangement the reactance must vary over the same range as the frequency.
Fig. 5. - The load ellipse of an electrostatic loud speaker varies in shape and inclination as the frequency changes.
The result of this characteristic is that the behaviour is strikingly different according to whether a triode or a pentode is used. Fig. 5 shows a diagram for a triode with an internal resistance (impedance) of about 2,500 Ω, and the three reactance eclipses have been drawn in to show what happens at the lowest, middle, and highest frequencies. The ellipses in each case are the largest that can go in without running into grid current on the left, or bottom-bend rectification at the foot.
At 80 Hz the full grid excitation is possible; that is to say, the grid voltage can be swung right from zero on one side to double the bias voltage on the other, without any possible risk of overloading due to rectification. The voltage developed across the loud speaker is the maximum possible.
Permissible Grid Volts
At 800 Hz it is just possible to give it the full grid, but the ellipse has opened out so much that it is approaching the danger zone along the foot. Still, it is working quite happily and developing practically the full voltage. At any higher frequency, however, the lower half would be flattened out, unless the whole ellipse were reduced in size by reducing the grid input. This is well shown at the upper extreme of 8,000 Hz, where the grid excitation must be reduced to less than a half in order to avoid rectification distortion, and the anode voltage developed across the loud speaker dwindles to about a tenth.
Fig. 6. - Curves showing maximum permissible grid Volts with capacity load.
These features are illustrated rather more concisely in Fig. 6. Curve A shows the maximum peak volts that can be applied to the grid without overloading, and curve B shows the voltage developed across the loud speaker with maximum grid volts as given. In each case the voltage is level up to a certain critical frequency, in this case 800 Hz, after which it rapidly falls. It must not be hastily concluded that the response suddenly falls off above 800 Hz, for it must be emphasised that curve B is strictly dependent on A, and shows the maximum output short of overloading the valve.
P P Eckersley has shown experimentally that in normal broadcasting the amplitudes to be expected at the upper frequencies are considerably lower, and would come well below curve A. In addition, by-pass capacitors, HF tuning characteristics, and other factors are almost certain to prevent the grid voltage from breaking the allotted bounds. So if the grid excitation is kept at a level six Volts throughout, as it can be without fear of overloading (curve Ab), the output is as shown by Bb.
Another point is that it is the falling part of curve Bb, rather than the level part, that is correct. The amplitude of diaphragm required to give a constant output of sound gets steadily less as the frequency rises. It is not possible to state exactly what is the ideal characteristic without knowing exactly how the particular type of diaphragm behaves at various frequencies; but a listening test with a constant-amplitude pure tone shows a satisfactory response down to about 800 Hz, and a falling response below that.
The Critical Frequency
We can calculate this critical frequency by measuring the distance in anode volts horizontally from the initial working point () on the valve diagram to the zero grid voltage point X, and dividing this by the vertical distance, in anode milliamps, from () to Y, the minimum current consistent with avoiding severe curvature. This gives the reactance in thousands of ohms, which is also equal to 1,000/2πfC where f is the frequency and C is the capacity in microfarads; so it is easy, knowing C, to calculate f. By altering C in either of the two ways already described it is possible to shift the critical frequency f, thus extending either the level or the falling part of the curve.
As it has just been stated that it is the falling part that is correct, it would seem that the sensible thing to do would be to make the critical frequency as low as possible. That means making either the actual capacity or the step-up ratio as large as possible. An additional allurement is that an increase in capacity means an increase in sound-radiating surface, and an increase in step-up means an increase in signal voltage, and in either case it looks as if the volume would be increased without any greater expenditure of power. But, while it is true that there is improved uniformity of response, the improved efficiency fails to materialise; for curve A is shifted to the left too, and necessitates severely cutting down the input. And, as the lower tones are less audible than the upper, there is an apparent falling-off in volume, and in endeavouring to restore it by the volume control the only result is rattling and distortion. If, on the other hand, the capacity is too small, the volume again drops, and what output there is consists almost entirely of extreme top.
Hence a compromise is necessary, giving a reasonable efficiency at the upper frequencies, the lower being augmented, if necessary, by a bass moving coil unit.
Fig 7. - Choke coupling provides the simplest connection between valve and loud speaker.
To descend for a moment from these theoretical reasonings to consider how the connection is made in practice, the simplest circuit is that of Fig. 7, where L is a choke of high inductance capable of carrying the valve anode current. In this way the loud speaker receives both the output voltage from the valve and the steady polarising voltage from the HT source. The choke behaves as a 1:1 transformer, and if its inductance is sufficiently high it by-passes a negligible proportion of the 'signal', and so the whole arrangement conforms very closely to the preceding theory. It may be of interest to realise that at one particular frequency the impedance of the anode load is almost infinite, due to the resonance of choke and loud speaker, and is also a pure resistance, and hence a nearly horizontal straight line instead of an ellipse. With normal components this frequency is about two or three hundred Hertz, but the phenomenon does not appreciably modify the performance as already described. The whole action is very beautifully confirmed by cathode ray oscillograph tests, using a variable-frequency oscillator. Beginning at the lowest frequencies, we see the 'lengthwise' ellipse become a straight line, and then open out into an 'upright' ellipse, which, unless the input is reduced, causes terrible overloading, the anode current rising much above normal.
Fig. 8. - Connections for applying polarising voltage with transformer coupling.
If a transformer is used to alter the ratio it is essential to allow for the polarising voltage (Fig. 8), and if it is a step-up ratio it may be desirable to add some auxiliary voltage, but only if the HT voltage itself is rather low. An old dry battery can be used, as no current is drawn. If the secondary is linked to the anode end of the primary the sum or difference of the voltages across both windings is obtained, giving a choice of three ratios altogether.
Push-pull Connections
Fig. 9. - Push-pull output using transformer secondary winding to energise the loud speaker.
A push-pull stage is much to be recommended, and Fig. 9 shows one method of connection. If there is no secondary winding, or if it is being used for another loud speaker, the connection of Fig. 10 is another of the many schemes. Although this looks a one-sided arrangement, it actually loads the whole transformer, which acts as a 2:1 step-down.
Fig. 10. - This circuit gives a 2:1 step-down and leaves secondary free for an additional loud speaker.
Summarising: the larger the area of diaphragm, or the higher the step-up, the lower is the frequency below which response falls off. But the lower also is the efficiency and the greater the tendency to rattle. So only when there is plenty of power available in the last stage is it possible to arrange these matters so as to go low down the scale. The smaller the number of milliWatts available the higher must be the critical frequency. A dual speaker combination is in any case the best for effective reproduction over the whole audible scale. Methods of obtaining this, together with tone control, and the conditions for pentode operation, will next be considered.
Circuits for Operation with a Moving-coil Unit
Fig. 11. - Critical reactance is given by OX in Volts divided by OY in Amps.
It has been shown that when an electrostatic loud speaker is connected to a triode output valve there is a certain critical frequency below which the response falls off. The critical frequency is that which makes the reactance of the loud speaker (which being in effect a capacitor is dependent on frequency) equal to OX (Volts) divided by OY (Amps) (Fig. 11) where Y is the lowest permissible anode current without serious bottom-bend rectification.
It is convenient to remember that this reactance is roughly equal to the optimum loacl for the valve. The critical frequency can be pushed lower to widen the effective response, but only at the expense of efficiency. So unless a very large output is available and used rather wastefully, it is better to obtain the bass response by means of a moving-coil loud speaker and to make the critical frequency quite high, say about 1,000 Hz. The following table may be useful.
The capacity of the loud speaker is the actual capacity multiplied by the square of the output transformer ratio, a step-up giving a larger capacity and vice versa.
Fig.12. - Practical circuit for combining the electrostatic with a moving-coil loud speaker following a triode output valve.
Fig. 12 shows a method of adding an electrostatic speaker to an existing triode and moving-coil (or inductor) speaker. The latter is likely to have a resonance somewhere about 2,000 or 3,000 Hz and as this comes within the province of the electrostatic it is desirable to cut it out by inserting a choke L1, with an inductance of about one Henry, preferably variable to allow adjustment to suit the working conditions. If L1 is used, it is important to avoid any capacity across the transformer, or even excessive self capacity, or there will be a series resonance and a worse state of affairs than ever. Another improvement is a variable resistance R of about 10 kΩ, which can be used to prevent the high-note overloading described earlier. A further refinement, with the object of preventing large low frequency voltages reaching the electrostatic speaker and causing it to rattle, is the choke L2, also tapped, with values up to 2 or 3 Henrys.
The Pentode Valve
Fig. 13. - Modified arrangement of Fig. 12, with step-down transformer to minimise overloading.
If R is in, the parallel resonance due to C and L2 may undesirably emphasise one frequency; if it is out, there may be both high and low-note overloading. But as L2 is of value mainly in high-power stages, the overloading point is not likely to be reached, and, in any case, the danger of overloading can be minimised by stepping down, as shown in Fig. 13. Remember that stepping down tends to reduce the electrostatic low-note response, and vice versa, while excessively wrong ratio causes loss of volume too.
Fig. 14. - Load diagram for the electrostatic loud speaker in conjunction with a pentode output valve.
The ease of the pentode is considerably different from that of the triode, and the situation is best explained while looking again at a valve load diagram (Fig. 14). Again we take the extreme and middle frequencies of 80, 800, and 8,000, giving with a loud speaker capacity of 0.02 μF reactances of 100,000, 10,000, and 1,000 Ohms. Neglecting for the moment any complications due to the necessary coupling arrangement, these three conditions are again represented by the three ellipses. Here the lowest frequency ellipse develops across itself the maximum possible anode voltage, but unlike the triode it does so with a very small grid swing. Any larger grid swing would result in excessive voltages being developed, besides severe distortion.
Fig.15. - Operating conditions for a pentode derived from the curves of Fig. 14.
The medium frequency condition is similar to that of the triode in so far as the full grid swing results in full output voltage with just tolerable distortion. The load ellipse only just succeeds in fitting in. The highest frequency condition is also similar in that the anode voltage developed is very low, but full grid swing is possible at all the upper frequencies. Transferring these results to Fig. 15, it is seen that curve B, which shows the maximum voltage that can be developed across the speaker without overloading the valve, is practically the same as that for a triode. But the actual state of affairs is very different, for the output shown is on the assumption of maximum allowable grid swing, which as indicated by curve A is just the opposite to that of the triode. Moreover as in average broadcasting the actual distribution of grid swing over the frequency scale is very different from curve A, being greatest at low frequencies, curve B must be interpreted accordingly. If the grid swing were distributed according to the intensity given at the various frequencies during normal broadcasting, the resulting curve B would slope off at the high frequency end very steeply indeed. A listening test confirms this by giving the rather surprising result that this type of loud speaker, noted for high-note response, is deficient in the same when run from a pentode.
At the lowest frequencies, however, there is a considerable divergence between theory and practice. It was explained in connection with the triode that some sort of coupling device is essential, and if this takes the form of a choke or a transformer, the loud speaker resonates with the inductance at some moderately low frequency. That causes little concern in the triode, for as long as the impedance in the anode circuit is greater than several times the valve impedance it makes little difference, and it is a series resonance, which causes the load impedance to become very small, which is to be guarded against. But in the pentode the output voltage increases almost without limit as the load impedance is raised, and the effect of a parallel resonance is to cause excessive emphasis of the resonant frequency, accompanied by severe overloading unless the volume is reduced so that all other frequencies are very feebly reproduced.
Practical Schemes
This serious objection can be counteracted by connecting a resistance in parallel with choke and loud speaker. If, for example, the resistance is 20,000 Ω, it is obvious that the load can never exceed this figure, and so a limit is not to the voltage that can he developed, and a much more uniform response obtained but somewhat at the expense of efficiency. Fortunately, the efficiency of the pentode drive starts off at a considerably higher maximum than that of the triode.
Fig. 16. - Dual loud speaker connections for a pentode output valve. The variable resistance provides tone control.
But the desirability of a dual loud speaker system is even more marked with the pentode, and so is the desirability of preventing the low frequency voltages from being set up across the electrostatic loud speaker. This object can be achieved by shunting it with a relatively low inductance and also a resistance to prevent a sharp resonance. A fairly large capacitor shunt across the moving-coil loud speaker completes the division of labour, and a resistance is necessary here also. All this sounds very complicated, but Fig. 16 shows how a single variable resistance can be made to serve not only as both shunting resistance but also as an effective tone control.
The ordinary transformer is used for the moving-coil loud speaker, and the upper frequencies are by-passed by the 0.1 μF capacitor. The electrostatic loud speaker coupling choke may be 1 or 1.5 Henrys on the primary side, and a secondary inductance to suit the size of loud speaker. The Varley 3H choke provides a variety of taps for obtaining the best balance. A 1:1 ratio is about right for a 0.01 μF speaker.
The tone-control potentiometer gives every tone from extreme low to extreme high preponderance.
Although any standard moving-coil loud speaker can be successfully used in a circuit of this type, it is, of course, preferable to use one specially intended for low-note reproduction. If there is no necessity to look after the high Volts, it is possible to do better justice to the bass by selection of suitable cone material and dimensions, and in other ways.
Fig. 17. - Simplified circuit for adding the electrostatic loud speaker to existing receivers. The loud speaker takes the place of the tone correction capacitor.
A simpler and less satisfactory circuit, but one which is more convenient to add to an existing receiver, and also gives some degree of tone control, is given in Fig. 17. It will be seen that in this the electrostatic loud speaker takes the place of the usual tone-compensating capacitor, but, of course, it is essential that the HT voltage should come across it.
When applying to existing sets, it is important to see that any tone-compensating system that may be left in circuit is not preventing the electrostatic loud speaker from pulling its weight. In fact, the whole receiver, whether with triode or pentode output, must obviously be capable of passing the higher audible frequencies to the loud speaker if it is to reproduce them. This point needs to be stressed, because it is customary to cut high frequencies down fairly drastically in the interests of selectivity and freedom from mush, scratch, and hiss. The best plan is to restrict such limitation to one particular part of the circuit, where it can be controlled.
Where conditions permit - as when a station is being received at short range in relation to its power the 'top cut' can be removed and the advantages of the electrostatic principle realised to the full. Attention should therefore be paid to bypass capacitor, anode filter chokes, grid 'stopper' resistors. and other devices which tend to side-track the high tones.
It is a disputed point whether phase distortion is an important imperfection in reproduced sound. It is generally agreed that the ear is not sensitive to displacement of phase, such as occurs whenever a transformer coupling is used, in the case of sustained sounds like those of the flute or organ. This matter, being so extremely difficult to handle either theoretically or experimentally, seldom goes beyond vague references to 'attack', yet it is undoubtedly of importance.
There seems little doubt that the electrostatic type of loud speaker is superior to any other in the reproduction of transients, hut it hardly has a chance if there is considerable phase distortion in the preceding stages. The substitution of well-designed resistance coupling for transformers helps, but even if transformers are eliminated from the receiver there are likely to be several in circuit before ever the wave strikes the receiving aerial. Careful test has established, however, that the removal of even one transformer of several can be detected by a sensitive ear. So the point is one worth considering, and the electrostatic loud speaker is the most useful with which to consider it.
The combination of moving-coil and electrostatic units gives full bass response with unusually good reproduction of high frequencies and transients.
|
