Let’s all come out of our socially distanced, online existence, away from our labs and offices, better off than when we started this stressful experience. We should take advantage of the wealth of new content recently added to the web to accelerate us up the learning curve.
We’ve all noticed the marked increase in the number of on-line webinars available. But with quantity does not necessarily come quality. I’ve spent many hours watching some of these offerings and have culled out the ones I think are particularly worth watching. This new regular column will highlight one or two webinars worth your time.
This week, I am featuring two events.
I watched the keynote address by Todd Hubing at the EMC Live Automotive EMC online conference. Readers of my column know I am a big fan of Todd Hubing. When he presents, chances are, we will all learn something new.
The electronic systems content in all vehicles is accelerating. Figure 1 is an example of the electronic systems in a typical car. Hubing’s main theme of his keynote is “The way we design automotive electronic systems is about to change radically.” The driving force (pun intended) is the migration to autonomous vehicles.
The problem is not just that there is exponential growth in the electronic content in vehicles, it’s that the reliability of the electronics must dramatically improve with the adoption of driverless cars.
In 2018, there were about 100 vehicles deaths a day in the US, and many more around the world. We accept this level of risk as part of the cost of our mobile culture. When there is no driver to assign blame, how many deaths are we willing to accept? Even 1 death a day would probably cause a worldwide shut down of all autonomous vehicles. There will be no driver to blame for accidents with autonomous vehicles.
The automotive industry must develop processes to increase the total reliability of systems. We can no longer optimize product design just to pass a compliance test. “It is not possible to ensure the safety of electronic systems in automobiles by testing alone. If you are in the automotive EMC business, you are going to have to learn to adapt quickly or you are going to be left behind,” Hubing says.
His final advice to automotive engineers is to keep learning. There is a lot of bad advice out there, with some poor design choices still being used. “This has to stop. It’s dangerous, expensive and time consuming.”
Todd’s perspective on the future of the automotive electronics industry will change the way you think about cars.
My second recommend for the week is Gustavo Blando, presenting “DC Block Capacitors Location (Does it matter?)” This is part of the series of gEEk spEEk webinars presented by Samtec. If you have not heard of this webinar series by the Samtec experts, you definitely want to check out the whole series.
In a high-speed serial link, does the location of a DC blocking capacitor matter? Should it be located closer to the TX or the RX? This is a hot topic of debate even among experts. Gus suggests some of the confusion arises from the different perspective of thinking of the problem in the time domain or the frequency domain.
The fundamental root cause of the problem DC capacitors create is as an impedance discontinuity due to vias and mounting pads. This question can really be expanded to the location of any discontinuity: does it matter? A typical geometry is shown in Figure 3.
The basic controversy is that when viewed in the frequency domain, it does not matter where the DC blocking capacitor is located because of the basic property of reciprocity in S-parameters, that S12 = S21. But, when viewed as the reflected signal in the time domain, the discontinuity will look bigger when closer to the TX than if located near the RX. There is less loss in the line for the reflected signal to get through the closer it is to the TX.
The real answer to this question, does location matter, is two-fold. To first order, it does not matter where the capacitor is located. You get the same insertion loss no matter where it is. But to second order, if there are discontinuities at the TX and the RX, or elsewhere in the channel, the presence of the discontinuity between them will create the possibility of a series resonance. These frequency dips in the insertion loss will depend on the spacing between the discontinuity and the TX and RX.
This second order feature is a little tricky to analyze because it depends on the spectral components of the signal, the specific frequency response of the series resonances and the type of equalization being used. Gus introduces the effective return loss as one metric to explore this interaction.
The analysis points out that for a specific data rate, a specific channel length and specific discontinuities, there is a slight difference in the effective return loss, as the location of the discontinuity moves around.
Generally, placing the capacitor in the center gives a slightly better return loss. But, be careful interpreting the relative benefits. Gus uses auto scaling on his plots and makes the difference on the plots look huge, when in fact they are a just a few percent maximum variation.
The final answer is: if you are worrying about the location, before you spend time to do a lot of detailed channel effective return loss simulation, it is worth spending the time to reduce the magnitude of the discontinuity by carefully sculpting the fringe fields to make the structure, whether a DC blocking capacitor or a via, more transparent.
Stay tuned for the next installment!