 
 
 
 
|
|
|
|
|
| Page 1 | Page 2 | Page 3 | Page 4 | Page 5 | Page 6 |
Issues in loudspeaker design - 6X - Spatial distortion in audio Y - Triggered burst measurements of tweeters -----------------------------------------------------------------------------------------------------------------------
X - Spatial distortion in audioSpatial distortion is a concept that needs to be considered in the context of accurate reproduction and recording of sound, since all sound exists in some form of space and carries therefore 3-dimensional properties. While some recording engineers seem to be aware of the issue, the majority does not. I guess most listeners have only a vague notion of the spatial dimension in sound and what is meant by distortion. Besides we have lived with these distortions for so long that we accept them as normal and often do not notice them, except when listening with attention to live, unamplified sounds. Furthermore, the influence of spatial distortion upon what we hear when listening to loudspeakers or earphones is rarely discussed in the technical or audiophile literature. Spatial distortion is obvious for earphones when music or voice are perceived between the ears and inside the head. Even when listening to a binaural recording that was done with microphones in a dummy head it is difficult to localize the orchestra in front and at some distance outside the head. Sources that were to the side, behind or above are more readily externalized, but are perceived closer than they were in reality. If head tracking is used during playback, then even a mono source is externalized. It stays at the same absolute location in space for a full 360 degree rotation of the listener. During head tracking the two ear signals are continuously changed to account for the expected change in ear signals between the direction of view and the virtual source at some distance in space [1]. Looking at the virtual source and then turning off the head tracking does not immediately pull the source back into the head. It takes several small left-right turns of the head before the virtual source slides smoothly from its perceived distance outside the head to its normal placement between the ears without head tracking. This experiment illustrates that the brain uses head rotation generated cues to place the source at some distance and azimuth by learning and uses memory to keep the source at that location even when it no longer receives confirming cues. It can stay there for several seconds before new cues take over. Frontal externalization is also possible without head tracking but it requires great precision in the measurement of the HRTF, which can show significant differences between individuals. Griesinger reports about the performance of a system where he records the sound pressure variations close to his own ear drums using probe microphones [2]. Here each microphone capsule is at one end of a soft, flexible tube and the other end is inserted into the ear canal and carefully placed close to the ear drum. The recording is then played back over earphones, which are equalized to produce the same sound pressure at his ear drum as was there during recording. In this system the perceived sound scene upon playback is externalized as it was during recording, but it turns with the head. With head tracking the scene remains stationary in space as would be natural. With earphone based systems the perceived distance to the source is usually too close or in the head. With loudspeaker based systems there is a minimum objective and perceived distance, which is the distance to the nearest loudspeaker. There is a correspondence to visual perception. The minimum objective distance from a viewer to an actor on a movie screen is the distance between viewer and screen. Perceptually, though, this distance is reduced when the scene zooms to a close-up of the actor. The effect is enhanced when the timbre and volume of the actor's voice change correspondingly. Yet we know that we are at a safe distance. This is not the case when watching a 3-D movie like Avatar or Up. Here the actor or an object can be in front of the screen and scarily close. Stereo sound systems rarely generate such closeness of virtual sources. Highly directional loudspeakers in low reflection environments can create the illusion of a center soloist that is in front of the loudspeakers. Usually, though, sound is perceived as coming from the plane between the two loudspeakers and from behind it. Even surround systems leave an acoustically empty space between the listener and the ring of loudspeakers. A recording of applause from the midst of an audience will not reproduce the hand claps - that were close to the microphones - as close to your ears. Instead you are in the midst of an artificial circle where all the hand claps occur on the outside of the circle. This is a form of spatial distortion. Ambisonics and Wave Field Synthesis hold the promise of much greater realism. Spatial distortion is introduced in the production of a recording by the placement and types of microphones used. Distortion can be reduced to some degree in the following mixing and processing stage. Spatial distortion is added during playback depending on the loudspeaker radiation pattern, loudspeaker and listener placement, and the reflective characteristics of the room. Spatial distortion in an audio system is perceived in comparison to one's experience and memory of the characteristics of real acoustic events. When, for example, trying to judge the spatial distortion of a playback system by listening to the recording of a symphony orchestra, then it is extremely helpful to have heard symphony orchestras live and from a variety of seats in the hall or in different halls in order to have memory of the different perceived sonic characteristics. A recording engineer should be in a better position to make spatial distortion judgments, because he has immediate access to the live event. The monitor loudspeakers used and their placement in the room limit what he can hear, if he is even interested in providing an accurate spatial rendering. More likely he is directed by the producer, the performers or by his own ideas as to what the recording should sound like in order to be successful in the market. He becomes artistic and creates a sonic painting that elicits a different reaction from the listener than a documentary recording of the same auditory scene. Hearing relies upon the interpretation of the continuously changing streams of air pressure variation at the two ear drums. Body and head movement add further cues. Together they lead to the perception of an auditory scene (AS), which may contain many different sound sources at different angles and distances in a unique space. That space is defined and bounded by the reflections that are generated by the different sources. Space is perceived to the extend that the reflections are heard. Of course with open eyes we can see the boundaries and particularities of space. Visual cues tend to dominate the perceived angular position and distance of sound sources. The AS that is perceived at an indoors concert has its own characteristic elements and many of those characteristics are different from an outdoor concert, as is readily heard [3]. Concert halls have been studied for a long time to understand which acoustic measurements correlate strongly with the AS preferences of listeners and performers in order to build venues that provide as pleasurable AS from a maximum number of seats. It would seem that any audio recording, transmission and playback system should be judged first of all on its ability to recreate an AS from venue A by playing back a corresponding recording inside venue B, which is a much smaller space. The resulting spatial distortion is in addition to the well known linear and nonlinear distortion contributions that are likely to occur in the process. The audio system must not necessarily transmit precisely the acoustic wave field from a point in venue A to a region in venue B in order to create a plausible AS. Our ear-brain hearing apparatus works efficiently with auditory cues, memory and learning. It fills in missing pieces as long as misleading cues are avoided. Thus it is absolutely important to understand, measure and minimize the generation of misleading cues along the path from A to B. It begins with the placement and types of microphones used for the recording and ends with the loudspeaker radiation pattern, loudspeaker and listener placement, and the reflective characteristics of the listening room. A large contributor to Auditory Scene distortion is the loudspeaker. You can test your sound system's accuracy by how readily it reveals artificiality in recordings. See Accurate performance of a stereo sound system. Linear and nonlinear distortion in audio systems can be readily measured and expressed in numbers, though the degree to which these distortions are perceived is often not proportional to the numbers. Spatial distortion of the AS can be perceived and described, but a method to measure the distortion and to express it in numbers has yet to be developed. I assume that spatial distortion has been a minor concern of recording engineers, because it is difficult to hear through the typical monitoring systems and because pan-potted stereo is such a convenient and popular recording method. Listeners to classical music expect to hear not only the direct sound from the instruments of the orchestra but also to have a sense of the venue's response. Consequently in classical music recording some attention is paid to rendering sources and space for a plausible AS. References:
-----------------------------------------------------------------
Y - Triggered burst measurements of tweetersARTA version 1.6.1 provides a well functioning external trigger mode that allows the use of a powerful external signal generator for safe high level testing of drivers. Such capability is essential for me in order to select drivers according to their distortion behavior and by how gracefully they degrade under overload. I like to use shaped tonebursts for such tests. I increase the amplitude of a single burst until I hear the onset of distortion. A short string of bursts tells me a lot about intermodulation distortion when I analyze its spectrum. Observation of the microphone output signal tells me about the time response of the driver to different amounts of applied power. I think that I can now
retire an old PC, which still runs Windows 3.1, DOS and the MLSSA card, that has
served me so well during the last 19 years. DRA Laboratories MLSSA gave
me trustworthy results and had unique features, but also lacked some features,
which made it desirable and also difficult to find a modern replacement. With
ARTA I have access to almost all the tools that I wanted for audio measurements
and analysis. Ivo Mateljan's User Manual is an excellent course in signal
analysis.
Here now is an example 1.5 kHz toneburst test signal, a single burst to show the waveform details, the 100 ms string of bursts that is applied to the tweeters, and the spectrum of the 100 ms burst. The spectrum shows distortion of the 33120A output signal, but for these tests they are low enough and of no concern.
Two tweeters are tested to illustrate the triggered burst measurement. The tests should serve as examples for what can be learned when large signals are applied safely. They are not complete tests. The first driver is a Seas 25TFFC (H-519) soft dome tweeter from a long time ago and the second is a BG Neo3W planar tweeter. The driver terminal voltage is set to 30 Vpp. The signal shape is the same as that of a 100% amplitude-modulated sinewave, where the carrier is at 1.5 kHz and the modulation frequency is 150 Hz. For 100% AM we can derive Vrms = 0.22 Vpp. Thus the 30 Vpp amplitude corresponds to 6.5 Vrms, which would deliver 5.3 W into 8 ohm. A 60 W amplifier, as used for the ORION tweeters, could apply 60 Vpp max to the driver terminals. I have certainly measured 30 Vpp during loud passages.
The microphone output trace appears clean to the eye and identical to the source signal.
A slight increase in the output envelope can be observed over the 100 ms time duration. This effect becomes more pronounced as the drive signal level is increased. It could be due to Ferro fluid in the magnet gap becoming less viscous as heat increases, which decreases the mechanical damping of the driver.
The driver shows primarily 3rd order distortion at -40 dB (1%). Note also the difference distortion products at low frequencies. They are not directly audible, but indicative of the nonlinear parameter at play.
Here we see considerable waveform distortion in the form of sharp peaks and the signal envelope not going to zero between the bursts. The 100% modulation has been reduced to a lower value. The driver has lost articulation. The effect is independent of applied signal level and indicates resonant behavior that is not electrically damped by the amplifier..
The microphone signal amplitude decreases over the 100 ms duration of the burst. It is probably due to the driver heating up and leads to a 2.3 dB signal amplitude change within this short time when about 5 W were applied. The tweeter shrieks audibly during the test.
The peak output at 1.5 kHz is about 5 dB higher than for the 25TFFC. So for a comparison at the same SPL either the Neo3W drive signal would have to be decreased to 17 Vpp, which will reduce distortion, or the drive level of the 25TFFC will have to be increased to 53 Vpp. In addition, for dipole application the back cavity of the Neo3W has to be removed, which will increase distortion. Partial cancellation of the front acoustic output by the rear acoustic output will require higher drive levels to maintain on-axis SPL. The lack of a rear chamber is likely to cause additional electro-mechanical problems. The large amount of distortion at relatively low power level would rule out this driver for my applications, which usually require a low crossover frequency.
Note that this is just an illustration of the triggered test method and not a complete evaluation of either driver. Both drivers have their applications, but probably not for the ORION+. The tests clearly show thermal effects. The signal voltage and the burst duration could be safely increased further, to show behavior under extreme conditions and eventually failure modes. These are important tests of the robustness and reliability of a product. Driver manufacturers give some indication of this by their power specifications, which I assume are for safe operation. Their low frequency power limitation is for cone excursion reasons, like distortion and damage. 25TFFC Neo3W For a 100% AM burst with identical power dissipation as a continuous sinewave signal, we have Vpp (AM burst) = 4.6 Vrms (sine) -----------------------------------------------------------------------------
| Page 1 | Page 2 | Page 3 | Page 4 | Page 5 | Page 6 |
|
|
