Markus wrote:

> I believe that the 5170 Hz line in my 63 uHz spectrogram was
> probably due to leakage of the 1pps reference signal into the
> left audio channel.

Could be.   No such problems here due to using LP smoothed
pulse for centroid measurement.

>  I also tried the 16k25a transmission today.

Needs about 15.9 Gbyte RAM when using the optimum list
length of 62k.   Would be a silly amount of memory except
that it works.

I feel better knowing that your workbench constructions look a lot
like mine!    One stray cat and the whole thing comes to pieces.

--
Paul Nicholson
--