Don Boomer wrote on Thu, 09 August 2007 17:17 |
I think you misunderstand how the power handling figures are derived.
First the tests use shaped noise ... not a single frequency ... that makes a huge difference.
|
No, I understand the test procedures, although many people do probably think that a X Watt rated speaker means you could apply X Watts of any signal to it. They also tend to look at the continuous rating, not realizing that the test signal used represents a 6dB crest factor and that many real world sources have higher crest factors.
Quote: |
Let me set the record straight about power testing.
The EIA (RS-326A) test that is used by JBL calls for shaped WHITE noise to be applied for 8 hours. White noise has a naturally occuring 3 dB/oct rise. First order filters (6 dB/oct) are applied at 40 Hz and 320 Hz. In this way the max power is delivered at about 320 Hz and is 13 dB down at 20 Hz and 12 dB down at 20 kHz. Here's the deal ... it doesn't stress a speaker as much as live music does in the extremes but it is longer.
The AES 2-1984 standard that we use here at Peavey uses PINK noise and runs for two hours. It is band limted at 20 and 20K with 2nd order Bessel filters.
The PINK noise (equal energy /octave) instead of WHITE noise (equal energy per frequency) means that the low frequency content is very strong by comparison. Also the highs aren't attenuated until a very high frequency the high content is also very strong.
|
I am having a little trouble understanding this. I believe that AES2-1984 (AES Recommended Practice Specification of Loudspeaker Components Used in Professional Audio and Sound Reinforcement) that you referenced is intended for rating loudspeaker components, not full range loudspeakers. The standard defines using pink noise bandwidth limited to one decade (from the LF point up one decade with the limits at the -3dB points and 12dB/octave Butterworth filters), a 2 hour period, a free air mounting and with the power calculated based on the measured RMS voltage and Zmin. It might be applied to driver sections, testing one 'section' from 100-1,000Hz and another from 1,000-10,000Hz would seem within the standard. However, the requirements of the standard, or at least the version I have, seem to limit the application to full range testing. Has the standard been revised recently or might it actually be a different standard or some derivative of the standard being applied? I am curious as a test with 20-20,000Hz bandwidth and Bessel filters appear to vary significantly from the standard.
I agree that the ANSI/EIA RS-426-A (EIA Loudspeaker Power Rating Full Range) and IEC 268-5 (International Standard - Sound System Equipment, Part 5: Loudspeakers) used by most manufacturers for reporting full range loudspeaker power ratings do use different shaped noise sources and longer periods along with several other differences in the procedures. But they are intended specifically for full range testing. It does seem that manufacturers who use the IEC or ANSI/EIA ratings for full range speakers do typically identify a passive full range rating equal to the rating of the LF driver, perhaps because they usually use the same test procedure for rating each section of the speaker as they do for rating the full range performance (which is different than the procedure used to test the drivers themselves).
Quote: |
The AES test is much more difficult at the extremes.
As far as the length of the test. In all honesty speakers reach an equalibrum after about an hour or so and while it sounds like 100 hours is more strenuous than two hours it hardly makes any difference in the real world.
|
I guess it may depend on whether you are looking at mechanical or thermal failure. The data I have seen from speaker manufacturers who publish test results for multiple procedures don't really support these statements. In fact, I've seen some products stating around a 10% reduction in the power ratings for IEC 100 hours versus IEC 8 hours.
Quote: |
Draw your own conclusions ... many experts have argued about it for a long time.
|
With the different spectra, bandwidth limiting, basis of calculated power, etc., between the different standard testing procedures, I'm not sure how much direct correlation can really be applied between devices using different rating procedures, especially comparing driver ratings to box ratings. Probably the one truth is that if the marketing department has their say, the test procedure used will be provides the biggest number as the result.