ERROR REPORT - 21 March 2003.
FINAL CORRECTED CSD plot of Bare RD-75 driver suspended in Free Air with Windowed Impulse plot from which it was computed.
ASSERTION: This CSD data is essentially "anechoic" given the short duration and the free air nature of the data.
RD-75 3m free air suspended CSD plot, at 1.66mS duration.
The 3m bare driver free air suspended CSD plot at left was also part of the e-mail I sent to Mr. Williams on January 24, 2003. Note that is is based upon the windowing of the 3m impulse to a very short duration of only 1.66mS. I had failed to include in that e-mail a corresponding plot showing the impulse windowed to 1.66mS. In addition I made comment about the regression sequence of windowing the 3m free air impulse to find the point where reflections were not an issue, as finding out how many angels could dance upon the head of a pin.
Mr. Williams rightfully wrote back that day:
"Whoa! How did it get so clean? I think I've lost the
thread as to what you did...We need to know so that we don't dance on pins either."
Mr. Williams' question, "How did it get so clean?", led me to author a very large HTML e-mail to a number of audiophile individuals in order to both clarify the situation and hopefully gain a concensus on the issue.
RD-75 3m free air suspended impulse at a Screen Gain of x5, and 1.66mS duration.
At left is the x5 Screen Gain display of the 3m free air impulse response windowed to 1.66mS. To my horror, 1.66mS of the impulse response was all that was left after the obvious reflections were windowed out of the magnified impulse data.
What can obviously be seen is that the RD-75 bare driver has a respectable CSD response,
with no untoward 'ridges' appearing - nor in particular, the suspicious 'cavity resonance'
decay ridge shown when the data was windowed to 2.59mS as shown on Page 3 of this report.
The conclusions are fairly clear: Garbage in - Garbage out! Furthermore, the FFT-based Cumulative Spectral Decay method is a rather fragile method, subject to both apparent misuse and misunderstanding by the unwary. Computer screen display of large dynamic range impulse data results in deceiving 'flat lines' indicating no acoustic reflections in the data. Yet when the screen display of the impulse data is magnified a much truer understanding of the quality of the data is seen, often revealing acoustic reflections which are shown to be the source of CSD 'ridges' and 'hash.'
Clearly 'ridges' in CSD plots can and do arise from acoustic reflections - that much has been demonstrated. Q.E.D. What of reflections from dipole baffle edges? Are such the source of "greyness" heard by the golden earred ones? Early reflections are considered by many to be the source of much muddiness and reduced clarity in physically large loudspeaker systems. This correction has been a painful revelation - and as we see, not without a social component as the motivation.
Following are the contributions of Mr. Alvin Foster, Boston Audio Society, and Mr. Tom Perazella, AudioXpress author, to the dialogue that this report represents.
Dipole Baffle Study Report#20.
Cumulative Spectral Decay Plots (waterfall) -
Free Air, Bare Driver, Page 1.
Cumulative Spectral Decay Plots (waterfall) -
Free Air, Bare Driver, Page 2.
Cumulative Spectral Decay Plots (waterfall) -
Free Air, Bare Driver, Page 3.
Cumulative Spectral Decay Plots (waterfall) -
Free Air, Bare Driver, Page 4 .
Cumulative Spectral Decay Plots (waterfall) -
Free Air, Bare Driver, Page 6 - NEXT PAGE.
Cumulative Spectral Decay Plots (waterfall) -
Free Air, Bare Driver, Page 7.
Dipole Baffle Study Report#22.
The RD75 Dipole Baffle Study - Table
of Contents
Acoustic Line Source Research -
Table of Contents.
