Ja, ich weiß....nicht schon wieder sagen einige, aber diese Messungen von Zaph sind eindeutig:
.Die Metallresonanzen sind völlig irrelevant mit Weiche!
Was diejenigen widerlegt, die das Gegenteil hier behaupten, Namen nenne ich lieber keine
August 1, 2007
Usefulness of the CSD...
As I've said several times before, the cumulative spectrum decay (CSD) is a highly overrated form of measurement and is merely a different way of looking at the frequency response. It's all generated from the same impulse and it's all linear distortion. Yet, the CSD is often misread. To make a point, I'd like to present two CSD's done under the same conditions - one for a poly cone driver with a well damped breakup, and one for a metal cone driver having a harsh breakup 6.5kHz.
http://www.zaphaudio.com/temp/Polycone-CSD-nofilter.gif
http://www.zaphaudio.com/temp/Metalc...D-nofilter.gif
Which looks better to you? To the person who does not understand linear distortion, the metal cone driver looks horrible. The huge ridge of energy storage looks scary. To the person who does understand linear distortion, the metal cone driver is the slightly better one. This is because the metal driver is perfectly smooth within the operating range that it will be used, while the poly cone driver has a "shelf" of energy storage at about 1500Hz. The metal cone driver is operating closer to a pure piston below 2kHz.
Aside from harmonic distortion which is a form on non-linear distortion, it's all about how easy the driver's response curve is fixed in the crossover. Don't overestimate the difficulty of metal cones - this one is easily controlled with only 3 components. Actually, both response curves for these drivers are very good and easily controlled in the crossover. Below we have CSD plots of each driver with a filter in place, giving us well shaped LR4 rolloffs at around 2kHz. They are not exactly the same, but they are close enough to make a point. The metal cone's breakup has been dealt with, and the poly cone's shelf between 1 and 2kHz has been smoothed out. The response curves, and thus the CSD plots, now look very similar.
http://www.zaphaudio.com/temp/Polycone-CSD-LR4.gif
http://www.zaphaudio.com/temp/Metalcone-CSD-LR4.gif
Basically, all we have left are artifacts of LR4 and the window setting of 10ms. To understand why the LR4 filtered CSD plots look almost the same is to truly understand linear distortion. Judging the quality of both a frequency response curve and a CSD plot only comes down to how workable a driver is with a crossover. Assuming that's not an issue, forms of non-linear distortion such as harmonic or intermodulation become the most valuable gauge of a driver's performance.
I hope this helps solve some of the misunderstanding centered around the subject. I'm not sure if I can think of any other ways to further clarify this.
.Die Metallresonanzen sind völlig irrelevant mit Weiche!
Was diejenigen widerlegt, die das Gegenteil hier behaupten, Namen nenne ich lieber keine
August 1, 2007
Usefulness of the CSD...
As I've said several times before, the cumulative spectrum decay (CSD) is a highly overrated form of measurement and is merely a different way of looking at the frequency response. It's all generated from the same impulse and it's all linear distortion. Yet, the CSD is often misread. To make a point, I'd like to present two CSD's done under the same conditions - one for a poly cone driver with a well damped breakup, and one for a metal cone driver having a harsh breakup 6.5kHz.
http://www.zaphaudio.com/temp/Polycone-CSD-nofilter.gif
http://www.zaphaudio.com/temp/Metalc...D-nofilter.gif
Which looks better to you? To the person who does not understand linear distortion, the metal cone driver looks horrible. The huge ridge of energy storage looks scary. To the person who does understand linear distortion, the metal cone driver is the slightly better one. This is because the metal driver is perfectly smooth within the operating range that it will be used, while the poly cone driver has a "shelf" of energy storage at about 1500Hz. The metal cone driver is operating closer to a pure piston below 2kHz.
Aside from harmonic distortion which is a form on non-linear distortion, it's all about how easy the driver's response curve is fixed in the crossover. Don't overestimate the difficulty of metal cones - this one is easily controlled with only 3 components. Actually, both response curves for these drivers are very good and easily controlled in the crossover. Below we have CSD plots of each driver with a filter in place, giving us well shaped LR4 rolloffs at around 2kHz. They are not exactly the same, but they are close enough to make a point. The metal cone's breakup has been dealt with, and the poly cone's shelf between 1 and 2kHz has been smoothed out. The response curves, and thus the CSD plots, now look very similar.
http://www.zaphaudio.com/temp/Polycone-CSD-LR4.gif
http://www.zaphaudio.com/temp/Metalcone-CSD-LR4.gif
Basically, all we have left are artifacts of LR4 and the window setting of 10ms. To understand why the LR4 filtered CSD plots look almost the same is to truly understand linear distortion. Judging the quality of both a frequency response curve and a CSD plot only comes down to how workable a driver is with a crossover. Assuming that's not an issue, forms of non-linear distortion such as harmonic or intermodulation become the most valuable gauge of a driver's performance.
I hope this helps solve some of the misunderstanding centered around the subject. I'm not sure if I can think of any other ways to further clarify this.
Kommentar