[hpsdr] HERMES schedule question

[Graham / KE9H]

Ray:

Please allow some thought for accurate clock distribution in the system, 
and time adjustment,
so multiple boards of whatever the end product looks like, at the end of 
various lengths of fiber,
can be timed/time-synced to a few nanoseconds and used in an HF phased 
array.

--- Graham / KE9H

==

Ray Page wrote:
>
> Phil,
>
> The more I think about it, I wonder if it would make more sense for me 
> to plan on using the Atlas architecture rather than the HERMES board 
> as the basis of the DSDR. I assume the Atlas backplane can handle a 
> large number of concurrent RX/TX baseband channels? If I took this 
> path then I would combine low-level RX/TX (Mercury/Penelope) DSP 
> functions into a single FPGA and put an FO transceiver on the board. 
> The endpoint RX/TX modules would be the same as I envisioned for 
> HERMES. So, the rig would comprise a PS, Atlas, Ozy I/II and my board. 
> The antenna side would either be a combo RX/TX, or discrete RX and TX 
> boards. At a later point, the lessons of HERMES could be applied to a 
> single board derivative that could also interface the the FO RX/TX 
> modules.
>
> The actual FO link speed would be 20 x 122.88MHz = 2.4576Gbps. With a 
> 97% payload efficiency I am left with 2.384Gbps, where 1.966Gbps is 
> for DDC/DUC data and the rest is for ancillary data. I like this 
> because it simplifies the clocking schemes. If a disciplined 10MHz 
> reference is available on the Atlas bus, then this reference will 
> propagate to the RX/TX modules via the embedded FO clock.
>
> Is this really all that appealing to anyone? I don't know if it is the 
> challenge or the end result that is driving me.
>
> -Ray
> Georgetown, TX
>
>
>
> On Fri, Jun 18, 2010 at 8:55 PM, Phil Harman <phil at pharman.org 
> <mailto:phil at pharman.org>> wrote:
>
>     Hi Ray,
>      
>     You are quite correct. The receiver board (Mercury) incorporates a
>     Digital Down Converter that takes the 1.966Gbps ADC data at 16
>     bits per sample and reduces it to 48/96/192ksps at 24 bits per
>     sample.  This slower data is passed over the Atlas bus to either
>     an Ozy board (USB interface to the PC) or shortly an OzyII  board
>     (Gigabit Ethernet).
>      
>     Hermes does the same except the functions of Mercury and Ozy are
>     combined on the one board.
>      
>     73's Phil...VK6APH
>
>
> ------------------------------------------------------------------------

 1276920070.0