Lots of mystery and misunderstanding surround target curves and their use. Everyone who’s any good at tuning cars that sound great uses one, even if they don’t know it. Someone who tunes by ear and adjusts and readjusts until he gets to the point at which he’s ready to deliver the car uses one. Someone who uses a microphone and an analyzer uses one, even if they’ve never actually seen one or thought about it.
“What? You’re crazy. I’ve tuned hundreds of cars and I’ve never even seen a target curve!”
Of course you have, whether you know it or not. A target curve is a frequency response graph that depicts the balance between low frequencies, midrange frequencies and high frequencies. That frequency response graph can be measured for every audio system in existence. If you tune by ear and at some point, it just “sounds right”, then you’ve come close to your target, even if you’ve never measured the frequency response with an analyzer.
The reason to use an analyzer isn’t because your ears aren’t good enough (even though they may not be). The reason to use a microphone and an analyzer is to make the tuning process faster and more precise. Think of the target curve as a template. It’s the difference between trying to cut a round hole without a guide and without a template. You may get close – and then have to trim and file and sand and trim and file and sand.
No one would claim that a process that could provide a perfectly round hole in five minutes is somehow worse than a process that provide a perfectly round hole in three hours. So why do we think an 8-hour tuning process without a guide is somehow better than a faster one that provides the same performance? I don’t know. I don’t get it.
Sometimes people say, “Sound is subjective” in an effort to defend lengthy subjective processes. To a point, I agree. Some people say, “All cars are different “. To a point, I agree. There are differences, but if we’re going to improve our ability to do a good job predictably, we have to focus on the similarities rather than the differences.
Here’s an example. What if you had chest pain and went to your doctor? He checked you out and said, “Probably something wrong with your heart. Here’s a referral to see a cardiologist, Dr Farfegbauer.” You call. You make an appointment. Eventually, you see the specialist. He check you out. Then he says,
“Something is wrong with your heart. All bodies are different and I won’t know anything until I open you up. I don’t know how long this will take or what will be involved. Sign here. We can start tomorrow.”
Seriously? All bodies are different? My body is remarkably similar to the bodies of other 55-year-old dudes. Sure, it’s uglier than some and less ugly than others. Bigger than some and smaller than others. But my heart isn’t in my arm and my nose isn’t in the center of my back. Bodies are far more similar than they are different. The similarity is what makes the doctor’s expertise possible. Cars are far more similar than they are different, too.
Expertise is having a vast understanding of the similarities, and the ability to recognize the importance or the unimportance of the differences. Your surgeon is an expert because he knows about hearts and bodies. He knows what matters and what doesn’t matter.
So, a target curve is simply a recognition that cars and systems are similar – and that if we can adjust them so their frequency response measurement is similar, they’ll sound similar. Choosing a target curve for equalization is simply a matter of choosing the point at which we stop focusing on similarity and begin to finish the process by focusing on what’s dissimilar, if that’s even necessary.
Why is this important? Because when we focus on similarity, we can define an objective process that gets us the result we’re after predictably – just like a router template makes it possible for us to know how long it will take to cut a circle.
To speed this process and to make this process easier, it’s really helpful if you can overlay this target curve in the RTA application you’re using. I recommend REW because this is a simple procedure.
Below is the Audiofrog target curve loaded into REW’s RTA. Notice that the dB scale on the left shows 5dB for each major deviation (each solid line).
One of the most common mistakes I see when people use an analyzer into which they cannot load a target curve and when they have a printed curve on a piece of paper, is that the vertical scales don’t match. If that’s the case, matching the curve shape isn’t actually hitting the target.
Below is the same target curve with the vertical scale set at 10dB/division. Same curve, different shape. So, if your piece of paper has the curve displayed at 5dB/division and your RTA is set to some other value, your results will be way off.
Overlaying the target curve in the analysis program fixes this problem, because the target will scale, just like your measurement.
Here’s the target on top of a measurement:
Now, with the target curve displayed along with our measurement, it makes it really easy to determine at which frequencies to cut and boost and at which frequencies – and by how much. In the graph above, we can see easily that we need to cut much of the midbass and the midrange by 6dB—we can read that from the graph easily. We can even see the graph change as we adjust and we can keep adjusting until the curves match—just like a router template for audio.
If your analyzer won’t allow you to do this, then I suggest getting a better one.
Until next time,