I perform real experimental tests for optical communications using DCO-OFDM. The transmitted signal is uploaded into an AWG (arbitrary waveform generator), then bias (2.5 V) is added before adopting the signal into the source light which includes a lens. The received signal is also focused by another lens and passed through the APD which is connected to the oscilloscope.
This structure is common and many people are using it; what I am asking about is the connection between the AWG and the oscilloscope. There is a cable of clock reference connecting the AWG to the oscilloscope. When I do not use it (the channel is always estimated and equalized), the constellation of the received signal becomes a circle (similar to the CFO effect). When using that cable, the signal becomes ok. I can not understand the meaning of that effect. Does that mean that when using that cable, we do not have any effect for the sampling offset, while when not used, I must estimate and equalize the frequency offset?