Basics of Good EMI/EMC Design

Click here to go to our page on magnetic materials

Click here to go to our page on permeability

Click here to go to our main page on skin depth

Click here to visit the largest online directory of EMC Test Labs

Our content on this topic is pretty poor... but the University of Missouri-Rolla site that has some great info on EMC:

This collection of insights is attributed to "John Electron".

Put everything in gold plated mu-metal boxes.

There must be an intimate connection between the box and the lid; weld them together. If you must economize on welding rods, then do a spot-weld every two inches or ¼ Lambda; whichever is smaller.

Don’t design anything electrical; use steam.

Don’t design anything with wires that connect to anything else. Use semaphores, hydraulics, or string. If you must use electricity, ground everything; including the signal lines.

Use metal connectors, metal backshells, metal inserts, metal switches, metal fasteners, etc. Solder them all together with 1½” silver-plated braided wire and ground everything.

Put ferrites on everything; even the string. The bigger the ferrite the better.

Don’t use motors; use mice in AlBeMet cage wheels and put ground straps on the mice.

Seal the mice in at the time of manufacture and subject the unit to environmental stress screening.

Don’t have ventilation holes; they let RF escape. It’s bad for the mice, but what the heck; there’s no MIL-STD on them.

Bond everything to the ground plane, including portable devices.

Copper or aluminum tape is no good; use solid gold clamps. You can always hock them for cash if the test fails and you are laid off.

If you must utilize electrical power supplies, use only large marine batteries; even for 3-phase AC - they look neat in the photos sitting on the floor in a wye. Use gold bars for connections; same reason as above. Drive a 12 foot copper stake through one of the batteries and the floor of the Faraday cage for a really solid earth ground.

Below, a "student" of John Electron describes his experiences in developing some hardware:

"Using your excellent guidelines, we developed a 3-phase power source consisting of three wye-connected marine batteries, interconnected by string encased in long ferrite tubes. To ground the batteries, we improved on your scheme by using, instead of copper, a gold stake through the battery into the ground. It was very difficult pounding it in without bending it, but anything worth doing is bound to take a little time. The results of the RE and CE testing were truly astonishing! ZERO emissions! Actually, it's a bit inconclusive, because we unplugged all the test equipment just to be sure. But we're still VERY encouraged by these results.

On the susceptibility side, however, we did eventually get to a level at which the ferrite burned through the string. I was thinking maybe we need some mice in AlBeMet cage wheels to pump coolant to the ferrite tubes. Perhaps surgical tubing running through corrugated cardboard fins, whaddaya think?"

What about skin depth?

Skin depth plays a big part in EMI shielding. If you have just one skin depth of metal. 36% of the electric field will penetrate it. Use a minimum of five skin skin depths to attenuate EMI by 40 dB.

Here the Microwaves101 Professor discusses thickness of metal shield with respect to its effectiveness, comparing metal foil to mesh at 40 GHz.

I think metal foil is the right approach - a mesh would let some of the RF through the small openings, even if they're less than a half-wavelength across.

Even at 40 GHz, skin depth in a non-ferromagnetic metal like aluminum is much less than the thickness of aluminum foil.  That says you should be able to get the 20 - 40 dB of shielding with a piece of foil.  And even at 1 GHz, skin depth is an eighth of a mil in aluminum.  So one mil thickness of aluminum should provide enough shielding.

One caveat - I'm told that the effectiveness of shielding depends on the angle of incidence of the E-field on the metal.


Author : Unknown Editor