Message-ID: <001301c27fc0$0f4c9240$0700000a@parissn2>
From: "Stewart Nelson" <sn@scgroup.com>
To: rsgb_lf_group@blacksheep.org
References: <001001c27fa0$78e71420$4e60063e@main> <8gfurucham8vvs0r2tp26c8qdaaffa9nb5@4ax.com>
Date: Wed, 30 Oct 2002 03:57:17 +0100
MIME-Version: 1.0
Subject: Re: LF: sound cards...well amybe if.....
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Sender: <owner-rsgb_lf_group@blacksheep.org>
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org

Hi all,

An S&H with 1 us sample time and 100 us hold time is very easy for
today's parts.  The state of the art is several orders of magnitude
better for both parameters.  Take a look at this A/D with S/H:
http://products.analog.com/products/info.asp?product=AD6645 .
It can sample at up to 105 MHz, and with decimation, a.k.a.
undersampling, works with band limited IF centered at up to
200 MHz.

Clocking from the sound card is also no problem.  Just tap its
internal sample clock and divide by 256 (or other oversample
factor).  The phase will be arbitrary with respect to the
output samples, but that should not matter.  Better, remove
the sound card's crystal and feed in your own stable main clock.

But it is much easier to design these systems when you can
set the sample rate to a value suitable for the input
frequency.  With a sound card, you may not have that option.
Paul's example, with one A/D channel, would be unable to
distinguish 143 kHz from 145 kHz; they would both produce
a 1 kHz audio signal, since the third harmonic of 48 is 144.
With stereo recording (using two S/H 90 degrees apart), you
can reject the image, and ideal filtering would give you up
to 48 kHz bandwidth, but maybe not the 48 kHz band you want.

IMO, the linrad approach is much cleaner and will perform better.

73,

Stewart KK7KA


> On Tue, 29 Oct 2002 23:08:21 -0000, "Alan Melia"
> <Alan.Melia@btinternet.com> wrote:
> 
> >Hi Alberto, we are all jumping on you for your over-enthusiasm........but it
> >occurs to me that this is a 24bit card....than may be more significant. It
> >has the potential to give the level of s/n we might need.
> >
> >Could not the dreaded aliasing be used to benificial effect, or am I missing
> >some subtle point. It occurs to me that a 96ksps sampler will 'fold back'
> >136kHz  to 40kHz....so if any anti-aliasing filter could be disabled (I
> >think, where used, these are normally passive rather than active ?).....it
> >might be possible to have a software 136kHz RX !!
> >THERE is a CHALLENGE for you software gurus !!
> >I will await my idea to be shot down in flames, before I conside buying one
> >of CL new audigy units !!
> >Cheers de Alan G3NYK
> >alan.melia@btinternet.com
> 
> One could think about using decimation, in which the band is limited
> by passive filters to say 125 kHz to 145 kHz, but as far as I
> understand, in order to use decimation, you need a sample & hold
> circuit ahead of the ADC with the sample time less than a half cycle
> time at 135kHz (or about 3 us, preferably even less), while the hold
> time should be the same as the ADC conversion time.  
> 
> In this example, even 48 kHz ADC sample rate would be sufficient, but
> how do you control an external sample&hold circuit from a sound card ?
> 
> Paul OH3LWR
>  
> 
>