In article , Bob Latham
wrote:
In article , Dave Plowman (News)
wrote:
Just for info I've been having a play with a Behringer Ultracurve Pro
DEQ2496.

Basically a dual channel equaliser. It has a built in pink noise
generator and mic input, and using those will auto eq a speaker
system. Quite remarkable given the low price.

I've often thought about this kind of digital processing and my
understanding of what happens is obviously a bit limited as digital
processing AFTER the volume control looks a bit dodgy to me. I'm not
saying I'm right, I don't know but it worries me because....

Assuming you set the sensitivity of the DAC input such that the dac
*just* does not clip when your playing the loudest recording you have at
the loudest you'll ever play it, what happens to the quality of
quantisation when you play a quietly recorded piece with volume turned
down?

Have to give the usual engineering answer "It depends..." :-)

e.g. The 851C I use employs a 'digital' volume and balance control. However
it converts all inputs to 384k / 24bit. So for most of the digital source
material most people will have, a 'good' system of this kind would allow
the output quantisation noise level to remain defined by the source
material for a modest amount of level scaling downward.

I expect some more modern systems will duly use a mix of higher internal
rates and either more bits per sample or heavily noise-shaped 'DSD' sic.
But the argument comes out much the same from a theoretical POV.

But of course in *reality* if you wind down the level more and more, then
eventually the signal/noise will start to suffer. Just a question of when
that starts to happen.

Bear in mind, though, that the chances are that even 24bit audio files you
may obtain will be likely to have a background noise or quantisation level
that is rather higher than implied by 24 bit range. And such a range is
likely to be well beyond what you'd experience in most normal homes.

The real problems with digital 'EQ' for room and speaker are more
complicated. It depends on the details of what you need to 'correct'. The
most obvious example being a room or speaker response that has a narrow and
deep dip somewhere. That implies needing to boost any signals at the
frequency quite a lot. Which may then clip the DAC... or clip the following
amps, etc. Consider the implications of a 20dB 'dip' for example. You'd
have to wind down the general volume by 20dB to avoid the clipping problems
which could be caused by boosting some components by 20dB before the room
and speaker which had a corresponding 20dB drop.

And some room/speaker problems are almost unfixable by such means as
"measure the frequency response, then 'invert it' to get a flat result".

So its reasonable to expect such an EQ to give you some improvements. How
much, and at what point you encounter problems, depends entirely on the
details of the case.

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html