The designs of high-frequency (HF) speakers are the most diverse. They can be ordinary, horn or dome. The main problem in their creation is the expansion of the directionality of the emitted oscillations. In this regard, dome speakers have certain advantages. The diameter of the diffuser or radiating membrane of HF tweeters lies in the range from 10 to 50 mm. Often the tweeters are tightly closed at the back, which excludes the possibility of modulation of their radiation by the radiation of LF and MF emitters.

A typical miniature tweeter with a cone diffuser radiates high frequencies well, but has a very narrow radiation pattern - usually within an angle of 15 to 30 degrees (relative to the central axis). This angle is set when the speaker output is reduced by typically -2 dB. Specifies the angle when deviating from both the horizontal and vertical axes. Abroad, this angle is called the angle of dispersion or dispersion (dispersion) of sound.

To increase the scattering angle, diffusers or nozzles for them of various shapes are made (spherical, in the form of a horn, etc.). Much depends on the material of the diffuser. However, conventional tweeters are unable to radiate sounds at frequencies much higher than 20 kHz. Placing special reflectors in front of the tweeter (most often in the form of a plastic grating) allows you to significantly expand the directivity pattern. Such a grating is often an elemental acoustic framing of a tweeter or other emitter.

The eternal topic of controversy is the question of whether it is necessary to radiate frequencies above 20 kHz at all, since our ear cannot hear them, and even studio equipment often limits the effective range of audio signals at a level of 10 to 15-18 kHz. However, the fact that we do not hear such sinusoidal signals does not mean that they do not exist and do not affect the shape of the time dependences of real and rather complex audio signals with much lower repetition rates.

There is a lot of convincing evidence that this shape is heavily distorted when the frequency range is artificially limited. One of the reasons for this is the phase shifts of various components complex signal. It is curious that our ear does not perceive phase shifts by itself, but is able to distinguish signals with a different form of time dependence, even if they contain the same set of harmonics with the same amplitudes (but different phases). Great importance has the nature of the decay of the frequency response and the linearity of the phase response, even outside the effectively reproducible frequency range.

Generally speaking, if we want to have uniform frequency response and phase response over the entire sound range, then the frequency range actually emitted by acoustics should be noticeably wider than the sound one. All this fully justifies the development of broadband radiators by many leading companies in the field of electroacoustics.

Placement of HF radiators There is a problem - the result to a large extent depends on where the heads are placed and how they are oriented. Let's talk about the HF head, or tweeter.

Features of RF heads From the theory of sound wave propagation, it is known that with increasing frequency, the radiation pattern of the emitter narrows, and this leads to a narrowing of the optimal listening zone. That is, you can get a uniform tonal balance and the right scene only in a small area of ​​​​space. Therefore, the expansion of the RF radiation pattern is the main task of all loudspeaker designers. The weakest dependence of the radiation pattern on frequency is observed in dome tweeters. It is this type of RF emitters that is the most common in automotive and household speakers. Other advantages of dome radiators are their small size and the absence of the need to create an acoustic volume, and the disadvantages include a low lower cutoff frequency, which lies in the range of 2.5-7 kHz. All these features are taken into account when installing a tweeter. Everything affects the installation location: the operating range of the tweeter, its directivity characteristics, the number of installed components (2- or 3-component systems), and even your personal taste. Let's make a reservation right away that there are no universal recommendations on this issue, so we cannot point a finger at you - they say, put it here and everything will be OK! However, today there are many standard solutions which are useful to know. All of the following applies to non-processor circuits, but this is also true when using a processor, it's just that its presence provides much more opportunities to compensate for the negative impact of a non-optimal location.

practical considerations. Let us first recall some canons. Ideally, the distance to the left and right tweeters should be the same, and the tweeters should be installed at the height of the listener's eyes (or ears). In particular, it's always best to push the tweeters as far forward as possible, because the farther they are from the ears, the smaller the difference in distance between the left and right drivers. The second aspect: the tweeter should not be far from the midrange or bass / midrange head, otherwise you won’t get a good tonal balance and phase matching (usually guided by the length or width of the palm). However, if the tweeter is set low, then the soundstage collapses downwards, and you are, as it were, above the sound. If set too high, due to long distance between the tweeters and midrange speakers, the integrity of the tonal balance and phase matching is lost. For example, when listening to a track with a recording of a piano piece, at low notes the same instrument will sound at the bottom, and at high notes it will fly up sharply.

RF head directivity. When we figured out the installation site of the RF head, we should decide on its direction. As practice shows, in order to obtain the correct timbre balance, it is better to point the tweeter at the listener, and to obtain a good sound stage depth, use reflection. The choice is determined by the personal feelings of the music you listen to. The main thing here is to remember that there can only be one optimal listening position.
It is desirable to orient the tweeter in space so that its central axis is directed to the listener's chin, that is, to set a different angle of rotation of the left and right tweeters. There are two things to keep in mind when orienting a reflection tweeter. First, the angle of incidence sound wave is equal to the angle of reflection, and secondly, by lengthening the sound path, we take the sound stage further, and if we get carried away, we can get the so-called tunnel effect, when the sound stage is far from the listener, as if at the end of a narrow corridor.

setting method. Having outlined, in accordance with the above recommendations, the location of the RF heads, it is worth starting experiments. The fact is that no one will ever say in advance exactly where a 100% "hit" with your components will be provided. The most optimal place will allow you to determine the experiment, which is quite simple to set up. Take any sticky material, such as plasticine, double-sided tape, Velcro or model hot glue, put on your favorite music or test disc and, with all of the above, start experimenting. Try different options for places and orientations in each. Before you finally install the tweeter, it's better to listen a little more and correct it on plasticine. to nowhere.

Creativity. Setting up and choosing the location of the tweeter has its own nuances for 2- and 3-piece systems. In particular, in the first case it is difficult to ensure close proximity of the tweeter and LF/MF emitter. But in any case, don't be afraid to experiment - we've seen installations where HF heads ended up in the most unexpected places. Is there any point in an additional pair of tweeters? For example, the American company "Boston Acoustics" produces sets of component speakers, where the crossover already has a place for connecting a second pair of HF heads. As the developers themselves explain, the second pair is necessary to raise the level of the sound stage. In test conditions, we listened to them as an addition to the main pair of tweeters and were surprised how much the space of the sound stage expands and the nuances are improved.

Theory of harmonics

Amplitude compression

What to do?

Overloading (clipping) power amplifiers is a common occurrence. This article deals with overload caused by an increased level of the input signal, as a result of which there is a clipping of the output signal.

After analyzing the "phenomenon" of this kind of overload, which allegedly causes damage to the speakers, we will try to prove that the true culprit of this is amplitude compression (compression) of the signal.

WHY DO SPEAKERS NEED PROTECTION?

All loudspeaker heads have a maximum operating power. Exceeding this power will damage the loudspeakers (SH). These damages can be divided into several types. Let's take a closer look at two of them.

The first type is excessive displacement of the GG diffuser. The GG diffuser is a radiating surface that moves as a result of an applied electrical signal. This surface may be conical, domed or flat. The vibrations of the diffuser excite vibrations in the air medium and emit sound. According to the laws of physics, for louder sound or more low frequencies the diffuser must oscillate with a larger displacement amplitude, while approaching its mechanical boundaries. If it is forced to move even further, then this will lead to excessive deflection. This happens most often with low frequencies, although it can happen with mid-ranges and even high-frequency ones (if low frequencies are not limited enough). Thus, excessive displacement of the diffuser most often leads to mechanical damage heads.

The second enemy of the GG is the thermal energy resulting from thermal losses in the voice coils. No device is 100% efficient. As for the GG, 1 W of input power is not converted into 1 W of acoustic power. In practice, most GGs have an efficiency of less than 10%. Losses due to low efficiency are transformed into heating of the voice coils, causing their mechanical deformation and loss of shape. Overheating of the voice coil frame causes a weakening of its structure, and even complete destruction. In addition, overheating can cause the adhesive to foam and enter the air gap, causing the voice coil to no longer move freely. In the end, the voice coil winding can simply burn out like a fuse in a fuse. It is clear that this cannot be allowed.

It has always been a major problem for users and developers to determine the power handling capability of multi-band speakers. Users who replace damaged tweeters are most likely to

convinced that what happened was not their fault. It would seem that the output power of the amplifier is 50 W, and the power of the speakers is 200 W, and, nevertheless, the tweeter fails after a while. This problem forced engineers to figure out why this is happening. Many theories have been put forward. Some of them have been scientifically confirmed, others have remained in the form of a theory.

Let's look at a few perspectives on the situation.

HARMONIC THEORY

Studies of the distribution of energy over the signal spectrum have shown that, regardless of the type of music, the level of high-frequency energy in sound signal far below the level of low frequency energy. This fact makes it even more difficult to figure out why tweeters are damaged. It would seem that if the amplitude of the high frequencies is lower, then the low-frequency speakers, and not the high-frequency speakers, should be damaged first of all.

Speaker manufacturers also use this information when developing their products. The idea of ​​the energy spectrum of music allows them to significantly improve the sound of tweeters by using lighter moving systems, as well as using thinner wire in the voice coils. In speakers, the power of the tweeters usually does not exceed 1/10 of the total power of the speaker itself.

But since there is more musical energy in the low-frequency (LF) range than in the high-frequency (HF) range, which means that, due to its low power, high-frequency energy cannot cause damage to high-frequency speakers. Therefore, the source of high frequencies powerful enough to damage the tweeters is somewhere else. So, where is he located anyway?

It has been suggested that if there are enough low-frequency components in the audio signal to overload the amplifier, it is likely that output clipping will create high-frequency distortion that is strong enough to damage the tweeter.

Table 1. Harmonic amplitudes 100 Hz square wave, 0 dB = 100 W

This theory became quite widespread in the early 70s and gradually began to be perceived as a "dogma". However, as a result of research on the reliability and protection of power amplifiers in typical conditions, as well as the practice of using amplifiers and speakers by typical users, it turned out that overloading is common and is not as noticeable to the ear as most people think. The operation of the overload indicators of amplifiers is usually delayed and does not always accurately indicate the real overload. In addition, many amplifier manufacturers deliberately slow down their response based on their own ideas about how much distortion must be generated in order for the indicator to light up.

More advanced and better sounding amplifiers, incl. amplifiers with soft clipping also damage tweeters. However, more powerful amplifiers damage the tweeters less. These facts further strengthened the theory that the source of damage to tweeters is still amplifier overload (clipping). It would seem that there is only one conclusion - clipping is the main cause of damage to high-frequency speakers.

But let's continue the study of this phenomenon.

AMPLITUDE COMPRESSION

With the amplitude limitation of a sinusoidal signal, the amplifier introduces large distortions into the original signal, and the shape of the received signal resembles the shape of a rectangle. In this case, the ideal rectangle (meander) has the highest level of higher harmonics. (see fig 1). A less clipped sine wave has harmonics of the same frequency but at a lower level.

Take a look at the spectral content of a 100Hz, 100W square wave shown in Table 1.

As you can see, the power delivered to the tweeter after passing this signal through a perfect 1 kHz crossover is less than 2 watts (0.83 + 0.589 = 1.419 watts). That's not a lot. And do not forget that in this case, a hard, ideal overload of a 100-watt amplifier is simulated, capable of turning a sine into a square wave. Further increase in overload will no longer increase the harmonics.

Rice. one. Harmonic components of 100 Hz square wave vs. 100 Hz sine wave

The results of this analysis indicate that even if a weak 5-10 W tweeter is used in a 100W speaker, it is impossible to damage it by harmonics, even if the signal takes the form of a meander. However, the speakers are still damaged.

So you need to find something else that could cause such failures. So what's the deal?

The reason is in the amplitude compression of the signal.

Compared to older amps, today's high-end amps have more dynamic range and better sound when overdriven. Therefore, users are more tempted to overdrive and clip amplifiers at low frequency dynamic peaks, as no major audible distortion occurs. This results in compression of the dynamic characteristics of the music. The volume of the high frequencies increases, but the bass does not. To the ear, this is perceived as an improvement in the brightness of the sound. Some may interpret this as an increase in volume without a change in sound balance.

For example - we will increase the signal level at the input of a 100-watt amplifier. The low frequency components will be limited to 100W as a result of the overload. As the input level increases further, the high frequency components will rise until they also reach the 100W cutoff point.

Look at fig. 2, 3 and 4. Graphs are graduated in volts. At an 8-ohm load, 100 W corresponds to a voltage of 40 V. Before limiting, the low-frequency components have a power of 100 W (40 V), and the high-frequency ones - only 5-10 W (9-13 V).

Let's assume that a music signal with low and high frequencies is fed into a 100-watt amplifier (8 ohms). We use a mixture of a low-level RF sinusoidal signal with a high-level LF signal (see Figure 2). The level of the high-frequency components supplied to the tweeter is at least 10 dB lower than the level of the low-frequency components. Now turn up the volume until the signal is clipped (+3dB overdrive, see Figure 3).

Rice. 2. A low level, high frequency sine wave mixed with a burst of high level, low frequency sine wave

Rice. 3. 100 watt amplifier output with 3 dB overdrive

Rice. four. Output from a 100-watt amplifier with 10 dB of overdrive

Note that, judging by the waveform, only the low-frequency components were limited, and the level of the high-frequency components simply increased. Of course, clipping generates harmonics, but their level is significantly lower than that of the meander we considered earlier. The amplitude of the high-frequency components increased by 3 dB in relation to the low frequencies (this is equivalent to the amplitude compression of the signal by 3 dB).

When the amplifier is overloaded by 10 dB, the amplitude of the RF components will increase by 10 dB. Thus, each 1 dB increase in volume causes an increase in the amplitude of the high-frequency components by 1 dB. The growth will continue until the power of the RF components reaches 100W. Meanwhile, the peak level of low-frequency components cannot exceed 100 W (see Fig. 4). This graph corresponds to almost 100% compression, since there is almost no difference between the high and low frequencies.

Now it's easy to see how the RF signal power exceeds the power of a 5-10-watt tweeter. Indeed, overloading generates additional harmonics, but they will never reach the level of the amplified original high-frequency signals.

You probably think that signal distortion would be unbearable. Don't fool yourself. You will be amazed to learn how high the overload limit is, above which it will no longer be possible to listen to anything. Just turn off the overload indicator on the amplifier, and see how far you turn the volume control of the amplifier. If you measure the output level of the amplifier with an oscilloscope, the level of overload will surprise you. 10dB of bass distortion is common.

WHAT TO DO?

If we can protect amplifiers from clipping, we can make better use of speakers. To prevent overload and the resulting amplitude compression in any modern amplifier, the so-called. clip limiters. They prevent the aforementioned amplitude compression, as when the threshold value is reached at any frequency, the level of all frequencies is reduced by the same amount.

In external limiters, the response threshold (threshold) is set by the user. fine tune

this threshold on the clipping level of amplifiers is quite difficult. In addition, the clipping level of amplifiers is not a constant value. It varies depending on the mains voltage, AC impedance, and even the nature of the signal. The limiter's threshold should continuously track these factors. The most correct solution would be to tie the threshold to the overload signal of the amplifier.

It is quite logical to build a limiter inside the amplifier. In modern amplifiers, it is easy to determine the moment of occurrence of overload with great accuracy. It is to him that the so-called built-in amplifiers react. clip limiters. As soon as the output signal of the amplifier reaches the overload level, the control circuit turns on the regulating element of the limiter.

The second parameter, after the threshold, inherent in any limiter, is the actuation and release times. More important is the recovery time after overload (release time).

There are two options for operating amplifiers:

• work as part of a multiband amplifying complex,

• work on broadband speakers.

In the first case, either only the low-frequency band, or the mid-range and high-frequency bands can be fed to the amplifier. When setting a long release time and operating the amplifier in the mid-high bands, the “tails” of the limiter recovery can be noticeable aurally. And vice versa - with a short release time and operation in the low band, distortion of the signal shape may occur.

When operating the amplifier on a broadband speaker, you have to look for some compromise value of the recovery time.

In this regard, amplifier manufacturers go in two ways - either a compromise release time is chosen, or a release time switch (SLOW-FAST) is introduced.

CONCLUSIONS:

If you ask me why this is necessary, then I will not answer you - then this article is not for you. If everything is in order with your motivation, then I offer for review some of the results obtained by me with the modest means and knowledge that I have available.

For starters - the guinea pig, who is he?

Our patient is a tweeter with a 3GD-31 cone diaphragm. The main claim to it is a significant unevenness and unevenness of the frequency response. Those. in addition to the unevenness of about 10dB between the maximum peak and dip, there are many smaller irregularities, as a result of which the frequency response is similar to a forest. I decided not to give the measured characteristics at the beginning of the article, because. it will be more visual to place them next to the final ones obtained after all design changes.
The main idea of ​​my actions, or rather the two main ideas, is, firstly, to add sound-absorbing elements inside the volume of the speaker in order to suppress resonances that arise in a closed volume with solid walls that easily reflect sound without noticeable absorption of its energy, which is the case of the specified speaker. The second idea is the processing of the diffuser material itself (no, not with A. Vorobyov's liquid ;-)), but with varnish, resulting in a composite material that is superior to the original (paper) in rigidity, but not inferior to it in damping its own resonances, which reduces bending deformation of the diffuser during its operation and thereby helps to reduce the resonant peaks-dips in the frequency response.

What's gotten into my head?

The fact is that I have been conducting similar experiments for a long time and received quite a lot of confirmation of the correctness and usefulness of my approach, but all the results were rather scattered. This was partly due to the lack of experience in acoustic measurements (and more so in the interpretation of the results), partly due to the incomplete formulation of the idea itself, the general plan of action. And when all this mosaic formed in my head into a more or less whole picture, I decided to conduct the experiment from beginning to end, simultaneously making all the measurements.

So what has been done?

For starters, the speaker was disassembled. To do this, the speaker coil leads were soldered from the terminals on the case, then, after soaking with acetone, the sealing cardboard ring was separated and the cone itself was peeled off in the same way from the metal “funnel” of the case. Next, the diffuser was removed from the housing and set aside for the time being.
First, the speaker housing was processed. Sectors were cut out of cloth, about 3 mm thick, exactly covering the inner surface of the body, which is a truncated cone. At the bottom (the smaller base of the truncated cone) a circle was cut out of the same material with a hole in the middle for the coil. After that, the inner surface of the body and the surface of the cloth blanks were smeared with one layer of Moment glue and almost immediately (because it dries very quickly and when I finished spreading the cloth patterns, the layer on the body was already dry) pressed against each other. Here is a photo of the resulting semi-finished product.

At that moment, the idea came to my mind that not only resonances in the volume of the case, but also in the walls themselves, could be to blame for the broken frequency response. the case is a kind of bell made of stamped sheet metal. To measure its resonances, I applied the following technique. Having placed the case on a soft base, with the magnet down, I installed the microphone directly above it, turned on the sound recording and hit the outside of the case several times with a plastic screwdriver handle. Then I chose the most successful (in terms of level) signal from the record and imported it into LspLab for analysis. Results a little later. Then, in order to dampen the body, it was glued on the outside with rubber from an ancient bicycle inner tube, using the same technology as the previous felting. Then, after complete drying - in a day, tests were again carried out, according to the same method as above. However, the sound from the impact was much weaker, so I automatically hit a little harder than during the first measurement - because of this, the signal level during the second measurement, in my opinion, turned out to be somewhat overestimated, but this does not play a significant role in this case . So, here are the first comparative results - the transient response of the speaker cabinet (in the form of a sonogram). Below is the original version.

It is clearly seen that after the revision, all resonances above 3 kHz were suppressed by a level value of more than 20 dB! From this image, it seems that the main resonance at 1200 Hz (by the way, interestingly, the main resonance of the speaker cone is located exactly at the same frequency) has become much stronger. This is not true, because the program normalizes the levels on the sonogram so that the most “strong” signals become red, however, this scale is valid only within one graph, and there are two of them on the image, so the red on the upper graph is 20 dB weaker than the red on the lower graph! Here is another - already more familiar graph - the frequency response of both measurements.

It can be seen that the damping efficiency increases with frequency and the suppression at frequencies of 3 kHz and above exceeds 30 dB! And this despite the fact that, as I said, in the second dimension I hit the body harder! You, lovers of "calm down" AC boxes, for the record - I give!

The diffuser was coated (not impregnated, namely coated) with nitro-lacquer (of all the materials tested for this purpose, it had the best effect on the properties of the speakers). On the inside, only one layer, on the outside, three. But, of course, these were not layers that paint not walls! When applied with a soft brush of the first layer, the surface is only moistened, and not much. The second and third layers are slightly thicker, but in total, the three layers are so thin that the fibrous structure of the paper is still visible from under them.

Before assembly, an additional “donut” of cotton wool was inserted into the cavity between the body and the diffuser in order to achieve maximum sound absorption in the volume as much as possible. In the following figure, the case prepared for assembly.

Another change was made to the coil leads. Initially, the thin wires of the coil winding itself were soldered to copper rivets on the diffuser (and hefty drops of solder were soldered!), Which should create a new resonant system from the mass of all this metal and the rigidity of the part of the diffuser on which it's all stuck. I did not like this state of affairs at all, so I decided to redo everything. I unsoldered the coil wires from the rivets, drilled them out and soldered the leashes connecting the coil to the external terminals directly to the voice coil wires. In the next picture, though not very good quality, the new state of affairs is shown. The remaining holes are sealed with paper circles.

Now I will give the summary result.

For starters, here is the frequency response of the original speaker and its after rework. Bold lines show the frequency response and phase response after rework.

At first glance, I did not achieve much success. Well, the dip at 4kHz decreased by about 3dB, the peak at 9kHz decreased by a couple of dB, and the frequency response leveled off from 12 to 20kHz. It can be completely attributed to random phenomena - the resonances in the diffuser were successfully redistributed. However, it should be said that this speaker was not very successful for the purposes of my experiment - it initially had an almost limiting quality for its design. For comparison, I will give a similar pair of frequency response for another sample - worse.

Here is all the miraculous effect of refinement on the face! However, I do not take this speaker as the basis of the article, because in this case this is all the data that I received, but I collected more information on the speaker described above.

Now I want to give the transient characteristics of the speaker. They are the same as for the body - in the form of sonograms, in my opinion, this is more clearly.

It is clearly seen that the original speaker has delayed resonances in the region of 5 and 10 kHz, reaching up to 1.3 ms in duration. After refinement, firstly, they are shortened by 1.5 times, and secondly, they break up into many smaller ones both in intensity and duration. Above 10 kHz, they do not exist at all - they have disappeared. In general, the impulse response has improved much more noticeably than the frequency response.
Based on this experiment, as well as several previous ones, I came to the conclusion that the varnish coating mainly affects the operation of the speaker in the highest frequency range, and various sound-absorbing materials work in the midrange.
Hull damping does not appear to have had a significant effect on the result.

In conclusion, I want to say that this article was written mainly with the aim of acquainting people who do not have the means of instrumental assessment of the objective parameters of speakers with the effect that specific actions have on a particular speaker sample.
As a result of these experiments, another idea arose to further improve the parameters. It will be the basis for further experiments and, if successful, the topic of the next such article.

http://www. /shikhman/arts/xe. htm

SAY A WORD ABOUT THE POOR BEEPER

Traditionally, the division of the MF and HF bands (or midbass-HF) is made by passive crossovers ( separation filters). This is especially convenient when using ready-made component sets. However, while the performance of the crossovers is optimized for this kit, they are not always up to the task.
An increase in voice coil inductance with frequency results in an increase in head impedance. Moreover, this inductance in the "average" midbass is 0.3-0.5 mH, and already at frequencies of 2-3 kHz, the impedance almost doubles. Therefore, when calculating passive crossovers, two approaches are used: they use the real value of the impedance at the crossover frequency in the calculations or introduce impedance stabilization circuits (Zobel compensators). A lot has already been written about this, so we will not repeat ourselves.
Tweeters usually lack stabilizing chains. In this case, it is assumed that the operating frequency band is small (two or three octaves), and the inductance is insignificant (usually less than 0.1 mH). As a result, the increase in impedance is small. In extreme cases, the increase in impedance is compensated by a 5-10 ohm resistor connected in parallel with the tweeter.
However, everything is not as simple as it seems at first glance, and even such a modest inductance leads to curious consequences. The problem lies in the fact that the tweeters work in conjunction with the high-pass filter. Regardless of the order, it has a capacitance connected in series with the tweeter, and it forms with the inductance of the voice coil oscillatory circuit. The resonance frequency of the circuit is in the operating frequency band of the tweeter, and a "hump" appears on the frequency response, the magnitude of which depends on the quality factor of this circuit. As a result, coloration of the sound is inevitable. Recently, many models of high sensitivity tweeters (92 dB and higher) have appeared, the inductance of which reaches 0.25 mH. Therefore, the issue of matching the tweeter with a passive crossover becomes especially acute.
The simulation environment Micro-Cap 6.0 was used for the analysis, but the same results can be obtained using other programs (Electronic WorkBench, for example). Only the most characteristic cases are given as illustrations, the rest of the recommendations are given at the end of the article in the form of conclusions. A simplified model of the tweeter was used in the calculations, taking into account only its inductance and active resistance. This simplification is quite acceptable, since the resonant impedance peak of most modern tweeters is small, and the frequency of the mechanical resonance of the moving system is outside the operating frequency band. We also take into account that the frequency response for sound pressure and the frequency response for electrical voltage- two big differences, as they say in Odessa.
The interaction of the tweeter with the crossover is especially noticeable for first-order filters, which are typical for inexpensive models (Figure 1):

crystal "color. An increase in inductance shifts the resonant peak to lower frequencies and increases its quality factor, which leads to a noticeable "click". A side effect of an increase in quality factor that can be turned to good is an increase in the slope of the frequency response. In the crossover frequency region, it is close to filters 2 orders of magnitude, although at a large distance it returns to its original value for 1 order (6 dB / octave).
The introduction of a shunt resistor allows you to "tame" the hump on the frequency response, so that some EQ functions can also be assigned to the crossover. If the shunt is made on the basis variable resistor(or a set of resistors with a switch), then you can even carry out operational adjustment of the frequency response within 6-10 dB. (picture 2):

DIV_ADBLOCK703">

https://pandia.ru/text/78/430/images/image004_61.jpg" width="598" height="337 src=">
figure 4

The third way is to introduce a resistor in series with the tweeter. This method is especially convenient for tweeters with an inductance over 100 mH. In this case, the total impedance of the "resistor-tweeter" circuit changes insignificantly during regulation, so the signal level practically does not change (Figure 5):

disc "> Stabilizing circuits are not necessary only for tweeters with low inductance (less than 0.05 mH). For tweeters with a voice coil inductance of 0.05-0.1 mH, parallel stabilizing circuits (shunts) are most beneficial. For tweeters with a voice coil inductance of more 0.1 mH, both parallel and series stabilizing circuits can be used.Changing the resistance of the stabilizing circuit allows you to influence the frequency response.For 1st order filters, changing the parameters of the stabilizing circuit has a noticeable effect on the cutoff frequency and hump parameters.For 2nd order filters, the cutoff frequency is determined by the parameters of its elements and depends on the inductance of the head and the parameters of the stabilizing circuit to a lesser extent.The value of the resonant "hump" caused by the inductance of the tweeter is directly dependent on the resistance of the shunt and inversely dependent on the resistance of the series resistor. cutoff frequency is in direct dependence filter quality factor. The quality factor of the filter is proportional to the resulting load resistance (HF heads, taking into account the resistance of the stabilizing circuit). The filter of increased quality factor can be calculated according to the standard method, but reduced by 2-3 times relative to the nominal load resistance.

The proposed methods for frequency response control are also applicable to filters of higher orders, but since the number of "degrees of freedom" there increases, it is difficult to give specific recommendations in this case. An example of changing the frequency response of a third-order filter due to a shunt resistor is shown in Figure 6:

home" three-four-way speakers had a switchable frequency response "normal / crystal / chirp" ("smooth-crystal-chirping"). This was achieved by changing the level of the midrange and treble bands.
Switched attenuators are used in many crossovers, and in relation to the tweeter, they can be considered as a combination of series and parallel stabilizing circuits. Their impact on the resulting frequency response is difficult to predict, in this case it is more convenient to resort to modeling.

DIV_ADBLOCK705">

After listening to music for a short time, I came to the conclusion that at an increased volume level, the HF sound pressure level prevailed over the rest of the frequencies to such an extent that discomfort arose. I had to either use the tone controls, or just turn off the music. By my nature, I didn’t want either one or the other, so I joined the struggle for a “comfortable” sound.

First of all, a resistance appeared in the crossover, connected in series with the speaker (Fig. 2). The capacitor had to be re-selected because the load resistance changed and the cutoff frequency along with it. The sound pressure has been reduced.

But "comfort" was not achieved. There was an opposite effect. At higher volume levels, the HF components were in moderation, but with a decrease in volume, the hand itself reached for the tone controls.

I had to try another option for regulating sound pressure - shunting the head with a resistance of 10-30 ohms (Fig. 3). This method is sometimes used. The smaller the value of the shunt resistance, the greater the suppression.

But the picture turned out somewhat different than it was intended. Basically, the resonant "hump" is suppressed, and the overall level change is negligible. The impact on the frequency response is also not bad, but the main task has not been solved. Nothing worked without tone controls.

Series and parallel resistors or circuits in this case are called dissipators. (dissipate means to scatter). They not only dissipate power, but also absorb the products of intermodulation distortion in the dynamics. So their influence on the character of the sound should be especially noticeable in inexpensive tweeters (Ed.)

Tone control is essentially an increase or decrease in sound pressure in a certain frequency band, depending on specific model head unit. Everyone has different adjustment options: on some devices they would be enough, on others they are not. There is also an opinion that the use of built-in tone controls worsens the sound of the system due to the correction of the frequency response of the head unit and additional phase distortions.
In addition, there are restrictions on the acoustic installation scheme used. When using a two-band front, when the adjustment band almost completely coincides with the tweeter's operating area, adjusting the sound pressure with the tone control is not so critical. But in systems with three bands, such an adjustment cannot give the desired effect, since when it is used, the frequency response of the midrange head will be distorted, part of the working band of which necessarily falls into the treble tone control zone.
As a way out, in these cases, the use of an equalizer with a sufficient number of control bands is justified. Using a simple 7-9 band EQ may not produce the desired effect. More advanced equalizers already cost a lot of money, which dramatically, one might even say - completely excludes their use in most amateur installations. Although, if we consider the system as a whole, the use of a multi-band equalizer will reduce the time when full customization the entire system. But that's not what we're talking about right now.

An idea arose - to use incandescent lamps to limit the level of high-frequency components at high volume. When heated, the resistance of the coil will increase and the power will be limited. Barreters are sometimes used in crossovers to protect against overload - the same lamps, but filled with hydrogen. Hydrogen promotes quick recovery low thread resistance. In this case, due to a sharp change in resistance, the dynamics of high-frequency reproduction will be disrupted. If, however, use ordinary lamp- there will be a smooth compression of the high-frequency range. The filament has a thermal inertia that depends on its mass. The more powerful the lamp, the greater the thermal inertia.

The use of a light bulb as a dissipator was initially simulated on a computer using the MicroCap program. The crossover circuit took the following form (Fig. 4):

A crossover circuit was simulated, the head was replaced with an equivalent circuit (to account for the effect of the inductance of the head itself). Then the frequency response graphs were obtained for all the options considered above.

The results of modeling the frequency response are shown in the graph (Fig. 8): At low volume, the resistance of the light bulb is about 0.5 Ohm. The frequency response of the crossover in this section is almost the same as the frequency response of the crossover without resistance.

From the frequency response graphs it can be seen that the decrease in pressure by -3 dB for all curves occurs at approximately the same frequency. For the option with a shunt resistance, the value of the capacitor was changed, since the cutoff frequency at the considered value went up.

Curve 1 - crossover frequency response without resistance. Curve 2 - Frequency response of a crossover with a series resistance of 1.2 ohms. Curve 3 - frequency response of a crossover with a shunt resistance of 16 ohms and a 3.5 uF capacitor. Curve 4 - frequency response of a crossover with a light bulb. The resistance of the lamp as a result of heating the spiral is 4 ohms. Curve 5 - frequency response of a crossover with a light bulb. The resistance of the lamp as a result of heating the spiral is 6 ohms.

After the "theoretical part" I moved on to practice. It was necessary to measure the resistance of the lamps at different voltages. By setting a different current with a rheostat, he measured the voltage on the lamp, the current strength and calculated the resistance according to Ohm's law. For three types of lamps, the following results were obtained (Fig. 9-11):

The graphs show the voltage value at which a weak heating of the center of the spiral begins.

results

After making changes to the scheme of his crossover, he began to listen. Let me remind you that the "comfort" of the sound was determined by ear. The use of the RTA analyzer was not supposed to be carried out due to the lack of it even on a citywide scale. Only by ear. If during prolonged listening there is no desire to use the tone controls, or turn off the source of "irritation", then I believe that the goal has been achieved.
In my system, the installation of bulbs from the interior lights, it seems to me, gave the expected effect. The "whistling" effect is gone, and there is no need to use the tone controls to increase or decrease the volume.

SIAMESE TWINS

Many modern installations use a double set of tweeters. The reason is the increased requirements for sound quality. The expansion of the dual-radiator pattern makes it easier to set up the sound stage, and the possibility of overloading the tweeters at high volume levels is reduced. External attractiveness also plays an important role, especially in exhibition works.
Another argument in favor of such a solution arises with channel-by-channel amplification. The well-known contradiction between the uneven distribution of the musical signal energy over the spectrum and the equal power of the amplifier channels is elegantly resolved when the tweeters are turned on in series. In this case, the maximum output power of the "beep" channels of the amplifier is halved compared to a normal load, which allows you to make full use of its dynamic range and reduce distortion.
However, all of the above implies the use of exactly the same tweeters. Another option is also possible - with different tweeters reproducing separate frequency ranges. The origins of this decision must be sought in home acoustic systems a quarter of a century ago. Reproduction of the entire frequency range above 3-5 kHz with one tweeter was then enough challenging task so it was split. The band from 3-5 to 10-12 kHz was reproduced by a small-sized cone tweeter, common for those years, and everything above was reproduced by a dome or ribbon horn super tweeter. With the development of technology, this solution has gone from mass home equipment, but has every chance of returning to the automotive one.
The problem of reproducing the entire high-frequency range with one tweeter was solved a long time ago, but a good broadband tweeter is a delicate and expensive product. At least in the lower and middle price range, no dome design and material can yet simultaneously meet all the requirements, which are mostly contradictory. Requires high rigidity, low weight, good internal damping. Therefore, for mass products, the results are disappointing:

The textile dome provides excellent elaboration of the upper mids and sound detail, but at the upper end of the range the sound is usually muffled (frequency response blockage). The metal dome provides excellent reproduction of the high-frequency part of the range. However, the low-frequency part of the range is not always reproduced adequately, the sound is often colored by the resonances of the dome itself (the tuning fork effect). A polymer or metallized dome provides a fairly wide frequency range, but, as a rule, with a significant uneven frequency response and radiation pattern. As a result, the sound can take on a different color.

Conclusion: advantages different materials must be united, and shortcomings must be compensated. Tweeters were the object of the study:

Prology RX-20s (Silk dome, 0.22mH inductance) Prology CX-25 (Metalized Mylar dome, 0.03mH inductance)

Listening showed that the silk tweeter, with all the detail of the sound, lacks "air", and the Mylar tweeter "clicks" beautifully, but when working with a first-order filter, it has a piercing "voice". Obviously, with an appropriate choice of the crossover frequency, they would make an excellent pair.
In order to simplify the design and facilitate the operation of the amplifier, it is most advantageous to use first-order filters. They create minimal phase distortion, which compares favorably with other designs. However, first-order filters provide too little attenuation outside the operating band, so they are only suitable for low input power or a sufficiently high crossover frequency (7-10 kHz). Therefore, in most serious designs, filters of higher orders are used, from the second to the fourth.
In this case, it was decided to use a quasi-second order filter using the inductance of the voice coil. The sensitivity of the tweeters turned out to be almost the same, and the inductance differed by almost an order of magnitude. This greatly simplified the design of the passive crossover, as the voice coil inductance entered the circuit.
The idea was inspired by the article "Say a word about the poor squeaker" ("Master 12volt" No. 47). It considered the interaction of the crossover and the tweeter, as well as methods of influencing the resulting frequency response. When working with a passive high-frequency filter, the inductance of the voice coil forms an oscillatory circuit with the filter capacitance, its resonance frequency is in the operating frequency band of the tweeter. As a result, a "hump" appears on the frequency response, the magnitude of which depends on the quality factor of this circuit. This can result in coloration of the sound and other artifacts. However, in some cases, these phenomena can be turned to advantage.

https://pandia.ru/text/78/430/images/image020_18.gif" width="420" height="320 src=">
figure 2

Capacitor C1 defines the lower limit of the reproducible frequency range of the entire system. The inductance of the voice coil BA1 is involved in the formation of the frequency response. In the crossover frequency region, the slope of the frequency response is close to filters of the 2nd order, although at a large distance it returns to the original value for the 1st order (6 dB / octave). The upper range limit for BA1 is formed acoustically. Since the return of a silk tweeter at frequencies above 11 kHz is noticeably reduced, it makes no sense to introduce additional signal attenuation. At the same time, the inductance of the voice coil and the capacitor C2 form a notch filter for a frequency of about 5 kHz. Suppression of this frequency range eliminated the "shrill" sound of the Mylar tweeter, leaving it to reproduce only the high-frequency portion of the range.
The frequency response of the voltage crossover is shown in Figure 3.

DIV_ADBLOCK711">

IMPROVE THE SOUND OF COAXIAL SPEAKERS

Component acoustic systems received wide distribution in car audi o, and with the advent of budget kits, their scope has noticeably expanded. The convenience of the layout, the ease of setting up the sound stage have earned them well-deserved popularity. However, in some cases it is more convenient to use coaxial speakers. There can be many reasons: the complexity of the cosmetic integration of component systems or additional tweeters, the desire to preserve the original appearance of the cabin, non-standard size, etc. In some cases, it is generally impossible to replace standard coaxes with other speakers without a radical alteration of the seats due to specific dimensions or design features . What to do in this case? Try to squeeze the maximum out of the available "raw materials".
Most often, coaxial speakers are installed in the dashboard and work in acoustic design"open box". Due to an acoustic short circuit, the reproduction of frequencies below 200-300 Hz is significantly weakened, regardless of the size of the cone and the frequency response of the radiator itself. All attempts to reproduce at least some semblance of bass without refining a regular place are meaningless. Therefore, we will consider the coaxial in the dashboard exclusively as a midrange-high-frequency emitter, and we will explore how its characteristics can be improved in this role.

Three sources and three components
(not Marxism, of course, but coaxial):

Main radiator Secondary radiator Crossover

The main radiator of mass constructions is equipped with a diffuser made of polypropylene of various modifications, and in standard coaxes it is often made of paper. In terms of sound quality last option preferred. Why is clear: smooth transition from the piston mode of operation to the zone mode, no overtones, low weight, rather high upper limit of the frequency range (7-10 kHz).
If we turn to statistics, then most of the "torpedo" caliber coaxes (10-13 cm) are equipped with one additional emitter. Most often it is a tweeter with a textile or plastic dome with a diameter of 13-18 mm, sometimes metallized. The natural resonance frequency of such emitters is 1.5-3 kHz, we will remember this for the future.
The crossover of most coaxes only works with a tweeter and is formed by a single capacitor with a capacity of 3.3-4.7 microfarads, most often an electrolytic one. Thus, this is the simplest first-order filter with a cutoff frequency of 6-9 kHz, so the suppression of out-of-band signals is insufficient, and the tweeter may be overloaded. The result is a "piggy squeal" and noticeable resonant overtones.

Where to begin

So, the first and most obvious way to improve the sound quality is to replace the oxide capacitor in the crossover with a more decent one, and at the same time reconsider its value. If the main emitter is paper, then it confidently wins back the mid-frequency range, and the help of the tweeter is required only in the high-frequency part of the range. In this case, the capacitance of the capacitor can be reduced up to 2 μF, this will shift the maximum return to the frequency range above 10 kHz. As noted at one time ("Say a word about the poor tweeter" - "Master 12volt" No. 47), the electrical resonance of the filter capacitance with the inductance of the voice coil of the tweeter forms a small hump on the frequency response, so we will "push" it up to improve the return in this frequency range. Increasing the frequency of the section will also increase the overload capacity of the tweeter, this will allow more power to be delivered to the speakers without risk.
Now let's deal with the main emitter. Since coaxials do not use "hard" diffusers prone to internal resonances, the transition from piston to zone operation occurs smoothly. Therefore, there is no need to additionally limit the frequency band from above.
An increase in voice coil inductance with frequency results in an increase in head impedance. Moreover, this inductance of the "average" coaxial is 0.2-0.4 mH, and already at frequencies of 2-3 kHz, the impedance almost doubles. The circumstance is unpleasant, but in our case it can be turned to good.
In the case of component acoustics, the crossover usually has an impedance stabilizer in the form of an RC circuit connected in parallel with the speaker. A number of works have shown that for mid-frequency heads it is more convenient to turn on a series resistor (dissipator). With this connection, the head is no longer powered from a voltage source, but from a current source, therefore, not only the impedance is stabilized in a wide frequency range, but also a significant reduction in intermodulation distortion, which is especially noticeable when using inexpensive wideband and mid-frequency heads.
Practice shows that it is enough to install a resistor with a resistance approximately equal to 0.5-1 of the nominal impedance of the head. For a crossover frequency above 300 Hz, the power dissipation of the resistor should be equal to 15-20% of the rated power of the head. Recoil reduction and damping degradation should also be taken into account, but we agreed not to consider the low-frequency region.
Now let's see what the inclusion of a resistor in series with a coaxial head will lead to. For modeling, as usual, we use the MicroCap environment and a simple model of a dynamic head with average Re and Le values ​​for coaxes.

mumbling" in the resonance frequency region of the main speaker (100-150 Hz). But, since the sensitivity has decreased by about 6 dB, you will most likely have to forget about connecting the modified coaxial to the built-in amplifier of the head unit. external amplifier there is an active crossover to limit the working frequency band from below.
As an experiment, several coaxial speakers were subjected to refinement different brands:

AUDAX (standard Renault) Prology PX-1022 JBL P-452

In all cases, the "enlightened" sound of the mid-frequency range was noted, the "hoarseness" of the tweeter disappeared with a large input power, and the overall tonal balance improved. Even the rough AUDAX with heavy cardboard cones and disgusting tweeters - and they have found a second wind.