Okay I finally got a moment to get back to this thread. And I want to get back to drivers and driver selection, and specifically woofers.
Many people start out looking and thinking I want a good woofer with low distortion. This kind of reminds me of back in the day when we looked at figures generated for advertising purposes as real measuring sticks. Distortion numbers can be as misleading as anything else ever mentioned. So I wanted to take a look at that aspect.
There are several problems with rated distortion numbers or figures.
1) Inconsistency of data and data taking methods.
2) Inability to get meaningful measured data.
3) Inability to distinguish between variations when listening.
Number one is simple. Measured data taken from one person is typically never taken the same way by the next person. There is no real standard for this stuff at all. By contrast we measure frequency response or SPL data based on a standard of 1 watt/1meter. So there is some level of comparison. This really doesn't exist with distortion measuring methods. So it is pretty useless to compare measured data for one source to another.
Number two is where it gets interesting. It has gotten to where anyone that buys a computer based measuring system thinks they can now measure distortion levels because the function is built into the software. The real truth is that the ability to get any useful measurements taken outside of an anechoic chamber is very difficult and spotty at best.
I think the Clio measuring system manual outlines the difficulties in getting any meaningful distortion numbers pretty clearly.
This is from the manual:
11.4 DISTORTION AND SETTINGS
Sinusoidal stimuli allow CLIO to evaluated distortion in its single harmonic form. If not
Set in Impedance Mode, CLIO always evaluates harmonics from second to fifth and
allows the display of each one separately via its own push buttons. While it is simple to
obtain meaningful distortion figures of electrical devices, measuring Loudspeaker
distortion in normal environments (without anechoic chamber) is not easy. We will only
give some advice here, relying on examples, as the topic is far beyond the scope of this
User Manual. To do this we will use CLIO’s FFT Menu in quite an advanced way. Distortion
evaluation is adversely affected by several parameters, two of which are the most
important.
Noise
50 dBSPL of ambience noise, a common figure, usually does not affect Amplitude
evaluation which is usually carried out at an average level of 90dBSPL. This is
particularly true using CLIO Sinusoidal Analysis capability which, by means of DSP
filtering, allows exceptional S/N Ratio. Unfortunately evaluating 1% distortion means
looking for signals that are 40dB lower than the 90dBSPL mentioned above, in the
same order of magnitude as environment noise.
Gating Effects
Device settling time, non perfect delay removal and reflections arriving within the
sampling time (Meter On) seriously affect distortion measurements, creating
artifacts.
Just to sum up the just of those two outlined problems. The noise floor is as loud or louder than the distortion levels trying to be measured. And room related reflections also add in measurement artifacts that disrupt the measurements.
There are two things that can be done to help overcome those two problems.
One is to raise the output levels. This creates a greater level of output verses room noise. Of coarse this raises distortion levels as well. The funny thing is that distortion levels can be different at different power levels. So what power level should be used? And room reflected effects get stronger as well. Again we get back to no reference for taking these measurements.
The second thing is to get the mic closer to the speaker. This also creates a greater difference between the output level and the noise floor. And it can easily overload the mic if the levels are not brought down to low levels. So you wind up with output levels that are fairly low. This gives no real world data either.
Furthermore, there can be some variation in the measurements because of fluctuations in the room noise levels. A refrigerator coming on in the next room over, a computer fan kicking in, or a heating or air conditioning unit coming on can greatly effect any measured responses at levels below 50db.
Then thirdly, is our ability to hear these claimed differences. So if someone says they measured woofer A and it had 1.2% total distortion and speaker B had 1.8% total distortion. Is one going to sound better than the other based on those numbers? What if one has a strong 3rd harmonic at 600Hz and the other doesn't. It has a strong 4th harmonic at 1.5kHz. But then there is a average.... There is little useful data.
Again these numbers are at or below the noise floor and almost impossible to tell if some even strong distortion problems are part of the music or not. Let alone small variations. And if we are listening at 80 to 85db levels then any measurable distortion will be in the 40db range and below our hearing threshold for any average room.
This reminds me of back in the late 70's and early 80's when most amplifiers or receivers pushed advertising the distortion numbers. Receiver A listed less than 1% distortion and receiver B listed only .1% distortion. Did it sound 10 times cleaner than receiver A? Then some companies started listing receivers with .01% distortion. And I remember seeing one listed at .005% distortion. I guess all you have to do is measure it at a small fraction of its rated power and somewhere you'll find a low number. The real truth was that there was many other things that determined how those receivers really sounded. And a distortion measurement taken at some unknown level really doesn't tell you much at all.
And the real truth is that there are many other factors that determine how a speaker sounds and a measured distortion figure really doesn't scratch the surface. In fact it really tells you nothing other than no major problems exist in that area. And that is only if the manufacturer can provide a good anechoic measurement.
Of all the measurements that can be taken involving acoustic output of any kind the distortion measurements tell you the least about how a driver will sound.
As soon as I have more time I will hit some things that do make a difference and talk about what to really look for in a driver.
