Andor wrote:

> Jim,
> 
> how fast is Newsfeeds? On Google Groups (former deja news), it takes
> about 6 hours for posts to show up - that is extremely slow, but it is
> for free. I wanted to check out Newsfeeds, but don't want to pay if I
> get the same speed that I can get for free.
> 

Hi Andor
	First of all I don't personally subscribe to Newsfeeds. My ISP used to
maintain a newserver, then several years ago started using Newsfeeds instead -
so it costs nothing additional, its just part of the monthly service fee
whether I use it or not. In general, I would say, the service was better
before. Lately, it seems to be more reliable. In the past I've had lots of
trouble with posts not showing up (including my own). 


> So, how long is the Newsfeeds propagation delay? :)

Its much better than 6 hrs.(when it doesn't lose the message). You might want
to try:

News.Individual.net 

Its free and seems pretty reliable.

-jim


-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

jim wrote:
...
> -jim
> 
> 
> -----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
> http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
> -----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

Jim,

how fast is Newsfeeds? On Google Groups (former deja news), it takes
about 6 hours for posts to show up - that is extremely slow, but it is
for free. I wanted to check out Newsfeeds, but don't want to pay if I
get the same speed that I can get for free.

So, how long is the Newsfeeds propagation delay? :)

Regards,
Andor

jim wrote:
> 
> Nigel Redmon wrote:
> > The fact that you get
> > obviously different results (beyond delay) depending on which set you
> > choose just shows that the input was not bandlimited.
> >
> 
> A) How can the input not be bandlimited? It's not analog.
> 
> B) And even if you mean the original analog source was not bandlimited - How
> can the difference (between odd and even samples) be anything beyond delay?
> 

OK, I had some coffee and woke up. You meant the input is not bandlimited to
the new reduced sample rate. Even so, the difference between the odd and even
channel (both will contain the aliased frequencies) should be nothing more
than delay. I suppose a frequency exactly at Fs/4 (of the original sample
rate) could be divided unequally between the 2 output channels. 

-jim


-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

Nigel Redmon wrote:
> The fact that you get
> obviously different results (beyond delay) depending on which set you
> choose just shows that the input was not bandlimited.
> 

A) How can the input not be bandlimited? It's not analog.

B) And even if you mean the original analog source was not bandlimited - How
can the difference (between odd and even samples) be anything beyond delay?

-jim


-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----==  Over 100,000 Newsgroups - 19 Different Servers! =-----

Regarding your concern over which is better 1 and 3, or 2 and 5, and you 
note on the difference between interpolating with the two sets:

Don't lose sight of the fact that this "pick the odd points or pick the 
even points--I don't care" decimation works precisely because the signal 
is assumed to be _bandlimited_ before you do the decimation--we arent' 
talking about decimating an arbitrary data stream. The fact that you get 
obviously different results (beyond delay) depending on which set you 
choose just shows that the input was not bandlimited.


In <40e57a6a$0$298$4d4ebb8e@read.news.de.uu.net> Piergiorgio Sartor  
wrote:
> Jerry Avins wrote:
> 
>>> The sequence: 1, 2, 3, 5 has more information in 1 and 3 or
>>> in 2 and 5? Or is _always_ the same?
>> 
>> Typo in your sample numbers? After the necessary filtering to halve 
>> the bandwidth, the samples 0, 2, 4, ... have exactly the same 
>> information as 1, 3, 5, ....
> 
> No, no typo, the 5 was intentionally used, in order
> to confuse a little bit the interpolator :-)
> 
> In case of 1, 2, 3, 4, just linear function can
> interpolate from 1, 3 or 2, 4 to the complete
> sequence. Same happens with 5 instead of 4, but
> we get an error here and there (in case of integer).
>  From 1, 3 we can get 2 and 4 (linear interpolation),
> the error is 1.
>  From 2 and 5 we get 0.5 and 3.5, so 0 and 3, the
> error is still 1, even if the real max error is 0.5.
> Probably it is possible to massage the number in order
> to get more close results in one case. Of course always
> with a given interpolator, linear in this case.
> 
>> Not unless a mechanism exists to make the rounding error depend on 
>> the parity of the sample number. Such a mechanism seems most unlikely.
> 
> OK.
> 
>> One reconstruction will be better, even if by a trivial amount. Of
>> course, had the sampling been delayed or advanced by 50 microseconds,
>> the odd and even sets would have been exchanged. About half the time,
>> the odd set will be better. About half the time, instantaneous 
>> Gaussian noise is positive.* Knowing that doesn't lead to an improved 
>> design.
> 
> Good.
> 
>> * Whatever the faculty or skill being considered, about half the 
>> people are below average.
> 
> Well, but the other half is always above... ;-)
> 
> bye,
> 

Jerry Avins wrote:

>> The sequence: 1, 2, 3, 5 has more information in 1 and 3 or
>> in 2 and 5? Or is _always_ the same?
> 
> Typo in your sample numbers? After the necessary filtering to halve the
> bandwidth, the samples 0, 2, 4, ... have exactly the same information as
> 1, 3, 5, ....

No, no typo, the 5 was intentionally used, in order
to confuse a little bit the interpolator :-)

In case of 1, 2, 3, 4, just linear function can
interpolate from 1, 3 or 2, 4 to the complete
sequence. Same happens with 5 instead of 4, but
we get an error here and there (in case of integer).
 From 1, 3 we can get 2 and 4 (linear interpolation),
the error is 1.
 From 2 and 5 we get 0.5 and 3.5, so 0 and 3, the
error is still 1, even if the real max error is 0.5.
Probably it is possible to massage the number in order
to get more close results in one case. Of course always
with a given interpolator, linear in this case.

> Not unless a mechanism exists to make the rounding error depend on the
> parity of the sample number. Such a mechanism seems most unlikely.

OK.

> One reconstruction will be better, even if by a trivial amount. Of
> course, had the sampling been delayed or advanced by 50 microseconds,
> the odd and even sets would have been exchanged. About half the time,
> the odd set will be better. About half the time, instantaneous Gaussian
> noise is positive.* Knowing that doesn't lead to an improved design.

Good.

> * Whatever the faculty or skill being considered, about half the people
> are below average.

Well, but the other half is always above... ;-)

bye,

-- 
   Piergiorgio Sartor

Piergiorgio Sartor wrote:

> Jerry Avins wrote:
> 
>> Would the original set of samples still be representative if the
>> sampling clock had started 50 microseconds later or earlier?
> 
> 
> I think delay does not matter.

I agree. Delay is the essential difference between the even and odd
sets.

> Actually I forgot one point, the samples are finite in time.
> 
> I mean there are, for example, 10 million samples to which
> this process should be applied.
> 
> My concern was related to the fact that samples are quantized,
> so the 5 million odd or even can have different quantization
> error, or maybe not...

That is relevant only if you know a priori which set of samples has the
lower mean-square error.

> The sequence: 1, 2, 3, 5 has more information in 1 and 3 or
> in 2 and 5? Or is _always_ the same?

Typo in your sample numbers? After the necessary filtering to halve the
bandwidth, the samples 0, 2, 4, ... have exactly the same information as
1, 3, 5, ....

> When reconstructing, I assume we will have to deal with rounding
> and so on.
> So is there any way to choose odd or even in order to minimize

Not unless a mechanism exists to make the rounding error depend on the
parity of the sample number. Such a mechanism seems most unlikely.

> Of course one can recontruct from odd and even and see what is
> better, I was wondering if there is a way to know this before.

One reconstruction will be better, even if by a trivial amount. Of
course, had the sampling been delayed or advanced by 50 microseconds,
the odd and even sets would have been exchanged. About half the time,
the odd set will be better. About half the time, instantaneous Gaussian
noise is positive.* Knowing that doesn't lead to an improved design.

> Maybe the question is not clear, since it is not clear even in
> my mind... :-)

The question seems clear to me. The answer, too.

> Thanks,

Welcome

> bye,

Jerry
_________________________________
* Whatever the faculty or skill being considered, about half the people
are below average.
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

"Piergiorgio Sartor" <piergiorgio.sartor@nexgo.REMOVE.THIS.de> wrote in
message news:40e5111e$0$291$4d4ebb8e@read.news.de.uu.net...
> Jerry Avins wrote:
>
> > Would the original set of samples still be representative if the
> > sampling clock had started 50 microseconds later or earlier?
>
> I think delay does not matter.
>
> Actually I forgot one point, the samples are finite in time.
>
> I mean there are, for example, 10 million samples to which
> this process should be applied.
>
> My concern was related to the fact that samples are quantized,
> so the 5 million odd or even can have different quantization
> error, or maybe not...
>
> The sequence: 1, 2, 3, 5 has more information in 1 and 3 or
> in 2 and 5? Or is _always_ the same?
>
> When reconstructing, I assume we will have to deal with rounding
> and so on.
> So is there any way to choose odd or even in order to minimize
> the effect of the finite precision in the reconstruction process?

The quantization error is typically shown as a white noise process.  I would
assume odd and even samples are equally prone to error so it doesn't matter
which you do.

> Of course one can recontruct from odd and even and see what is
> better, I was wondering if there is a way to know this before.
>
> Maybe the question is not clear, since it is not clear even in
> my mind... :-)
>
> Thanks,
>
> bye,
>
> -- 
>    Piergiorgio Sartor

Steve wrote:

> No, in theory they are NOT the same when you are taking even and odd
> samples. Because the odd sequence is one sample delay(essentialy), so
> its spectrum in frequency domain is expanding by 2 and then shifting
> by pi from the original sequence spectrum. Its even sequence spectrum
> is expanding by 2 from original sequence spectrum.

I meant in "reconstruction theory".

As far as I know, in order to reconstruct the signal,
regardless of delay, odd or even should be the same.
But also regardless of finite precision math.

bye,

-- 
   Piergiorgio Sartor

Jerry Avins wrote:

> Would the original set of samples still be representative if the
> sampling clock had started 50 microseconds later or earlier?

I think delay does not matter.

Actually I forgot one point, the samples are finite in time.

I mean there are, for example, 10 million samples to which
this process should be applied.

My concern was related to the fact that samples are quantized,
so the 5 million odd or even can have different quantization
error, or maybe not...

The sequence: 1, 2, 3, 5 has more information in 1 and 3 or
in 2 and 5? Or is _always_ the same?

When reconstructing, I assume we will have to deal with rounding
and so on.
So is there any way to choose odd or even in order to minimize
the effect of the finite precision in the reconstruction process?

Of course one can recontruct from odd and even and see what is
better, I was wondering if there is a way to know this before.

Maybe the question is not clear, since it is not clear even in
my mind... :-)

Thanks,

bye,

-- 
   Piergiorgio Sartor