Вы находитесь на странице: 1из 22

Click to edit Master title

style
A Quantitative Study into the Errors in
Measuring an Automotive Vehicle
Radiated Emissions Signature Using the
CISPR 12 Method
February 2019

© HORIBA MIRA Ltd. 2015

© HORIBA MIRA Ltd. 2015


What is Needed to
Determine the Level of the
Emissions ?

▪ Perform measurements to record the amplitude of the


electric field – Requires expensive measurement system / facility
▪ Perform EM Simulations – Requires very detailed model of the source device, this
information may not be available and powerful computer to run the simulations.

▪ Both measurements and simulations will still only record


an approximation of the actual emissions signature of the
device

© HORIBA MIRA Ltd. 2015 March 17, 2020 2


Radiated Emissions
Measurements Setup
(Non-Automotive)
▪ Performed in SAC or OATS

▪ DUT on non-metallic table 0.8 m high

▪ Antenna to DUT separation distance


typically 10 m

▪ Antenna scanned in height between 1 m


and 4 m

▪ DUT rotated through full 360o in small


increments (usually 1o)

▪ Horizontal and vertical receive antenna


polarisations

▪ “Maximum” emissions recorded over


each frequency over the range of interest
© HORIBA MIRA Ltd. 2015 March 17, 2020 3
Why Scan Antenna in Height

▪ Direct radiation path from DUT to Antenna


▪ Reflected path from DUT to antenna via conductive ground plane
▪ Path Length difference between direct and reflected signals causes a phase
shift and cancellation of signals (amount depends on phase difference)
▪ Antenna is scanned in height to find point at which direct and reflected signals
are in phase (i.e. no cancellation) - hence highest emissions

© HORIBA MIRA Ltd. 2015 March 17, 2020 4


Why Scan DUT in Azimuth

▪ If DUT was electrically small signal would radiate


isotropically (equally in all directions)
▪ Electrically large DUT will have a radiation pattern
that has peaks and troughs (exhibits directivity).
▪ Vehicles will be electrically large at all frequencies
covered by the relevant test standards

DUT rotated through 360o maxima and minima


will be recorded at each frequency

© HORIBA MIRA Ltd. 2015 March 17, 2020 5


CISPR 12 Automotive Radiated
Emissions Measurements

▪ What makes automotive emissions measurements


different?
▪ CISPR 12 is the reference standard details how Automotive emissions
measurements are performed
▪ - Single receive antenna height of 3m used
▪ - Measurements performed at two azimuth angles only (referred to as CISPR A,
and CISPR B for the rest of this presentation)

▪ What are the reasons for the differences ?


▪ - Reduced costs
▪ - Shorter time

© HORIBA MIRA Ltd. 2015 March 17, 2020 6


Frequency Range of Interest

▪ CISPR 12 specifies 30 MHz and 1 GHz


▪ Typical emissions from a commercial vehicle begin to fall off above ~400 MHz

© HORIBA MIRA Ltd. 2015 March 17, 2020 7


Issues with Current Vehicle
Measurements

▪ Maximum emissions
potentially not recorded
▪ Errors of up to 30 dB
possible -based on
simulated data of “simple”
vehicle model [1]
CISPR A CISPR B
▪ Automotive Industry resist
change to current
procedure (based on cost,
time and perceived risk)

© HORIBA MIRA Ltd. 2015 March 17, 2020 8


Possible Alternative
Measurement Methods &
Their Disadvantages

▪ Full Spherical Scan – Very costly and time consuming, requires turntable /
motorised antenna mast.

▪ Reduced Spherical Scan – Still costly and time consuming and doesn’t record
maximum, equipment requirements still same as full spherical. Method used to test
almost all other electrical items

▪ Reverberation Chamber – Still costly and time consuming and doesn’t record
maximum, records mean amplitude. With knowledge of the directivity of the DUT, it is
possible to convert the mean amplitude to an equivalent maximum amplitude

▪ “Test Wire” – A system originally designed for testing large industrial machines in-
situ in the factory. To be investigated for automotive measurements

© HORIBA MIRA Ltd. 2015 March 17, 2020 9


Test Wire Method - Setup

▪ Test Wire is positioned 100 – 200 mm above DUT surface


▪ 150 W Termination Resistors at each end of test wire between wire and
ground (100 W at the measurement receiver end).
▪ Measure voltage across resistor using EMC receiver / spectrum analyser
▪ Measured voltage is the used to determine “correction” factor
▪ Multiple Test Wire positions were used

© HORIBA MIRA Ltd. 2015 March 17, 2020 10


Correction or ‘K’ Factor

▪ ‘K’ Factor is used to generate equivalent E field value from the voltage
measured at termination of test wire
▪ The ‘K’ Factor can be calculated using the following:

𝑣
𝑑𝐵 𝐸
𝐾 = 20 ∗ log 𝑚 … … … . . [𝐸𝑞 1]
𝑚 𝑈 𝑉

■Where E = maximum E field recorded over a full spherical scan and


U = maximum voltage recorded across termination resistor. An individual
‘K’ Factor will be calculated at each frequency of interest

© HORIBA MIRA Ltd. 2015 March 17, 2020 11


1/3 Scale Simple Vehicle
Test Case (SVTC) Physical
Model

© HORIBA MIRA Ltd. 2015 March 17, 2020 12


Measurement Polar
Plots (cont’d)

▪ The polar plots on the previous slide show how from relatively low
frequencies (200 MHz+) the polar emissions data shows significant (up to
30 dB) nulls in pattern
▪ Field recorded at CISPR 12 positions 10 - 15dB lower than maximum
recorded over full azimuth scan
▪ Fairly coarse azimuth increment possibly “masking” deeper minima
▪ At each frequency of interest the difference between the E field recorded at
the CISPR A and CISPR B positions (900 and 2700) and the maximum
E field over the entire 3600 was calculated. This was performed for each
noise source position.

© HORIBA MIRA Ltd. 2015 March 17, 2020 13


Error Comparison Between
CISPR 12 and Test Wire
Methods

© HORIBA MIRA Ltd. 2015 March 17, 2020 14


Results Discussion

▪ Based on initial data, the average Test Wire method “Error” was, at worst,
comparable with CISPR 12 method at best 5dB lower.
▪ Current results based on “Test Wire” positioned along centre line of two axis
of model

© HORIBA MIRA Ltd. 2015 March 17, 2020 15


Possible Issues with
Test Wire Method
Investigations
▪ Impedance hard to control over wide frequency range
▪ Near Field effects and directivity of test wire itself – due to its physical length
▪ Tests only performed on a single model, how repeatable are the results?
▪ Can a single ‘K’ factor be applied to multiple models?
▪ Limited number of frequencies currently investigated
▪ Are the Test Wires recording the highest levels present?
▪ Could other Test Wire positions offer better results?
▪ Required Test Wire would need to be terminated to the metallic body of the
vehicle (invasive)

© HORIBA MIRA Ltd. 2015 March 17, 2020 16


Surface Currents on Model

▪ Simulations have been used to record


surface currents induced on model due to
source antenna in multiple positions inside
model
▪ Models of both SVTC and “real” vehicle show
areas of high surface current around windows
and doors
▪ Current Test Wire positions not in areas of
high surface current.

© HORIBA MIRA Ltd. 2015 March 17, 2020 17


Surface Current Simulations /
Measurement Comparisons

▪ ‘Normalised’ surface current data for


each test point from measurements
and simulations.
▪ The data shows that the main ‘hot
spots’ are in similar position in both
sets of data

▪ Slight differences were expected as


the data sets are not from the same
size vehicle (small saloon (simulated)
and large 4x4 vehicle (measured)
▪ Results were poorer below 150 MHz

© HORIBA MIRA Ltd. 2015 March 17, 2020 18


‘Test Strip’ Development

▪ Due to problems noted earlier (impedance,


test wire directivity, termination to vehicle)
the Test Wire was re-designed
▪ Based on ‘Flexmstrip’, originally suggested
by Pissort, Catrysse & Vanhee[1].
▪ Electrically small below ~300 MHz
▪ ‘Test Strip’ was 300 mm long x 10 mm
wide above a 4mm sheet of Perspex – the
whole arrangement was then placed
above a copper ground plane
▪ The ‘Test Strip’ does not rely on being
terminated to the body of the vehicle.
▪ Impedance a much closer overall 50W (
average 85W recorded, better match to
measurement receiver).

© HORIBA MIRA Ltd. 2015 March 17, 2020 19


‘Test Strip’ to CISPR 12 Method
Error Comparison

▪ Comparison of CISPR 12 method to


‘Test Strip’
▪ 10 dB mean average error for CISPR 12
method
▪ 0.7 dB mean average for ‘Test Strip’ – all
test points considered
▪ 2 dB mean average for ‘Test Strip’ – 4
test areas considered

© HORIBA MIRA Ltd. 2015 March 17, 2020 20


‘Test Strip’ to CISPR 12
Method Error
Comparison

© HORIBA MIRA Ltd. 2015 March 17, 2020 21


Future Work

▪ Measurements to be performed on multiple vehicles (different models styles)


▪ Comparison repeated to confirm results apply
▪ Validate / check surface current measurements below 150 MHz
▪ Build final version of test wire using ‘N’ connectors to try and improve higher
frequency impedance match

© HORIBA MIRA Ltd. 2015 March 17, 2020 22

Вам также может понравиться