Harness influence in Bulk Current Injection testing Frédéric Lafon1, François de-daran1, Laurent Caves2 1
VALEO, EMC Department, 2 Rue Fernand Pouillon 94042 Créteil Cedex, France, Frédé
[email protected] 2
VALEO, Center of Electronic Expertise, 2 Rue Fernand Pouillon 94042 Créteil Cedex, France
Abstract — Bulk Current Injection test is one of the most typical test performed in the automotive industry. This test is often criticized because of a bad reproducibility of results (Due to misunderstanding or missing knowledge of some phenomenon). Some studies have been performed to estimate the influence of a certain number of parameters, such as the probe location on the harness or the uncertainty of the measurement chain. We propose in this paper to discuss on the influence of the harness structure in term of positioning of each wire regarding the others.
II. SIMULATION MODEL FOR ANALYSIS In this part we develop and describe the different models of the BCI set-up, starting from the general set-up as defined on figure 1
Injection probe
I. INTRODUCTION We focus in this paper on the BCI test set-up according ISO 11452-4 automotive standard [1], which is reminded on fig.1. The principle of this test is to use an injection probe (acting as a transformer), to inject a current on a 1m length harness separating the loads and the equipment to test. A current probe (on the right side of fig.1.) can be used to monitor the current, but the method consists in injecting the same power level that we need to have a certain current on calibrated load (JIG). The power is applied in an open loop way. When realizing BCI tests on equipment according this method, most of parameters identified as influent are typically well defined. We can consider for example the impedance of the loads and the location of the injection probe. Some experiences have demonstrated that the harness configuration could also have an important influence and we propose to focus on this parameter. This work can be realized especially with Spice models [2] developed in the past years for the injection probe. This open new perspectives for the comprehension of some phenomenon and for the evaluation of tendencies by using the simulation tools, especially Spice like tools.
Device under test
200mA
LISN/ Loads Fig. 1.
Bulk current injection General set-up
The global simulation schematic will include the models of the following parts: 1) The LISN (Linear Impedance stabilized Network). This model is based on the theoretical model as defined in the CISPR standards [3]. To have a correct model on the frequency range between 1 MHz and 400 MHz, this model includes parasitic elements such as the parasitic capacitance of the inductance which can be measured using a network analyzer. The final model used is given below (placed in a Spice black box). C16 20e-12 L8 5e-6
After an introduction of the models used for the different elements which constitute a BCI set-up, we will exploit the complete BCI model to estimate the influence of the harness in term of location of the wires ones in function of the others. We will consider in our approach the structure of the harness as a multi transmission line model where elements will depend on the geometry which will be modified to be representative of realistic configurations.
C15 100n
C17 1e-6 L10 5e-9
0
Fig. 2.
R65 50 L9 5e-9
0
LISN Equivalent schematic for 1-400 MHz
2) The injection probe model.
Reference axis
This is issued from a previous work presented in [2]. This work allowed building a Spice equivalent model of the probe usable on the frequency range between 1 MHz and 400 MHz.
1 mm Insulator for wires
εr = 1.5
3) The harness model. If we consider that the transverse structure is globally constant along the harness, we can propose some equivalent models of the harness using coupled transmission lines models. We propose for this study to consider a harness with 4 wires. The location of the wires between each other can change and we propose to estimate the influence of this parameter for the following realistic configurations.
5 cm above the ground plane
Wood : εr = 2
Fig. 5.
Configuration 3 for the harness
Reference axis
1 mm Insulator for wires
a.
εr = 1.5
Initial configuration
In this configuration we consider that the 4 wires are at the minimum distance of each other.
5 cm above the ground plane
Wood : εr = 2
Reference axis Fig. 6.
Configuration 4 for the harness
Insulator for wires
εr = 1.5
Wood : εr = 2
Fig. 3.
5 cm above the ground plane
initial configuration for the harness ( configuration 1 )
b.
For each configuration, we consider that the transmission lines are without losses and calculate with a finite difference code the values of the elements of the capacitance and inductance matrix [4]. From this information we build the Spice model of coupling lines which will be presented as a black box for the global simulation as follow:
Other configurations U68
In these configurations one or several of the wires are separated from the others by 1 mm. This implies a lower capacitance between the conductors due to an higher distance between them.
M
M
LMT4FILS_1M
Reference axis 1 mm 0 Insulator for wires
Fig. 7.
0
Coupled transmission lines model representation
εr = 1.5
One model is developed for each configuration
Wood : εr = 2
5 cm above the ground plane
4) The device under test For this study we consider for the device under test the following equivalent structure:
Fig. 4.
Configuration 2 for the harness
Input studied L6
R41
100e-9
Lbreak
1
C11 100n Cbreak
Input 2 and 3
R44 500e5 Rbreak
R48 1e6
R50 100
5) Parasitic elements
Ground wire Fig. 8. Loads configuration considered for the study and for the 4 wires harness.
We consider that the input 2 and 3 are connected to respectively high and low impedance which won't change in function of the frequency. Our study will focus only on the behavior of one input on which we have place a typical RLC structure representative of what can be found on an electronic part. To consider several configurations in term of equipment we propose to take into account variations of the impedance of this structure. To realize it we have develop a model with parametric R-L-C elements and the following values and tolerances: Default Value 500 kΩ 100 nH 100 nF
Rbreak Lbreak Cbreak
Using Monte Carlo analysis with a uniform distribution will allow covering a large range of values for these different elements. For example for a resistor, values will be between 0 and 1 MOhms.
Tolerances 99 % 99 % 99 %
The main parasitic element to take into account which is at the origin of the creation of a differential voltage is the parasitic capacitance between the printed circuit board and the ground plane. This model enough until 500 MHz has been checked by [5]. The value of this capacitance can be evaluated considering a two plate capacitor structure which implies that:
C =ε⋅
S h
(1)
With a 10 cm x 10 cm PCB 5 cm above the ground plane we estimate a capacitance around 2 pF which be used in the model. The complete model for our evaluation is given fig.9.
TABLE I DEFAULT VALUES AND TOLERANCES FOR THE INPUT MODEL
VAMP = 40 U2 BAT
R2
DUT R_1 R6 50
LISN_R_ext U4
R4 1k VAMP = 40 U6
1
C2 100n Cbreak
U68
M M
LMT4FILS_1M
BAT
R10 1e6
R12 100
VAMP = 40 R14 U12 1k DUT
0
GND
100e-9
Lbreak
R_2
0
VAMP = 40 U10
VBAT
L2
0
R_1 R16 50
LISN_R_ext U14
BATTERIE12V
0
C4 2e-12
R_2
0 0
Fig. 9.
Complete simulation schematic
R8 500e5 Rbreak
III. EXPLOITATION OF THE MODEL We can now use the simulation capabilities and make an evaluation of the influence of this harness structure. We have defined that two parameters are interesting to analyze: -
The common mode voltage on the input studied The differential voltage between this input and the ground of the product under test (local ground).
The common mode voltage will be representative of the common mode current measured during the test. In conclusion, using simulation to estimate the impact of the harness on this common mode voltage will provide information about the impact on the common mode current. This current is information available during a BCI test and this will allow having the point of view of the final user of a BCI test bench. He may effectively use these current fluctuations to estimate the influence of the harness structure.
Common mode voltage ( V )
The differential voltage will provide information about the real stress level of a device under test. Some previous studies have demonstrated that Integrated
Circuits are more commonly disturbed by the differential voltage across their inputs than by the common mode voltage [6]. The differential voltage will correspond to the real perturbation criteria of a device under test. The influence of the harness on this parameter will be obtained only with simulation. A BCI test bench doesn't allow having access to this kind of information. The objective of this study will be also to evaluate if the influence evaluated using the common mode current is enough or not to estimate the real influence of the harness structure in the stress level of the test. Simulations are performed using a Monte Carlo Analysis with 100 runs and a uniform distribution. IV. RESULTS FOR COMMON MODE VOLTAGE In nominal configuration (defined on fig. 9 with default values for Rbreak, Lbreak, Cbreak) we obtain the results given below. The voltage is measured between the input of the load and the ground plane. It allows in particular to identify the resonance frequencies due to the harness length and to correlate these phenomenons with the dispersion values between the different harness configurations.
180 160 140 120 100 80 60 40 20 0 1
10
100 F( MHz )
Fig. 10.
Nominal results for the common mode voltage
1000
For each sets of values for Rbreak, Lbreak and Cbreak, we calculate the differences between each harness configuration. The maximum values are recorded and allow building the envelop. The results are given on Fig.11. 1.6
Difference ( dB )
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 1
10
100
1000
F( MHz ) Fig. 11.
Maximum difference between the different harness configurations.
V. RESULTS FOR DIFFERENTIAL MODE VOLTAGE The most important differences are in the resonance areas and maximum levels identified are around 1.4 dB. It means that on the common mode current there won't be significant modifications due to the harness set-up. These tendencies are in the same range than uncertainty of the measurement system typically calculated around 1-2 dB.
On the following example, we can identify that the differential voltage globally increases with the frequency, especially due to the effect of the parasitic capacitance. We can also remark that the levels are the most important where resonance appears (like for common mode results)
Difference voltage ( Volts )
30 25 20 15 10 5 0 1
10
100
1000
F( MHz )
Fig. 12.
Typical result for differential mode voltage
Considering all the results for the different configuration s of the harness and for the loads, we can
calculate the maximum dispersion we can expect on this differential voltage.
Difference ( dB )
8 7 6 5 4 3 2 1 0 1
10
100
1000
F( MHz ) Fig. 13.
Maximum dispersions on differential voltage
Before 10 MHz, differences are important but no significant because the differential voltage level is very low (< 0.1 Volts). After 10 MHz as presented in fig.13. This level is significant, and we can observe that the dispersions will be very important on the complete frequency range. Maximum levels are around 8 dB.
Some tests will be realized to confirm experimentally these tendencies.
REFERENCES [1]
ISO 11453-4 Standard – " Road vehicles — Component test methods for electrical disturbances from narrowband radiated electromagnetic energy — Part 4: Bulk current injection (BCI)"
[2]
F. Fabrice DUVAL, " Bulk Current Injection Test Modeling and creation of a test methodology ", EMC Zurich 2003, pp.493-498
[3]
CISPR25 Standard - Second edition - 2002– "Radio disturbance characteristics for the protection of receivers used on board vehicles, boats and on devices – Limits and methods of measurements "
[4]
FD2D – Jan Carlsson
[5]
S.Egot and M.Klingler, "Caractérisation expérimentale de charges appliquée a la modélisation CEM d'équipements électroniques automobiles", CEM04, Toulouse France, pp.261 – 264
[6]
O.Maurice, J.Pigneret, " Susceptibilité numériques ". Congrès CEMCOMPO 1999
VI. CONCLUSION This work presents some tendencies concerning dispersion in BCI tests due to the harness configuration for a four wires harness. It has been demonstrated that on the common mode voltage maximum difference is around 1.4 dB. This doesn't seem critical and on common mode current criteria as it is generally realized, influence won't be expected on results. On the differential voltage these differences can reach 8 dB around resonance and most of results are higher than 3 dB. This means that to manage the reproducibility during BCI tests, the harness must be always in the same configuration (using straps for example) and should be considered as an unconditional part of the system under test. This work also tends to demonstrate that the length of the harness is a critical factor. Increasing the length at 2 meter for example, will amplify this phenomenon. It is recommended to reduce the length of the harness for BCI tests. At least, it's important to consider that this has been realized on a 4 wires harness, and we could expect more important tendencies with a high number of wires. Real harnesses have generally some twists due to process, and this structure will reduce the influence of the location of the wires between themselves. But in comparison with a harness developed for laboratory test, not twisted, the influence will be higher.
des
composants