Almost all EMC engineers these days are familiar with MIL-STD-461 and its various revisions because it has influenced the procedures in virtually all present day EMC standards. The original document and the A, B and C revisions all had requirements for measuring both broadband (BB) and narrowband (NB) emissions. When MIL-STD-461 was first released (July 31, 1967), the Empire Devices NF-105 receiver was the workhorse of the RFI/EMI/EMC industry, and the early standard releases show line-art diagrams of the receiver setting on the shielded enclosure ground plane. Yes, folks, we actually sat in the shielded enclosure when we did the measurements – turned the knob, recorded three frequencies per octave, repeated on and on until we were finally finished.
The NF-105 had a number of manually tuned plug-in units to enable coverage of the frequency range from 14kHz to 1GHz. Each of these tuning units had a single bandwidth. It might seem as if making a narrowband/broadband determination with only one bandwidth would be an impossible task. As I indicated in the previous post, a narrowband signal has zero bandwidth (or nearly so) whereas a broadband signal’s spectral occupancy extends beyond the bandwidth of the receiver. Armed with this knowledge all that needs to be done is to apply the Tuning Test. Just tune up and down by one receiver impulse bandwidth. If there is essentially no change in the measured amplitude, the signal is BB; otherwise it is NB. This is great if the tuning is being done manually but does not lend itself to being easily automated.
If the measurement receiver has multiple bandwidths, the BB/NB determination is really easy. Tune in the signal, then change the bandwidth, and if there is no change in measured amplitude or it’s less than or equal to the NB bandwidth criteria (usually either 3 or 6 dB), the signal is NB; otherwise it’s BB. Select the appropriate bandwidth and abracadabra all broadband signals disappear. They can then all be called narrowband. The same thing holds for coherent broadband. If the bandwidth is made small enough to encompass a single harmonic, then the signals are narrowband.
If you are still using a specification that requires measurement of broadband signals, there is a problem determining whether an intermittent signal that appears to be CW is really a NB signal or a transient, which is a BB signal. The only way to know for sure with a single measurement is to simultaneously measure its peak and average levels or analyze its modulation characteristics if it has any. The peak vs. average test can also be used for those weird signals that seem to defy categorization, provided they exist long enough to apply the test. A BB signal will have a significant change in amplitude when switching from Peak to Average while a NB signal will not.
Those that live in front of a spectrum analyzer will be annoyed that I didn’t discuss the sweep time test. It’s great if you have repetitive pulses, but it isn’t worth a hoot for determining randomly occurring NB emissions or pulses. That holds for most of the tests. Besides that, the displayed amplitude and frequency on many spectrum analyzers changes with a change in sweep time.
That’s one of the many reasons why MIL-STD-461D gave up measuring broadband signals! All current revisions now specify the receiver response characteristics and demand that the sweeps be slow enough to capture any signals that may repeat. If measurements are made from DC to daylight in a nanosecond or two, all EUT’s will pass! That’s also why many of the spectrum analyzers (or the way they are used) aren’t fully compliant with MIL-STD-461.
These problems and the opportunity for signal detection manipulation were first described by a pioneer EMC engineer named William (Bill) Cipperly in an article titled “EMC Engineer, Scientist or Witch Doctor?” that appeared in Frequency Technology magazine circa 1968.
– Ron Brewer
