Clock recovery

from Wikipedia, the free encyclopedia

The aim of clock recovery (also symbol clock synchronization ) in digital transmission technology is to determine the transmission clock of the transmitter from a received digital signal and thus to enable the precise sampling of the received signal or to correctly align ( synchronize ) the digital signal sent back in the opposite direction . The clock recovery is necessary on the receiver side in order to determine the periodic sampling times of the received data stream. Without this time-accurate alignment, the digital received signal cannot be correctly evaluated, which would result in massive bit errors. If a transmission method offers clock recovery, the complex parallel clock transmission using a separate transmission channel for the purpose of synchronization can be dispensed with.

In order to be able to obtain the send clock from the received signal, the received signal must have a sufficient number of signal edges. This means that long sequences of '1' or '0' without edges should be avoided in the transmission. This can be achieved through appropriate channel coding and the use of appropriate line codes such as the biphase mark code or the 4B5B code . Furthermore, statistical methods such as scramblers can be used on the transmission side , which reshape the transmission signal in a pseudo-random manner and thus on average provide a sufficient number of signal edges for clock recovery.

Procedure

Basically, two different methods are used for clock recovery:

They are briefly presented in the following sections.

Clock recovery through non-linear distortion

In this method, the signal edges are obtained by non-linear distortion of the input signal, which then result in the symbol clock with a bandpass or PLL and subsequent threshold value decision. In principle, various non-linear functions can be used, such as squaring or differentiation with subsequent formation of the absolute value (rectification) of the input signal.

In practice, however, it is almost only the differentiation with formation of the amount in combination with a PLL that is essential. This means that the changes in the received signal over time are rectified and used as a reference frequency for a PLL that is roughly tuned in the capture range. In the steady state of the PLL, the clock obtained in this way is then at the same frequency and phase-locked to the transmission clock.

The disadvantage of this method is that the exact phase position cannot be regained with it. In practical systems, the exact sampling time of the received data is then usually made by means of a phase offset (time offset) set in the receiver. This time offset represents the exact sampling time within the duration of a symbol. In multi-frequency systems, this time offset must therefore be derived from the respective frequency, for example using tables.

Clock recovery through decision feedback

In this method for clock recovery, use is made of what is known as the first Nyquist condition . This condition means that the temporal symbol crosstalk, also called intersymbol interference, disappears. The prerequisite for this is the use of adaptive channel equalizers for compensation for practically all transmission channels and their distortions .

If the transmission channel is correspondingly equalized and fulfills the freedom from intersymbols, the transmission clock can be exactly recovered by synchronizing to the zero crossing of the received signal before and after the individual transmission pulses. This information is averaged over time and is used as a control variable for a controllable oscillator .

The advantage of this method is that it not only recovers the sampling frequency exactly, but also the sampling phase. The expense associated with adaptive filters is disadvantageous. In particular, the necessary training sequence at the beginning of a transmission must be provided within the framework of the transmission protocol.

Another form of clock generation by means of a feedback control loop is the use of a modified Costas loop . Usually this loop is used for the coherent demodulation of digital angle modulations such as phase shift keying . Instead of synchronizing to the carrier frequency, a correspondingly modified Costas loop can also be used to recover the symbol clock. Here, too, the phase position is correctly recovered.

See also

literature