What's the use of a 192 kHz sample rate?

On May 5, 3:04 am, nos...@nospam.com (Don Pearce) wrote:
> On Sun, 4 May 2008 19:54:57 -0700 (PDT), rickman <gnu...@gmail.com>
> wrote:
>
> >I am not an RF guy and the figure I actually remember was 150 dB.  I
> >hedged it a bit as 150 sounded rather extreme to me.  I dunno if 150
> >is anything outside of ordinary or not.
>
> OK 150 - different bandwidth, then. The basic thermal noise at the
> front end of a receiver is -174dBm + 10 log(bandwidth) + noise figure.
> A noise or interference level that desensitizes that by 1dB will be
> about 10dB lower than that level. You need to do the know the
> bandwidth and noise figure to know the significance.
>
> What is significant is that this receiver has a desensitization limit
> for interference that it throws at itself. I have never, ever come
> across that before - it is always an external spec; you design the
> internals so it doesn't interfere.

I have *never* seen a commercial receiver that did not generated self
interference.  They normally are much more concerned with cost and are
willing to live with a few spurs of noise in the spectrum.


> >When you say that it costs too much to go around the design cycle many
> >times, you clearly don't know much about the military procurement
> >process.  On the last radio that they built while I was there, they
> >were up to rev 14 of the board that had nothing but the UI controller
> >and external interfaces.  You need to remember that often the
> >development process is cost plus and the customer is *asking* for
> >tough specs.  It is only when they can't be delivered that they back
> >off.
>
> Iteration 14 and the company was still in business? Anyone can afford
> to do it again if they can't do it right. Was this a "cost-plus"
> contract? Three or perhaps four was more the number I had in mind.

You still don't understand the government procurement process.  The
company doesn't pay for this stuff.  The government pays for
development as well as the product.  That is one of the reasons that I
don't work there anymore.  I have a strong drive to get it right the
first time, literally.  But the politics meant that anyone in the
company could raise a stink about any silly imagined potential problem
and I would get pressure to change the design, *after* the board was
built.


> >What is really funny is that you are getting wrapped around the axle
> >about my use of this figure when that was really just an aside to an
> >aside of my original point.  Funny how these discussions get so far
> >off topic.
>
> If you like. It was clearly a number you threw in because it sounded
> impressively big, even though you hadn't a clue what it meant.

You don't have to be insulting.  I may not know RF by chapter and
verse, but I have a feel for the issues.  I remember very clearly the
issue that was being discussed when I was given that number, I just
don't remember the units.  Lets face it, we all talk about dB in
context and know exactly what is meant.  But at some point *which* dB
is being discussed can be forgotten.  I still get mixed up when
figuring dB in phone lines, but when I have to I can go back to
references and figure out exactly what is correct.


> >Did you read my post which used the original 140 dB figure?
>
> Yes. It was nonsense.

Then why are you bothering to discuss this with someone who posts
nonsense?  Are you compulsive-obsessive?  You are still discussing an
issue that has nothing to do with the original point.  Do you want to
discuss the original point or are you just going to harp on this one
tiny detail?

rajesh wrote:

> There is no gurantee that the audio signal is bandlimited. If we cant
> percieve freq higher than 20k doesnt mean that they are not present

No guarantee? We regularly use low-pass filters to guarantee that the 
signal is bandlimited.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

On May 5, 9:44 am, rajesh <getrajes...@gmail.com> wrote:
> On May 5, 5:54 pm, dpierce.cartchunk....@gmail.com wrote:
> > > it does help the signal to get reconstructedl more accurately.
>
> > Not if the waveform is sampled at grater than 2 times the
> > bandwidth, it does not. Once you sample greater than the
> > Nyquist limit, NO extra information is gathered, not matter
> > HOW much more your sample. The resulting output waveform
> > is the same whether you sampled at 2..01times the bandwidth
> > or 2000000 times the bandwidth. The output waveform DOES
> > NOT get any more accurate.
>
> There is no gurantee that the audio signal is bandlimited. If we cant
> percieve freq higher than 20k doesnt mean that they are not present

This statement, all by itself, tells us you have no clue
what you are talking about, and that your attempts to
invoke Shannon are hollow.

For proper sampling and reconstruction to occur,
IT IS A PREREQUISITE THAT THE SIGNAL
BANDWIDTH BE LIMITED TO LESS THAN
1/2 THE SAMPLE RATE. Any sampling system in
this domain that does not follow this struct requirement
IS BROKEN.

Therefore, it can be taken as axiomatic, that once a
signal has been sampled at faster than twice its
bandwidth, sampling at even a higher rate WILL NOT
gain you any accuracy in reconstruction.

> > You've invoked Shannon. How about going and actually
> > READING and UNDERSTANDING it not?
>
> i will do that

But you already acted as if you had.

Have you?

On May 5, 9:44 am, rajesh <getrajes...@gmail.com> wrote:
> If we cant percieve freq higher than 20k doesnt
> mean that they are not present

As the lawyers say, true but irrelevant.

If we can't perceive them, then their presence or
absence is irrelevant.

On May 5, 9:56 am, rajesh <getrajes...@gmail.com> wrote:
> but at what cut off frequency is it being filtered in case of 192
> kHz ?

THEORETICALLY, such a sampler is capable of a
bandwidth of < 192/2 or less than 96 kHz.

But...

> I dont think its 20...its much much higher.

PRACTICALLY speaking, no. Recordings are NOT made
with B&K 1/4"  or 1/8" laboratory condenser microphone:
they are made with microphones of FAR less bandwidth,
like 20 kHz or so.

Look at the attenuation of air at ultrasonic frequencies,
and such.

On May 5, 10:09 am, rajesh <getrajes...@gmail.com> wrote:
> On May 5, 7:05 pm, Oli Charlesworth <ca...@olifilth.co.uk> wrote:
> > If we can, then of course a higher sampling rate will sound better.
> > But that goes against the premises of the OP, and is nothing to do
> > with the ECC or interpolation that you've been going on about!
>
> > --
> > Oli
>
> I said we cant percieve, but i didnt say they arent there..

Again, true but irrelevant.

> i will continue the dicussion on ECC tomorrow.

Hopefully, you will be much better prepared.

As a hint: the issue of proper sampling vs bandwidth
is a topic COMPLETELY separate from ECC. You
might want to keep that in mind during your preparations.

On May 5, 10:09 am, rajesh <getrajes...@gmail.com> wrote:
> On May 5, 7:05 pm, Oli Charlesworth <ca...@olifilth.co.uk> wrote:
> > If we can, then of course a higher sampling rate will sound better.
> > But that goes against the premises of the OP, and is nothing to do
> > with the ECC or interpolation that you've been going on about!
>
> > --
> > Oli
>
> I said we cant percieve, but i didnt say they arent there..

Again, true but irrelevant.

> i will continue the dicussion on ECC tomorrow.

Hopefully, you will be much better prepared.

As a hint: the issue of proper sampling vs bandwidth
is a topic COMPLETELY separate from ECC. You
might want to keep that in mind during your preparations.

On May 5, 10:09 am, rajesh <getrajes...@gmail.com> wrote:
> On May 5, 7:05 pm, Oli Charlesworth <ca...@olifilth.co.uk> wrote:
> > If we can, then of course a higher sampling rate will sound better.
> > But that goes against the premises of the OP, and is nothing to do
> > with the ECC or interpolation that you've been going on about!
>
> > --
> > Oli
>
> I said we cant percieve, but i didnt say they arent there..

Again, true but irrelevant.

> i will continue the dicussion on ECC tomorrow.

Hopefully, you will be much better prepared.

As a hint: the issue of proper sampling vs bandwidth
is a topic COMPLETELY separate from ECC. You
might want to keep that in mind during your preparations.

On May 5, 3:09 pm, rajesh <getrajes...@gmail.com> wrote:
> On May 5, 7:05 pm, Oli Charlesworth <ca...@olifilth.co.uk> wrote:
>
> > On May 5, 2:56 pm, rajesh <getrajes...@gmail.com> wrote:
>
> > > On May 5, 6:50 pm, Oli Charlesworth <ca...@olifilth.co.uk> wrote:
>
> > > > On May 5, 2:44 pm, rajesh <getrajes...@gmail.com> wrote:
>
> > > > > There is no gurantee that the audio signal is bandlimited. If we cant
> > > > > percieve freq higher than 20k doesnt mean that they are not present
>
> > > > If the audio signal has not been appropriately bandlimited before
> > > > sampling, then of course there will be problems.  But in the real
> > > > world, the signal is explicitly filtered before it hits the sampling
> > > > circuitry.
>
> > > but at what cut off frequency is it being filtered in case of 192
> > > kHz ?
> > > I dont think its 20...its much much higher.
>
> > But you've just acknowledged that "we can't perceive freq higher than
> > 20k".
>
> > If we can, then of course a higher sampling rate will sound better.
> > But that goes against the premises of the OP, and is nothing to do
> > with the ECC or interpolation that you've been going on about!
>
> I said we cant percieve, but i didnt say they arent there..i will
> continue the dicussion on ECC tomorrow.

You're going round in circles again.  If we can't perceive frequencies
>20kHz, then it doesn't matter whether or not they are present in the
original audio source, nor whether or not we reproduce them with our
digital audio system.

However, if they are still there at the point where sampling takes
place, then the anti-aliasing filter has been incorrectly implemented.

What exactly are you trying to say that differs from any of this?  At
the moment, it's impossible to tell what your argument is, because you
appear to be changing your mind on every post.


--
Oli

On May 5, 1:27 pm, dpierce.cartchunk....@gmail.com wrote:
> On May 5, 9:44 am, rajesh <getrajes...@gmail.com> wrote:
>
> > If we cant percieve freq higher than 20k doesnt
> > mean that they are not present
>
> As the lawyers say, true but irrelevant.
>
> If we can't perceive them, then their presence or
> absence is irrelevant.

Are you guys still arguing over this???  The issue is not whether
ultrasonic signals can be perceived, it is about the sample rate.  The
microphone may well have a cutoff at or below 20 kHz so that there is
nothing in the inaudible range.  Still, a sample rate higher than 40
kHz can be a good thing.