[LAD] Getting the most out of onboard HDA

[Jens M Andreasen]

On Fri, 2008-09-12 at 12:49 -0700, Justin Smith wrote:
> I am not sure if I understand what you aare saying herer at all.
> 
Go on?

> What I imagined was this: instead of sending one floating point
> output, you send two, out of phase with one nother, except ot perfetly
> out of phase, 

Yes, as in: L = mono, R=mono * -1; // this is the baseline

> so tht you retin the potential of the full bit depth (if
> each signal were just the inverse of the other, you are wasting half
> of the bits you send). Now that I think about it more, this woul be
> most useful if the full signal chain used this format,

I disagree unless you have other anolog signals in mind.

>  .. and to really
> double your bit depth sending -29 on one side and +1 on the other
> would have to be different from sending -30 on one side, and 0 on the
> other. I think I was just confused, and maybe you had something in
> mind other than my erronious speculation.

With a 24bit signal of +/- 15 integer values, we would be drowned in
noise. But since this is foremost a theoretical discussion, let us stick
to those numbers.

Let us say that our programs current examination of the incoming float
leaves us with the impression that i'15' is the the nearest intger
approxination, then we could also naively multiply by -1 and send both
signals - that is to say L == 15 and R == -15 - down the balanced line.
This should give us the advantage of reduced noice from external
sources, since we at the receiving end will do the inverse, that is to
say that for the signal we care about we would dol: NewMono = L + (R *
-1) which will come out just the way planned and expected. The noise we
encountered on the balanced line affects both wires equally and are
hence eliminated since - following the equetion above - it will collopse
into: x-x == 0


Whatever ...


What I had in mind regarding the +6dB notion was that a signal can have
an (integer) amplitude of 0.499.. which we would normally flush to zero
because that is all we'we got, but if you have two of them - one on each
side of the sending line - then we could average out your signal and
pretend that we have a higher (actual) resolution than speified.

/j