EEVblog Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: Gregory on September 01, 2014, 05:51:36 pm

Title: QPSK Demotulator: frequency and time recovery
Post by: Gregory on September 01, 2014, 05:51:36 pm
Hi guys!

In a QPSk demodulator I need to recover carrier frequency and symbol sample time. How can I do both if frequency error generation needs correct time lock and Gardner time recovery needs frequency correct lock?

Thanks
Title: Re: QPSK Demotulator: frequency and time recovery
Post by: David Hess on September 01, 2014, 06:32:36 pm
In a QPSk demodulator I need to recover carrier frequency and symbol sample time. How can I do both if frequency error generation needs correct time lock and Gardner time recovery needs frequency correct lock?

Alex, what is Costas Loop?
Title: Re: QPSK Demotulator: frequency and time recovery
Post by: scatha on September 02, 2014, 09:44:01 am
If you've got access to Simulink,  Dick Benson of Mathworks has written some pretty nice models showing how to perform symbol timing/carrier recovery of QPSK in the face of various impairments.

Pretty much the standard text in this area is 'Digital Communication Receivers: Synchronization, channel estimation, and signal processing' by Meyer. The symbol timing error estimator on pp. 289-295 can be used in conjunction with a loop filter and fractional delay filter to correct for symbol timing, and from here it's pretty simple to correct for carrier offset.



Title: Re: QPSK Demotulator: frequency and time recovery
Post by: danfo098 on September 02, 2014, 03:04:14 pm
A Costas Loop run completely independent without any feedback from the downstream demodulator and symbol timing recovery scheme and it can aquire a lock on the carrier frequency and phase if the initial frequency offset is not too large (half or a quarter of the signal bandwidth, not sure but something like that).

If the initial frequency offset is too large the carrier frequency can be estimated (one method if FFT) and then you initialize the Costas Loop with the estimated frequency which should then be in the pull-in range.
Title: Re: QPSK Demotulator: frequency and time recovery
Post by: coppice on September 02, 2014, 03:23:22 pm
A Gardner TED does not require that the carrier recovery is locked. You can make it the first step in recovering the signal, and nail down the carrier phase and frequency by monitoring the symbol locked information that comes out of the TED. Other totally different strategies, based on estimating the carrier first are also possible. A lot depends on whether you have a distinct training sequence to work from, or if you need to recovery the signal blindly.