low frequency alignment

[jlo]

hi all,
I often see discussions about the best low freq alignment of speakers. But has alignement (magnitude, Q, group delay, aso,...) been really audibly tested ?
Is there a real audible difference between a Q of 0,5, 0,707 or 0,9 ?
It is not so easy to change a Q just by a switch or a remote control so who really did it ?

Some points may add difficulties to listen to the real effect of alignment :
- in a normal room, there is a great chance that the speaker cutoff frequency is quite near from a room mode (less than 10Hz apart). Room modes behave like second-order biquads and generally have Q between 2 to 20 or more. So I suspect that you cannot really listen to the effect of the low Q of a speaker when a room eigen tone is so near.
- you cannot use two speakers each with its own alignment, because they won't be at the same physical place and also because one speaker will influence the other
- every recorded music has been processed through many high-pass filters. M. Leach pointed that, to have no influence on phase, those cutoffs should be 20 times lower than the lowest frequency (ie 2Hz for a clean response at 40Hz). So is it worth to ajust Q, phase, to lower group delay if when you do your listening tests, the source itself has allready 10 times more group-delay !

Does all this mean that the only way to test has to deal with following constraints :
- if you cannot use a room, you need to use headphones. But a headphone without its own LF problems ! Maybe some electrostatics have very low cutoff frequencies or an electromagnetic like an Etymotic ?
- if you cannot use a loudspeaker but you need to simulate a low frequency alignment, you need to do an auralization (a computer real-time simulation) of the effect
- if you cannot use recorded music, you have to find music that passed only through highpass filters of very low cutoff frequencies (you'll have to do the recording yourself) or use special test signals

That's becoming a bit complicated, but I suspect there is no simpler way.

[forr]

Hi JLO,
"it is not so easy to change a Q just by a switch or a remote control so who really did it ?"

Q of a driver can be manipulated, using either a transform, for loads having a 2nd order hi-pass behaviour (it can easily be done by a numeric processor), or an amplifier with positive or negative output resistance, for any kind of loads.

[jlo]

Q of a driver can be manipulated

You're right : I'm not saying you cannot do it, just said that it is not so easy. You can preprocess the incoming signal so to modify the efffective alignement. The typical Linkwitz transform is known for second order highpass but other preprocess could also be done for various alignements (ie 4th order BR)
So who really did controlled listenings with such a preprocessing ? Where are the results ?

[djk]

The KUBE from KEF had a variable Q control that worked quite well. Adjusting the effective Q of the speaker system over a range of about 0.5~1.0 changed the sound of a string bass from 'lean' to 'fat'.

http://www.marchandelec.com/wm8.html

[sreten]

Originally posted by jlo

But has alignement (magnitude, Q, group delay, aso,...) been really audibly tested ?

Hi,

The short answer is yes. Over a number of years you will hear
all sorts of alignments and their interactions with typical rooms.
You just then join the dots .....

/sreten.

[forr]

JLO,
Manipulating Q with a numeric processor is a very easy way.
Making an amplifier having an adjustable output resistance requires just a bit of electronic knowledge, it can be done even without opening its box using an awfull amount of three resistors.

http://www.diyaudio.com/forums/showt...793#post725793

Results : as you said, mainly dependant from the room and the frequency response. Colloms suggested some starting values for closed boxes in his book.

[jlo]

The short answer is yes

it changed the sound of a string bass from 'lean' to 'fat'.

please, is there a "long answer" with references to controled tests, (blinded, ABX,...) or AES papers,...? I found none yet. Thanks

[jlo]

Making an amplifier having an adjustable output resistance

I know that many products used some kind of feedback to adjust output resistance (ACE, Studer-Revox, PSI, Yamaha, Mackie,...) but I don't know if it is so easy to adjust some parameters (Ro) without modifying others (gain, ...) so I'm not sure that it is a simple solution for DIYer tests.

Manipulating Q with a numeric processor is a very easy way.

I think that I'll have to do it myself
I already did it just for the group delay (changing group delay without changing magnitude) but maybe I'll do it also with a variable preprocess to change alignement just with a mouse

[djk]

By definition, the difference between a Q of 0.5 and a Q of 1.0 is 6dB, you don't need to do a double blind test of that.

If you can't hear 6dB in the 30hz~40hz region, you are truely hearing impaired. At around 85dB the contour lines of equal loudness in the 30hz~40hz region are only about 3dB~4dB apart. They are about 10dB apart in the 2Khz~4Khz region. One contour line is subjectively twice as loud (after Fletcher and Munson).

[joje]

hi everyone!! im a little curious about bridging the output stage of the amplifier..but i was wondering if why only few uses this topology even though it is capable of quadrupling the power output...just want to ask what are the disadvantage using this configuration. please if anyone??/