Editor’s note: This question was asked in response to Interference Technology’s recent webinar by Keith Armstrong. To view the webinar, click here.
Q1. What did it take to fix the supercomputer immunity?
A1. Can’t remember now, long time ago (22 years or so). Did I say immunity? I think it was actually emissions. But I do remember that it wasn’t very difficult, really, just down to basic poor design of interconnections (they were putting high data rates down flat ribbon cables).
Like most EMC problems, it’s not the application that is important, it’s just down to basic EMC design techniques, which is the point I was trying to make in the webinar.
The problem is that good EMC design techniques are necessary for cost/time-effectiveness, but generally taught at universities and so everyone makes the same mistakes unless their education in this necessary area of design is provided after graduating, usually by their employer.
Q2. Can an LRU (Line Replaceable unit) chassi be grounded through a Pin ground?
A2. As I understand it, this chassis ground is for safety only, and possibly also for providing a path for surge currents (e.g. from lightning, having quite low frequencies).
Don’t confuse safety grounding with the RF Reference. RF-bonding and the RF Reference may use elements of the safety grounding system (most typically the chassis, frame, enclosure, etc.) and this is partly why the confusion has arisen.
As you know, I strongly recommend never using the words ‘grounding’ or ‘earthing’ in connection with EMI or EMC, because they just cause confusion with safety grounding and safety earthing – see my lengthy reply on this from a previous webinar in this series.
Q3. How do you filter out harmonics in a system that’s composed of different supplier units?
A3. This question has nothing to do with the webinar material. There are passive and active filter units from many suppliers that can do what you want. If you want more discussion, contact me directly.
Q4.When intra-EMC is served, will inter-EMC be satisfied too ?
A4. Well, one can always find exceptions, but in general if we achieve intra-EMC by using good EMC design practices from the start of a project, in circuit design, component choice, PCB design and layout, etc., and with a view to achieving the inter-EMC specifications required, then if inter-EMC is not completely complied with, at least it is made much easier, quicker and less costly overall.
It is always the case that leaving EMC to the end of a project adds unpredictable delays and costs and suffers from unquantifiable financial risks – as a result it is always bad project management.
Doing good EMC design from project start takes care of SI and PI, which is why it saves time and cost overall. Even if there were no EMC standards to comply with, good EMC design would be necessary to save time and money and improve competitiveness whilst reducing financial risks!
Q5. Some OEMs require electrical isolation of a few 1000 volts between high voltage return (HV GND) and low voltage return (LV GND) and this requires a slot or split in the return plane. I know solid plane is recommended, but how do we address this functional requirement of electrical isolation and yet design the proper return plane?
A5. This is typical of auto, marine and aerospace because before anyone had invented radio or electronics they had got into the bad EMC habit of using their chassis as their power returns, to save the cost and weight of having return conductors. Now it often causes big problems for EMC. Given that they use the chassis for battery return currents for high-power electrical systems, they are concerned that direct connections to the chassis can allow heavy power currents to flow through the PCBs and fry them, even set them on fire. The typical solution for electronics in such situations is to apply all of the following:
a) Never use the chassis for any of your electronic unit’s signal or power returns.
b) Use twisted-pair conductors (or triples, quads, etc. if necessary) for the sends and returns in each cable – not straight conductors.
c) Fit CM chokes on every power or signal input or output to help encourage the RF return currents to flow in the cables that you want them to, and not in the chassis.
d) To deal with the remaining stray currents that are being carried by the vehicle chassis, RF-bond your PCB 0V plane to the local chassis through multiple capacitors rated at the appropriate voltage, spread uniformly apart all around the perimeter of the plane. We usually find that 2 caps are better than 1, 4 caps (one at each corner) is better than 2, 8 caps (one in each corner and one in the middle of each side) is better than 4, and so on.
Q6. Is immunity testing with reverberation chamber more real world testing than the standard single tone testing according to IEC61000-4-3?
A6. This question is not related to the webinar material.
Two issues are being mixed up in this question – typical reverberation (or stirred-mode) chamber RF immunity testing also uses a single test frequency.
To deal with the basic question – for almost all applications, the EM environment is reverberant with even a single RF source causing radiation to imping on other items of equipment from multiple angles and with varying polarizations at the same time. So a reverb chamber is a better simulation of most real-life EM environments.
Q7. Given the ability of a conducting sheet is proportional to its skin effect, what is the relationship between skin effect and caustic atmosphere such as pollution over time.
A7. As the surface of the metal corrodes, it becomes less conductive and its skin depth increases. Some corrosion products are insulating (e.g. aluminum oxide) and some can be semi-conducting (e.g. copper oxide), so the real-life effects can be more complex than a skin-depth analysis could deal with alone. It is reasonable to say, however, that the surface conductivity, overall conductivity, and shielding effectiveness of a metal sheet are all degraded by the corrosion of its surfaces.
Q8. What is most important thing keep in mind to reduce interference frequency difference or power level?
A8. This question doesn’t seem to relate to the webinar material.
Suppression techniques such as filtering, shielding, surge/transient suppression, galvanic isolation, etc., can all be regarded as reducing the power level, but I don’t understand the question well enough to attempt any better reply.
I suggest that the questioner contacts me directly and describes what problem he or she is trying to solve.
Q9. Which of your books covers conducted immunity 10V/m testing?
A9. I have written about how to do ‘proper’ emissions and immunity testing according to many of the usual IEC standards, including conducted immunity to IEC 61000-4-6, in the free guides available from www.reo.co.uk/knowledgebase.
But if the question was about how to design to comply with conducted immunity tests, then the necessary design techniques are covered in my textbooks, which are only available from www.emcacademy.org/books.asp (and not from Amazon or any other distributor).
Q10. Should power harnesses be shielding?
A10. Another question that has nothing to do with the webinar material!
The answer is: it depends on whether whatever the power harness is connected to is sourcing excessive RF currents, or is susceptible to RF noise currents that can be coupled from the EMC environment (as common-mode or differential-mode currents in both cases).
Q11. Why would a board fail line conducted immunity tested at only 4MHz ?
A11. Yet another question that has nothing to do with the webinar material!
In general, if something fails an immunity test only at one specific frequency, it is probably because it is either operating at that frequency and so is suffering direct interference with its operation as a result; or else because it has a resonance at that frequency.
A 4MHz resonance could be caused by a Power Distribution resonance in the PCB itself or in its power supply (if separate), or it could be caused by the resonance of one or more long cables attached to the PCB.
But due to the lack of information I have no idea if any of the above actually applies to the board in question.
-Keith Armstrong
