[hpsdr] coherent dual mercury receivers

[Georg Prinz]

Hello Evryone,

after I setup METIS successfully, it is time to start discussion on a 
coherent dual mercury receiver system. The target should be to link 
HPSDR to linrad and maybe to other e.g. radioastronomy applications.

For, I am not a profesional programmer, I need some support.

First I would like to collect ideas and informations, what we will need 
and how it could be solved. Maybe, there is anybody out there working on 
this problem. In that case we don't need to re-invent the wheel.

Why do we need coherent receivers?

In case of linrad, the orthogonal signals received by a cross-yagi, are 
processed through an auto-corelation algorithm. The reason is, that both 
signals are orthogonal transformed to turn the system virtually so far, 
that one signal will show noise, only, and the other the complete 
signal. To determin the angle, we have to auto-corelate the two signals. 
This operation makes sense, only, if both receivers are locked together, 
in such a way that we are able to process the equivalent samples of each 
input signal. Otherwise we will get GIGO (garbage in garbage out). This 
is different from the diversity system made by Joe (K5SO).

Looking at the data sheet of LTC2208, we realize, that there is no sync 
input.
There are two clock-pairs which shall be taken into consideration: ENC 
(differential encode input) and CLKOUTP (data valid output).

Talking to Kirk, he made following assumptions:

- to designate a board that sends a sync pulse that all the other boards 
watch for and sync to. You have to take care that all delays are met 
from source to destination within 1 clock cycle.
- there are following delays to be taken into consideration:
-- rising clock edge to sync-pulse out of the final flip flop to an 
output pin on the chip
-- bus delay from one board to another
-- input pin on the destination board to the flop input that grabs the 
signal.

Out of the last team-speak session I learned that Phil is working on a 
system using time stamped sampels.
This would be exact the version to be sure to compute samples from both 
signals taken at the same time.

I know that this kind of data processing is often done in process 
control systems, like state estimation, to compute Loadflow in 
power-distribution systems. I can imagine, that time stamped data will 
meet other aplications in radioastronomy.

Maybe, we need a solution combined out of both proposals.

Ok, so far, I am curious about your responses!

Vy 73, Georg - dl2kp