--- Jim Reid <jimr.reid@verizon.net> wrote:
> Another fact of heterodyne rigs is that the noise in
> the frequency
> range being converted to some other range, goes
> right along
> with the signals, hi! So noise is NOT being
> eliminated with
> heterodyning.
And this is why the cumulative contribution of noise
by a multitude of mixer stages is such a thorn, not
only do you have the original noise you are
incrementally adding more in each mixer stage. And as
you also pointed out IM products, .... a new set of
those for each mixer stage. Luckily with today?s solid
state devices the noise contribution by the amplifier
stages can be held to a minimum, mixers on the other
hand tend to be a bit more troublesome. Bottom line is
there is nothing you can to do to in the receiver
itself to ?improve? on the original S/N ratio of the
signal coming into the feedline. What you best hope to
achieve is to not degrade the precious S/N ratio while
filtering, amplifying and demodulating that signal to
make it useable for our ears (or data mode
processing).
>
> So, the question: what is "conversion"?
>
> Answer: It is another way to WAY complicate the
> receiver design,
> and must be VERY carefully planned and implemented;
> otherwise,
> more mischief will result than desired outcome! If
> the frequency
> plan is well done/implemented, and excellent rcvr
> results!
True indeed, unfortunately it is a necessary evil (aka
design trade-off) that we must live for now, and to at
least some degree for awhile yet.
>
> Final point: just because, in a DSP core rig the A
> to D converter,
> or the DSP processor itself is spec'd with more
> bits; that is
> 32 bit vs. 16 bit processor or 24 bit vs. 16 bit A/D
> converter,
> does NOT mean better performance. What results in
> better/
> worse performance is the accuracy to which the bits
> represent
> real sample values of the analog signal amplitude!!
> Now that
> IS a horse of an entirely different color and design
> problem!
> What determines when a bit "flips" from 1 to 0??
> That is the
> question, hi. The REAL accuracy of such is
> somewhere
> near to 16 to 18 bits.
Also very true, while 24 or 32 bit sampling makes the
job of dealing with the signals easier it isn?t in and
of itself ?better? than 16 bits. What more bits buy
you more than anything else is a wider AGC/dynamic
range that can be implemented/handled via software vs.
other analog means. However with the ever decreasing
price point of 24 bit A/D?s today there really is no
reason to not use them in any new design effort. The
more critical specification of the A/D as you point
out is its accuracy. Also another important parameter
is the A/D?s own S/N ratio. For us in radio land where
we frequently chase signals at very low S/N thresholds
we do care about that specification a lot!!.
Duane
N9DG
__________________________________________________
Do You Yahoo!?
Yahoo! Tax Center - online filing with TurboTax
http://taxes.yahoo.com/
|