A Brief History of EMC Measurements

Even in light of these advances, the human interface is still a critical part of EMC instrumentation. And fortunately for us humans, the drudgery of data collection and reduction has been greatly reduced.

I performed my first EMI measurements in 1965 using an Empire Devices/Singer Metrics NF-105 receiver, consisting of the basic unit and four plug-in tuning units that covered 150 kHz to 30 MHz, 30 to 200 MHz, 200 to 400 MHz, and 400 MHz to 1 GHz.

Antennas for measurements below 30 MHz consisted of a loop antenna for magnetic-field measurements and a monopole high-impedance antenna for electric-field measurements. For measurements from 30 MHz to 1 GHz, a tuned dipole was used.

Since instruments were calibrated only over a very limited frequency range at one time, actual measurements of a device were usually performed using the "signal substitution" method–a slow, but accurate, process. At a particular measurement frequency, the signal was matched with a known source from a calibrated signal generator. The signal generator would be either a narrowband, tunable generator or an impulse generator, depending on whether the measurement was broadband or narrowband. Neither the U.S. military services nor the FCC used the CISPR detector. In most military standards, narrowband limits were expressed in either dB/µV or dB/µV/m, and broadband limits in dB/pµV/MHz or dB/pµV/MHz/m. These measurements were recorded at each frequency of interest. Since a tuned dipole was used as an antenna, it also needed to be readjusted at each frequency.

If measurements were desired in both the idle mode and the operating mode of a product, this again doubled the number of data points. Since this was before the days of pocket calculators–and long before PCs–each datum had to be reduced by hand or by an electromechanical calculator.

One of the problems with the measurement instrumentation just described was that it employed vacuum-tube technology and plug-in tuning units. When changing frequency ranges, much time was spent waiting for the warm-up and stabilization of the next tuning unit.

But late in 1965, Fairchild Electrometrics Corp. announced the EMC-25 Interference Analyzer, and we immediately placed an order. Our unit arrived in 1966. This was the first all-solid-state radio-interference meter–an entirely self-contained band-switching instrument that covered the frequency range from 150 kHz to 1 GHz in approximate octave increments. Because it needed no external plug-in tuning units, the warm-up time between frequency band switching was eliminated, along with several hours of waiting associated with each measurement series.

The gain flatness across each band was much better than that provided by vacuum-tube instruments, which facilitated tuning through frequencies and obtaining a general idea of the compliance of the EUT. Provisions were also available for electrically scanning each band and plotting amplitude versus frequency. Rechargeable battery power was standard, allowing portability.

This measurement technology, however, was not without problems. VHF and UHF transistor technology was relatively new, and several of the frequency bands used MOSFET transistors that were very sensitive to ESD. The first RF amplifier stage could also be damaged by large signals and/or transients if the input attenuator was on a low setting. As in earlier instruments, the analog meter reading and attenuator setting were added together mentally or on paper to obtain the signal or noise measurement reading. This, as well as old habits, caused many engineers and measurement technicians to continue to use signal substitution for measurement.

About 1970, Hewlett-Packard introduced the 141S/8552/8553 spectrum analyzer. This instrument allowed a sweep of its entire frequency range to be displayed on a variable-persistence storage CRT. Although this analyzer greatly improved EMI measurements, its major shortcoming was that it scanned only from 10 kHz to 100 MHz. A short time later Hewlett-Packard rolled out an improved unit that would scan to 1250 MHz.

The innovations in spectrum analyzers that improved EMI measurement productivity were a switchable-resolution bandwidth that remained constant throughout the sweep, virtually constant system gain, data storage via CRT phosphor (which was photographed), and measurement instrument accuracy across the spectrum of interest.

While measurement personnel began moving away from signal substitution and manual data recording by frequency, there was still no automated method of data reduction and plotting. To facilitate data reduction from a Polaroid photograph, we designed the label shown in Figure 1 and attached it on the back. This label allowed all the necessary information to be manually recorded on the photograph, and it was always available when data reduction took place. Since these spectrum analyzers displayed a linear frequency scale, the measurement spectrum was usually divided into three or more scans, and multiple photographs were taken. This allowed the final data to be plotted manually on a logarithmic scale.

At the 1976 IEEE International EMC Symposium we observed the demonstration of the Hewlett-Packard 8568A spectrum analyzer. This instrument offered definite improvements over the previous models, one feature in particular being the use of an IEEE 488 connection that allowed the transfer of the measured data to a desktop controller or high-function calculator. We immediately placed an order and received a unit.

This instrument's controller used a magnetic tape cassette to store programs and data, and it was connected via the IEEE 488 interface to a plotter. We wrote a software program that would plot the reduced data directly from the spectrum analyzer, simplifying and quickening the measurement process.

Simultaneously, we constructed our first absorber-lined shielded room. While all previous radiated EMI measurements had been performed either in an open area or in a 6 x 6­m shielded room, we had the space and the budget to construct a 10 x 9­m, 3.5-m-high-overall chamber with 1.8 m absorber on the walls and 1 m absorber on the ceiling. Although our first measurement attempts provided useful information, our results would be considered unusable by today's standards.

The FCC introduced its First Report and Order on the rules for computing devices in 1979. Shortly thereafter came the rules and measurement procedure MP-4. The FCC had determined that in the interest of international harmonization, a CISPR quasi-peak detector would be desirable for performing both radiated and conducted EMI measurements.

Fortunately, we had purchased Schwarzbeck VUME 1520 and FSME 1515 measurement receivers to measure our products to the limits specified in VDE 0871 and VDE 0875. These receivers contained quasi-peak detectors that would allow measurements according to the new FCC requirements. Because of the manual nature of these receivers, however, we were forced once again to reduce data by hand.

By this time, Rohde & Schwarz among other manufacturers had introduced CISPR receivers that had the capability to store antenna factors and process and transmit measurement data over an IEEE 488 connection. We upgraded our measurement system and software to include these capabilities and again were able to shorten our entire measurement process.

Figure 1. Information label.

1989

In mid-1989, the European Union EMC Directive was made known to all those involved in developing, producing, and marketing products internationally. Fortunately, CISPR 22 was chosen for the EMI requirement–a standard very similar to the German requirements that had been in existence for decades. Therefore, very few changes in measurement procedures were required.

Measurements that took 20 hours in 1965 take less than 15 minutes today. And in those 15 minutes much more data is collected–and archived for future reference–than was ever possible in 1965. I doubt that there will be a proportionate time evolution over the next 20 years, but I am sure that there will be an increase in accuracy, resolution, and convenience.

Some measurement sites have increased automation to the point where the measurement technician has virtually no decisions to make. Nevertheless, my experience has been that a knowledgeable, experienced measurement technician or engineer who makes appropriate decisions during evaluation can significantly decrease the time required to do the work. Our current system still demands that these decisions be made by humans.