Prev: Problem VisualBasic exe file to project and form data....
Next: biological query
From: MoiInAust on 8 Dec 2009 13:03
> And to answer the other question, the sweep need only be fat enough to
> give a good stable trace on the oscilloscope/screen, say 50 Hz?
That was meant to be *fast* enough!


From: Dr J R Stockton on 8 Dec 2009 12:27
In comp.lang.basic.visual.misc message <7ZSdnfmPFtac0YHWnZ2dnUVZ_r2dnZ2d
@giganews.com>, Sun, 6 Dec 2009 19:27:45, Jim Mack <no-uce-ube(a)mdxi.com>
posted:
>Dr J R Stockton wrote:
>> In comp.lang.basic.visual.misc message
>> <4b1a2eba$1(a)dnews.tpgi.com.au>, Sat, 5 Dec 2009 20:58:14, MoiInAust
>> <user(a)user.com> posted:
>>>
>>> Thanks Olaf. I wanted to read 455 KHz +- 10 KHz so it sounds like
>>> no go.
>>>
>>
>> If it is bandwidth-limited to 10 kHz then one can recover the
>> waveform by sampling at only about 10 000 samples/second.
>
>Nyquist says you need at least two samples per cycle to adequately
>recover the waveform.

A little knowledge is a dangerous thing; and that you have. You should
read <http://en.wikipedia.org/wiki/Nyquist-Shannon_sampling_theorem#Shan
non.27s_original_proof> and <http://en.wikipedia.org/wiki/Nyquist-
Shannon_sampling_theorem#Sampling_of_non-baseband_signals>, for example.

Nyquist said what you say, but in conjunction with a condition that I
expressly ruled out.

You should read all of the pages cited by <http://en.wikipedia.org/wiki/
Nyquist> in connection with Harry Nyquist and sampling.

--
(c) John Stockton, near London. *@merlyn.demon.co.uk/?.?.Stockton(a)physics.org
Web <URL:http://www.merlyn.demon.co.uk/> - FAQish topics, acronyms, & links.
Correct <= 4-line sig. separator as above, a line precisely "-- " (SoRFC1036)
Do not Mail News to me. Before a reply, quote with ">" or "> " (SoRFC1036)
From: Schmidt on 12 Dec 2009 14:48

"Dr J R Stockton" <reply0950(a)merlyn.demon.co.uk> schrieb im Newsbeitrag
news:eRFTqnDzxoHLFwM7(a)invalid.uk.co.demon.merlyn.invalid...

> >>> ... I wanted to read 455 KHz +- 10 KHz so it sounds like
> >>> no go.
> >>>
> >>
> >> If it is bandwidth-limited to 10 kHz then one can recover the
> >> waveform by sampling at only about 10 000 samples/second.
> >
> >Nyquist says you need at least two samples per cycle to
> >adequately recover the waveform.
>
> A little knowledge is a dangerous thing; and that you have...

Hmm, at least it was a recommendation which would have
worked (for sure) in the given case.

If you sample a given 455kHz (+-10kHz) signal at only
10kSamples/sec, then you place the Nyquist-frequency (zone)
of the ADC at 5kHz - and (sub-)sampling the input-signal
this way, would give you backfolded frequencies at multiples
of 5KHz max only - therefore not covering the +-10kHz-input-
range fully.

So your recommendation of "about 10 000 samples/second"
would not work. ;-)

Olaf







From: Dr J R Stockton on 13 Dec 2009 12:57
In comp.lang.basic.visual.misc message <ufET1V2eKHA.5020(a)TK2MSFTNGP02.ph
x.gbl>, Sat, 12 Dec 2009 20:48:52, Schmidt <sss(a)online.de> posted:
>
>"Dr J R Stockton" <reply0950(a)merlyn.demon.co.uk> schrieb im Newsbeitrag
>news:eRFTqnDzxoHLFwM7(a)invalid.uk.co.demon.merlyn.invalid...
>
>> >>> ... I wanted to read 455 KHz +- 10 KHz so it sounds like
>> >>> no go.
>> >>>
>> >>
>> >> If it is bandwidth-limited to 10 kHz then one can recover the
>> >> waveform by sampling at only about 10 000 samples/second.
>> >
>> >Nyquist says you need at least two samples per cycle to
>> >adequately recover the waveform.
>>
>> A little knowledge is a dangerous thing; and that you have...
>
>Hmm, at least it was a recommendation which would have
>worked (for sure) in the given case.
>
>If you sample a given 455kHz (+-10kHz) signal at only
>10kSamples/sec, then you place the Nyquist-frequency (zone)
>of the ADC at 5kHz - and (sub-)sampling the input-signal
>this way, would give you backfolded frequencies at multiples
>of 5KHz max only - therefore not covering the +-10kHz-input-
>range fully.
>
>So your recommendation of "about 10 000 samples/second"
>would not work. ;-)

There is a difference, of a factor of two, between 455 +- 10 kHz,
bandwidth 20 kHz, and 455 kHz bandwidth 10 kHz. Also, "about" is quite
an elastic term.

The need is to take samples at more than twice the rate corresponding to
the bandwidth, and to do so long and accurately enough to obtain
sufficient resolution.

In one case I used to deal with, the bandwidth was as near to zero as
the finest known oscillator makers could do it (i.e. a good sine wave),
the sample rate was either faster or slower than twice the wave
frequency, and (leaving out other considerations) the need was to
recover in effect the amplitude and phase.

--
(c) John Stockton, near London. *@merlyn.demon.co.uk/?.?.Stockton(a)physics.org
Web <URL:http://www.merlyn.demon.co.uk/> - FAQish topics, acronyms, & links.
Correct <= 4-line sig. separator as above, a line precisely "-- " (RFC5536/7)
Do not Mail News to me. Before a reply, quote with ">" or "> " (RFC5536/7)
From: Schmidt on 14 Dec 2009 08:29

"Dr J R Stockton" <reply0950(a)merlyn.demon.co.uk> schrieb im Newsbeitrag
news:XbUx43CMsSJLFw9X(a)invalid.uk.co.demon.merlyn.invalid...

[SubSampling a given input-signal at a lower than "Nyquist-rate"... ]

> >If you sample a given 455kHz (+-10kHz) signal at only
> >10kSamples/sec, then you place the Nyquist-frequency (zone)
> >of the ADC at 5kHz - and (sub-)sampling the input-signal
> >this way, would give you backfolded frequencies at multiples
> >of 5KHz max only - therefore not covering the +-10kHz-input-
> >range fully.
> >
> >So your recommendation of "about 10 000 samples/second"
> >would not work. ;-)
>
> There is a difference, of a factor of two, between 455 +- 10 kHz,
> bandwidth 20 kHz, and 455 kHz bandwidth 10 kHz. Also,
> "about" is quite an elastic term.
Sure... ;-)

> The need is to take samples at more than twice the rate
> corresponding to the bandwidth...
That's not the only condition - you would also need to take
care, to place the "starting-point" of such a "SubSampling-
Nyquist-zone" exactly at the "center-point" (in our
example the 455kHz) - or choose a twice or three times
as large sample-rate - and place the "starting-point" below
the lowest frequency of the interval ...(here: 445kHz) and
do a "center-correction" in software.

In the OPs case a "correct centering" would be achievable
with a sample-frequency of:
70kHz
(which would give a Nyquist-zone-window of 35KHz) ...
and multiplied: (13 * 35 = 455) this would meet the condition
(this is the lowest frequency which fulfills the requirements, as
long as one wants to calculate with "whole numbers only").

For not that easy to "produce" sampling frequencies of:
30,333333333333KHz
Nyquist-zone-window = 15.1666666666666KHz
(the multiplier to reach 455kHz exactly would be 30 then)
or
a sample-frequency of:
20.222222222222kHz
Nyquist-zone-window = 10.111111111111KHz
(the multiplier to reach 455kHz exactly would be 45 then)
....the approach would also work, but these are unusual,
difficult to ensure sample-frequencies.

Otherwise (placing the "Nyquist-zone-backfolding-point" not
exactly at the center-frequency, but somewhere "within") the
approach would require a lot of corrections in software - really
don't know, if the OP wants to go there and is able to handle
the approach correctly in these "special cases".

Also required (although being lower) would be a very *stable*
sampling-frequency (and a very short sample-and-hold interval
IMO, working at least "at the level" of 455kHz), to reduce
jitter and make that backfolding-approach a:
"reliably working success without surprises". ;-)

So, the "brute-force-approach" is not a bad recommendation,
since the potential surprises are lowered (at the cost of
somewhat higher harware-requirements). So it boils down
(as always) to the wellknown:
larger hardware-fixcosts (bought 3rd-party stuff)
vs.
knowledge+time (to grow "your own, cheaper thing")

But we're getting more and more off-topic now -
a hardware- (or electronics-) forum would really be a better
place to discuss such things.


Olaf


First  |  Prev  | 
Pages: 1 2 3 4 5
Prev: Problem VisualBasic exe file to project and form data....
Next: biological query