Evaluation Engineering - 18

COMPLIANCE

more significant than just the obvious
accounting figures.

How hard is it?
If we consider just emissions, there are two
basic measurements required: Conducted
emissions and radiated emissions.
Conducted emissions is easy. For
power cables, an LISN connected to an
EMC analyzer or receiver is required. For
other cables (data, sensors etc.), these only
need checking if they are more than 3m
long. No need for screened rooms, tents
or anything special. Do the test on your
own workbench.
Radiated emissions need more consideration. In principle, it's easy. Just set up an
antenna 3m from the product and measure
the results on your EMC analyzer or receiver. In practice, there are two problems
with this: Ambient noise, and test site calibration. These are often mistakenly linked,
but they are entirely different and separate
issues which require different solutions.
For ambient noise, a "screened" environment such as a test cell or anechoic chamber provide the obvious solution. Note that
"screened room" is not suggested as a solution. Screened rooms should not be used
because of the internal reflections which
will create massive measurement errors. A
cost-effective alternative is to use an "open"
site with a system which can measure and
then subtract the ambient noise from the
emissions from the device-under-test
(DUT). This system works well on moderately noisy sites, but strong local sources
such as transmitters may overwhelm the
receiver or analyzer. As a solution: If the
DUT is portable, find a quieter site.
Test site calibration is the major source
of measurement error in any radiated
emissions measurement. Manufacturers
are inevitably restricted in their choice of
sites due to space and budgetary restrictions. Typical sites are a manufacturing
area, loading bay, conference room, car
park, or something similar. These sites
will all suffer from the proximity of metal
surfaces which will reflect RF. Reflected
RF will interfere with the measurement
and the result is a site that does not behave
like a true OATS. If, however, a known and
calibrated source of RF is substituted for
the DUT, the characteristics of the test site

EVALUATION ENGINEERING JULY 2019

An MQ-4C Triton is lifted inside Patuxent River's
(Maryland) anechoic chamber during August 2015
for EMC testing.
U.S. Navy

can be measured. A correction factor can
then be calculated so that when the DUT
is measured, the results are corrected so
that they appear as though they had been
measured on a good 3m OATS site.

The process
The ERS is a comb generator, with a frequency interval of 2MHz. Each 2MHz
peak is measured and documented, a
total of 485 points in the range 30MHz
to 1000MHz. Each ERS is traceable to
the master EMC OATS in the UK (NPL,
Teddington). It is supplied with the radiated emissions data as measured at
3m. If "your" site was as "good" as NPL,
the measurements would agree with the
supplied data. Inevitably, the results will
differ. At each 2MHz point, the emission
level from the ERS is measured and the
effect of any ambient noise is also taken
into account. This measured level is then
compared with the calibration data, and
a correction value can then be calculated.
If for example, the measured value was
49dBuV, and the calibration data from
that frequency was 57dBuV, then your site
is reading 8dB low, so a correction factor
of +8dB should be added to any DUT measurements at that frequency. The process
is applied to all 485 frequencies.
When measuring the DUT, the ambient noise is first canceled, then the site
correction factor is applied.

Measurement uncertainty
A properly configured test bench and a decent EMC analyzer or receiver should be
able to deliver conducted emission results
with the 4dB uncertainty margin that any

accredited test lab would apply. There are
no requirements for screened rooms, just
a ground plane under the nonmetallic test
bench, LISN bolted to the ground plane
and short cables from the LISN to the DUT.
Radiated emissions are not so easy.
Note that even accredited test labs will
not quote better than 6dB uncertainty.
Inherent variances are due to DUT cable
positioning and orientation, coupling to
chamber or test cell structure and (for
mains-powered DUTs) local mains supply impedances and imbalances. However,
the aim of any self-test process is to match
OATS setup conditions as far as practicable. The ERS is a very significant aid in
this respect. Without some site calibration,
the uncertainty is "off scale", but by using
an ERS, and rigorous process, uncertainty
should be better than 10dB. An excellent
alternative is to use a similar product
which has been measured at a test lab
and to use this to characterize your site.

In summary
EMC measurements are often portrayed
as difficult, specialist, and even obscure,
but in reality, they are just another topic
that we engineers need to "learn." There
are many sources of good information
on the web related to test techniques.
Setup, configuration and procedures
are important, but these are straightforward. It helps to have receiver/analysis
software that is simple and intuitive to
use, designed for occasional users, rather
than full-time test lab technicians. It also
helps to have an instrumentation supplier
prepared to advise and provide guidance
on the test procedures.
The changes in the FCC rules do provide a fresh opportunity to save on time
and budget if the new advantages are
grasped.
David Mawdsley is managing director of U.K.-based
Laplace Instruments. He has
been involved with the development of EMC compliance
test and measurement techniques since 1995.
His particular focus has been on fully integrated systems that can be used by manufacturers to achieve self-test, self-certification
of commercial products.

Evaluation Engineering

Table of Contents for the Digital Edition of Evaluation Engineering