| NARTE News Selected On-line Articles Volume 18 Number 3 Fall 2000 |
Background--Test Sequence
The tests described were conducted at five different sites, designated herein as test sites "A', "B ~', "C" "D" and "E". Tests initially requested by the manufacturer were conducted on test site "A". Following these tests, the customer for the equipment then had the equipment under test (EUT) re-tested in his own 3m anechoic chamber (test site "C"), and then on an independent test facility's 10m Open Area Test Site ("B'). After these subsequent customer tests showed a large difference when compared to the tests conducted at test site "A", the manufacturer then had the tests repeated in his own GTEM (test site "D'). The fifth test site ("E") is an open area weather proof test site almost identical to test site "A".
Test Site "A" Set-Up
A small piece of electronic equipment had been tested for radiated emissions on numerous occasions for the manufacturer, throughout the development of the product and also on the production version.
These tests were performed by at least three test personnel. The requirements were derived from the FCC Part 15 and the EN55022 Class B requirements.
The software mn by the manufacturer was considered "worst case."
The EUT was powered through a single cable that also contained a serial data interface.
The cable was taken down to the power supply located on the Open Area Test Site (OATS) ground plane and then brought back to the tabletop.
The cable orientation was as close to that shown in 9(c) of ANSI C63.4, with the power/signal cable brought back up to the non-conductive tabletop and then looped back down again in a serpentine fashion, with the loops at least 40cm above the ground plane.
ANSI C63.4 mentions that, "In order to replicate emission measurements, it is important to carefully arrange, not only the system components, but also, system cables, wires and AC power cords." In discussing cable orientation, ANSI C63.4 continues with: "It is essential to recognize that the measured levels may be critically dependent upon the exact placement of the cables and wiring. Thus preliminary tests may be carried out while varying cable positions in order to determine the maximum or near maximum emission level."
These tests at site "A" were made using a turntable. The radiated emissions were found to vary dramatically, depending on the rotation of the table. After a maximum emission was found, the cables were then separated or moved within the confines of the recommended ANSI configuration. Again, at least three different personnel manipulated the cables ! This reorientation was also only made at one (worst case) frequency, whereas different cable orientations maximize the emissions at other frequencies.
Taking this approach limits the amount of time required for a test, but by limiting the basic cable orientation to the ANSI C63.4 recommendation, the absolute worst case emissions may not be found.
Test site "A" was a 3m OATS that had been tested in accordance with ANSI C63.4 for Normalized Site Attenuation (NSA). The NSA test had been conducted in 1997, and the measured site attenuation was within +/-4dB of the theoretical NSA provided in ANSI C63.4 with a safety margin (i.e., the measured NSA was lower than the +/-4dB tolerance). In all of these tests on site "A", the EUT met all of the radiated emission measurements with a safety margin of at least 8.8dB below either the FCC or ENS5022 limit.
In near field measurements on all of the different EUTs, close to the EUT and the cable, the measured levels were also consistent with a low level emitter.
Testing at Sites "B", "C" and "D"
The same type of product, but not with the same serial number, was then tested in a 3m semi-anechoic chamber test site (test site "C"), which reported emissions up to 3.65dB above the limit (i.e., 12.46dB higher than the 3m OATS measurements). At exactly the same frequency the difference in measurements could have been as high as 25dB. The EUT was then tested on a 10m OATS (test site "B") and also in Gigahertz Transverse Electromagnetic (GTEM)--test site "D".
The measurements made on a 3m OATS and a 10m OATS often do not correlate to the far field inverse linear distance extrapolation factor of 20 log 10m/3m = 10.5dB. It is common that, in moving the EUT location from 3m to 10m, the reduction at some frequency is less than 10.5dB. Thus, an EUT may be just within specification when measured at 3m and out of specification at 10m. However, this does not account for the 14dB difference between sites "A" and "B". In the GTEM measurements the test engineer made the cable "dormant" by the addition of three ferrites. Therefore, if cables were the main source of emissions, the GTEM measurements should have been the lowest. Also the test on site "C" was repeated, using a short cable and with the power supply placed on the top of the turntable, with the same high levels of emissions. This again led us to believe that cables were not a factor.
Verification of Test Site Equipment
These test results did not look good for test site "A", and so extensive measurements were made on the site and test equipment. One possibility for the difference was that an EMI receiver was used on sites "B" and "C" and a spectrum analyzer (S/A) was used on site "A". A variation in the measured level of broadband noise can be expected due to the different shape of the impulse bandwidth used in the receiver, and the resolution bandwidth used in the S/ A. The analyzer typically requires a 4dB correction for broadband noise. However, the measurements using the S/A did not vary significantly with change in Resolution Bandwidth (RBW) as the emission was predominantly narrowband.
In the final measurements made on site "E", described later in this article, a receiver was used and a good correlation was seen between sites "A" and "E", not consistent with a 4dB systemic error.
The test site "A" measurements were made using two different spectrum analyzers that had been calibrated and checked against a calibrated signal source. Quasi peak and peak measurements were the same and so the quasi peak detectors were not a problem. Cable attenuation and preamplifier gain were checked and found to be acceptable. Before using the test site, a "sanity check" was made of the antenna, cables, and preamplifier using the ambient from a number of FM radio stations (about the only advantage of the OATS, apart from accuracy). The levels during all tests on the EUT were within 1 dB of the values measured when the site was first constructed.
The only remaining possibility for the difference was the antenna calibration and the NSA characteristics of the site. A log periodicPoiconical antenna and a Roberts dipole were both used during measurements on the EUT, and it seemed unlikely that both would be equally out of specification. Nevertheless, two of the log periodic biconical antennas were tested on a free space range and the coupling between the antennas was exactly as predicted, based on the input signal level, their gain, and antenna factor. Also, as expected, the free space measurement failed the NSA requirements with a difference of 5.7dB, due to the absence of the ground plane.
The Equation for AN is:
AN = VDirect - Vsite-AFT - AFRM (all terms dB)
Where VDirect is the direct measurement via the cables, Vsite is the radiated
measurement using the transmit and receive antennas and the same cables, AFT is the transmit antenna, antenna factor and AFR is the receive antenna factor.
As the direct measurement was made with a calibrated output signal generator,
this measurement also tests that the spectrum analyzer was measuring accurately
the amplitude of the frequencies of interest, with the expected cable attenuation·
Please note: During radiated measurements, the same spectrum analyzer reference level
was used as for direct measurements. The radiated measurement used the calibrated
antenna factors, which are critical in meeting the NSA requirements. Thus, if we
assume a very good OATS performance, the NSA measurement is also a way of reconfirming the accuracy of the antenna calibration. This was further tested by replacing the biconical log periodic receive antennas with two reference dipoles and repeating the NSA measurement.
Table 1 shows the NSA over the frequency range of interest using two log periodic biconical antenna;
Table 2 shows the use of two dipoles. These measurements demonstrate that the site is at a worst case within 2.1dB of the predicted; and at the critical frequency of the EUT (220MHz) is within approximately 1 dB of the value for NSA, interpolated between 200 and 250MHz.
| f(MHz) | NSA (dB) Δ |
|---|---|
| 200 | 0.82 |
| 250 | 1.68 |
| 300 | 2.10 |
| f(MHz) | NSA (dB) Δ |
|---|---|
| 200 | 0.31 |
| 250 | 1.69 |
| 300 | 1.90 |
Final Testing at Site "E" and Focus on Different Site Cable Configurations
Even though site "A" met all of the requirements, the EUT was tested yet again, but on a 3m test site "E". These measurements showed the EUT passed the requirements, hut by only 1.6dB, unlike the 8.8dB found on site "A". However, one difference in the cable configuration used on test site "E" was that the cable was taken down from the center of the turntable, rather than from the edge. Then it was brought back from the battery and coiled in the center of the turntable in a large loop, instead of in small vertical loops.
The exact same cable configuration was tested at site "A" and the difference in emission measurements between site "A" and "E" was a worst case 3.8dB, which is an acceptable difference based on antenna factor calibration errors, NSA errors, measuring instrument errors and the difference in receiver to S/A bandwidth shape.
This final test showed that cable orientation does play a role in the level of emissions, by as much as 2.9dB in this particular instance.
In the tests conducted at test sites "B" and "C", which showed the highest levels of emissions, the length of cable after the power supply was coiled on the ground plane. This cable orientation seems far away from the spirit of the ANSI recommended layout and may explain the large difference in measurements. The measurements, however, were apparently the same with the short cable and the power supply placed up on the top of the turntable.
Table 3 shows the emissions in dBgV/m compared to the ENS5022 Class B limit.
| f(MHz) | Test Facility and Antenna | Level (dBµV/m) | Limit (dBµV/m) | a(dB) |
|---|---|---|---|---|
| 212 | Site E log/bicon | 31.4 | 40 | - 8.6 |
| 212 | Site A log/bicon | 31.6 | 40 | - 8.4 |
| 220 | Site A dipole | 31.2 | 40 | - 8.8 |
| 220 | Site E dipole | 35.0 | 40 | - 5.0 |
| 228 | Site A log/bicon | 35.7 | 40 | - 4.3 |
| 228 | Site E dipole | 38.4 | 40 | - 1.6 |
| 236 | Site A log/bicon | 36.2 | 47.5 | - 11.3 |
| 236 | Site E log/bicon | 37.1 | 47.5 | - 10.4 |
©2000 NARTE Incorporated