The search for a practical reference in EMC emissions measurements may parallel Jason’s quest for the golden fleece. Just like Jason, in attempting to realize a noble goal, we keep running into labyrinths of complexity and detail.
With most measurements, you use an instrument calibrated by comparison to a standard or reference. But in EMC measurements, the seemingly simple task of achieving repeatable and reproducible results still forms the bulk of today’s research.
So what’s the problem? Well, the reference for EMC emissions measurements isn’t something you’ll find on a laboratory shelf. It’s the behavior of free space. That’s the place where all waves are plane waves, the characteristic impedance is 377 W , and the E- and H-fields are inversely proportional to the distance from the source.
This ideal behavior exists at a minimum distance from the radiating source known in antenna theory as the far field and is approximately equal to 47.7(106)/frequency. This distance is 1 meter at 47.7 MHz.
Utopian EMC radiated emissions measurements take place in free space, using an antenna with no directional bias. All events are independent, linear, superimposable, or otherwise mathematically predictable. The Friis equation, for example, is used for site-attenuation calculations.1 The ideal test site is a free-space simulator, and the perfect antenna behaves the same in all directions.
Radiated EMC emissions testing depends on antennas and test sites, but both are imperfect. How, then, do you calibrate the antenna? ANSI C63.5 describes a method of antenna calibration known as the standard site method or three-antenna method, but it assumes an ideal test site.2
A round-robin method of antenna calibration has been described in which the errors are identifiable and correctable. The testing would be conducted at three sites and the average value used as the perfect site. These perfect-site numbers then would be used to calibrate the antennas.
Aiming Higher
Upper-frequency EMC test limits may go beyond 1 GHz, increasing the frequency spectrum over which the behavior of all equipment must be characterized. Expect a greater deviation from the ideal under these circumstances.
Clark Vitek, an EMC engineer at CKC Laboratories, said that CPU speeds have increased from 100 MHz to near 1 GHz, and some wireless devices have fundamental frequencies to 38 GHz. He indicated that interference over 1 GHz will be a problem until we test at higher-frequency limits.
The FCC Part 15 and CISPR 22 standards apply to EMC testing of information technology devices like computers and telecommunications equipment.3,4 According to Dennis Handlon, an EMC engineer at Hewlett-Packard, CISPR proposes testing to 18 GHz. It will be a challenge, he pointed out, to provide receivers, antennas, and preamplifiers with the required sensitivity at that frequency.
EMC emissions test equipment already covers a broad frequency range. By continually increasing the upper frequency limit, it becomes more difficult to control test-setup impedances. But, reflections occur when impedances don’t match, and this creates a problem. By not adhering to a well-defined pattern, reflections are difficult to correct. Ground-plane reflections are a good example of this.
Getting Faster
The actual task of performing EMC measurements is a burden because of the sheer number of readings required. As a result, there have been efforts to automate testing to save time and money and reduce errors. By increasing the upper-frequency test limit, this situation will become even more severe.
Most people are interested in speed, quality, and cost, according to Chris Kendall, an EMC engineer at CKC Laboratories. He said that certain types of test chambers, like mode-stirred-rooms and free-space rooms, are becoming popular because you get more data faster.
Expeditious test methods may provide some relief, but sometimes this approach can create problems. Mr. Handlon of HP pointed out that an open area test site (OATS) in most urban areas may contain ambient RF signals.
Even though EMI receivers and spectrum analyzers offer expedient emissions testing with frequency sweep and other capabilities, they have difficulty in distinguishing ambient RF signals from the intended measurement. In noisy outdoor environments, according to Mr. Handlon, the normal procedure is to test the product in a chamber and identify problem frequencies or a suspect list. Then, you select those same frequencies for outdoor testing.
Dr. Michael Foegelle, the senior principal design engineer at EMC Test Systems, said that a modern vector network analyzer will maintain an OATS to help establish viable calibration. EMI receivers and spectrum analyzers, unfortunately, do not benefit from the same ambient signal rejection. There is some question about whether this approach can be accomplished in a reasonable amount of time, prompting the expediency issue to resurface. Of course, the ideal instrument will do it all.
To arrive at good EMC measurements, you must translate the results to ideal, or at least reference, conditions. This is a problem as evidenced by the difficulty in achieving repeatable and reproducible results. Because of uncertainty in EMC measurements, standards have arisen to analyze and quantify them.
Mr. Kendall of CKC Laboratories indicated that it’s a challenge to correlate one type of test site to another. It is even difficult, he said, to get one OATS to agree with another OATS within the ±4-dB uncertainty limit stated in ANSI 63.4.5
CISPR 16 allows ±3-dB uncertainty for receiver, antenna, cable loss, and signal mismatches.6 But the amount of uncertainty in test equipment is nil compared to the OATS and antennas calibrated on the OATS using the three- antenna method in ANSI C63.5, according to Mr. Kendall. Another source of error, he said, is the lack of impedance control of the power leads as they feed through the OATS ground plane to the turntable.
Test equipment is not the only source of error, noted CKC’s Mr. Vitek. He cited test-site construction and operator-induced errors, such as not maximizing cables during testing as required by the standards, as more significant problems.
The uncertainty in EMC emissions measurements that relates to equipment is a combination of the receiver, antenna, and cabling, said HP’s Mr. Handlon. Calibration and the capability to characterize cable loss, he continued, are important in determining uncertainty figures. He also pointed out that reference generators are sometimes used to measure the attenuation of cables and test sites.
EMC Test Systems’ Dr. Foegelle suggested using a comb or white noise generator for comparing the attenuation of various test sites. Unlike an ordinary generator which sends out one frequency at a time, this device sources many frequencies at once to do the job faster and more inclusively.
Manufacturers of antennas and other EMC emissions test equipment are beginning to quantify the uncertainties associated with their products. Dr. Foegelle said that these manufacturers and calibration laboratories want to specify lower uncertainties as a marketing tool.
Current and future work on the standards committee will continue to drive this trend. But, Dr. Foegelle warned, some equipment and methods promoted as expedient come at the expense of greater uncertainty.
Be careful. Uncertainty figures could turn into a numbers game. Rely on the reputation of the manufacturer that stands behind these figures.
Dealing With Uncertainty
Where do the uncertainty figures come from? NIS 81 outlines statistical methods to derive uncertainty figures for random errors and recommends using historical data to estimate uncertainty figures for systematic errors. Standards relating to uncertainty are listed in Table 1.
References
1. Friis, H.T., “A Note on a Simple Transmission Formula,” Proceedings IRE, Volume 34, pp. 254-256, May 1946.
2. ANSI C63.5: Electromagnetic Compatibility—Radiated Emission Measurements in Electromagnetic Interference Control-Calibration of Antennas.
3. FCC Part 15: Limits and Methods of Measurement of RFI From Information Technology Equipment.
4. CISPR 22: Limits and Methods of Measurement of RFI From Information Technology Equipment.
5. ANSI C63.4: Methods of Measurement of Radio-Noise Emissions From Low- Voltage Electrical and Electronic Equipment in the Range of 9 kHz to 40 GHz.
6. CISPR 16-1: Specifications for Radio Interference Measuring Apparatus and Measurement Methods.
Standard
Title
NIS 81
The Treatment of Uncertainty in EMC Measurements, NAMAS
ISO Guide 25/Draft Standard 17025
Guide to the Expression of Uncertainty in Measurements
EN 50222
Standard for the Evaluation of Measurement Results Taking Measurement Uncertainty Into Account
Table 1.
EMC Emission Test Products
GTEM Offers Emissions and
Immunity Compliance Testing
Measurements made by the Model 5405, the Gigahertz Transverse Electromagnetic (GTEM) Cell, are accepted for demonstrating compliance with FCC Part 15 and 18 radiated emissions testing and IEC 61000-4-3 radiated immunity testing. It is suitable for use with compact electronic devices such as notebook computers, pagers, or portable phones. The cell provides a test volume of 25 × 25 × 25 cm; larger models are available. A typical radiated emissions test can be completed in 15 minutes, as compared to an outdoor area test site which can take all day. EMC Test Systems, (512) 835-4684.
Antenna Performs Emissions
And Immunity Testing to 3 GHz
The EM-6917A-1 Biconicalog Antenna eliminates the need for making adjustments or switching antennas for emissions testing, and it can handle the power capacity for immunity applications. Each portion of the combined industry standard biconical and log periodic antenna is calibrated for field-strength measurements from 26 MHz to 3 GHz. An integral balun minimizes the effect of cable placement on measurements. The nominal impedance is 50 W , the VSWR is <2.0:1 typical, and the power-handling capability is 1,000 W at 1 GHz. Electro-Metrics, (518) 762-2600.
Portable Receiver Has
Three Tuning Modes
The SCR 3101 EMI Receiver weighs 7.3 lb, including batteries which run for three to four hours. The unit measures to IF bandwidths specified in CISPR 16-1 and has ±2-dB accuracy from 9 kHz to 1 GHz. There are three tuning modes. The parallel detectors are peak, quasipeak, and average. Features include AM/FM demodulators and an LCD. The receiver runs the company’s EMI software. A PCMCIA card stores data and measurement routines. Schaffner EMC, (973) 252-8001.
Precompliance Receiver
Tests Emissions to 1 GHz
The PMM 7000 is a precompliance receiver that connects to the PC via RS-232 to provide EMC emissions testing from 150 kHz to 1 GHz. Windows-based software contains all commands needed for testing on one screen. Also included is a pre-loaded list of standards used for the CE marking. The unit features an internal line-stabilization network and automatically performs measurements for all tests according to CISPR 16. Antenna Research, (301) 937-8888.
Compact Test Chamber
Operates to 1 GHz
The S-Line Series of shielded TEM cells offers a compact alternative to anechoic chambers for precompliance emissions or immunity testing. For emissions testing, filtered feedthroughs are provided for connecting the EMI receiver or spectrum analyzer. The VSWR of the empty cell is 2.5 maximum, the frequency range is 150 kHz to 1 GHz, and the shielding effectiveness is ³ 75 dB at frequencies <500 MHz. The S-Line 700 accommodates an EUT of 13.7 in.3, and the S-Line 1000 increases the volume to 18.9 in.3. Tektronix, (800) 426-2200.Copyright 1999 Nelson Publishing Inc.
February 1999