|
Arabic
Bengali
Chinese
English
French
German
Hebrew
Hindi
Italian
Japanese
Korean
Malay
Polish
Portuguese
Spanish
Turkish
Ukrainian
Vietnamese|
ENCYCLOPEDIA OF RADIO ELECTRONICS AND ELECTRICAL ENGINEERING dbx noise reduction system - past and present. Encyclopedia of radio electronics and electrical engineering
Encyclopedia of radio electronics and electrical engineering / Audio equipment In the proposed article, the author considers the features of the device, operation and application of one of the most effective noise reduction systems - the dbx compander system, which at one time competed with the well-known Dolby-A system. Moreover, based on a thorough analysis of the shortcomings of such systems, he created a compander UWB, practically devoid of their main drawback - a noticeable distortion of the fronts of musical signals. Many people are well aware of the name of Ray Milton Dolby, at least by the names of the most common noise reduction systems - Dolby-B, Dolby-C and Dolby-S, designed for use in home appliances. He also created Dolby-A (the first commercial noise reduction system) and Dolby-SR companders for professional use. Suffice it to say that the word "dolby" is sometimes used in the most general sense to refer to noise reduction systems in general, and not a specific type. To date, in professional recording, due to the transition to digital technology for multi-channel recording and the displacement of analog tape recorders, noise reduction systems have lost their former importance. The only noise reduction system that is currently used in high-quality analog technology is Dolby-S / SR. However, a quarter of a century ago the situation was different. Ray Dolby's firm was just "getting on its feet" with its four-way system.1, which allowed to reduce the noise by only 10 dB. Dolby was quite complicated, expensive ($300 per channel), and most importantly, it required precise adjustment of tape recorders (±0,2...0,3 dB). Only first-class studios could afford this (London-Decca. Deutsche Grammofon Gesellschaft, etc.)2. It is no coincidence that the trial operation of the Dolby system began precisely at the Decca studio in England, and not in the USA. At the same time, there were many places where, along with less criticality to the accuracy of the equipment settings, a noise reduction of more than 10 dB was required. The first success in solving this problem fell to the American David Blackmore. The dbx compander noise reduction system he created in 1971 (US Pat No. 3,789,143)3 was easy to use, inexpensive and provided noise reduction up to 30 dB. But its main advantage turned out to be non-critical to the spread of transmission coefficients and frequency response of recording and playback channels.
It is worth recalling that most of the noise reduction systems proposed by that time (and even later ones) turned out to be of little use for practical use. Their main drawbacks were either excessive sensitivity to defects in the recording medium (magnetic or film tapes), or the introduction of unacceptable distortions into the sound. Dolby managed to stand out against this background at the cost of using a complex multi-band device, the visibility of distortion was reduced by limiting its regulation (0 ... 10 dB in the range of input signal levels from -40 to -20 dB). Naturally, the noise suppression in this case turned out to be small. Blackmer thought otherwise. Since the criticality to the uneven frequency response in the Dolby system is caused by the division of the signal spectrum into bands, therefore, the compander must be made broadband so that it processes the entire frequency band at once4. And since the criticality of level matching in the Dolby system is caused by unequal processing of signals with different levels, the compander must be designed in such a way that its operation algorithm does not depend on the signal level5. Based on this, the noise reduction system was designed, which laid the foundation for the company dbx (written in lowercase letters) - from David Blacmer Excellence (according to other sources, Experience). Now this company is one of the "giants" in the market of studio equipment. In addition, the successful VCA (voltage-controlled amplifier) design developed by Blackmer is still used in most studio sound processing devices to this day. A block diagram of the main version of the dbx noise reduction system is shown in the figure, borrowed from proprietary materials. The noise suppressor consists of two parts: the main channel through which the processed signal passes, and the control channel. The input signal during recording, having passed through the input band-pass filter of the PF, the frequency pre-distortion generator of the main channel (corrector 1) and the voltage-controlled amplifier (VCA), simultaneously arrives at the output of the device as a whole (i.e., at the input of the recording amplifier) and at the input of the control channel. The control channel consists of an input frequency corrector (corrector 2), a phase splitter, two r.m.s. Thus, with an increase in the level of the output and, accordingly, the input signal, the transmission coefficient of the UNU decreases. thus the signal is compressed. During playback, the input of the control channel receives the same signal as the input of the main channel, the polarity of the voltage that controls the UNA is reversed (to obtain expansion, not compression) and. finally, the frequency response of pre-distortion in the main channel is changed to mirror the one that was during recording. The frequency corrector in the main channel during recording is in front of the UNA and reduces the level of low-frequency signals by 12 dB (inflection points 370 and 1590 Hz). During playback, it is turned on after the UNU and restores the level of low-frequency signals. In the control channel, the signal passes through a second frequency equalizer, which raises the level of high-frequency signals by 20 dB (inflection points at 1600 Hz and 16 kHz). A phase splitter of the second order (Phase Splitter) is connected to the output of the frequency corrector. Two signals are taken from its outputs, the phase shift between which in the frequency range of 20 ... 200 will fluctuate about 90 ° (quadrature signals). Further, this pair of signals is fed to two quadratic rectifiers operating on a common smoothing capacitor. The smoothed voltage is used to control the gain of the UNA. The slope of the rectifiers is chosen so that the compression ratio during recording is 2:1. In other words, the output level changes by 5 dB when the input level changes by 10 dB. The purpose of using a phase splitter is to eliminate the main disadvantage of a broadband compander: due to the need for a fast response to high-frequency signals, the response time of the rectifier should be as short as possible (tens of microseconds). But then it turns out to be less than the period of the low-frequency signal itself, and, consequently, the low-frequency signal will modulate itself, which will lead to a harmonic coefficient of the order of 20 ... 40%. In order to avoid pulsations in the control signal, Blackmer took advantage of the fact that sinzx+cos2x=1. That is, when using two quadratic detectors and the phase shift of the input signals by 90 °, their output ripples cancel each other out. It should be noted that the rectifiers work with the logarithm of the absolute value of the input signal, since the UNU has an exponential regulation characteristic. In addition, the charging time constant of the integrating capacitor is made inversely proportional to the slew rate of the input signal. This results in good smoothing for slow changes in the input signal (high time constant), while for fast signal rises the rectifier responds faster (the gain "reset" rate can be as high as 90dB per millisecond!). The gain recovery rate at the loss of the input signal is 140 dB per second. This value is approximately one and a half times higher than the rate of recovery of ear sensitivity after the end of a strong signal, as a result of which the noise at the onset of a pause is attenuated faster than a person is able to hear it. Due to the use of RMS rectifiers, phase distortions in the transmission channel practically do not affect the operation of the compander in steady state. The assignment of frequency correction is non-trivial. The first frequency equalizer (in the main channel) is intended for a relative rise in high frequencies during recording (during playback, they are mirrored attenuated along with noise). In addition, the attenuation of low-frequency signals, on which most of the signal power is concentrated, allows you to partially "unload" the recording channel from them, thereby reducing distortion and modulation noise. It is curious that Dolby applied a similar correction ("Spectral-skewing") only fifteen years later, when developing Dolby-SR. The second frequency corrector (in the control canapes) performs three functions at once. First, it to some extent protects the control channel from inaudible low-frequency noise, which in its absence would cause chaotic signal modulation. Secondly, the phase shift in this corrector shifts the phase of the control voltage ripples in such a way that their fronts fall approximately at the moment when the useful signal passes through zero. Due to this, the effect of control voltage ripples is reduced at those frequencies where the phase splitter no longer provides quadrature (above 500...800 Hz). Finally, boosting the high frequencies in the control channel reduces the level of steady-state high-frequency signals at the output of the compressor (starting at about 5 kHz), which prevents tape and recording channels from being overloaded. This is how the classic dbx or dbx-l denoiser works. In addition to the structure described above, other companies under licenses also produced its variants, similar in characteristics. I must say that with all the elegance of this design, donkey ears of a technocratic approach to development stick out of it. The fact is that when working with constant or smoothly changing in level sinusoidal signals, everything was in perfect order, but the processing of pulsed signals was accompanied by large distortions in the processes of their rise and fall. This significantly changes the timbre of the sound of many instruments.6. Therefore, sound engineers who recorded classical and jazz music avoided the use of the dbx compander, especially when recording drums. In addition, level spikes during compressor operation (arising due to the delay in reducing the gain when the signal was growing), reaching up to 12 ... 18 dB, forced to reduce the average recording level by the same amount. As a result, the effectiveness of noise reduction decreased.7. In other words, the signal-to-noise ratio with a large signal turned out to be less than in the absence of a noise suppressor at the very 12 ... 18 dB. In professional reel to reel tape recorders, this went unnoticed. In cassettes, with a loud signal, you can hear the "breathing" of noise, while the sound is "muddy", while in a pause - deathly silence! So if the recording level on the tape is set equal to -15 ... -20 dB (so that the emissions pass undistorted), then the signal-to-noise ratio in the cassette recorder will not exceed 30 ... In a good reel-to-reel tape recorder operating with a high tape speed and wide tracks, the first of these figures can be obtained at a recording level of -40 ... -50 dB, but hardly in a conventional cassette recorder. Further, the use of phase splitters and a pair of quadratic rectifiers did make it possible to sharply reduce ripples when rectifying harmonic oscillations ("sine"), but turned out to be almost useless in detecting real signals. Accordingly, the intermodulation distortion of low frequencies during compression turned out to be fair (2 ... 10%). Another problem was caused by the fact that the frequency response of the control channel in the dbx system has a form that is far from mirror in relation to the spectral noise density of tape recorders. Therefore, when playing weak signals, the mutual correspondence between the operation of the compressor and the expander is violated. This is due to the fact that the control circuit is overly sensitive to the highest frequency (and low frequency) noise, which, if not audible, causes parasitic modulation of the signal due to detection in the control channel. As a result, the real noise reduction turns out to be less than the theoretical one, and in real conditions, in terms of pause noise, it is only 18...25 dB (if we take into account the margin for overload by emissions), and not 40...60 dB. By the way, parasitic modulation causes trouble in almost all noise suppressors, which is why a band pass filter is needed at the input of the noise suppressor, which attenuates signals with frequencies that go beyond the audio frequency band (especially from the RF side). To reduce spurious modulation of the signal, Blackmer later introduced a 10th-order low-pass filter with a steep rolloff and a cutoff frequency of 35 kHz (in addition to the 6 Hz high-pass filter to suppress low-frequency noise) into the control channel. In addition, the characteristics of the frequency corrector in the control channel have been changed. Its frequency response is semi-chip sloped at +440 dB per octave below 4,8 Hz and above 10 kHz (up to XNUMX kHz), with a flat section in between. The processing of impulse signals after refinement became even worse (due to the delay introduced by the filters)8, and the risk of tape overload at the highest (and lowest) frequencies has greatly increased, This version of the device was called dbx-ll. And finally, in the early eighties, a consumer version of dbx-ll was released, in which a conventional full-wave rectifier was used, the filter in the control channel was simplified and the phase splitter was eliminated9. It is this truncated version that is implemented in the well-known AN6291 chip. Despite the noted shortcomings, unpretentiousness and good noise suppression earned the dbx compander a good reputation in mid-level studios, especially after the release of a number of multi-channel tape recorders (Tascam, Otari, Fostex) with built-in dbx. (The competing system - Dolby-A was cumbersome to implement and therefore was always issued as a separate device, and besides, Dolby was in no hurry to sell licenses for its production). Nevertheless, it must be said that dbx, in an effort to overtake Dolby Laboratories, at one time sold licenses for their noise suppressors without restrictions. This led to the appearance on the market of versions simplified to inoperability (most often they saved on the input filter), and evil tongues joked that dbx is "Dolby for the poor". The main reason for the appearance of level spikes during operation and the appearance of dynamic errors was a subtle error in the construction of the control channel. The fact is that the phase splitter delays the signal at both of its outputs, in other words, the control signal inevitably lags in relation to the input signal. That is why, despite all the tricks with increasing the speed of the detector (variable response time constant), emissions were formed when rapidly increasing signals were applied10. A comparison with the High-Corn noise reduction system, proposed by Telefunken specialists in the mid-seventies, is appropriate here. High-Corn is in many ways similar to dbx: the compression ratio is the same (2:1), both systems are wideband, both use frequency equalization with high-frequency boost during recording and attenuation during playback. But there are also differences. Firstly, the law of compression in the High-Corn system is obtained in a different way, by connecting in series two identical controllable amplifiers (CLAs) with a common control. In this case, the operation of the compressor is based on the fact that if the signal level at the output of the second UNA is maintained constant by adjusting the gain of both simultaneously, then the signal at the output of the first UNA will be compressed in a ratio of 2:1. As already mentioned, when building a broadband compander, there is a problem associated with an increase in distortion at low frequencies due to the insufficient inertia of the detector. Therefore, the signal level detector in the High-Corn system is built in such a way that, after a very fast response, it has a certain "hold" time, during which the control voltage remains unchanged, and after it can quickly drop. As for the dynamic characteristics, due to the short response time (about 200 µs), the compression surges were small. Distortion at lower frequencies was significantly reduced by choosing the dwell time (25 ms) equal to half the period of the lowest frequency signals (20 Hz). Those are his good points. The bad news was that due to the relatively quick recovery of the compressor gain, audible "squishes" were sometimes generated after the dwell time. They became more frequent if the signal arriving at the expander had a noticeable parasitic amplitude modulation (more than 5...10%). For household tape recorders, such a PAM value is more a rule than a defect, and as a result, clicks followed one after another. Another drawback of the HighCorn system was that the frequency response of the detector, like edbx, turned out to be far from specular with respect to the noise spectrum of the playback channel. With the compressor and expander operating over the entire range of input signals (as in dbx), this would lead to a large parasitic signal modulation by noise. The developers of the High-Corn system solved this problem, as they say, "on the forehead": they refused to use a constant compression ratio (and expansion) at all signal levels, introducing a threshold below which the compressor did not work. As a result, there was a problem of matching levels, as in Dolby systems. Later, a two-lane version was developed by the joint efforts of specialists from Telefunken and Nakamichi, called High-Corn II. The crossover frequency was about 5 kHz. It didn't work much better and was soon forgotten. Soon the same fate befell the original version - High-Corn. This was probably due to the fact that due to the excessive rise in high frequencies during compression (up to 17 dB) and the lack of measures to reduce the level of the recorded signal at high frequencies, there were problems with overloading the tapes. In addition, noise pops annoyed during exposure after the passage of the fronts of the pulse signals. But back to the dbx compander. Unfortunately, Blackmer did not have time to figure out what was the reason for the large emissions and reduce them. As a result, the market for professional noise reduction products remained in the hands of Dolby.11. Therefore, dbx (already without Blackmer) attempted to introduce its system into household appliances. I must say that she succeeded in this: in the early to mid-eighties, most high-end cassette decks (Technics, Akai, Aiwa) "were armed" with one or another version of the dbx compander, and record manufacturers released a number of discs, the sound track on which was compressed with its help, dbx for records is distinguished by the lack of frequency correction in the main channel. Nevertheless, by our time, dbx has practically disappeared from household tape recorders. Probably, along with the shortcomings discussed above, the fact that a recording made with Dolby-B, with a certain blockage of high frequencies, plays tolerably without Dolby, but a recording compressed by the dbx system sounds terrible without decoding. In addition, the Dolby-B expander, unlike the dbx expander, can also play the role of a dynamic filter when playing noisy recordings. However, as the author's research has shown, the disadvantages of the dbx compander can be relatively easily minimized. The only drawback remains - the incompatibility of recordings with regular and compressed UWB Dolby. The advantages - good noise reduction, unpretentiousness, acceptable complexity and good repeatability - remain. The most important thing is that the degree of "damage to the sound", i.e., the visibility of distortions, in the version of the dbx-like compander developed by the author turned out to be less than that of any domestic Dolby, including Dolby-S, especially with an imperfectly tuned tape recorder. "Achilles' heel" of the prototype - emissions during compression - is practically "cured". To achieve this result, it was necessary to make four significant improvements to the original version of the compander (dbx-l). First of all, the phase splitter was replaced by a phase shifter, to the output of which one of the rectifier channels is connected (the other channel is connected bypassing the phase shifter). Secondly, the frequency responses of the pre-emphasis circuits in both the main and control channels have been changed to match the characteristics of the compact cassette format. Thirdly, in order to reduce the distortion of signal dynamics, weaken the influence of parasitic amplitude modulation and noise modulation ("breathing"), the compression ratio was reduced to 1,5:1 (as in the Telcom system). Fourthly, a forcing circuit was introduced into the detector, which accelerates its reaction with a sharp increase in high-frequency signals (such as a blow to a cymbal, metallophone, or triangle). Finally, the detector's time constant has been made composite to better match the properties of human hearing. These measures made it possible to practically eliminate both surges during operation and parasitic signal modulation. As a result, the subjectively perceived degree of noise reduction in comparison with the prototype has increased significantly despite the reduction in the compression ratio. This is especially noticeable when recording "live", unprocessed signals. The real dynamic range of a good cassette recorder reaches 85...90 dB, which is more than enough for most applications. Dynamic range measured according to a toughened technique as the ratio of the maximum signal with a frequency of 1000 Hz (at 1% distortion!) to the IEC-A weighted pause noise, in the author's tape recorder layout12 exceeded 90 dB using BASF Chrom Super tape at 4,76 cm/s. As for the overload capacity, the frequency response of the through channel at a signal level of +6 dB is uniform in the range from 20 Hz to 20 kHz (according to the criterion +0 ... -1,5 dB), and "0 dB" of the noise suppressor is reduced to a tape magnetization level of 185 nWb/m. Notes
Author: S. Ageev, Moscow
Machine for thinning flowers in gardens
02.05.2024 Advanced Infrared Microscope
02.05.2024 Air trap for insects
01.05.2024
▪ Created the world's strongest electromagnet ▪ TP-Link EAP610 Ceiling Access Point ▪ Biodiversity conservation with vanilla
▪ section of the site Normative documentation on labor protection. Article selection ▪ article of the past life the meanest traits. Popular expression ▪ article Why do tombstones exist? Detailed answer ▪ Astra article. Legends, cultivation, methods of application ▪ article Imitator of sounds of a drop. Encyclopedia of radio electronics and electrical engineering ▪ article Cuts friction. Encyclopedia of radio electronics and electrical engineering
Home page | Library | Articles | Website map | Site Reviews
www.diagram.com.ua |
See other articles Section 
Latest news of science and technology, new electronics:
All languages of this page