http://abelian.org/vlf/tmp/161102b.gif

Stefan wrote:

> The new error map is more informative. The errors
> are quite equally spread i find.

Yes.  There are thinner bits and thicker ones, but I suppose
if just random red lines were plotted, the eye would be drawn
to random bunches and voids which actually have no significance.

Difference between success and failure is only a few bits, eg

 Eb/N0 = +1.7 dB  bit error rate = 501/1312  Virtually certain decode
 Eb/N0 = -0.2 dB  bit error rate = 531/1312  ~95% chance
 Eb/N0 = -0.5 dB  bit error rate = 535/1312  ~80% chance
 Eb/N0 = -1.0 dB  bit error rate = 541/1312  ~50% chance
 Eb/N0 = -1.7 dB  bit error rate = 551/1312  Almost no chance

Something like 20 bits between a good chance and no chance. Cliff
effect.

Still working on 2016-11-03 06:00 16K25A S30 C11, so far no decode
but a couple of settings remain to try.

--
Paul Nicholson
--