To Doug Schneider,
I read a lot of very positive speaker reviews, yet the test measurements of some of them point to high peaks or plateaus of distortion.
A rule of thumb from my experience is that for really clean-sounding speakers the distortion should measure at least 45dB less than the clean signal. Yet I see some speakers with distortion as high as 25dB less than the signal at critical frequencies like 2kHz that have glowing review commentary and no mention of the distortion.
Why is this?
I agree that the distortion measurements we do are valuable; in fact, other than a couple publications in Germany, I don’t know any other publication in the world that supplies this information. What’s more, I know that our distortion measurements have raised the eyebrows of many designers whose products we’ve measured, which, in turn, has had a profound impact on future products that have come from them. All in all, it has been a good thing.
That said, the distortion issue is not as it seems. While on the surface it seems obvious that you should push distortion as low as possible, it is not clear how low is low enough, and at what frequencies. For example, it’s well accepted that higher levels of distortion are tolerable, even undetectable, at some frequencies but not others. The problem is, it has not yet been determined as to what frequencies and at what levels. Many feel that the character of the distortion also plays a role. For example, second-harmonic distortion is considered benign, and sometimes even pleasant. Work still needs to be done. For a bit of information on that, I recommend reading an article Sean Olive wrote in 2009 called “What Loudspeaker Specifications Are Relevant to Sound Quality?” where he stated the following:
The relationship between perception and measurement of nonlinear distortions is less well understood and needs further research. Popular specifications like Total Harmonic Distortion (THD) and Intermodulation Distortion (IMD) do not accurately reflect the distortion’s audibility and effect on the perceived sound quality of the loudspeaker.
He wrote that over six years ago, but I’m pretty sure that not much has yet changed. What I also know is important is that that designers have to balance lowering distortion with other factors. For instance, I was talking to a driver designer who said a small amount of distortion was noticeable in a certain tweeter they were producing, but only in measurements, not in listening tests. He said that they found the source of the distortion, eliminated it, but the tweeter ended up not sounding as good. Most likely, their fix for distortion introduced another problem.
I also want to address your point about distortion being 45dB below the main signal. A distortion level 40dB below is 1%. A level of 45dB below is 0.56%, which is pretty ambitious for a loudspeaker, since most consider 1% clipping for an amplifier. More important, what’s missing is at what sound pressure level you’re talking about. I know of many loudspeakers that have distortion components 45dB below the main signal across critical frequencies when you measure them at a sound pressure level of 80dB from a distance of 1 meter away; however, when you measure speakers at 90dB at a distance of 2 meters away, which is what we do when we test, things turn much different, since that requires the speaker to play far louder. . . . Doug Schneider