Column: JESD204 standard


Figure 3: Eye diagram going into a long ISI PCB


caused by the filtering effects of a transmission line. Mathematically, it can simply be modelled as a low-pass filter. When sending high-speed serial data down a transmission line, the filtering results in a distorted signal. Pre- emphasis and equalisation counteract the filtering effects of ISI with the goal of bringing the response at the end of the channel back to as close to flat frequency as possible, resulting in a signal that is not distorted by ISI. With a basic understanding of


emphasis and equalisation and ISI, the next step is setting them. What many people ask first is how long a trace can be driven with and without emphasis/ equalisation. Real-world PCB designs have too


many variables that can affect ISI to be able to specify the channel in terms of trace length. Variables like trace length and width, vias vs. no vias, choice of dielectric material, connectors vs. no connectors, trace material, corners, passive components and distance to ground-plane can all affect channel performance. So, how can channel characteristics ever be correlated with pre-emphasis/equalisation? The solution is to specify the channel


in terms of insertion loss. Insertion loss is described in the JESD204B specification as a measure of the power loss of a signal over frequency. Pre- emphasis, equalisation and PCB channel can all be related in terms of insertion loss (and gain). Using a relevant


frequency (the JESD204B specification lists a three-quarter baud rate) and an insertion loss limit (JESD204B lists -6dB), the gain provided by pre- emphasis and/or equalisation can be selected to bring the frequency response at the selected frequency up above the loss limit. For example, a PCB channel with -12dB of loss at +9GHz would need +6dB of pre-emphasis/equalisation gain to bring the total back up to -6dB. Alternatively, converter manufacturers


can provide a table of pre-emphasis/ equalisation settings vs. PCB insertion loss. Tis method can result in a better solution, as it does not depend on as many assumptions. To build such a table for a transmitter (and to emulate end system designs), a set of test evaluation boards can be built with varying trace lengths. The eye diagram at the end of the


PCB trace can be directly measured and compared against the JESD204B receiver mask. By trying various PCB trace lengths, there will be one that results in the eye just barely passing the receiver mask. Since the insertion loss of that specific trace can be measured, the drive capability for a specific pre-emphasis setting is known. Compare Figure 2, which shows an eye diagram at the end of an ISI PCB, to Figure 3, which shows the eye diagram going into an ISI PCB. In this case, the data rate is 5Gbps, the ISI PCB has 8dB of insertion loss at 4GHz, and pre-emphasis is off. Repeating this process vs. pre-


emphasis settings will result in a table 16 November 2021 www.electronicsworld.co.uk


of pre-emphasis settings vs. insertion loss. A similar approach can be used on a receiver with equalisation; start with a BERT generator that is outputting the maximum allowed total jitter (except for ISI jitter). Using the same set of ISI test boards with varying trace lengths, test with longer and longer traces until the receiver starts to get errors that exceed the target bit error rate (1 × 10–15


). Measure the insertion loss of the


PCB trace. Repeat for every equalizer setting. In summary, if a JESD204B device manufacturer provides only pre-emphasis/equalisation gain, the first method can be used to pick settings. The best method is for the manufacturer to provide a table of settings vs. channel insertion loss. Should pre-emphasis or equalisation


be used? From a frequency response correction standpoint, there’s no clear reason to use one instead of the other. However, in most cases, pre-emphasis can generate a certain amount of gain with less power. If system power is important, that could be a reason to choose pre-emphasis over equalisation. Another advantage is that the effect on the signal can be directly measured with an oscilloscope. It can be common to have both a


JESD204B transmitter with pre-emphasis and a receiver with equalisation. How would you determine when to turn on both? Simply, if the insertion loss of the channel cannot be overcome by just pre-emphasis or just equalisation,


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46