In GNURadio's PSK tutorial they place the costas loop after clock sync and linear equalizer. This seems odd to me.

I'm thinking that it's easier to lock onto a carrier frequency at the higher RF sample rate rather than on the clocked symbols. And similarly it seems to me that it should be easier to to clock recovery on the frequency-synced post-costas signal.

It will also presumably be easier to use a correlation estimator after the costas loop, too. And correlation estimator has to be done before clock sync, since it uses the clock sync information to sync.

Indeed, my experiments decoding data always seem to work better when I start with the costas loop.

But I claim no expertise, so I assume the GNU Radio tutorial is correct. But why is it correct?


1 Answer 1


The job of the costas loop is to "derotate" the constellation points.

But for that you first need to look at the constellation points at the symbol instants. If you look at the signal between these instants, you don't get a constellation point, and hence, the costas loop doesn't converge as nicely, or, not at all, if your SINR becomes too bad when you're far away from the optimal sampling point.

  • $\begingroup$ Thanks. With your answer and reading a bit more I think I now understand it better. But I don't understand then how correlation estimator can work against an off-frequency signal. Seems to me like locking to the carrier should be the first thing one does, no? $\endgroup$
    – Thomas
    Jul 3, 2022 at 13:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.