[hpsdr] hpsdr] About the Dynamic Range of Receivers

[jeff millar]

John B. Stephensen wrote:
>  
> There are also intermodulation distortion products generated by the 
> ADC. The levels are never specified for audio ADCs and only 
> occasioally specified for high-speed ADCs. The IMD levels are always 
> higher than the quantitization noise and are critical if most of the 
> filtering is done via DSP.
All the new High Speed ADCs (including LTC2208) have one and two tone 
SFDR specification for a variety of RF input test cases.  One tone spurs 
arise from harmonics of the input folding back into the Nyquist band.  
Spurs from two tones tests come from both IMD and harmonics.

It's possible to select a frequency plan for narrow band applications 
(such as ham applications) such that the low (louder) harmonics fall 
outside the portion of the band that the DSP cares about.  I'm new to 
this list and haven't looked, does the frequency plan attempt to dodge 
around IF harmonics???

Two tone SFDR for the LTC2208 looks like it exceeds 90 dB...certainly 
above the noise floor, but not bad at all. 

It's actually quite difficult to design the RF chain such that the A/D 
is the limiting factor.  The A/D driver design has to use a very high 
IP3 buffer and consume a lot of power. 

The approach of using a single QSD mixer helps a lot.  I didn't see what 
the Mercury board intends for the buffer between mixer and A/D, but with 
up to 0 dBm input level, it seems to need more than +45 dBm OIP3 to keep 
up with the A/D.

This is a very cool project.

jeff, wa1hco
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20060808/4682fccc/attachment-0004.htm>