AltiumLive 2022: Stopping Your EMC Problems Before They Start

Reading time ( words)

Todd H. Hubing has been working with EMC for decades. He began as an engineer at IBM, then served as a professor at the EMC lab at the University of Missouri-Rolla (now Missouri Science and Technology), and later moved to Clemson, where he founded the Vehicular Electronics Library. Now, Todd is focusing on his most recent endeavor, LearnEMC.

At AltiumLive 2022, Todd shares his thoughts on how to avoid EMC issues before they have a chance to wreak havoc on your board, simply by making smart decisions early in the design process. As he points out in this interview, this is much better than waiting until you have EMC issues later on and possibly redesigning a board until you pass EMC.

In this interview, Todd explains some of the common errors he sees that lead to EMC troubles later in the design cycle, as well as how to avoid them, and the need for designers to educate themselves on EMC design techniques.

Andy Shaughnessy: How's it going, Todd?

Todd Hubing: It's going well, thanks.

Shaughnessy: Todd, why don't you give us a bit of your background, since you’ve got a long EMC history. Tell us about yourself, your previous work, and now your current company.

Hubing: I got started in EMC 40 years ago, in 1982. I went to work for IBM and was in their EMC group for about seven years. Then I accepted a faculty position at the University of Missouri-Rolla. While I was there, I teamed up with Tom Van Doren and then later, Jim Drewniak and Dick DuBroff. We had an EMC laboratory, and we basically did our research looking at real problems with real products, and trying to understand what was happening, how to model it, and what to do about it.

I was there for 17 years, and then I accepted an endowed professorship at Clemson University, and continued to work in EMC, mostly automotive. I was with CU-ICAR (Clemson University International Center for Automotive Research) while I was there. I retired from Clemson in 2015, and since then I’ve been doing consulting and teaching short courses, and a couple of college courses.

Shaughnessy: Sure, and some of us remember that when you and Tom Van Doren were at the EMC lab at University of Missouri-Rolla, you were doing this EMC work that was real life, not typical academia, if you know what I mean.

Hubing: That's right. When I was at IBM, we had a lot of university faculty come work with us. And I have to admit, I had a relatively poor impression of university faculty because they were so theoretical and they would always change the problem into something that could be solved, but they changed it into something that wasn't our problem anymore. So, yes, going to UMR was my opportunity to focus on what was really happening.

Shaughnessy: And now you’re getting to share a lot of that with the attendees at AltiumLive. And I see you’ve got a presentation on how layout can create more problems with the EMC. Give us a rundown.

Hubing: I decided to focus on "Common PCB Layout Mistakes That Cause EMC Compliance Failures." And of course we could talk all day about that, but I decided to focus not so much on the obvious mistakes. Most of the people that I work with in design, as I’m consulting, are not making the obvious mistakes, the things that are the subject of lots of courses and presentations, but I am seeing a lot of the same mistakes over and over again. So, these are not necessarily the most common mistakes that an absolute beginner would make, but they’re the most common mistakes I’m seeing in the boards that come to me.

Shaughnessy: Give us some examples of these common errors that can have big effects on EMC.

Hubing: The one that I start out with, the main one, is grounding problems. As much information as there is about grounding out there, people still tend to get it wrong. And it's almost like the more you’ve heard about grounding, the more likely you are to get it wrong. People are confusing ground with current return, which are both very important concepts.

I will often see products where they’ve gone to great lengths to isolate themselves from a ground that they really should be connecting to. And the opposite can happen too, where they should be isolated, but they haven't properly isolated. And it's really not that complicated. And so I’m hoping to kind of clear things up a little bit with that.

Shaughnessy: Do you have a few horror stories in your presentation?

Hubing: In the presentation, sure, I have examples. Of course, I can't show examples from my consulting work, but I was able to find some examples of good and bad designs in publicly available board layouts.

Shaughnessy: So what advice would you give to somebody who is grappling with EMC problems?

Hubing: Well, I think the best advice is you’ve got to educate yourself. There's a lot of very good information available on the internet, and like anything else, there's also a lot of really bad information on the internet. But the more you’re exposed to, the more questions you ask, the more you’re going to become comfortable with this. And then of course, I’m probably biased, but I would highly recommend taking EMC courses.

There are many different short courses out there that people can take. And that gives you an opportunity to interact with an instructor and ask questions like, "Why is this the way it is?" or "You’re saying this and I heard from somebody else that it was this other way, so which is correct and why?" Do everything you can to educate yourself.

Shaughnessy: What are some of the takeaways that an attendee would have from your presentation? What do you hope they walk away with?

Hubing: Well, I’m going to talk about six common layout mistakes, and I’m hoping that they’ll be less likely to make these mistakes on their next design. We will see, but that's what I would like attendees to take away.

Poor grounding is the most common mistake, but inadvertently bypassing the filter is a close second. We see people design really nice filters, and then they provide a path for the noise to go right around it. Those coupling paths are not always obvious, but if you systematically look, they pop right out.

We’ll talk about power converter layout as well. A well-designed power converter really shouldn't be a noise problem. The new converters are actually very good compared to what we had 20 years ago, but we are still seeing people who are getting power converter noise in their radiated emissions or conducted emissions. That's almost always due to a poor layout. And that's almost always because they followed the manufacturer's recommendations, which are often very poor. So we’ll talk a bit about good versus bad power converter layout and show some examples.

Power bus decoupling was an issue back when we first met 20 years ago, and it's still an issue. Proper techniques for power bus decoupling are different for different kinds of products. People will often take a class or watch a seminar and learn something that applies very well to some kinds of products, but it doesn't apply to their product. When they do it in their product, they end up with power bus noise. Your digital devices shouldn't put noise on the power bus that causes that causes an EMC problem.

Another problem that's become really important more recently happens when people are paralleling large and small ceramic capacitors. It's probably never been a very good idea, but people got away with it in the past because the large one was an electrolytic capacitor and the small one was a ceramic capacitor. But capacitor technology has gotten to the point now where you can actually get some very large ceramic capacitors. They have a low ESR and a large one will resonate with a small one. And so we’re seeing more products where everything looks great, except for one narrow band spike typically above a gigahertz.

It appears like it comes out of nowhere and it's a resonance that they inadvertently created. And there's no reason for that to happen. We really shouldn't be putting these large values in parallel with small values ever. There's no need to anymore.

The last common mistake is that people are putting common mode chokes on boards that shouldn't have a common mode choke. And it's hard to say exactly why that's happening. But some of that, I think, is due to app notes and guidance coming from filter component manufacturers, the people selling these common-mode chokes.

There are some situations where common-mode chokes, of course, are perfectly appropriate, but we’re seeing a lot of designs where they’ve got them in places where they really shouldn't have them. They’re causing a problem rather than solving a problem.

Shaughnessy: It seems like everybody was talking about common-mode five or six years ago.

Hubing: Well, certainly common-mode currents are the source of radiated emissions, yes. That's true. But, whether it's appropriate to use a common mode choke is a different story.

Shaughnessy: Interesting. And you’ve done a lot of work with automotive? I guess that's likely to increase with EVs and autonomous vehicles, plus infotainment systems in general.

Hubing: Yes. I would say at least half of the boards that I’m reviewing are automotive boards. They’re the ones usually getting the grounding wrong. And I’m a bit concerned because the grounding problem is not only an EMC problem. If it's a grounding problem on a steering control module or a brake control module, that's a safety problem as well. So I see that as a large problem in the industry. It's not that we don't know what to do. It's that people are not always doing what they should do.

Shaughnessy: Right. Everybody's been talking about signal integrity for 20 years, but some designers still look at designing for EMC as black magic.

Hubing: I would agree. There's a lot of apprehension about EMC, but there is a lot of confusion too. Really, if you back up 20 years, the things that worked 20 years ago are not working today. They’re just not the right things to do anymore, and that's because things have changed. The circuits are different. The frequencies that we have problems with are different and even some of the tests are different. And yet there's a reluctance to adapt. If somebody's been working in the industry for more than 20 years, they feel like they know what works and what doesn't work. And I think that's a large part of the problem.

I personally noticed that when I’m working with a brand-new company, a startup, they generally get things right pretty quickly with EMC. Often the worst problems are either with the older companies or OEMs that are supplying a product to the older, more established companies.

Shaughnessy: Does it seem like a lot of OEMs don't have the subject matter experts on staff anymore, so they’ll just let somebody like you, a third party, deal with the EMC stuff?

Hubing: That is often how we end up winning that argument. The customer will bring in the OEM and myself and basically get us in the same online meeting to negotiate what's going to be done. And in the end we ask the question, "So, is the OEM guaranteeing that this is going to comply? Or am I guaranteeing that it's going to comply?" I guarantee that it will comply if they design it right, and the OEM is usually not willing to do that.

So generally, when we win that argument, that's the way we win it, although we don't always win. Sometimes the OEM insists on doing things in a way that, in my opinion, are not necessarily the safest or most reliable way, but they are the customer.

Shaughnessy: Right. And they’ve often designed it for respins, and they’ve even budgeted for respins. Do you see that happening a lot?

Hubing: Absolutely. And that's a big point I make in my classes. When you keep redesigning it until you pass the test, you’ve designed a product to pass a test—not to be a good product, and not to comply when it's out in the field. And there is a lot of that happening, not just in automotive, but in most industries. People will get into the test chamber, redesign the product until it passes, and then they have problems in the field. Or if it's component, they have problems when they put that component in a system.

Shaughnessy: Yeah. Cool. Well, it sounds like you have a really interesting job, and it sounds like you’re having fun.

Hubing: It is lots of fun. I love it.

Shaughnessy: Thanks for speaking with me.

Hubing: Thank you, Andy.

View Todd Hubing's presentation "Common PCB Layout Mistakes that Cause EMC Compliance Failures" below.