Issue link: https://iconnect007.uberflip.com/i/691539
64 The PCB Design Magazine • June 2016 magnetic (TEM) mode of propagation, where the electric and magnetic field lines are restrict- ed to the direction of propagation (normal to the direction of propagation), to demonstrate where field lines are prevalent. Transmission lines can be broken down into stages, when the signal and return path are a short fraction of the rise time away from the reference plane. The returning signal current path is the key to understanding mutual inductive problems in connectors, crosstalk between parallel transmis- sion lines and also EMC. Signal current does not flow down to the end of the signal trace and then return back to the source. But rather, as the signal wave front propagates, the return current builds up simultaneously creating mul- tiple paths back to the source. Howard dem- onstrates this very effectively with animation showing charged particles as little balls moving in slow motion down a signal trace, while corre- sponding particles on the returning signal trace move in the opposite direction. It is this type of demonstration that really sticks in the mind and creates a lasting impression. There are also many other animations that clearly demon- strate examples throughout the three seminars. A discussion of skin depth, where current flows only in a shallow band at high frequencies, and dielectric losses that need to be considered at high frequencies, follows. A non-uniform distribution of current, around the periphery of a conductor, creates a non-concentric field. At high frequencies, magnetic lines of force will not penetrate a conductor but rather flow tan- gential to the conducting surface. The trick is to ensure that the radiated field lines do not flow between the victim signal trace and reference planes, inducing current and hence crosstalk. Howard also points out that dielectric loss is not a function of the conductor geometry but depends on the loss tangent of the surround- ing materials. This is why a low loss dielectric material (Df = 0.002) is more effective, at high frequencies, compared to standard FR-4 (Df = 0.02). Dielectric loss is also frequency depen- dent, so the maximum frequency of operation is also significant. Finally, the section I have been waiting for: PCB traces, connectors, vias and differential sig- naling. Microstrip and stripline stackup config- urations are discussed, in detail, with the effects of surface roughness, nickel plating and solder masks. This necessitates the use of a 2D field solver such as the ICD Stackup Planner. The time-space diagrams of transmission line reflections are very enlightening. This is a simple way of analyzing reflections from down- stream loads on the transmission line. A load at the end or multiple loads, along a multi-drop line, look like capacitors (IC input capacitance). These delay the rising edge and reflect back down the transmission line as near-end cross- talk (NEXT), creating "potholes" in the signal. Howard has a unique solution to this crosstalk by deliberately creating an equal and opposite reflection to neutralize the pothole. This tech- nique is ideal for the fly-by routing of multi- drop loads. I will definitely implement this strategy on my next DDR3/4 design. Types of PCB connectors and the use of proper grounding, to alleviate high inductance and EMI, are also discussed. Vias are basically a type of connector that transfers a signal from layer-to-layer in a multilayer PCB. So, ground stitching vias need to be utilized in order to reduce loop inductance. Removal of unused inner layer via pads reduces capacitance. Dan- gling vias create reflections above 1GHz but can be improved by back-drilling, using truncated vias or special antipad clearances. Howard's ap- proach for differential vias is unique. Differential signaling also has its challenges and both differential-mode and common-mode noise are discussed in detail. Reducing common mode noise is the key to good differential de- sign. There are also special issues with crosstalk, MASTERING "BLACK MAGIC" WITH HOWARD JOHNSON'S SEMINARS " Transmission lines can be broken down into stages, when the signal and return path are a short fraction of the rise time away from the reference plane. "