Speaker system electrical damping for orchestral music reproduction

 

  

By now, as far as I can judge, a paradoxical situation has emerged in the field of sound reproduction. Though there are amplifiers and acoustic systems with very good characteristics and even with moderate prices, there is a considerable number of the consumers not satisfied with the result. All the users can be provisionally split into four groups.

   The first group: “Why bother? If the apparatus perfectly reproduces the electronic music, modern pop and vocal performances, then what the hell do you still need? The symphony orchestra is no different from other sources of the musical signal, and if you don’t like its sound, then make claims to the sound producers, not the manufacturers of the speaker systems and amplifiers.”

  The second group of the “witnesses of warm tube sound” is convinced that apart from the tube devices, nothing can come closer to the ideal, and the ideal is unattainable in principle.

  The third group is convinced that only the silver-plated wires connecting the amplifier to the speakers can save the day. At the same time, the power supply should be a masterpiece by definition, the amplifier should be mounted, and its price should start from ten thousand dollars.

  The fourth group states carelessly: “Is there a need in a tube sound, a good stage and a high fidelity of instrument timbre? Connect 2Ohm resistors in series with the speakers and get what you want. Not enough? Turn on the feedback through the transformer.”

  There is also a numerous fifth super group with a non-blunt argument similar to “there are no fools in the west, and if you have an advanced technique and you are not happy with it, then clean your ears.”

   Perhaps all this reasoning is a common fluctuation in social groups, like the fans who are in favour of Jaguar only, and in no case of Mercedes.  Or vice versa.

But if, for a change, one seeks the opinion of the professional musicians, one will find out that they still have remarks to the modern sound-reproducing equipment. Moreover, the disadvantages are noted while reproducing the orchestral music and it happens more often with a symphony or chamber orchestra. The tube amplifiers are also preferred by the professionals with absolute hearing and extensive experience. But musicians, alas, cannot formalize their objections, and usually the comments sound like this: "... but in this fragment something is wrong" or "the first violin sounds dull.” Of course, it is almost impossible to translate these comments into the language that the engineer understands...

   Let us try, however, to solve the problem of the fidelity of the reproduction of the orchestral music not using empirical radio amateur practice, but using the method that is used for the work in the conditions of the lack of the information and high uncertainty of the input data. It means that each hypothesis corresponding to the Occam's criterion will be confirmed experimentally. And let us move step by step, not trying to get everything at once.

  So, first let us take the budget class equipment and begin the test. Everything works, the frequency response is close to the stated, the power too, but there is some strange sobbing in the high-frequency domain. If used as the “boombox”, it seems to be even not bad, but with the orchestra there is some obvious discomfort. We disassemble, examine critically the Chinese assembly, and see that the electrolytic capacitors hang on high-frequency heads. One piece per speaker (!). It turns out that the innovators can be met not only in our country; the Chinese cost optimizer also does not waste time. Let us replace the capacitors with nonpolar ones.  

   The grunting at high frequencies, of course, disappears. But some sort of dirt still remains, an incomprehensible haze in the mid-frequency range. We continue to check and record the oscillograms at the output of the amplifier when operating on an active load - everything is in order, the monochromatic signal looks decent in the entire frequency range and at all power levels.  We turn on regular speaker systems, and ... profanity flows in a free stream, interspersed with complete foul language. We look: the apparent ultrasonic excitation of the amplifier is on the oscillogram, due to the feedback operation.

 

Fig.1. Waveform of the output signal of the audio power amplifier. Ultrasound excitation.

 

  

Obviously, the reactive components of the impedance of the speaker systems cause reverse current pulses, the duration of which is comparable to the time delay of the inverse feedback. Thus, there is a classical phase lag in the control loop and, as a result, the excitation. Since the whole process is on ultrasonic frequencies, it is not audible, but the distortions on medium and low sound frequencies also appear - it is they that create acoustic discomfort.

   We eliminate the disgrace and check the whole set "by ear". We get a typical standard non-transparent "transistor" sound with a completely "lost" scene. And, as expected, the variety phonogram is reproduced quite decently, but there are problems with the chamber and symphonic orchestras.

  Let us try to find out the reason. We also take into consideration that there are special programs for the assessment of the quality of the amplifiers. In the result of their work, a very detailed spectrum is obtained with a high frequency resolution and the detection of spurious components to -100 dB or less. But often the amplifier with nominally worse parameters sounds subjectively better.  In this case, checking the equipment only with the help of a curve tracer and/or spectrum analyzer can be unpromising. Since the system obviously works better with monochrome than with musical signals.

   Let us estimate the sound of the orchestra, which is to be reproduced. This is the broadband signal (the overlap in frequency is up to three orders of magnitude), modulated in frequency and amplitude (up to 80 dB or more), and it is an additive mixture of modulated signals of many different musical instruments. Since the instruments work in sync, the sum of their sounds is the very complex superposition, giving out emissions, the envelope of which can be of almost any shape.

  However, a priori known signal is required for the probing analysis of the sound-reproducing equipment. But, as it was already mentioned, the slowly varying sinusoidal signal of the curve tracer will be an unacceptable simplification, and it can be assumed that some of the reproduction defects will be missed. For example:  we fix the normal transmission of sinusoidal signals in the entire frequency range with the maximum admissible value of the signal from peak to peak. But replacing a sinusoid, say, by a trapezoid, lowers the peak factor of the signal, and the power required will increase in inverse proportion to its magnitude. As a result, distortions are possible, which are guaranteed not to be seen when testing with a monochromatic signal.

  It is also obvious that the choice of the location for the delta pulse or the white Gaussian noise limited in the band will require the correlation and the spectral estimation, and this will greatly complicate the research procedure. Therefore, Heaviside function can be a good option - it will allow to observe the process in real time on the oscillogram.

  We will use the square wave as the probing signal and choose a realization that is shorter than the duration of its period. Obviously, there should be the picture of the rapid rise of the beam, followed by the exponential decay due to the discharge of capacitors. The decay may not be quite exponential, because it also depends on other differentiating circuits in the preliminary cascades, but the smooth asymptotic approximation to zero should be obligatory.

 

Fig. 2. Current and voltage during capacitor discharge

 

  

Now we will conduct an experiment: we apply a square wave to the input of the amplifier, synchronize the oscilloscope, let us supply voltage to the upper beam from the speaker system input, and current to the lower beam in the inverted switch. This scheme is convenient for its visibility - it turns out to be actually the eye diagram, in which the distance between the beams on the ordinate axis is proportional to the square root of the instantaneous power.

 

Let us limit the bandwidth of the amplifier from above. In accordance with the theory we should expect the relatively rapid rise, both in current and voltage, as well as the smooth decline close to the exponent.

 

Fig.3. Speaker input current/voltage diagram. The lower beam (current in the circuit) clearly deviates from the smooth decline. The loudspeaker is in the standard speaker system (NB!).

 

   Let us suppose that such behavior of the graph of the current is due to the direct and inverse power conversion due to the movement of the diffuser and the coil. In order to test this hypothesis, we will forcibly stop the diffuser and repeat the experiment.

 

Fig.4. The diffuser is stopped, pay attention to the graph of the current

 

   We will carry out an additional measurement with the amplifier's open passband and compare the current/voltage diagram with the free and locked diffuser.

 

Fig.5. Diagrams for the free (top) and locked diffuser (bottom) with the amplifier's bandwidth open (there is an additional RF rise).

 

  

As it can be seen, distortions of the current form appeared in the region of low times, i.e. high and medium sound frequencies. In this case, there is no doubt that the detectable effect is due precisely to the direct and reverse conversion of electrical energy into mechanical energy. Let us also pay special attention to the fact that the voltage graph remains unchanged in all cases, i.e. the amplifier negative feedback system correctly processes the input waveform. But the instantaneous power (the square of the distance between the beams) does not change proportionally to the voltage, and this should inevitably lead to distortions in the sound reproduction.

    Naturally, the question arises: why is this effect not detected in the spectral evaluation of the quality of the sound reproducing equipment?

The problem is that it will show up very weakly while applying monochromatic signals and only in the phase spectrum, which in itself will be sufficiently non-monotonic due to the presence of differentiating and integrating circuits and/or various equalizers in the amplifier.

    When using non-sinusoidal signals, the detection of the observed effect in the frequency domain will also be difficult, since it is concentrated in time, and in accordance with the basic principle of uncertainty, is distributed throughout the spectrum. And, if the spectrum of the applied non-harmonic signal is complex, then catching the small differences of the most different spectral components is difficult and unpromising. A spectrogram with ultrashort overlapping realizations is needed, but at the same time we should say good-bye to our resolution in the frequency domain. Thus, in this case, the work in the time domain is more preferable than in the frequency domain.

   If we talk about the quantitative assessment of the introduced distortions, then the measurements of the post-convolutional lobe of the Barker sequence, in my opinion, may be the best solution. However, such measurements require the development of the special technique.

    Thus, suppression of the observed effect can be considered to be absolutely necessary for the high-quality reproduction of the complex musical pieces. Indeed, while it is not suppressed, the amplifier together with the speaker system actually represent a musical instrument with its own overtones: the input music composition piece is an excitation signal (something like a violin bow), and the multiplicative (NB!) component of the effect shown provides an acoustic picture in the form response to the excitation. The resulting distortions are, of course, small, but they are captured by a musical ear on the complex signals of orchestral music.

   Let it be necessary to achieve suppression of the undesirable effect by 10 ... 12 dB. Deeper suppression is desirable, but it is hardly achievable in practice. The ways to solve the problem can be different: for example, the provision of the effective negative feedback that will monitor the instantaneous power. However, this has already happened - EMF, EDF systems (electromechanical and electrodynamic feedback). Fast and good results can not be expected; any such inverse feedback will also have a phase lag, and it is necessary to use a non-linear forward approximation for the efficient operation, i.e. The prediction of signal values, or at least to work on changes in the first derivative, and it will immediately increase the sensitivity to the noise and spurious high-frequency components. 

   Another option that seems to be more practical is to supply the “correct” instantaneous power to the speaker, and let it do with it what it can. After all, the good speaker system is the offspring of the hard work of the professionals who designed, developed and tested it. So, let it show itself in all its glory.

   For the implementation of this approach, it is necessary to match the impedance of the speaker with the active resistance of the damper, and to apply the output signal of the amplifier to the active load, which transmits some of the energy to the same damper. Thus, the direct connection of the amplifier with speakers through active loads will be relatively small. The amplifier will operate at almost purely resistive load. The speaker system will be electrically damped as much as possible, and its power will be provided by the source that has a predominantly active impedance component. But we will have to sacrifice the coefficient of the efficiency, and the output impedance of the amplifier is to be raised up to the value at least double of the speaker impedance.

  The final circuit diagram will be very simple and reducible to a resistive divider.

 

Fig.6. Speaker system connection via resistive divider

  

Let us check the result using the equipment which we started with. This freak has not become a decent product, but the “transparent” sound was obtained, and the “scene” appeared ... Thus, listening to music (and not just pop music) has become possible.

 

Fig.7. Diagram at the speaker system input. The goal is achieved, the picture pleases the eye. R2=7 Ohm, R3=5 Ohm, R4 (speaker system)=4 Ohm.

   The calculation of the resistors is simple: we assume that R1 and R4 are purely active, and the impedance at the input and output of the resistive divider must be respectively equal to R1 and R4. If it is possible to adjust R2 and R3, then it makes sense to make the final adjustment of the resistors “by sound” within + -20% relative to the calculated values, since the creative process is always more productive than any dogma. In the case of the deliberately low impedance amplifier output, one can first try R2=7 Ohm, R3=5 Ohm, but this is a palliative. The resistors are to be used with minimum inductance, and their power is selected taking into account the power of the amplifier (trimmings the heating elements from the toaster are well suited).

  

 

Conclusions:

1. The effect of additive and multiplicative interference caused by mechanical processes in a speaker system can be detected while analyzing a priori known signals in the time domain.
2. The active load of the amplifier together with the effective electrical damping of the speaker system makes it possible to achieve the suppression of the described effect and the acceptable sound of complex musical pieces.
3. The inevitable power loss of the proposed solution does not potentially pose a serious problem, although, most likely, the conceptual diagram of the high-impedance output stages of the amplifier should be developed in a new way.
4. The described method was also tested on the personal amateur-built equipment. While listening to the references of the professional musicians the author got nothing to be ashamed of

 

P.S. And for those who fancy doing something on their own: some of my additional considerations after 40 years of experience

 

1. Creative disputes similar to “we listen not to the amplifier or acoustics, but to the power supply capacitors (or other elements, carelessly chosen by the authors)”are, by definition, wrong. If we want to keep to this topic, then we listen to the interference pattern in the premises. That is why the precise orientation of the speakers in azimuth and elevation is very important, and the experimental work will certainly pay off with a good stereo effect. The importance of the correct wave front of the pair of speakers is enormous.

 

2. Digital signal sources have a bad habit of supplying DAC by-products in the line. And nobody can cancel the multiplication on the non-linearities of the current-voltage characteristics of amplifiers ... Therefore, a passive RC (19KHz, at least -6dB) filter at the amplifier input is absolutely necessary when playing orchestral music.

 

3.  Shielding the amplifier from RF interference is necessary - if you can hear characteristic frames when using a GSM phone, you can expect other RF signal sources to affect your equipment both through direct detection and due to other more complex processes.

 

4. All digital equalizers are evil. And, if they are made on all-pass filters, then this is the absolute evil Any spatial perception will be lost. All tone controls should have a linear phase response.

 

5. The acoustics for music in the “boombox” style and for symphonic music is different by definition: the second one is also suitable for the first variant, but the first one is not suitable for the second one in principle.

 

6. It is not clear why, rather for the sake of aesthetics, the manufacturers do not cover speaker systems with sound-absorbing material from the outside. If you ignore the requirements of beauty (and of your spouse) and cover the parts of the wall in the immediate vicinity of the speaker, then the result will be the significant improvement in the “scene”.

 

7. Since company Technics began installing weights on the woofer magnet, there have been many upgrades of the speaker cases. The demand of the aircraft designer Ilyushin remains the most appropriate for orchestral music: "Always let the power along the shortest path.” It means tightening speaker front and rear planes. I would add for myself: the design must be preliminary strained, i.e. front and rear walls are constantly tightened with effort.