The main drawback with using an a = 1 data system to achieve symbol timing is the sacrifice in bandwidth that such a large a entails. Also, when the received signal is accompanied by noise, individual zero-crossings may be corrupted by the noise and some form of averaging is desirable over many zero-crossings if accurate timing is to be achieved. |
One circuit that can average the effects of noise as well as the non-perfect zero-crossings caused by smaller values of a is the feedback timing control loop shown here. The circuit operates by using a mono-stable to create pulses of duration Ts/2 at each zero-crossing, which are then compared in a digital mixer with a locally generated clock running close to the symbol rate. The output of the mixer is integrated and filtered in order to produce a smoothed dc control voltage that is used to nudge the clock on to the correct symbol rate. The loop filtering process also serves to average out the noise accompanying the input signal and zero-crossing variations due to the small values of a.
|
![]() |
In practice, there is a compromise between wishing to quickly acquire the symbol timing for rapid data decoding, and having a long averaging time to minimize 'timing jitter'. Two excellent references on timing and carrier recovery circuits are Lindsey and Simon (1972) and Gardner (1966). |