Acceptable margin depends on what the measurement is for. Are you checking out a unit that is going to be sent to FCC Lab for sampling? Are you trying to qualify subassemblies or components from new vendors? Is this test another in a series of identical tests on identical equipment?
Most US and EU limits are just that, limits. Stay under them and you pass. Statistical analysis might show that to maintain a certain confidence level that the nth unit will also pass, you may need to maintain a 6 or even a 10 dB margin. In a sampling situation, the regulatory agency or the party with the power to approve (maybe a customer) must account for normal site to site variations, say +/- 2 dB per site, so a measurement margin of 2 dB may be adequate, but risky. Over the years I have seen customers require margins of 10 dB, 6 dB, 4 dB, 2 dB, and 1 dB for a given test to be a "pass". The higher the margin, the better the EMC engineering, the greater the confidence level, and the better the perceived quality of the product. All this comes at a cost, and it's ultimately up to those who pay the bills to decide the level of risk they can afford. If it's one number you need, I would reluctantly put forth a 2 dB margin, subject to the caveats discussed above. I appreciate your willingness to collect this info and look forward to the results of your poll. Best regards, Tom Cokenias RFI/EMC Consultant
