[hpsdr] Mercury SSB Phase Noise Testing

[Graham / KE9H]

Larry and all:

Actually, there is a "DDS" in Mercury, it is a CORDIC implemented in the
FPGA, so no physical chip to see, other than the FPGA, and no
manufacturer's spec sheet to read, other than the FPGA source code
on the SVN.  It generates a "local oscillator" signal and associated 90 
degree
shifted signal anywhere from 1 Hz to 60 MHz, used to do the mix-down 
conversion
to the baseband.  It is locked to the 10 MHz reference signal on the 
Atlas Bus.

--- Graham

==


Larry Gadallah wrote:
> ***** High Performance Software Defined Radio Discussion List *****
>
> John Miles wrote:
>> ***** High Performance Software Defined Radio Discussion List *****
>>
>> Typically the phase noise of any SDR will simply be that of its DDS, 
>> which
>> can be extrapolated from the DDS's spec sheet for the clock and output
>> frequencies being used.  You can't measure it with an 8656B, as the
>> generator is far noisier than any DDS.
>>   
> Just to be clear, there is no DDS in the Mercury. Mercury is a direct 
> sampling SDR, where the RF signal is directly converted to digital 
> data by the LTC2208 ADC. The selection of a specific signal from the 
> 0-54 Mhz spectrum happens later in the digital domain processing by 
> using a DDC (digital down-converter) module, which executes in the 
> FPGA. This results in a lower bandwidth signal (192 kHz?) that is then 
> sent to the PC over the USB and further processed and demodulated in 
> the PC. The clocking of the ADC in Mercury is handled by a low-noise, 
> low-jitter oscillator that drives the ADC at roughly 130 million 
> samples/second (122.88 Mhz to be exact), but this is a fixed 
> frequency. The only variable frequency components in Mercury are 
> purely software.
>
> Contrast this to the direct conversion approach used by Flexradio, 
> Soft-Rock etc. where a DDS (or a crystal oscillator) is indeed used to 
> down-convert the signal(s) of interest to baseband (0 Hz IF) and this 
> signal is then sampled and the baseband samples are sent to the PC for 
> processing (USB or via a Soundcard, depending on where the sampling is 
> done) at somewhere between 44 and 192 Khz.
>
> The interesting difference between these approaches is that since the 
> entire HF spectrum can be sampled at once with the direct sampling 
> approach, the selection of signals of interest is done by code 
> contained within the Mercury FPGA. This means that it is possible to 
> receive multiple signals simultaneously, limited only by code space in 
> the FPGA. For example, Alex, VE3NEA's Skimmer Server uses the QS1R 
> receiver (very similar in architecture to the Mercury) to 
> simultaneously receive and decode all of the CW portions of 7 HF bands 
> at once with a single QS1R SDR receiver!
>> The best approach would be to simply use the receiver and FFT 
>> software to
>> look at a single tone from a clean crystal oscillator.  The result is 
>> the
>> phase noise response, less 10*log(BW) where BW is the bandwidth per bin,
>> plus a dB or two for the window function's noise response.
>>   
> This approach would work well, particularly if the phase noise 
> characteristic of the crystal oscillator were known in advance and 
> could be subtracted from the measured results.
>
> Cheers,
>


 1256589910.0