Speaker measurement - Does it represent reality

[oon_the_kid]

Hi,

I am starting a thread to discuss how speakers are measured and if the way they are measured truly represent how they will reproduce music. As a person in fullrange proponent, one could easily see that most multi ways speakers outclasses fullrangers by a mile. But when you put on some music, you must wonder why the fullrangers quite often sound better. There is of course phase accracy etc and losses through croosover etc. But most of these properties have not yet been defined. So I have put together some physical measurements that ought to be done on drivers to better represent the sound it reproduces.

1) My first bone of contention is the usage of sinewaves in the measurement of speakers. Although this has been argued quite often about discussed that music is not sine wave, one factor people forgot is in a forced oscillation system, the system would naturally want to result in a sine oscillation. For example, if you applied a square/triangle wave to a pendulum, the moment the force is stop and the pendulum will go back into a sine oscillation. So I feel measurement of distortion from a true sinewave is not very relevant, since they want to oscilate as a sine anyway. I feel a more accurate way would be to put a few sudden impulse/anomaly at various points of a sine waveform. Maybe by edititng a .wav file. Then measure the resultant waveform as picked up by the microphone and measured in an oscilloscope. In such scenarios, a speaker with a heavy cone would reproduce the sound very differently, since it has all the momentum of the cone. It would produce a sinewave without the impulse. A lgither cone as the one typically found in wide rangers would have less momentum and as such more likely to pick it up.

2) the measure of low level signal. A signal generator is used to power the amp/speaker. A microphone is positioned as close as possible to the speaker. Leve is reduced until the signal cannot be picked up anymore. One thing I noticed about high quality tweeters and low quality tweeters from boom box, is while their specs is the same, the major difference is high quality tweeters are more distict. In other words, you can hear the treble even if the sound is very soft. Whereas low quality tweeters only reproduce loud trebles, soft trebles are completely lost. I believe this might have to do with static resistance of the speaker, ie how much force you will need to push a speaker to overcome the friction before the speaker start to move. Of course for full rangers this number tends be much smaller, to ensure higher efficiency. But good tweeters should be even smaller than full rangers. I believe this is the main drive for people to push the crossover frequency as low as possible, because the tweeter produces low level sound better than midbass. If the static resistance is too high it wouldn't be able to reproduce small signal treble at all.

I would like to hear your thoughts/comments on this matter.

Thanks.

Oon

[Conrad Hoffman]

In the old days, you'd measure a speaker with a pink noise signal, and a 1/3 or 1/10 octave analyzer. Something like a GR system that connected the analyzer filter to a chart recorder and printed out a nice bar chart. Sine waves have never been popular or useful for speaker measurements because the resulting graph, if well resolved, is so full of system and room resonances, that one can't make much sense of it. Outdoors or in an anechoic chamber is better, but most speaker testing is done with noise, multiple tones or narrow band sweeping tones. With modern equipment, impulse tests can tell all, or close to it. IMO, if you're looking for resonances, sine waves are much easier to use. If you're looking for response, they're the last choice. Thus I'd say your sine wave complaint is a straw man argument because people mostly don't do it that way. It comes back to the equivalency of time domain and frequency domain measurements. They really are, but it's not intuitive and has led many astray. As for the pendulum argument, even though a speaker has a fundamental resonance, it had better have decent behavior over a wide frequency range- see the previous text.

The next part about friction should better be discussed as linearity. If there's some type of compression going on, and all speakers suffer from this to one degree or another, we need to measure it. How much non-linearity occurs at the lowest levels I've no idea, but usually that's where transducers perform best, not worst. IMO, the main difference between a good tweeter and a bad one is smoothness of response over a defined angle and a lack of diaphragm breakup. There's a danger in listening to speakers and theorizing why they sound the way they do. The answer will be wrong about 99% of the time. The only way to go about it is to combine measurements and listening sessions, then work out what might be going on. It's never fast as acoustic measurements have many facets and the results are rarely straightforward.