Originally appeared in the September/October 1990 issue of Car Stereo Review magazine.
Time and time again Im asked the same old question: "What can I add to my system to make it sound better?" At times, Im tempted to reply, "Earplugs." (Fortunately, Im usually able to restrain myself.) Seriously, dramatic improvements in sound quality can be achieved without adding anything – and by spending little or no cash. In fact, I recommend that you forget about bringing new components or accessories into the fold until youre sure each existing component is working to its maximum potential – both individually and as part of a team. And this advice applies to simple satellite/subwoofer systems as well as complex, bi- and triamplified packages.
How do you insure that each link in the audio chain is as strong as it can be? Assuming that youre using high-quality components and theyre properly linked together, there are two key areas to address: signal-to-noise ratio (S/N) and frequency response.
Expressed in dB, S/N compares the strength, or level, of the audio signal (music) to residual noise such as hiss. For example, an S/N of 80 dB means the audio signal is 80 dB above the systems noise level. The goal is to achieve the highest possible S/N and, as youll see later, this is a matter of carefully adjusting the input and output levels of each component in the system. In general, an S/N of 90 dB or higher is good, while anything above 100 dB is outstanding.
The goal in adjusting frequency response is to achieve a smooth response curve – one thats devoid of abrupt peaks and valleys. Note that the curve doesnt have to be ruler flat, but the transitions between peaks and dips should be gradual and flowing. "Flat" curves often sound bad because the response characteristics of our ears are far from flat, and most listeners like to boost bass or treble. Knocking your systems response into shape usually involves tweaking its crossover and/or equalizer, among other things, and these tasks can be accomplished by ear and/or with the help of sophisticated test equipment (more on that later).
To maximize S/N (and minimize noise), you must carefully adjust the input sensitivity of each signal processor and amplifier in your system.
For a better understanding of S/N, consider the hypothetical systems in Figures 1A and 1B, each of which comprises a head unit (such as a CD player), an equalizer, and an amplifier. In Figure 1A the input sensitivity of the EQ is set too low, and its output voltage is a lowly 0.1 volt at full volume. As a result, the input-sensitivity control on the amplifier must be cranked way up to produce usable volume from the speakers. Lets assume that a 16-foot cable runs between the dash-mounted EQ and the trunk-mounted amplifier (the longer the cable, the more potential for noise). Given this typical scenario, its fair to assume that 0.0001 volt of noise will be induced into the signal cable by a power source like the alternator, the cars electrical wires, or perhaps the air conditioners fan motor. To determine the S/N of this system, we can work through the following textbook formula (LOG stands for "logarithm"):
S/N = 20 x LOG (Signal Vrms / Noise Vrms)
S/N = 20 x LOG (0.1 / 0.0001) S/N = 20 x LOG (1,000) S/N = 20 x 3 S/N = 60 dB
Just for the record, a system with an S/N of 60 dB is virtually unlistenable.
In Figure 1B, the input sensitivities of the systems components are properly adjusted, resulting in a healthy EQ output of 4 volts at full volume. This means that the input-sensitivity controls on the amp can be kept to a minimum. Assuming, again, that 0.0001 volt of noise has been induced into the signal cable, we can substitute 4 volts for "signal" in the above equation to find the improved S/N:
S/N = 20 x LOG (Signal Vrms / Noise Vrms)
S/N = 20 x LOG (4 / 0.0001) S/N = 20 x LOG (40,000) S/N = 20 x 4.6 S/N = 92 dB
As you can see, simply turning up the EQs input sensitivity while turning down the amps input sensitivity produces a 50-percent gain in S/N. The noise level at 92 dB is practically non-audible.
Maintaining a reasonable amount of dynamic headroom is a secondary objective in the level-adjusting process. Like S/N, dynamic headroom is also expressed in dB, but it describes a systems ability to handle very loud transients (sudden crescendos) without clipping (see Figure 2A) – the more headroom, the less likely the system will clip. Clipping is a harsh, raspy form of distortion that occurs when an output circuit is overloaded or its input is overdriven – its called clipping because the peaks of each waveform are flattened, or "clipped" (see Figure 2B). So while you want to maximize the signal level as it passes from one component to another, you must also make sure that its level does not exceed the electrical capabilities of the signal processor or amplifier its passing through.
The first step in adjusting system level is to review each components owners manual to make sure you fully understand the units nuances. Then set the bass, treble, balance, and fader controls on the head unit to the "flat" position and its volume control to its minimum position. If a preamplifier is in the chain, set its input-sensitivity controls for minimum gain, which is accomplished by rotating the trim pots (usually on the rear or bottom of the unit) counterclockwise; in addition, set the preamps volume control to minimum. If an equalizer is being used, set its input-sensitivity controls for minimum gain. Then set each of its frequency controls to the maximum "boost" position – doing so allows room for adjusting the EQ later. (If the level of each frequency control is left in its "flat" position, clipping will occur whenever the listener boosts one or more frequency band.) If an active crossover is being used, set its input-sensitivity controls to minimum; if the component has output-level trim pots, set them for three-quarters volume. Finally, set the input-sensitivity controls on the amplifier(s) for minimum gain.
Once these preliminary adjustments have been made, turn the system on and play a tape or disc that youre familiar with. Then set the volume control on the head unit for three-quarters volume. Dont worry if you dont hear the music at this time – all this means is that the minimum-gain settings previously made are too low to drive the speakers. To bring system output up to an audible level, turn up the input- and output-level controls on each of the remaining signal processors and amplifiers to half volume; you should now have a moderate output level. If you have a preamplifier, turn its volume control to three-quarters volume. If the unit has clipping indicators (usually in the form of red LEDs), simply turn the input-sensitivity controls up until the LEDs begin to wink on and off; when this occurs, the preamp has reached its maximum output and any further increase in signal level will cause clipping.
If there are no clipping indicators, youll have to set the input levels by ear. Slowly turn the input-sensitivity controls on the preamp clockwise. As you do this, decrease the input sensitivity of the signal processor or amplifier thats next in the signal path; the idea is to keep system volume at a moderate level. Continue this "back and forth" action until the sound becomes distorted or fuzzy. When this happens, you have driven the preamp into clipping; slowly back its input-sensitivity controls down until the sound is clean.
Follow the above procedure to set input levels for the equalizer and active crossover (if your system has these components and they have input-sensitivity controls). The final step in setting system signal levels is to adjust the input sensitivity of each amplifier. Proceed by slowly increasing the input controls until the speakers powered by that amp begin to sound fuzzy. Then slowly nudge back the controls until the distortion disappears. If a signal processor in the chain doesnt have input-sensitivity controls, dont worry – simply skip it and proceed to the next device. Chances are that the component sports a "unity-gain design," which means its output level is approximately the same as its input level.
As Ive mentioned, the primary goal of adjusting your systems frequency response is to achieve natural sound, which is a function of a smooth frequency-response curve. To smooth out your curve, youll need a real-time spectrum analyzer, or RTA – theyre available from such manufacturers as Alpine, Audio Control, and Cetec Ivie; prices range from $700 on up to $7,000. If you dont own or have access to an RTA, most car stereo shops will analyze your systems response for a nominal fee. The RTA measures sound output across the audible spectrum, and it can provide a graphic representation of your systems frequency response, allowing you or your installer to visually pinpoint the source of sonic problems so they can be corrected. The RTA also measures sound-pressure level (SPL), which tells you how loud your system can play.
To use an RTA, first set the head units bass, treble, balance, and fader controls to their flat positions and turn the units volume off. Then set the preamps tone controls to flat and its volume control for three-quarters volume. Next, set the equalizers frequency controls to flat or "0 dB." (Be careful not to disturb the input-sensitivity adjustments made earlier.) Now youre ready to place the RTAs microphone inside the vehicle. Mic placement will greatly affect the frequency-response measurement. Ideally, the mic should be placed at ear level on the drivers side of the vehicle, which will insure measurements that closely approximate what the driver hears; take all measurements from this same location.
Once the mic is in place, set the RTAs controls:
Decay: This switch controls how fast the RTA s display responds. Setting it to Slow will make it easier to analyze your systems response characteristics.
Detector: When given the choice of RMS or PEAK, set this switch to RMS, which will give you a more accurate indication of the systems output.
Weighting: Set this one to OFF or FLAT so that the RTA does not factor in the nonlinearities of the human ear; this is the only way to get a true indication of your systems frequency response.
Reference: Set this switch to "90 dB" so you can view the systems entire output through the RTA s display window (while listening at a moderate volume).
Resolution: Set this one to "3 dB," which will cause the RTA to display response variations in 3-dB increments. Once the response of the system begins taking shape, you can reset this switch to the finer "1-dB" setting and tweak the response.
To obtain an accurate reading of frequency response, you must use program material that contains an equal quantity of sound at each of the measured frequencies. Fortunately, the special test signal called pink noise meets these requirements. Pink noise sounds like the hiss you hear in between channels on your radio, and test tapes or CDs containing pink-noise tracks are pretty common. (The test disc produced by the dB Drag Racing can be obtained for $15 from the organization.
The first step in real-time analysis is to adjust system output. Turn on the system and play the pink-noise track on the test disc or tape. Slowly increase the head units volume control while monitoring SPL on the RTA s display. Increase the volume until you get an SPL reading of about 90 dB. Once this level is reached, the frequency display should indicate output across the audio band. Then shut all of the doors to the vehicle and roll up the windows. If possible, have a friend sit in the drivers seat while you take the response measurement. This is important since the presence of a body has a substantial effect on frequency response. After measuring the response of the system for 15 or 20 seconds, press the STORE button on the RTA, which will save the response curve in memory. If the unit is equipped with a printer, print out a hard copy of the response curve. If it isnt, you can transcribe the response data from the units display onto a piece of graph paper.
Shaping System Response
There are several common frequency-response anomalies youre likely to encounter. The first occurs in multi-amp systems whose output is improperly balanced. This is one of the easiest problems to diagnose and correct. Figure 3 shows the frequency response of a biamplified system comprising a head unit, a two-way active crossover set at 200 Hz, and two amplifiers – one for bass and one for high frequencies. Notice how the frequencies above 200 Hz are offset from the frequencies below 200 Hz – that is, the overall output level is significantly lower. This means that one of the following conditions exist: the woofers are more efficient (and louder) than the tweeters; the tweeters output is obstructed; the high-frequency amplifier has a lower output level than the low-frequency amp; or some combination of the above.
The first step in correcting this problem is to verify that nothing is obstructing the tweeters output. Then perform another spectrum analysis and either increase the input sensitivity on the high-frequency amplifier or decrease the input sensitivity on the low-frequency amp (preferred) until the "shelf" separating the low and high frequencies disappears.
Crossover misalignment is another common source of installer headaches. A hump or dip in the frequency response near the crossover point indicates a problem. Humps are usually caused when the high- and low-pass crossover points overlap. Figure 4 depicts the response of a two-way system with a 300-Hz low-pass crossover point and a 100-Hz high-pass crossover point. Since the crossover points overlap from 100 to 300 Hz, both amplifiers simultaneously produce output in this frequency band, resulting in a response hump. There are two ways to alleviate this problem: Reset the high- and low-pass crossover points to the same frequency, or simply narrow the overlap region until you get a smoother response curve.
Dips are typically the result of a gap between the high- and low-pass cross- over points. Figure 5 depicts the response of the same two-way system, but this time the low-pass point is 100 Hz, the high-pass point 300 Hz. Since neither the high- or low-frequency amplifiers are producing output between 100 and 300 Hz, response droops in that region. Again, you can either reset the low- and high-pass crossover points to the same frequency or narrow the width of the gap between them until the response curve smoothes out.
Phase errors are one of the most challenging and frustrating problems to overcome during the response tweaking process. While the majority of these problems occur in the mid-to-high-frequency range, errors at low frequencies are most noticeable and result in thin-sounding bass. Phase errors are either electrical or acoustical in nature. The electrical variety can be corrected by simply reversing the polarity on one of the drivers (switching the positive and negative leads on one speaker). Acoustical phase problems are more difficult to alleviate because they typically occur when the listener is not centered between the systems speakers. The result is that some frequencies are multiplied while others are canceled; this can result in an overly prominent set of frequencies, which will make the music sound unnatural. Known as the "comb-filter effect," this is one of the toughest problems to correct.
Figure 6 depicts the response of a system plagued by the comb-filter effect. Notice the sharp dip at 100 Hz. This is the frequency at which the output from one speaker is exactly 180 degrees out of phase with that of the other driver. Also note that additional dips occur at the exact harmonics of 100 Hz (200 Hz, 400 Hz, and so on).
Fixing this problem is more a matter of experimentation than anything else. First, try reversing the polarity of one of the offending speakers to make sure the problem isnt electrical in nature. If the problem still exists, try relocating one of the suspect speakers. Another option is to insert a passive highpass crossover with a cutoff frequency slightly below the frequency where the largest dip occurs – 100 Hz in the case of the system shown in Figure 6. The idea is that the phase shift caused by the crossover will fix the problem.
Setting your equalizer is the final step in adjusting the response. Remember, your EQ is not a crutch. It should be used only to fine tune response – not to compensate for gross response deficiencies. Figure 7 shows a system that is bothered by minor frequency-response glitches. To improve such a curve, you should monitor system output on the RTA and adjust the equalizers boost/cut controls until the curve becomes smooth.
Hint: Split the difference in large band-to-band transitions. For example, when you have a 10-dB difference between 125 and 250 Hz, increase one band by 5 dB and decrease the other by 5 dB instead of simply adjusting one control by 10 dB. This will allow room for further EQ adjustments in this frequency range.
Tremendous improvements in sound quality are possible without spending a dime – provided you take the time to identify response irregularities and correct them. Patience is an important virtue here. So remember to take a close look at your systems response curve before you reach for your wallet. If youre lucky, the only thing youll have to buy is a pile of new CDs or tapes.