[linrad] Re: Experience of fast A/D cards?

["JPM_LIBERTY" <jean-pierre.mallet1@xxxxxxxxxxxxxx>]

----- Original Message ----- 
From: "Bob Stiles" <b06.s@xxxxxxxxx>
To: "Linrad mailinglist" <linrad@xxxxxxxxxxxxxxxxxxxxx>
Sent: Thursday, June 16, 2005 10:28 AM
Subject: [linrad] Re: Experience of fast A/D cards?


Hello David, and Linraders
I have a home-built card which uses an Analog Devices 14-bit A/D,
running at 40MHz, and feeding an Intersil HSP50016 digital
down converter.  The DDC was originally made by Harris, but they
sold the design on.  The bandwidth can range from about 200Hz to
200kHz, depending on the decimation rate in the DDC.  Output is I/Q
pairs, 16-bit resolution, although it's configurable up to 38 bits
(!).  These are delivered serially over re-purposed Ethernet patch
cord, using LVDS drivers, to a SHARC DSP.  The 40MHz clock comes
directly from a OCXO in a TTL package from Vectron.
(Please pardon my jargon)

The board was designed originally as a  medical imaging MRI
receiver, where the center frequency and bandwidth (and hopefully
the SNR) of the data are well known beforehand, but of course the
board is just a general-purpose digital receiver.

(There's also a 90MHz low-pass filter on the
input, and two cascaded AD603 variable gain amplifiers to provide
a nominal 80dB dynamic range on the front end)

As far as experience goes - well yes, sampling a 70MHz signal at
40MHz seems to work fine.  You just have to tell the DDC which
'Nyquist quadrant' you're in, and it will invert the polarity of
the imaginary channel for you.  The hardest part seems to be
determining the contribution of the different noise sources
(roundoff error in the DDC, thermal noise in the analog parts,
power supply noise, pickup of ambient noise, others sources yet
to be discovered...)  I reckon I can see harmonics at 10MHz
intervals coming from my frequency synthesizer - a PTS 160.  This
PTS generates its output using a 12-bit D to A, so I am
injecting truncation noise from this as part of my test!  It's just the best
thing I have.

With undersampling, clock jitter becomes more and more significant
a source of error, as the frequency of interest increases.
However, since you are subsequently decimating the data (in my
understanding - not just throwing the samples away, but averaging
them together) it tends not to matter so much.  Undersampling and
decimation are a very powerful combination.

I don't have much experience of commercially available A/D cards,
although I've been looking at the data sheets on-and-off for a long time.
They're expensive, proprietary, don't do exactly what you need etc.
     I would tend to avoid anything that plugs into a PC, since it's a
really noisy environment for high resolution RF measurements.
I hope this is of use to you all...
Best regards
Bob Stiles


#############################################################
This message is sent to you because you are subscribed to
 the mailing list <linrad@xxxxxxxxxxxxxxxxxxxxx>.
To unsubscribe, E-mail to: <linrad-off@xxxxxxxxxxxxxxxxxxxxx>
To switch to the DIGEST mode, E-mail to 
<linrad-digest@xxxxxxxxxxxxxxxxxxxxx>
To switch to the INDEX mode, E-mail to <linrad-index@xxxxxxxxxxxxxxxxxxxxx>
Send administrative queries to  <linrad-request@xxxxxxxxxxxxxxxxxxxxx>




#############################################################
This message is sent to you because you are subscribed to
 the mailing list <linrad@xxxxxxxxxxxxxxxxxxxxx>.
To unsubscribe, E-mail to: <linrad-off@xxxxxxxxxxxxxxxxxxxxx>
To switch to the DIGEST mode, E-mail to <linrad-digest@xxxxxxxxxxxxxxxxxxxxx>
To switch to the INDEX mode, E-mail to <linrad-index@xxxxxxxxxxxxxxxxxxxxx>
Send administrative queries to  <linrad-request@xxxxxxxxxxxxxxxxxxxxx>


LINRADDARNIL