RTA: Friend or Foe?

Meet the car audio enthusiast’s greatest ally — the real time analyzer.

Red Tomato Affliction! Those words strike fear in the hearts of most seasoned tomato fight competitors. But not car audio competitors — tomatoes don’t scare us. (One thing for sure though, real car audio competitors don’t like raw tomatoes!) What scares many of us are Real Time Analyzers — red ones, blue ones, black ones, or otherwise.

For many years, it has been very common to see competitors spending the last few precious minutes before having their systems judged, tweaking the system for that elusive, perfect, flat RTA curve. It seems kind of funny to get so wrapped up in a supposedly “simple and objective” frequency response test of an audio system. But, when it’s entirely possible (at IASCA competitions) to win or lose a show because of a poor RTA score, you can be sure the serious competitors will sacrifice whatever it takes to ensure they get those RTA points. (BTW, the sacrifice usually includes some type of offering to the RTA gods!)

Why is RTA such a big deal? Is there a better way to do it? Do we really need to keep it around? My answers to those questions would be 1) I don’t know, 2) Yes, and 3) Yes! The RTA is one of the most useful and reasonably affordable test and measurement instruments available to anyone in the pursuit of high-quality audio — car, home, studio, or otherwise. However, it’s also one of the most (if not the most) misunderstood things we do in the judging lanes.

Why? As the owl in the Tootsie Pop commercial says, “Let’s find out.” In the first half of this two-part article (it’s going to be really long, so we had to split it up), we’ll cover the basics of how an RTA functions, some simple measurement examples, some of the basic capabilities and limitations of the RTA, and the (often disregarded) care and maintenance of an RTA. In the next issue, we’ll cover specific techniques on taking useful RTA measurements in a car, some data interpretation techniques, competition-specific techniques to help ensure maximum points on the score sheet, and we’ll show some examples of RTA curves from some of the top competitors’ sound systems on the circuit.

The How...
The basic operation of an RTA is this: It measures the average energy in octave-fractional bands across the audible spectrum from 20 Hz to 20 kHz. The bands can be divided in 1-octave, 1/2-octave, 1/3-octave, etc. Some RTAs allow 1/12-octave divisions. Of course, the one we in car audio are most familiar with is the 1/3-octave RTA that has been standard issue in the competition lanes.

When testing system frequency response, a pink noise signal is fed to the system and reproduced through the speakers. Pink noise is a pseudo-random noise signal that contains a constant amount of energy in each octave band, as opposed to white noise that has constant energy per Hertz. It correlates better with the human hearing system’s perception of loudness because it’s a constant percentage of bandwidth, and we tend to hear pitches on a logarithmic frequency scale. In pink noise, the actual instantaneous signal level in each frequency band is varied pseudo-randomly, but the average intensity over time is the same level in all bands. The varied level is the reason the individual band level jump around so much when the RTA response time is set to fast, but averages out to a steady state, consistent level when a longer time average is taken.

If the pink noise from the source unit is fed directly into the RTA, it should result in a flat response “curve” when averaged over time. Figure 1 is a photo of just such a measurement. The CD player output is fed directly into the RTA. This is the “reference” with which the RTA compares a system’s measured response when using a microphone. If the RTA and microphone are accurately calibrated, the resulting curve will be representative of the total energy in each fractional-octave band at the microphone tip location.

Figure 1

Figure 2

The total energy includes everything in that frequency band, regardless of whether it is generated by the system, an engine running near by, or otherwise. So, performing RTA measurements in a relatively quiet area where the system output is at least 10 dB above the ambient noise level is a necessity for solid, repeatable results.

Doing Some Basic Tests...
When actually measuring a system’s frequency response, there are a number of considerations to have in mind. We’ll use a simple system, consisting of a CD player, two internally powered near-field monitors, and an RTA. The system is set up in a large, open space so that the only significant hard, reflective surface anywhere near the system is the floor, and it’s covered with thick acoustical foam to minimize its effects on the measurements.

The first measurement is simply a reference for everything else we’ll cover. Pink noise is played through a single speaker. The microphone is placed one meter directly in front of and on axis with the speaker. Figure 2 shows the results of this set up using 1/3rd-octave band-width filters. Figure 3 shows the same test, but is the result of using Time Delay Spectrometry, a much more precision, detailed measurement system that yields results with much finer resolution than RTA. Note that in both figures, the response is, overall, very smooth with the typical bass roll-off below 50 Hz (it only has an 8-inch woofer) and above 16 kHz. The slight rise in the 100 to 160 Hz range is due to the reflection off the floor. Keep these “curves” in mind as we change the system set up and make more measurements.

Figure 3

Figure 4

The next measurement is made using the same set up, except that a single piece of glass (18-inches square) is held 12 inches from the microphone, creating a single reflective surface near the mic. (Sounds kind of like a car, huh?) In this case, the direct sound from the speaker arrives at the microphone first. The sound wave also continues towards the glass, and is reflected back to the mic, traveling an additional 2 feet. Figures 4 and 5 show the resulting measured response. Figure 4 uses a 1/6th-octave RTA measurement, while Figure 5 uses a 1/12th-octave RTA measurement. Both of these settings yield better resolution than a 1/3rd-octave response test, and, hence, allow us to see better what’s actually happening under these conditions. Note the significant dips in the response at 500 Hz and at several frequencies above that. Also note the general overall un-even response throughout the spectrum compared to Figure 2. This is the result of comb filtering.

Comb filtering occurs when two combining sound waves have different amplitudes, phases, and frequencies. In this case, the first wave is the primary arrival sound directly from the speaker to the mic. The second is the reflected wave from the speaker, to the glass, then to the mic. When the two waves meet at the mic, the actual signal level at each discrete frequency is the sum of the two waves’ amplitudes and phases at that frequency. If the waves are in phase, the result will be higher output (by 6 dB, theoretically). If the two waves are 180 degrees out of phase, the result is theoretically zero at that particular frequency. The dip in response at 500 Hz is the result of the direct and reflected waves being out of phase at the mic tip, thus creating a null in the response curve.

Figure 5

Figure 6

Figure 6 shows the same test with the TDS measurement system, set up to show very fine detail in the response. The resolution of the TDS system allows us to see the combs in much more detail than the RTA, simply because the width of the RTA filter bands does not allow it. (This in no way means that the RTA isn’t a good tool. It simply illustrates an example of one of the things we need to keep in mind when making an RTA measurement.) Note that the dip at 500 Hz is actually much deeper than displayed on the RTA curves, and there are actually many more deep dips from there up to 20 kHz. There’s also a large, narrow band peak at 750 Hz, not visible on the RTA response curve, resulting from the two waves being in-phase at that frequency.

This example illustrates one of the basic limitations of any measurement system. Early reflections, like that off the glass panel, wreak havoc on the overall measurement results. This is one of the fundamental problems we face when trying to gain useful measurement data in a car.

Care Is Needed
I would like to cover one more item before signing off though: basic RTA care and maintenance. An RTA, like any other audio measurement tool, is a sensitive device. The most delicate part of the system is the microphone. It must be handled very carefully. Don’t drop the microphone. Don’t tap it on any surface. Don’t blow directly into the tip. Don’t just throw it unprotected in the bag with the RTA. Any of these things can easily damage the diaphragm. Even with some of the more durable mics available, it doesn’t take much to degrade the frequency response. If you’ve ever seen two totally different RTA curves when simply changing between two different mics, then one is probably damaged. If you think a mic is damaged, just send it to the manufacturer and have it checked out. Trying to use an RTA with a damaged mic will yield very few, if any, good test results. Take care of the equipment. It’s easy, and worth the extra effort.

Until Next Time...
We’re about out of time for this issue, but, rest assured, we’ll go over more of the why’s, how’s, and what if’s in the next issue. If you want to get a head start, check out the CAR SOUND & PERFORMANCE 12-Volt Forums over the past few months, and maybe look into the reference books below.

The more we know about Real Time analysis, the better our test results and interpretation of those results will be. And, in the long run, the more we’ll know about tuning our systems to sound better. That’s what it’s all about!