Design007 Magazine

PCBD-June2016

Issue link: https://iconnect007.uberflip.com/i/691539

Contents of this Issue

Navigation

Page 57 of 89

58 The PCB Design Magazine • June 2016 strangely enough, he is always the one to give bad feedback, probably because he didn't learn anything. But I guarantee that you will be on the edge of your seat throughout Howard's entire seminar series. His dynamic teaching style ensures you feel like you are not just a part of the audience—you are actually partici- pating in the demonstrations. The picture that Howard paints leaves a lasting impression on how electro- magnetic fields propagate and how they induce voltages and current (crosstalk) into nearby signals. The following is a section-by-section dis- cussion of the course contents. 1. High-Speed Digital Design Engineers and PCB designers need to understand electromagnetic theory, appreciate how coupling oc- curs and why energy moves to un- intended, sensitive parts of the circuits. A logic schematic diagram masks details crucial to the operation of unintentional signal pathways vi- tal to your understanding of signal performance, crosstalk and EMI. To realize these factors, one must uncover the hidden schematic, operating behind the logic diagram, to reveal the parasitic elements that affect the circuit. These parasitics are invisible to the uninitiated, but become very clear once skillfully explained in detail. You will gain new insight into what really happens in the circuitry. Also, understanding the frequency band that really matters for digital design is very im- portant. Traditionally, we used 0.35/Tr (where Tr is the rise time in ps) for the upper bandwidth. However, Howard recommends using an upper knee frequency of 0.5/Tr, which forms a crude, but useful, translation between time and fre- quency domains. So for instance, if the rise time is 500ps, which is typical these days, then the upper bandwidth is actually 1GHz regardless of the clock frequency. Furthermore, the constant improvement in the IC process reduces die size which speeds-up the rising edge. This in turn pushes the knee frequency up, causing signal overshooting and ringing. Howard also cites the difference between "lumped element" and "distributed systems." In Circuit Theory 101, we are taught using lumped element assumptions where the system delay is much less than the signal rise time. However, Howard points out that when the system delay is much larger than the signal rise time, a more complex distributed analysis is required. This system is characterized by distributed delay and reflections; this is the real world of high-speed design. Here, capacitance draws surge current causing reflections, inductance causes ground (supply) bounce and noise, mutual capacitance causes crosstalk between high impedance cir- cuits and mutual inductance produces crosstalk in connectors especially where the layout is questionable. I was impressed by the way that Howard ad- dresses bidirectional signal terminations. I have simulated the position of a series resistor on a bidirectional data trace and it doesn't make any difference whether the resistor is placed at ei- ther end or in the middle. But, having a resistor at both ends is an elegant solution, as the resis- tor and input capacitance, of the tri-state load, basically form an AC termination—I would not have thought of that! MASTERING "BLACK MAGIC" WITH HOWARD JOHNSON'S SEMINARS Figure 2: Dr. Johnson measuring crosstalk in his giant scale BGA model.

Articles in this issue

Archives of this issue

view archives of Design007 Magazine - PCBD-June2016