[hpsdr] Todays Webinar on CORDICs / dual mercury

[Georg Prinz]

Hello Kirk,

the webinar dated Apr 8, 2009 was superb! The fantastic part was to show
the way how to proceed from the theory into real code. This is often not
as simple as it sounds.

I have the impression, that the cordic algorithm is similar to the
representation of a sin x - oszillator in the z-domain:

f(t)=sinwt; F(z)= (z*sin wt)/(z*power2 - 2*z*cos wt +1). At the
z-parameter you can look like a delay element.

Further more as a comment, the cordic algorithm is not restricted to
oszillators, but can be used for fft-algorithms, too. The reason is that
fft needs some matrix rotation, too.

Regarding the remarks on an implementation of two mercury boards I am
still looking for a reasonable solution:

According to my investigations in the internet, the only existing
solution is done by Leif in his linrad system. It is not sure, if he
will adept linrad to hpsdr because of licence issues.

Nevertheless I would like to find a way to do it with mercury, because I
don't like to buy an additional hardware system (WSE).

My idea is to use an additional board with a FPGA to implement the
algorithm for the two channels/mercuries and communicate via one
OzyJanus with the PC. This would reduce the data amount to the PC and
the changes on the PC-software.

Up to now I am not clear about all the necessary theory equations to
think about an implementation. The way Leif is doing it is based upon
the ortogonal system. If you take the two complex orthogonal input
signals (v- and h-polarization) you can create two other ortogonal
signals which will match exactly the polarisation plane. So far Leif
didn't describe which orthogonal transform function he used, but
described the computation of the complex parameters of the new functions
only. The transformation will be done in such a way that one of the two
functions will have the searched signal and the other the noise. To
compute the coefficients an autocorelation between the vertical and
horizontal signal is done to find if there is any signal available at
each plane. All signals and the autocorrelation result have to be
avereaged over time and everything in real-time.

As there is no detailed description existing I have to go into the code
to understand it completly. Maybe someone can help me. But for the
moment I have to exercise your lectures, Kirk.

To pack the additional code into one of the mercury fpga could be a
possibility, if there is enough space, but does not look to be a very
good solution from the point of software structure.

Vy73, Georg, dl2kp

  


On Mi, 2009-04-08 at 12:12 -0700, Kirk Weedman wrote:
> ***** High Performance Software Defined Radio Discussion List *****
> 
> Here's the link for the webinar for Wed Apr 8, 2009 at 3:30PM PST. 
> Todays webinar will be on CORDICs
> 
> https://www2.gotomeeting.com/register/862494972
> 
> 
> _______________________________________________
> HPSDR Discussion List
> To post msg: hpsdr at hpsdr.org
> Subscription help: http://lists.hpsdr.org/listinfo.cgi/hpsdr-hpsdr.org
> HPSDR web page: http://hpsdr.org
> Archives: http://lists.hpsdr.org/pipermail/hpsdr-hpsdr.org/


 1239284483.0