Do you ever wonder what the most common EMC failures are so that you can (hopefully) avoid them? Well I do, so I brought together 5 EMC consultants who work hands on with EMC troubleshooting to see what their experiences have been. I asked them to outline what the one or two most common EMC failures are that they’re called in to help out with, and also to give us some tips for ways to avoid them.
Here’s the great advice that they had to give….
I’m a consultant and most of my work is in the military/aerospace segment. By far the most the most common EMC problem I see is radiated emissions (RE) outages followed closely by conducted emissions (CE) outages.
RE problems are most often caused by poor shield termination techniques or poor printed wiring board (PWB) routing. The root cause of these is often poor design strategy or, should I say, lack of a design strategy. For example, if your strategy is to minimize shielding in your interconnect cabling, you must focus strongly on PWB design to control RE. Alternatively, you can implement an excellent cable shielding design and get away with a little less attention to PWB design. Typically I find there is no strategy other than “just do a little of this or a little of that and we’ll hopefully pass the test.” For designs that must meet strict requirements like the Army Ground requirements of MIL STD 461, good shielding and good PWB design are essential.
A particular RE problem I often see is disregard of lines/signals that carry steady-state or infrequently changing signal such as those that indicate on/off, pass/fail, transmit/receive, or a slow/non-changing analog voltage (called ‘discrete’ inputs/outputs in the mil/aero vernacular), as well as passive interfaces like thermistors. The associated circuits may not generate energy that would cause an RE failure, but that does not mean they are not an EMC problem waiting to happen. You must ensure that such circuits are routed away from noisy signals that could couple energy onto the lines and cause RE problems. These circuits are usually easy to filter heavily with a few simple parts. Remember that in the low state, these signals may effectively couple the circuit signal return (‘ground’) out of the unit. This is likely the lowest impedance common mode noise source in the unit. A similar situation exists when the signals are in the high state and are effectively connected to a voltage supply rail.
The frequent causes of CE problems are poor control of return current paths (particularly common mode current) or poor filter high frequency response due to parasitics. Most power supply designers do a good job of controlling differential mode current and designing filters. Unfortunately, it is often assumed that a low-pass filter’s nearly ideal low and mid frequency response continues upward in frequency. Due primarily to parasitic capacitance, filter attenuation can drop off significantly at higher harmonics of switching frequencies (on the order of several megahertz). Common mode return currents are a bit more insidious. The designer must ensure that there is a low impedance return path for CM current that stays local to the switching element. This requires careful attention to where the currents are generated and where they will flow, remembering that current will always favor the path of lowest impedance. CE problems likely occur if that low impedance path is outside of the unit.
As a consultant, primarily working in the industrial, scientific and medical realm, I see radiated emissions as the number one issue. This is followed closely by ESD and radiated susceptibility. The issues causing radiated emissions are just as described by my colleagues above, however, I would add specifically that high speed digital signals crossing gaps in return planes are what keeps the bills paid here.
As for the immunity issues of ESD and radiated susceptibility, I believe these are cropping up more and more largely due to the fact that supply voltages to ICs have been decreasing from 5V to 3.3V to 1.8V to 1.2V, and even lower. Thus, the noise margins are substantially lower than in the past. Couple this with smaller die sizes and the proliferation of wireless, mobile phones, and other high-power communications systems and we’ve greatly increased the risk of interference and circuit disruption from nearby RF fields. The solutions are as described above.
Francesc Daura Luna
The 3 main issues that I encounter are:
#1 Radiated emissions. The solutions that I find usually involve good PCB design and good shielding at box and cable level.
#2 ESD problems. The main solutions for this involve good PCB design, good shielding at box level and using the correct immunity protection components.
#3 Conducted emissions. Main solutions: good PCB and switching power supply design. Correct filtering.
I would say in general it is Radiated Emissions. However, I do have clients who struggle with other issues more than RE problems. For example, those who make switching power supplies have conducted emissions issues. A few of my aerospace clients have concerns about lightning. Interestingly, even the clients who have radiated susceptibility tests in the thousands of V/m don’t seem to worry about radiated susceptibility testing. The people who have that issue are groups like medical companies, where they have to measure extremely small voltages (biological levels) and they get subjected to radiated levels 10 orders of magnitude greater than what they are trying to measure. Fortunately, the frequencies are different enough to allow some good filtering to occur.
Most of my clients (about 80%) are aerospace manufacturers. So that might account for my answers.
Systems Integration/EMC Engineer Sr. at Lockheed Martin
I am an EMC Engineer that has worked in a test environment for 9 years. All the information from the previous contributors are great but you could also look at EMI problems and solutions from a test stand point. A good percentage of problems in RE (radiated emissions), as everyone has pointed out, are issues related to test setup. These issues include but are not limited to, shields of cables being less than adequate (due to customers not providing actual cables because of size limitations etc.), bonding of connectors to chassis being well into the 100’s of milliohms if not higher due to rubber (non-EMI) gasketing for moisture purposes or non torqued connections, passing cables through waveguide tubes instead of creating patch panels with bulkhead connectors (support equipment noise couples in really easily with large setups, when using things like signal generators or amplifiers, make sure to take this into account. On the flip side a patch panel will also help when using sensitive measurement devices for data collection during radiated immunity. A scope or analyzer is going to do its job well and show you the transmit signal, so if you’re monitoring your device in the transmit frequency band be sure steps are taken to assure your test setup isn’t causing you to think your device is failing.) and standards knowledge (often times test setups require kluged cables or connectors in order to meet a test setup which may cause unwanted EMI).
All in all, a sit down with your local lab to find out facility requirements and how to optimize your chances of passing or at least a few pre-scans would assist in passing your test during formal scans.
Related Post: How to Do an EMC Design Review
Thanks very much to the contributors of this valuable information. If you want to contribute to the conversation, I’d love it if you could leave a comment below, or if you want to contribute a case study (anonymously), then just tell me about it using this short form here. Everyone benefits from this information, so sharing your experience will help us all to improve our knowledge and skills. If I get more input from EMC consultants or engineers, I may come back and add it to this article, or publish more similar articles.