"The EggKing" wrote
I'm not eactly sure how to word this but I must as it has bugged me for some
time now:

If an encoder has a cutoff frequency (eg. 15KHz in NICAM), to prevent
aliasing or any other problems, is the 'shape' of he wave preserved within
the data produced? I mean to say; The higher frequencies that most of us
have no chance of hearing as pure sin waves (15K+) do shape the waveform as
a whole. If the higher frequency cannot be sampled is therefore the shape of
the wave altered in encoding?

You need to read up on signal theory.

You seem to be wondering whether, if you have a periodic signal that
repeats at 15kHz (or a bit less), but whose shape is not sinusoidal,
e.g. square, sawtooth, etc., then the reconstituted signal (after
going through a perfect A-D and then D-A process), will be identical
to the original signal.

The answer is NO, because the fact that a signal is periodic with a
frequency of 15kHz does NOT mean that the maximum frequency "present"
in it is 15kHz. If the 15kHz signal is not sinusoidal, it WILL have
higher components, and in the case of waveforms with sharp edges,
then MUCH higher frequency components will be present. These will ALL
be lost in the A-D, D-A process. In the ideal case, the BEST you
can do, mathematically, would be to end up with a 15kHz sinusoid.

In the real world, the output from the D-A converter running at 30kHz
(remember Nyquist) will not be a 15kHz sinusoid, but will be a 15kHz
square wave. However this does NOT mean you're getting some higher-
frequency information "for free". The output ought to be filtered to
remove ALL components above 15kHz, and when that is done, you will
find that you ARE left with a 15kHz sinusoid again. The stuff that was
removed in the filtering was junk, garbage, effectively noise. It does
you no good, it's unwanted, and you should get rid of it.

All of the above assumes infinite sampling resolution (in the other axis,
the one that isn't time (usually voltage in reality)). In reality,
A-D converters do not have infinite resolution. CD has 16 bits of data,
corresponding to 65536 distinct levels. This limitation makes things
even worse, by adding what can be regarded as another noise signal
to the sampled signal. It is predictable, but it's still like noise.

If you haven't understood this, ok, I may not have worded it ideally,
and it's hard without pictures, but you will need to study some
signal theory / Fourier / sampling.


Martin
--
M.A.Poyser Tel.: 07967 110890
Manchester, U.K. http://www.fleetie.demon.co.uk

"Fleetie" wrote in message
...

In the real world, the output from the D-A converter running at 30kHz
(remember Nyquist) will not be a 15kHz sinusoid, but will be a 15kHz
square wave.

Absolutely and totally wrong. The output of any proper DA is a smooth wave
with no square waves related to the sampling frequency.

However this does NOT mean you're getting some higher-
frequency information "for free". The output ought to be filtered to
remove ALL components above 15kHz, and when that is done, you will
find that you ARE left with a 15kHz sinusoid again.

A DAC that lacks such filtering is not a proper DAC, which is why its wrong
to say that a DAC has a square wave output.

"Arny Krueger" wrote
"Fleetie" wrote in message
...

In the real world, the output from the D-A converter running at 30kHz
(remember Nyquist) will not be a 15kHz sinusoid, but will be a 15kHz
square wave.

Absolutely and totally wrong. The output of any proper DA is a smooth wave
with no square waves related to the sampling frequency.

ARE left with a 15kHz sinusoid again.

A DAC that lacks such filtering is not a proper DAC, which is why its wrong
to say that a DAC has a square wave output.

I disagree, but it's semantics.

Depends what YOU CHOOSE TO CALL A DAC. I'm talking about it
stopping at the digital part that switches the resistors/
voltage divider/whatever (not over-familiar with the innards
of DACs).

I suggest you review your zeal to "look cool" by disagreeing
with people as forcibly as you can, in favour of an attitude
more closely allied to the facts and theory being discussed.

If (as may well be the case, and as makes perfect sense)
real-world "DAC" chips in fact do incorporate such filtering,
then well and good. It removes the need for the user to add
this filtering themselves.

BUT STILL, WHAT I WROTE ABOUT THE SQUARE WAVEFORM STILL APPLIES
SOMEWHERE INSIDE THE DAC, I WOULD IMAGINE, SO MY EARLIER STATEMENTS
ABOUT THE ADDED HIGHER-FREQUENCY "NOISE" NEEDING TO BE REMOVED
ARE CORRECT.

Oh, and I'll have fries with that, cheers Arny.

--
M.A.Poyser Tel.: 07967 110890
Manchester, U.K. http://www.fleetie.demon.co.uk

In article , Fleetie
wrote:


In the real world, the output from the D-A converter running at 30kHz
(remember Nyquist) will not be a 15kHz sinusoid, but will be a 15kHz
square wave.

This assumes the DAC has a "sample and hold" method for asserting the
output in between the sampled instants. Probably a good
assumption/approximation in many cases, but the actual waveshape will
depend upon the DAC.

As you indicated, though, the final output will depend upon the filtering
employed to help reconstruct the waveform.

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

Jim Lesurf wrote:
In article , Fleetie
wrote:



In the real world, the output from the D-A converter running at 30kHz
(remember Nyquist) will not be a 15kHz sinusoid, but will be a 15kHz
square wave.


This assumes the DAC has a "sample and hold" method for asserting the
output in between the sampled instants. Probably a good
assumption/approximation in many cases, but the actual waveshape will
depend upon the DAC.

As you indicated, though, the final output will depend upon the filtering
employed to help reconstruct the waveform.


A DAC without a reconstruction filter is as daft as an ADC without an
anti-aliasing filter.

Ian