WINNING EDGE:MARCH/APRIL 2002

“On the planet Earth, if you jump up, you will come down is a universal experience. Knowledge about the law of gravity is not. One does not have to understand the law of gravity in order to come down after jumping up. One does need to know the law of gravity if he or she is to accurately predict how high one will jump and how fast one will come down. A measuring stick in the hand of one who is totally unaware that there is a law of gravity would result in few meaningful measurements about the effects of gravity because it would not occur to them what measurements to make or why to make them.”

Don and Carolyn Davis
Syn-Aud-Con

So the law of gravity is not one of our primary concerns in car audio. (Just be glad we have it!) But the quote brings forth a very appropriate analogy. Without an understanding of the basic principles of acoustics and the proper application of an audio analyzer like an RTA, any measurements we might take would likely be not very useful. So, armed with what we covered in the last article, let’s dive deeper into the meat and potatoes of how to use an RTA to help us dial in our high-performance mobile audio systems.

When making an RTA measurement in any space, but especially in a small space with many reflective and absorptive surfaces very near the measurement area, it’s very important to understand what the environment will do to the measurement results. And, in a car, it’s very important to use a few special techniques to minimize these environmental effects so that the best possible, most accurate, and most helpful test results can be achieved.

Get In The Car
Let’s take a look at some in-car measurements, all made with the mic approximately 26 inches above the seat, facing forward, and the TEF-20 RTA sample average time is set to long/slow. Figure 1 was taken according to IASCA RTA measurement standards, with the mic 26 inches above the center of the seat, and at the center of where the driver’s head position would be. Several observations can be made regarding the curve. As is common with most car audio systems, there are several significant “measured” variations in the frequency response. Note that the word measured is in quotations, and for a good reason. There can be a large difference between what is heard by the human ear to sound properly balanced, compared to what is measured using an RTA.

Some of the variations shown are common in many car audio systems. For example, in a very small space, the lowest frequencies must be amplified relative to the overall spectrum in order to sound properly balanced. There are also peaks at 160, 1k, and 12.5 kHz. Dips in response are also evident at 100, 400, and 5 kHz. These are due factors such as the physical dimensions of the vehicle interior, the different path lengths, the different materials in the interior, and those pesky reflective surfaces we talked about last time. Every system installed in every vehicle has its own set of variations in the measured frequency response. As system tuners, if we’re going to rely on the RTA to aid us in our task, we must be able to distinguish between those measured variations that represent problems we need to correct, and those that are simply the result of acoustical properties within the listening space that are not representative of what actually sounds properly balanced to our ears.

Move The Mic...
In a large room, it’s normal to expect that the measured frequency response to be relatively consistent across a wide range of positions within the listening area. The walls are far away, relative to the direct sound emanating from the speakers, and the speakers are generally far enough away to allow a coherent wave front to develop before reaching the microphone. In this type of situation, the microphone can be moved several inches, or maybe even several feet, without significant variations in the response curve. The assumption that an RTA measurement is representative of the overall listening area is likely valid in this situation.

But now get back in a vehicle. Trapped inside a small box with reflective and absorptive surfaces within a foot or less of our ears, tweeters here, mids there, and path lengths be damned, we can only expect the frequency response to vary with microphone position.

Figures 2 through 7 are measurements taken with the microphone in six different positions within the area where a listener’s head would normally be. The sequence of mic positions corresponding to the figures is as follows: 3 inches behind the left ear position, at the left ear position, 3 inches in front of the left ear, 3 inches in front of the right ear, at the right ear, and 3 inches behind the right ear.


Figure 2	Figure 3

Figure 4	Figure 5

Figure 6	Figure 7

The key to gaining information from these figures is not to look at the shape of the curves, but, rather, look at the differences between them. All of these curves were made within a few inches of each other in the area where a listener’s ears might be, yet they’re very different. We might ask why, and the answer is pretty simple. The plotted RTA curve is an indication of the frequency response only at the tip of the microphone. It’s NOT necessarily representative of the frequency response 6-feet, 1-foot, or even an inch from the microphone tip. The curve is only as good as the pea-sized area around the mic tip — that’s all! The assumption that an RTA curve is representative of the overall response in a larger space should be confirmed by moving the mic around the space.

Technique Only...
So, if we want to use the RTA to help us tweak our system’s frequency response, but we can’t rely on a single mic placement to do it, how do we go about it? Is it even possible? Absolutely! How? Take an average of multiple measurements in the space around where our head would normally be. An average frequency response within this area can reduce effects of the pinpoint response anomalies that are apparent in Figures 1 through 7. It’s really easy to do. And the best part is that it won’t cost any more to do it than a single location measurement.

To take an average RTA response measurement, you’ll need to sweep the mic tip around in a spherical area where the driver’s head would normally be so that the curve will represent an average frequency response in that volleyball-sized area. Sweeping the mic can be done by sitting either in the back seat or in the opposite-side seat from where the measurement is being taken. Figure 8 shows just such an average measurement. Note that it’s much smoother in overall shape than Figures 1 through 7. While it’s not “flat,” the narrow band peaks and dips are not nearly as pronounced.

Sweeping the mic greatly reduces the effects of position-specific measured frequency response anomalies, and allows the RTA measurement to be a much more effective aid in our system tuning.

The Analysis
So, now we’ve got a good average frequency response curve. How can it help us? Let’s take a look!

When working with any frequency response analyzer, it’s a good idea to start the equalization process by finding those areas of the measured response that respond linearly with equalization changes, and those that do not. To find them, start with the EQ settings flat, if possible. You’ll probably need two people to do this one. Have one person sweep the mic around the listener’s head area while the other makes EQ changes while watching the display. Take each EQ band, one at a time, and pull it down 6 dB below flat and check what happens to the display. Then push it up 6 dB above flat and watch what happens. If there’s a corresponding 6 dB drop, and a respective 6 dB rise, you’re in good shape for that band. If it doesn’t change by the same amount, then this is an area where the response is not minimum phase, and the RTA should not be relied on when making changes to this band. You’ll have to rely exclusively on your ears here. (For a definition of minimum phase and what causes it, check out the bibliography in the last issue. We simply don’t have enough room in this article to go into detail.)

Once you know what areas can be controlled while relying on the RTA, start your equalization process by first attenuating any large peaks. Pull them down only until they’re near the same level as the bands around them. Now, put in some music and compare the sound with that of your reference system. (You know, that high quality reference system we talked about several issues ago!) Now, with the EQ in your lap while you listen, take those bands and move them between flat and the setting you determined using the RTA as your guide. Compare what you hear to what you see, and use your ears as the final determinant.

Once you’re happy with these bands, go back to the RTA and keep working on any obvious problem areas. Be very careful when trying to bring up dips in the response as they’re often caused by cancellations and may not be a minimum phase area.

Dial It In!
Keep working back and forth between the RTA and your ears. After some time and practice, you should be able to make the car audio system sound a lot like your reference, at least as far as tonal balance goes. Don’t be surprised, however, if the final RTA curve doesn’t look flat. That’s O.K. Often, what’s heard to sound correct doesn’t always look flat on the display. For example, the curve in Figure 8 is the average RTA measured frequency response of the 4-Runner system in its sound quality mode. There are several variations in the measured RTA response, but the audible spectral balance of the system is extremely close to that of the Genelec 1031A Reference Monitors that are used as the reference for the system.

As a personal observation, I haven’t heard any car audio system that sounds exceptional and also measures truly flat with an RTA. They all have their own set of measured variations. However, I have heard many systems that measure very flat, yet sound just awful.

Always use your ears to confirm the changes you make using the RTA are correct. And, of course, always use your ears as the final determinant as to how the system EQs are set.