Sometimes what you think you know just isn’t so, and the reality is more than you care to accept.
To keep up with new components, architectures, materials, design techniques, and various developments, I read a lot of technical articles written by credible authors. Usually, they are both informative and straightforward.
But I recently can across an article that both scared and worried me. No, it was not a sensationalist article about how even a low voltage can be dangerous (it actually can be, under the right circumstances), how electromagnetic radiation from smartphones will cause all sorts of physical ailments (no credible studies support that contention), or how we are doomed since we are running out of something (could well be accurate, but there’s little I can do about it right now).
Instead, it was a very serious, fully detailed, and highly credible article, Getting the Best EMC from Shielded Cables Up to 2.8 GHz, Part 1; my fear factor was amplified when I got to the second part, Getting the Best EMC from Shielded Cables Up to 2.8 GHz, Part 2. As the title indicates, most articles dealt with shielded cables, although they did touch on non-cable shielding in a small way.
Here’s why I was scared by what I read. Cables and their shields are very important topics — no news there. Consequently, there are many guidelines and so-called “rules of thumb” about how and where to terminate the ground shield in these cables. Unfortunately, not only are these guidelines often in conflict, but the right or best answer may depend on the specifics of the arrangement.
The basic rules I’d learned to follow are:
- terminate (ground) both ends of the shield
- terminate one end only, at the source
- terminate one end only, at the receiver
In other words, we have contradictory guidelines. The article’s author makes the right answer very clear; for effective shielding into the gigahertz range; you always need to terminate the shield at both ends, with full 360-degree coverage. (The rules are somewhat more flexible at much lower audio frequencies.)
The author’s statements were not just based on his hands-on personal experience, however useful that might seem. Instead, both parts of the article were replete with diagrams of test setups and results using sophisticated EMC sniffers and other test equipment in various configurations. You may not like what he said, but it would be difficult to argue with the unambiguous conclusions.
OK, so maybe what I was taught or learned about which end of the cable shield to terminate was wrong; it was somewhat of a shock, but I could get over that. Next, the article showed, again with detailed test results, how the easy, widely used short “pigtail” termination for the shield was often ineffective (Figure 1). Even if only a few millimeters long, its small inductance hurt its performance at higher frequencies and could negate much of the shield’s performance (Figure 2). Instead, what is needed is 360° physical coverage at the shield termination.
Then the news got even worse. That innocuous pigtail terminal could actually act as a radiator of electromagnetic energy and create more EMI rather than merely being ineffective at attenuating it. Now I was really scared — I’d been doing crisp, short pigtail terminations for many years, admittedly only up to a few hundred megahertz, but that still might be a high-enough frequency to be worrisome.
This pigtail reality is known to experienced EEs working at the higher frequencies, but seeing it presented so dramatically and unambiguously really hit hard. Reading through this article carefully, I had that sinking feeling that many of the guidelines I followed or assumed made sense about cable shields and termination were wrong, and that’s a creepy feeling. It’s the written version of being told by someone who really knows their stuff, “I hate to tell you this, but you are doing it all wrong.”
After all, what else did I think I knew but about which I was actually wrong? Or was it that as circuits moved up the RF spectrum, many of the rules and assumptions I subconsciously used were no longer applicable? It’s very worrisome when the assumptions you build up over the years are no longer valid.
This brings me to a related concern. Today’s highly advanced, dense, high-speed, high-performance gigahertz-class designs are simultaneously complex and unforgiving. So many things have to be done right, and rules and guidelines for doing them properly often conflict with each other. It’s like this cartoon in the 1952 book “How to Get to First Base: A Picture Book of Baseball” by Marc Simont and Red Smith (Figure 3) (and no, I didn’t buy it when new — I picked it up at a yard sale a few years ago).
.
(Incidentally, the book also has another clever cartoon that I keep in mind when I see those “talking heads”, know-it-all commentators on politics or sports (Figure 4). But that’s a story for another day.)
The engineer’s dilemma is that are so many marketing, performance, and regulatory mandates pulling in different directions, and pretty soon, some (or many) of these are in conflict. While there may be flexibility with some of them, there’s little or none for others. That’s especially so for industry standards or regulatory mandates. After all, you can’t “almost” meet an IEEE 802.x standard or “not quite” meet an IEC mandate for minimum creepage and clearance distances.
I think I may have to re-assess what I think I know well before proceeding too far into the project swamp next time. How designers work through, over, and around the various requirements is both a skill and an art. The requirement that shielding and terminations be done a certain way to ensure it actually provides shielding just adds to the list.
Related EE World Content
With so many mandates, can successful designs still be achieved?
Why I’m divided on “Right to Repair”
The connector is often the EMI problem
Rudiments of radiated EMI/EMC
What are some common EMI/EMC tests?
External References
- In Compliance, Keith Armstrong, “Getting the Best EMC from Shielded Cables Up to 2.8 GHz, Part 1”
- In Compliance, Keith Armstrong, “Getting the Best EMC from Shielded Cables Up to 2.8 GHz, Part 2”
- Dana Bergey and Nathan Altland, DesignCON 2008, “EMI Shielding of Cable Assemblies”
- Kenneth Wyatt, Interference Technology, “HDMI Cables and EMI”
What was presented in this article should be intuitive, which is that the shielding required depends primarily on the frequencies involved.
Shielding serves dual purposes, both to keep undesired signals out and to keep the internal signals in.
These are rather different requirements that usually use similar materials and arrangements. For keeping signals in, the shielding usually is effectively a portion of the transmission line, participating in the actual signal path. That is why the continuous conductive path is important. Transmission line theory and equations verify this.
Keeping undesired signals out is more complex because the realm of undesired signals is much broader, and so the acceptable gaps in shielding must be understood based on the anticipated interference frequencies, and the origin of the interfering undesired signals.
Unfortunately, as shown in the image of the HDMI cable, cost of implementation often outweighs the need for effectiveness, and the implementation is only symbolic of actual shielding. This is common in consumer products.
In addition, it is often true that just because a writer has a literate style does not mean that they have an adequate understanding of the subject, or any applicable information at all.