Imaging Tutorial: By Eric J. Holdaway & Rob Greene
When listening to a pair of stereo speakers, do you hear the singer in the center and the band spread out, in the positions that they were in when the recording was done? What is a Stereo Imaging?
Imaging is what is called a psycho-acoustic effect. It is the ability of your sound system to fool your brain into recreating the original placements (either acoustical or conceptual) of the singers and instruments that are being played back from a recording.
The recording format that we deal in is called stereo. Stereo means that all of the information (music, vocals, ambience, special effects, etc.) from the performance is being recorded on a 2 channel (left and right) recording format. And as we at Speaker Works are fond of saying to people when they are asking about center (3rd) channels, IT'S STEREO NOT TREREO.
Imaging from a stereo playback system is experienced by the listener due to their perception of the sound characteristics of frequency, phase (relative arrival times of the sound wave), and energy (amplitude/volume/loudness).
Now, because our recording format is a 2 channel format, the ideal reproduction system would be 2, single speakers that have identical and perfect sound reproduction capabilities. They might be placed with the same spacing as the microphones that recorded the performance and would not be affected by their surroundings (environment). The listener would be in front of and facing the speakers, so that the speakers are equally distant from the listener.
I can hear all you people saying, "Yeah right. There is no such thing." You're right. About the closest thing to the "ideal" that I have outlined is a set of headphones. Music played though headphones is not acoustically affected by the surroundings, because the signal plays directly into the ear canal, which is a very small volume. It cannot add any late or early reflection, and will not affect phase or energy output of the headphones, because the headphones are equally distant from each ear drum. Headphones essentially recreate only what is already in the recording. (For the sake of this discussion we will not go into all the aural effects of the ear. It is enough to say that the brain's built in equalizer compensates for these "effects" automatically).
I feel that the car offers many of the same advantages as headphones. Because the listening area in the auto is small, it does not add late reflections (echo). With careful selection of speaker location and aiming you can increase the amount of direct radiated sound and minimize the undesirable early reflections. The idea is to minimize the impact that the environment has on the sound reproduction. Remember that headphones do not interact with the listening environment. Headphones only interact with the "room" inside your ear. Headphones maximize the ratio of direct to reflected sound. Please, at this point, do not confuse these added room reflections with the reflections that are part of the recording and hence part of the "music" we want to accurately reproduce. I am referring to environmental (room) added time aberrations, sometimes called the "Room Signature."
End of theory. Now for practice. How do you create imaging in the car? Can speaker location promote imaging and if so, what are the correct locations? What other factors contribute to imaging?
The common wisdom on speaker placement to encourage imaging is to place the midrange wherever it would fit in the doors, and to put the tweeters as high on the door or dash as you can get them. This is thought to be the way to give the sound height. The thinking goes that the higher the tweeters are mounted, the higher the image will be, and vice versa. This is not necessarily true! In fact, the physical locations of the speaker have little to do with the listeners' perception of sonic height. Also, by separating the tweeters from the midranges, the sound quality will be degraded. Speaker Works had an old "rule of thumb" that we use. It states, "The closer together the midrange and tweeter are, the farther apart they will sound." In other words, if the tweeter and midrange are together, you are more likely to achieve spatial effects like imaging, sound staging, and ambience. On the other hand, if you separate the tweeter and midrange, you are most likely to "achieve" easily detectable speaker locations with no imaging (point source localization).
So we find that the tweeter and midrange speakers need to be very close to one another to sound good and to help achieve your goal of imaging. Further, we need to concern ourselves with the arrival time of the sound from our tweeter and midrange pairs. The easiest way to manipulate the arrival time of the sound from the t/m, or any speakers, is by changing the position in a manner which affects distance (more to come on this).
There are many positions where the t/m could be mounted. Each position affects the arrival time of the sound from the t/m. (The scope of the following discussion is limited to the relationship of speaker position to arrival time only, and not to get into the speakers' off-axis energy output, which is important, but as you will see, takes care of itself). This effect on arrival times is caused by the distance that the sound travels from the left or right speaker array to the listener's ears. This distance is called the Acoustic Signal Pathlength. We will have a right pathlength and a left pathlength, and the difference between these two distances is called the offset. All of the speakers in your system will have a signal pathlength and offsets.
Here is the key to creating imaging in any stereo system: Make the offset in the pathlengths from the left and right T/M (tweeter & midrange pairs) to the ears as small as possible.
If you mount the T/M high in the door of the vehicle, you will have a pathlength difference of up to 3 feet between the left and right speakers and your ears. Assuming that you are in the driver side seat, the speaker(s) that is in the driver side door (nearest to you) will have a pathlength of about 2'3", and the speakers that are in the passenger side door (far side) will have a pathlength of about 4'7" to 5'7" (depending on the size of the vehicle). At this point, the near side speaker will become the predominant imaging cue that you will hear. This is because the closest speaker is heard first and is also usually loudest (Due to being closer to your ears). You will not hear a centered image without turning up the level of the "far" speaker (or by turning down the level of the "near" speaker) to compensate for the difference in arrival time. This compensation is done with the use of the balance control which creates some unique problems all its own. The sonic results will only be better for the driver side and will be a disaster for the passenger side. And the "Balance Method" can only correct for small arrival time differences. With the testing that we have done at Speaker Works, we have found that you can compensate with the balance control for a maximum pathlength difference of somewhat less than 2 feet. In our example system we have a pathlength difference (offset) of at least 2'4". You will not be able to achieve any semblance of imaging with the speaker(s) in the high door locations. So trying to fix an arrival time problem by adding a large sound level imbalance problem won't lead to good imaging.
The pathlength difference improves to around 1'8" by moving the tweeter and midrange to the bottom of the door. (I can hear, "The tweeters to the bottom of the door?!? Heresy, heresy! The sound will be coming from the bottom of the door!" - Remember, the height of the image does not necessarily follow the physical location of the speakers). With this offset we can start to hear an imaging improvement. We are able to get some real center image, and by using a small amount of balance control correction, we are able to improve that even more for the driver side or passenger side only.
We are seeing that by installing the t/m farther from the listener, we are achieving an equalization in the Pathlength offset. Our first example had an offset of 2'4" and our second example got the offset down to 1'8". The reduction in Pathlength offset was accomplished by increasing the overall Pathlength of both t/m. The near side speaker array now has a Pathlength of around 3'5" and the far side array has an overall Pathlength of about 4'11".
So the next logical question is, how do we increase our overall Pathlength farther to help improve the imaging? At Speaker Works, we have had great improvement in imaging and overall sound quality by moving the t/m array into the kickpanels of the vehicle. The kickpanel is the area directly in front of the doors, next to your feet. (I hear the rumbling out there that that cannot work. "Tweeters on the floor that's crazy!". I've been hearing that for the last 13 years that I have been doing this). Speaker Works first tried the kickpanel location way back in late 1979, at the insistence of one of our customers by the name of Rob Greene. It turned out that he is an Acoustic Engineer and is now a long time friend and collaborator. When we first tried it, we installed the speaker flat onto the kickpanels. The improvement in imaging was immediate. The pathlength offset is now down to around 12 inches!!! and we need very small balance control correction to achieve a good center image.
Pathlengths
Now we have very equal arrival times to the listener from the t/m. But, (and the message ALWAYS follows the "but"), our perceived arrival energy is not equal because of the extreme off-axis of the near side speakers. To correct this problem, we needed to bring the speakers back on axis by aiming them back at the listener. Remember, the reason we are now going to aim them back toward the listener is to equalize the perceived arrival energy of the speakers.
Pathlengths 2
We have found that by building the kickpanels out with the USD Kick fiberglass Kickpanel enclosures that we get from USD Audio, (retail was $180.00 a pair), and aiming the t/m array in the general direction of the ceiling dome light of most cars, you can achieve at worst, a 30 degrees off-axis on both the left and right speakers for the driver and passenger. (In most cases we can get that down to 20 degrees). Once this is done, your system will have advanced into a new world of sonic holography.
In most cars, by mounting the T/M down in the kickpanel and aiming them toward the listeners, you can achieve pathlength offsets between 9 to 12 inches (keeping in mind that you want to minimize the pathlength offset and equalize the off-axis offset). The sound arrival times and sound arrival energy from the left and right t/m will be almost equal.
The latest evolution in the quest for improved imaging has lead to the introduction of an all new, unconventional class of speaker that is being applied to car audio. This new speaker once again increases the overall pathlength and there by shortens the offset. They were featured in AS&S in Nov. '92. They are call the USD WaveGuide
. The WaveGuides
take a unique approach in mounting location, speaker design, and construction.
Like the kickpanel, the WaveGuide
is mounted to face the listeners, equalize the energy arrival. Here's were they differ, instead of being mounted on the kickpanel, they are mounted along the bottom of the dash, above your feet and legs. And you want to talk about L O N G pathlengths, the speaker extends all the way back to the firewall on most cars. With the WaveGuide
, we are seeing offsets of only 5 to 8 inches.
The vastly improved offset is only the start of the reasons that these new speakers will out perform conventional t/m. The WaveGuide
consists of one speaker per channel. It is able to accurately reproduce a very wide bandwidth because of it's compression driver design. By being a single speaker per channel it is what we call a single point source. When multiple speakers are in use, we are relying on mutual coupling, (that's why you need to keep conventional tweeters and midranges close together), proper innerchannel polarity between the t/m, and correct phase at the x-over point to help in the creation of imaging (As well as pathlength. The rules have not changed). Being that the WaveGuides
are only one speaker per channel, polarity between the separate t/m and the phase at the t/m x-over point is eliminated and the system is simplified.
Imaging is created by a stereo system, when the listener's perception of the sound characteristics of frequency, phase, and energy meet certain criteria. If you take to heart the lessons within this article, you will enjoy improved imaging on an existing system and you will be able to start your future system on the right foot.
Now, all you "Techno-factoids" that want to nit-pick all the little points of this article, please consider the exclusions "rounding error." This is a broad overview of this subject, and if I can spend more time in front my word processor, we will continue farther into this and other interrelated areas of sound reproduction.
