Date: Wed, 03 Jun 1998 04:36:43 -0700
To: bass@mcfeeley.cc.utexas.edu
The Corrected Report and how it came to be.
The Cumulative Spectral Decay (waterfall) plots for the bare RD-75 driver, measured in free
air, and previously displayed in DBS Report #21 here after June 1998 have been removed as
being in gross error.
New data analysis has shown the RD-75 to have significantly better decay response than was
shown. Simply put, a microphone stand reflection in the impulse response data was allowed to
contaminate the CSD plot.
The recognition, selection and windowing-out of offending reflections in "semi-" or "quasi-anechoic" impulse data is partly subjective in nature. This subjectivity is based as much upon prior understanding of the sensitivity of the FFT technique to reflections in the data as to the interpretation of a computer screen display attempting to display a potential range of 8-24 bits of the amplitude data in often much less than a thousand pixel points of vertical display resolution.
Correcting public mistakes is important, especially when an amateur's errors might influence the
marketplace. This record of the mistake's discovery and the subsequent confirmation of its
correction from fellow audiophiles provides an instructive tutorial on the sensitivity of FFT
data processing to flawed data, and upon incorrect assumptions regarding data processing. The
next time you see a CSD plot with lengthly decay 'ridges', be sceptical.
There are at least three important lessons I have taken from this:
1. The RD-75 frequency response decay spectrum in nominal "anechoic" conditions is good.
2. The Culmulative Spectral Decay (CSD) FFT-method can be frightfully contaminated by reflections in the impulse data. Low-frequency plot response is exaggerated due to the DC offset introduced when succeeding time-sliced FFT computations start after the impulse beginning.
3. When one is defining "semi-anechoic" or "quasi-anechoic" in terms of a computer screen display of the impulse response, screen gains of 30 or more maybe needed to discern 'reflections' or other problems relating to the assumption of a semi-anechoic or anechoic condition of the data.
Below is the tale of events that led to the above understanding - disguised as a painful data processing tutorial combined with the growing comprehension that others are making the same error. The question is: What is the bare Bohlender-Graebener RD-75's measured frequency response decay profile under anechoic conditions?
On January 29, 2003 I sent a draft HTML e-mail of the materials below to the following members of the Dipole Forum: Leslie Williams, Rob Gold, and Doug Purl. Additionally, Alvin Foster (Boston Audio Society) and Tom Perazella (AudioXpress author) were sent a copy with a request for help and confirmation of the problem. Leslie, Alvin and Tom have provided CSD plots which will be included in the materials below.
On May 3, 1998 I made a number of free air measurements on one of my RD-75s with my good
friend John Nipps in the Gardena High School Auditorium. Rudi Blondia was assisting and
observing. On June 3,
1998 I sent a "Waterfall Plots - Free Air, Bare Driver Dipole Baffle Study Report #21" to the
Bass List internet forum and archived it and the plots at this webpage.
The RD-75 Dipole Baffle Study effort was clandestine. The third and final day of measurements
was a bit chaotic due to the knowledge that it would the last
access to the large-room venues at the high school. Too much was attempted that last day. As it
turned out, the auditorium was an inferior measuring environment compared to the much
larger gymnasium. After two more RD-75 Dipole Baffle Study posts to the Bass List my life
entered a new phase as a veteran activist at the university where I worked until July 2001.
Dipole Baffle Study Report#20.
Cumulative Spectral Decay Plots (waterfall) -
Free Air, Bare Driver, Page 2 - NEXT PAGE.
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 5.
Cumulative Spectral Decay Plots (waterfall) -
Free Air, Bare Driver, Page 6.
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.
