Pre-emphasis technology - RF knowledge that digital engineers should master (Part 5)

    In the chip at the transmitting end, the simplest way to implement de-emphasis is to delay the output signal by one or more bits, multiply it by a weighting coefficient, and add it to the original signal. As shown in the following figure:

 

    After the pre-emphasis or de-emphasis signal is observed directly at the signal transmitter (TX), it is not an ideal eye diagram. The following figure shows an eye diagram of a 10Gbps signal with -3.5dB pre-emphasis seen at the transmitter, from which we can see an obvious "double eyelid" phenomenon. 

 

    If the pre-emphasis setting is approximately matched to the loss caused by the transmission channel, the signal may not look good at the transmitter, but after being transmitted through the transmission channel to the receiver, the signal eye diagram is still good. This is because after the signal is transmitted through the PCB or cable, the high-frequency component will attenuate, and the amplitude attenuation of the transition bit will be much greater than that of the non-transition bit. Therefore, the obvious pre-emphasis or de-emphasis effect in the previous figure is usually not seen at the receiving end (RX) of the signal, but an improved eye diagram. Therefore, in essence, pre-emphasis or de-emphasis is also a signal pre-distortion technology.

 

         Another point to note is that the parameter settings of pre-emphasis or de-emphasis need to match the loss characteristics of the signal transmission channel to achieve a better signal improvement effect. The following figures reflect the eye diagram of a 10Gbps signal after it is transmitted through an ordinary 5m long SMA cable. The upper figure is the signal eye diagram seen at the receiving end when no signal processing is performed at the transmitting end. It can be seen that the signal has been greatly deteriorated after transmission; the middle figure is the eye diagram seen at the receiving end after -3.5dB de-emphasis is performed at the transmitting end. It can be seen that although the amplitude of the eye diagram is reduced (the low-frequency component is compressed) through de-emphasis, the overall eye diagram opening is larger (the eye height is increased), which is a suitable de-emphasis setting; the lower figure is the eye diagram seen at the receiving end after -6dB de-emphasis is performed at the transmitting end. Because the de-emphasis compensation is a bit too much, there is obvious overshoot in the eye diagram, and the improvement of the eye diagram opening is not as good as when -3.5dB de-emphasis is used. In this case, if you continue to increase the de-emphasis, the resulting eye diagram may be even worse than when no de-emphasis technology is used.