I think one of the keys things is Dr. Ralph Morrison's assertion that we need to get engineers to leave the lumped parameter idea of circuits and think in terms of fields. Maxwell really did get it right. Too many engineers are thinking lumped systems and circuits; not EM fields and transmission lines. These are the folks who think it is black magic.
For those who received an 'A' in EM fields class, and still can apply it, successfully designing high speed circuits and passing EMI tests is just a matter of fighting management to start doing it on day one.
well, if you start a design from the beginning with EMI requirements in your mind - in that case you are a wolf in the sheep's skin one RF ingenieur who designing not RF circuits - you will not have big issue to met FCC, but unfortunately RF is considered here as black magic, and in the last half century digital only was "inn". After clock rates got higher and rise times shorter, the devil is here and that is RF, you need to deal with it with RF design practices, which are opposite, to general assumption not black magic, but a dfferent set of logic
All the fundamentals of EMI are in classic undergraduate and graduate electromagnetics.
The things they don't teach are pragmatic problems like
- coils couple, so how to estimate the coupling and minimize it
- when is a capacitor not a capacitor any longer
- a resistive wire is bisected by a capacitor with some series resistance to ground. Given the wire resistance and capacitor to series resistance ratio, what value of cap will provide X amount of decoupling at Y frequency
- electrons accelerating along a wire radiate, what can be done to minimize the radiation to keep the coupling to some acceptable level
- what is the inductance between two points on a ground plane and at what frequency does this inductance make the ground connection no longer a good ground
It is all very practical applications of classical electrodynamics. It is just easier to teach that components are contained things with a particular characteristic and here is how to analyze circuits, now go build something useful (never mind about the man named Maxwell behind the curtains).
Now we can no longer ignore these effects of making circuit faster and denser.
Even the random bits of advice on decoupling digital logic is not always fully understood or explained to the engineer doing the board design. I'm sure you've experienced that puzzled look you get when you tell someone you need a couple different types of bypass caps, and they wonder why one is a much smaller value than the other, and how that could possibly help!
While this can make it easier to pass FCC Part 15 testing when good EMI design practices were not followed in the first place, all that is happening is the RF is getting spread around and made to appear "thinner" within a narrow bandwidth. Similar to what happens when one steps in something typically found in a cow pasture - it doesn't go away, just spreads out more and sticks to the bottom of the boot.
One possible solution is to use a Low EMI Spread Spectrum Clock Oscillator. Mercury United Electronics offers various drop-in package sizes and the average EMI Reduction is about -12dB. It has proven to be an excellent long term fix. http://www.MercuryUnited.com
EMI was never taught in school back when I was getting my EE degree. Oh, I was given random bits of advice on decoupling digital logic, but that was as far as it got. My first real introduction to EMI was on my first assignment. It was an eye opener to find out we had to 'waste' precious I/O pins by supplying differential signals from chassis to chassis in a military system, that twisted pair wire would be much, much more immune to noise than single ended wire, and clock signals had to be terminated to prevent double clocking and clock skews.
Decades of experience and a few week-long training courses later I now instinctively consider these effects so early in a design that rarely do these age-old problems creep in anymore. Murphy's law has taught me to leave nothing to chance anymore. Everything, and I mean EVERYTHING must be considered before embarking on a design.
Sounds like something I built for a friend - an electric fence. Used an old valve radio transformer. On one half cycle the HV winding charged a cap of a few uF. On the other half cycle the filament winding triggered a SCR that put the charged HV cap across the primary of a car ignition coil. We could get one inch sparks out of it - in fact it flashed over a couple of ignition coils before we started using only half the centre-tapped HV winding. It also caused lots of problems with the neighbourhood TVs.....
My friend was called Frank, so we called it.....Frank's Zapper.
I'm horrified now at the crudeness of it, and have often thought about how I'd redesign it to give off less EMI. But valve radio transformers are not as common as when I was a kid, unfortunately...
I wasn't a ham, Daryl, but my teenage Tesla coil wiped out the whole street's TV reception!
EMI engineering is actually really basic stuff; the problem has been bad teaching over the years, and engineers have been encouraged to ignore the advice of their RF (and in many cases pro-audio) colleagues. But the materials are there now, you can read, study, understand and apply EMI techniques just like all the others you've learned.
A Book For All Reasons Bernard Cole3 comments Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...