Dear Peter Sorry to be so late replying. One cannot in general predict the fields very close to an item of equipment from the fields measured at 10 metres. But maybe in your particular case you might be able to have some confidence in the calculation you used, especially since you don't intend to put the items any closet than 1 metre.
Take a look at the useful Appendix C in Tim WIlliam's book "EMC for Product Designers" (Third edition, Newnes 2001, ISBN 0-7506-4930-5). On page 329 and 330 you'll see the full set of field equations for the emissions from a small current element. All networks of real conductors can have their emissions predicted from these equations, if you have enough computing power and enough time. As you will see, there are terms in each of the expressions for the emissions of electric and magnetic fields which relating the reciprocals of r, r squared, and r cubed, r being the distance from the tiny element. When we measure at 10 meters we generally only measure the "reciprocal of r" terms, because all the others have decayed away to nothing (or very little, anyway). Mostly, this is common-mode noise. The "reciprocal of r squared" and "reciprocal of r cubed" terms are more usually associated with differential mode emissions and with induction - both magnetic and capacitive coupling. The point at which the "reciprocal of r" terms start to dominate over the others is about one-sixth of a wavelength, so it is frequency dependant. The region where the 1/r terms dominates is known as the far field, and where the other terms dominate is called the near field. Generally speaking, the near field effects tend to increase the coupled levels of interfering signals into the victim circuit, although phase cancellation can and does occur in any real (and therefore complex) product with reflecting surfaces nearby. At 30MHz, the near-field far-field boundary is about 1.5 metres, so one might expect to get anomalous results by simply factoring the 10 metre measurement by the ratio of the distances when one gets closer in than 1.5 metres. If you are not going to place the items any closer than 1 meter, I reckon you can use the simple 'ratio of the distances' multiplier with reasonable accuracy above 50MHz. But there is no way to predict the near field intensity from the far field measurements, unless you measure the amplitude and the phase of the emissions over the surface of a sphere, and know the radiating structure of the product, and once again have a large and powerful computer and lots of time. Regards, Keith Armstrong > on 1/9/02 11:18 PM, [email protected] at [email protected] > wrote: > > > > > Hi Folks. > > > > At the moment I'm examining as a generic case, the potential for > > interference with Item A (tested to comply with 3V/m radiated immunity) > > caused by Item B (tested to comply with FCC or EN Class A [industrial] > > emissions). > > > > Using simple inverse distance ( E2 = E1 x d1/d2 ) extrapolation (assuming > > dominant interfering frequencies will be in the far field), I come up with > > a required separation distance of approximately 75cm to ensure the 3V/m > > immunity limit of Item A isn't exceeded by the 47dBuV/m emissions from > Item > > B. > > > > Based on this, I'd expect then the risk for EMC problems should be > > relatively low provided: > > 1. A minimum separation of 1m was used between Items A & B; > > 2. No direct interconnection of A to B via cables; > > 3. Use of a mains filter and/or separate power supply sources for A & B; > > 4. The nature of Item B is such that no significant low (eg.power) > > frequency magnetic fields are emitted; > > > > Does anyone have any experience to suggest that the minimum separation of > > 1m under theses conditions would not be adequate? > > > > Thanks, > > > > Peter Poulos > > Design Engineer > > Foxboro Transportation > > (Invensys Rail Systems Australia) >
